Visit ebookball.

com to download the full version and

explore more ebook or textbook

Academic Emergency Medicine 1st edition by Jeffrey

Kline

_____ Click the link below to download _____

https://ebookball.com/product/academic-emergency-
medicine-1st-edition-by-jeffrey-kline-25956/

Explore and download more ebook or textbook at ebookball.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.

Academic Emergency Medicine 4 1st edition by Jeffrey Kline

https://ebookball.com/product/academic-emergency-medicine-4-1st-
edition-by-jeffrey-kline-25962/

Academic Emergency Medicine 3 1st edition by Jeffrey Kline

https://ebookball.com/product/academic-emergency-medicine-3-1st-
edition-by-jeffrey-kline-25960/

Academic Emergency Medicine 2 1st edition by Jeffrey Kline

https://ebookball.com/product/academic-emergency-medicine-2-1st-
edition-by-jeffrey-kline-25958/

Statistics in Medicine 4th edition by Statistics in

Medicine Academic Press 9780128153291 0128153296

https://ebookball.com/product/statistics-in-medicine-4th-edition-by-
statistics-in-medicine-academic-press-9780128153291-0128153296-4336/
Pediatric emergency medicine An Issue of Emergency
Medicine Clinics of North America 1st edition by Sean Fox,
Dale Woolridge ISBN 0323583504 978-0323583503
https://ebookball.com/product/pediatric-emergency-medicine-an-issue-
of-emergency-medicine-clinics-of-north-america-1st-edition-by-sean-
fox-dale-woolridge-isbn-0323583504-978-0323583503-4550/

Atlas of Emergency Medicine 5th edition by Kevin Knoop,

Lawrence Stack 9781260134957 1260134954

https://ebookball.com/product/atlas-of-emergency-medicine-5th-edition-
by-kevin-knoop-lawrence-stack-9781260134957-1260134954-4386/

The Emergency Medicine Trauma Handbook 1st Edition by Alex

Koyfman, Brit Long ISBN 9781108597654 1108597653

https://ebookball.com/product/the-emergency-medicine-trauma-
handbook-1st-edition-by-alex-koyfman-brit-long-
isbn-9781108597654-1108597653-2812/

Prepare for the Pediatric Emergency Medicine Board

Examination 1st edition by Muhammad Waseem 9783030283728
3030283720
https://ebookball.com/product/prepare-for-the-pediatric-emergency-
medicine-board-examination-1st-edition-by-muhammad-
waseem-9783030283728-3030283720-4358/

Pediatric Emergency Medicine 2nd Edition By Alisa McQueen,

Margaret Paik ISBN 1482230291 978-1482230291

https://ebookball.com/product/pediatric-emergency-medicine-2nd-
edition-by-alisa-mcqueen-margaret-paik-
isbn-1482230291-978-1482230291-4494/
Academic
Emergency
Medicine
A GLOBAL JOURNAL OF EMERGENCY CARE

Vol.31 No. 6
June 2024
ISSN 1069-6563
 15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14526 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Academic Emergency Medicine
CONTENTS

Volume 31 · Number 6 · June 2024 · www.aemj.org

ORIGINAL ARTICLES
Conflict in emergency medicine: A systematic review
Timothy Edward Tjan, Lee Yung Wong, Andrew Rixon et al. 538

Clinically significant differences in self-reported pain scores in children with headaches

Daniel S. Tsze, Christian Thiele, Gerrit Hirschfeld et al. 547

Timing and causes of death to 1Փyear among children presenting to emergency departments
Stefanie G. Ames, Apoorva Salvi, Amber Lin et al. 555

The emergency department trigger tool: Multicenter trigger query validation

Richard T. Griffey, Ryan M. Schneider, Keith E. Kocher et al. 564

A qualitative assessment of emergency physicians’ experiences with robust emergency department

buprenorphine bridge programs
Jessica Marie Heil, Jonathan M. Lassiter, Matthew S. Salzman et al. 576

Trends in pediatric emergency department transfers from Indian Health Service and tribal health systems
Fiona A. Pirrocco, Hamy Temkit, Cherisse Mechem et al. 584

SYSTEMATIC REVIEW
Assessing the one-month mortality impact of civilian-setting prehospital transfusion: A systematic
review and meta-analysis
David W. Schoenfeld, Carlo L. Rosen, Tim Harris et al. 590

RESEARCH METHODOLOGY: EMPIRICAL RESEARCH ҃ METHODOLOGY

Identifying signs and symptoms of urinary tract infection from emergency department clinical notes
using large language models
Mark Iscoe, Vimig Socrates, Aidan Gilson et al. 599

RESEARCH LETTERS
Odds ratios versus risk ratios in intensive care research: Using frailty in intensive care as a case study
Bernhard Wernly, Michael Beil, Bertrand Guidet et al. 611

Agreement of electronic health record–documented race and ethnicity with parental report
Monika Goyal, Elizabeth R. Alpern, Michael Webb et al. 613

#StopTheStigmaEM: Building a social media–based movement to support emergency medicine

mental health
Amanda J. Deutsch, Al’ai Alvarez, Stephanie Balint et al. 617
 15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14526 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Academic Emergency Medicine
CONTENTS

Volume 31 · Number 6 · June 2024 · www.aemj.org

THE BRASS TACKS: CONCISE REVIEWS OF PUBLISHED EVIDENCE

Paxlovid for nonhospitalized patients with COVID-19
Fatima Johari, Rajesh Verma 621

INVITED COMMENTARY
Reframing conflict in the emergency department as an expected and modifiable source of moral injury
James R. Miner, Adam Rieves, Rebecca Nahum et al. 624

LETTERS TO THE EDITOR

Spinal care again
Mark Hauswald 626

Response to “Utility of electroencephalography in toxin-induced seizures”

Marlis Gnirke, Mark K. Su 627

REPLY
Response to the letter: “Utility of electroencephalography in toxin-induced seizure”
Alexander M. Sidlak, Brent Dibble, Brian Schultz et al. 629

REFLECTION ARTICLES
Doctors with never-ending borders
Zhaohui Su 631

Finding success in failure

Huma Farid 632
 15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14526 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Academic Emergency Medicine
Official Journal of the Society for Academic Emergency Medicine | saem.org

JEFFREY A. KLINE, MD KEN MILNE, MD, MSC ZACHARY F. MEISEL, MD, MPH, MSHP
Editor-In-Chief Senior Associate Editor for Social Media Senior Associate Editor for Health Communica�on
Wayne State University School of Medicine South Huron Hospital Associa�on Perelman School of Medicine, University of
Detroit, Michigan Exeter, Ontario, Canada Pennsylvania
[email protected] Philadelphia, Pennsylvania
DEBORAH B. DIERCKS, MD, MSC
CHRISTOPHER R. CARPENTER, MD, MSC Senior Associate Editor for Cardiovascular JILL STOLTZFUS, PhD
Deputy Editor Medicine Senior Associate Editor for Sta�s�cs
Mayo Clinic Rochester, Rochester, MN University of Texas Southwestern St. Luke’s University Health Network, Temple/
Medical Center St. Luke’s School of Medicine
JAMES R. MINER, MD Dallas, Texas Bethlehem, Pennsylvania
Senior Associate Editor for Electronic Publishing
Hennepin County Medical Center
Minneapolis, Minnesota

ASSOCIATE EDITORS Brian C. Hiestand, MD, MPH Robert F. Reardon, MD RESIDENT EDITORS
Wake Forest University Hennepin County Medical Center
Beau Abar, PhD Jasmine Thompson, MD
Health Sciences Minneapolis, Minnesota
University of Rochester Medical Center LSU Spirit of Charity
Winston-Salem, North Carolina Aaron Robinson, MD, MPH
Rochester, New York New Orleans, Louisiana
Ula Hwang, MD, MPH Hennepin Healthcare
Harrison J. Alter, MD, MS Yale University School of Medicine, Minneapolis, MN Courtney Wechsler, MD
Highland Hospital - Alameda New Haven, CT Michael S. Runyon, MD, MPH Detroit Receiving Hospital
Health System Oakland, Carolinas Medical Center Detroit, Michigan
Timothy B. Jang, MD
California Harbor-UCLA Medical Center Charlotte, North Carolina
Kamna Balhara, MD, MS Torrance, California Elizabeth Schoenfeld, MD DEPARTMENT EDITORS
Johns Hopkins Medical Institution Peter C. Jenkins, MD University of Massachusetts School of
Associate Social Media Editor
Baltimore, Maryland Medicine, Baystate Health
Indiana University School of Medicine Michael Gottlieb, MD, RDMS
Worcester, Massachusetts
Indianapolis, Indiana Rush University Medical Center
Brigitte M. Baumann, MD, MSCE Richard Sinert, DO
Cooper University Hospital Alan E. Jones, MD Chicago, Illinois
SUNY Downstate Health Sciences
Camden, New Jersey University of Mississippi Medical
University The Biros Section on Research Ethics
Center
Steven B. Bird, MD Jackson, Mississippi Peter Sokolove, MD Shellie L. Asher, MD
University of Massachusetts University of California Albany Medical College
Damon Kuehl, MD
Medical School San Francisco Albany, New York
Virginia Tech Carilion
Worcester, Massachusetts School of Medicine Michelle D. Stevenson, MD, MS Director, Resident Member of the
Roanoke, Virginia University of Louisville Editorial Board Program
Bernard P. Chang MD, PhD Louisville, Kentucky
Columbia University Irving Medical Michelle Lin, MD, MPH, MS Mark B. Mycyk, MD
Center New York, New York Icahn School of Medicine at Mount Andrew Taylor, MD, MHS Cook County Hospital
Yale School of Medicine Chicago, Illinois
Makini Chisolm-Straker, MD, MPH Sinai
New Haven, Connecticut
Icahn School of Medicine New York, New York Editor of Infographics
Daniel L. Theodoro, MD, MSCI Kirsty Challen, BSc, MBChB, MRes,
Mount Sinai Michelle L. Macy, MD, MS Washington University
Lurie Children’s Hospital, PhD
Robert Cloutier, MD, MCR St. Louis, Missouri
Northwestern University Lancashire Teaching Hospitals
Oregon Health & Science Jody Vogel, MD, MSC, MSW United Kingdom
Chicago, Illinois
University Stanford University
Portland, Oregon Brandon Maughan, MD, MHS, MSHP Palo Alto, California Media Reviews
Oregon Health & Science
Wendy C. Coates, MD University Stephen P. Wall, MD Peter E. Sokolove, MD
Harbor-UCLA Medical Center New York University School of Medicine University of California San Francisco,
Alice Mitchell, MD New York, NY San Francisco, California
Torrance, California
Indiana University
D. Mark Courtney, MD, MSCI School of Medicine Kabir Yadav, MDCM, MS, MSHS Reflections and Resident Portfolios
University of Texas Southwestern Harbor-UCLA Medical Center Brian Zink, MD
Indianapolis, Indiana
Medical Center Torrance, California Brown University
Paul Musey, Jr., MD Providence, Rhode Island
Dallas, Texas Lalena M. Yarris, MD, MCR
Indiana University
Oregon Health & Science University
Latha Ganti, MD, MS, MBA Indianapolis, Indiana Portland, Oregon
University of Central Florida Peter D. Panagos, MD
Orlando, Florida Shahriar Zehtabchi, MD
Washington University State University of New York
Richard T. Griffey, MD, MPH St. Louis, Missouri Brooklyn, New York
Washington University Jesse Pines, MD, MBA, MSCE Mark R. Zonfrillo, MD, MSCE
St. Louis, Missouri US Acute Care Solutions (USACS) Alpert Medical School of Brown
Kennon Heard, MD, PhD Canton, Ohio University
University of Colorad o School Michael A. Puskarich, MD, MSCR Providence, Rhode Island
of Medicine Hennepin County Medical Center
Aurora, Colorado Minneapolis, Minnesota
 15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14526 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EDITORIAL OFFICES

Peer Review Coordinator Journal Manager SAEM

Taylor Bowen Laura Giblin 1111 E. Touhy Ave., Suite 540
Des Plaines, IL 60018
[email protected] [email protected]
Tel 847-257-7236
Tel 434.218.3366 Tel. 847-257-7223 Fax 847.813.5450
[email protected]

Copyright and Copying (in any format)

Copyright © 2024 Society for Academic Emergency Medicine. All rights reserved. No part of this publicaঞon may be reproduced, stored or transmi‚ed
in any form or by any means without the prior permission in wriঞng from the copyright holder. Authorizaঞon to copy items for internal and personal
use is granted by the copyright holder for libraries and other users registered with their local Reproducঞon Rights Organizaঞon (RRO), e.g. Copyright
Clearance Center (CCC), 222 Rosewood Drive, Danvers, MA 01923, USA, www.copyright.com, provided the appropriate fee is paid directly to the
RRO. This consent does not extend to other kinds of copying or use such as copying for general distribuঞon, for adverঞsing or promoঞonal purposes,
for republicaঞon, for creaঞng new collecঞve works, for resale, or for arঞCcial intelligence tools or technologies. Permissions for such reuse can be
obtained using the RightsLink “Request Permissions” link on Wiley Online Library. Special requests should be addressed to [email protected]
Disclaimer
The Publisher, the Society for Academic Emergency Medicine (SAEM) or Editors cannot be held responsible for any errors in or any consequences
arising from the use of informaঞon contained in this journal. The views and opinions expressed do not necessarily reYect those of the Publisher, the
Society for Academic Emergency Medicine (SAEM) or Editors, neither does the publicaঞon of adverঞsements consঞtute any endorsement by the
Publisher, the Society for Academic Emergency Medicine (SAEM), Editors, or Authors of the products adverঞsed.
For submission instrucঞons, subscripঞon and all other informaঞon visit h‚p://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1553-2712
Received: 13 November 2023 | Revised: 4 January 2024 | Accepted: 7 January 2024

DOI: 10.1111/acem.14874

ORIGINAL ARTICLE

Conflict in emergency medicine: A systematic review

Timothy Edward Tjan BBioMed1 | Lee Yung Wong MBBS, GradDipClinEd, FACEM2,3 |
Andrew Rixon PhD4

1
Melbourne Medical School, The
University of Melbourne, Melbourne, Abstract
Victoria, Australia
Background: The emergency department (ED) is a demanding and time-­pressured envi-
2
Emergency Department, Austin Health,
Melbourne, Victoria, Australia
ronment where doctors must navigate numerous team interactions. Conflicts between
3
School of Business, Law and health care professionals frequently arise in these settings. We aim to synthesize the
Entrepreneurship, Swinburne University individual-­, team-­, and systemic-­level factors that contribute to conflict between clini-
of Technology, Melbourne, Victoria,
Australia cians within the ED and explore strategies and opportunities for future research.
4
Department of Business, Strategy and Methods: Online databases PubMed and Web of Science were systematically
Innovation, Griffith Business School,
searched for relevant peer-­reviewed journal articles in English with keywords relating
Griffith University, Brisbane, Queensland,
Australia to “conflict” and “emergency department,” yielding a total of 29 articles.
Results: Narrative analysis showed that conflict often occurred during referrals or
Correspondence
Lee Yung Wong, Emergency Department, admissions from ED to inpatient or admitting units. Individual-­level contributors to
Austin Health, Melbourne, Australia.
conflict include a lack of trust in ED workup and staff inexperience. Team-­level con-
Email: [email protected]
tributors include perceptions of bias between groups, patient complexity, commu-
Funding information
nication errors, and difference in practice. Systems-­level contributors include high
Swinburne University of Technology;
Australasian College for Emergency workload/time pressures, ambiguities around patient responsibility, power imbal-
Medicine
ances, and workplace culture. Among identified solutions to mitigate conflict are
better communication training, standardizing admission guidelines, and improving
interdepartmental relationships.
Conclusions: In emergency medicine, conflict is common and occurs at multiple levels,
reflecting the complex interface of tasks and relationships within ED.

KEYWORDS
communication, conflict, consultations, emergency department

I NTRO D U C TI O N Failure to address conflicts effectively can result in costly med-

ical errors, litigation, decreased job satisfaction, reduced pro-
Conflict and its management and resolution is a fundamental ductivity and turnover, incivility, and generally strained health
aspect of improving health care outcomes, patient satisfac- care systems with significant strategic importance for clinical
1
tion, and the well-­b eing of health care professionals globally. governance. 2,3

See related article on page 624

LYW is the recipient of a Swinburne University Postgraduate Research Award (SUPRA) scholarship, jointly funded by Swinburne University of Technology and the Australasian College
for Emergency Medicine.

Supervising Editor: Richard Sinert

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2024 The Authors. Academic Emergency Medicine published by Wiley Periodicals LLC on behalf of Society for Academic Emergency Medicine.

538 | 
wileyonlinelibrary.com/journal/acem Acad Emerg Med. 2024;31:538–546.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14874 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TJAN et al. 539

Emergency department (ED) doctors operate within an in- clinicians' perspectives and decision-­making processes around pa-
creasingly demanding, stressful, and time-­p ressured healthcare tient care.11 Importantly, conflict may differ across health care con-
environment, where they must manage numerous interpersonal texts; a physician's experience and perceptions of workplace conflict
and team interactions.4,5 Conflicts over clinical decisions and ac- may differ significantly from a surgeon's (e.g., personality clashes vs.
tions often arise from this complex setting, leading to potential ad- communication and performance issues, respectively).10,12 Hence, it
verse patient events and exacerbating access block issues. 6 While is vital to examine conflict specific to the ED environment, to un-
individual studies have identified contributing factors to conflict, cover the nuances of how and why it is experienced as well as spe-
a comprehensive review that explores and summarizes these is- cific strategies for addressing it. This review identifies contributing
sues is either lacking, or not specific to the ED (cf. general health factors along individual, interpersonal, and organizational lines, in
1
care, see Kim et al. ). This review aims to provide a comprehensive tandem with existing frameworks on conflict in health care.1 By in-
examination of individual, team-­level, and systemic factors that corporating multiple levels of analysis, this additionally aligns with
contribute to conflict between clinicians within the ED setting. the literature on the various workplace identities that people hold,
Ultimately, we hope to provide a more nuanced understanding of be they personal (as an individual), relational (within a group), or col-
the complex ED environment and valuable insights to aid ED clini- lective (as part of an organization).13
cians in effectively managing conflict. In summary, these are the questions addressed by this review:
Conflict is defined as a “process that begins when an individual
or group perceives differences and opposition between oneself and 1. What constitutes conflict in the ED?
another individual or group about interests, beliefs or values that 2. What factors contribute to conflict in the ED?
matter to them.” (p.8)7 In the workplace, conflict often encapsulates 3. What strategies are employed to address and resolve conflict in
task issues and/or relationship (socioemotional) issues.8 Task issues the ED?
arise from disparities in procedures, priorities, or resource alloca-
tion when performing work-­related tasks and also involves process,
whereas relationship (socioemotional) conflict revolves around M E TH O D S
breakdowns in interpersonal interactions, often resulting from mis-
communication or personality clashes.7 Search strategy
In health care, conflict has been attributed to multiple factors,
including incompatible personal motivations, high workload, stress, The first two authors (TET, LYW) performed a systematic search
role ambiguity, and poor leadership.9 Clashing personality traits through the online databases PubMed and Web of Science from
and work-­related factors such as team size and tenure are also as- June to August 2023. The methodology adhered to PRISMA guide-
sociated with high-­intensity conflict between physicians and their lines14 (with the exception of comparing effect measures, which
10
superiors. In a field where effective multidisciplinary teamwork were not relevant to this review). Search terms included “emergency
is essential, unresolved conflicts extend beyond the immediate dis- medicine,” “emergency department,” and “emergency care” as well
agreement, shifting the focus from patient care,11 hindering cohe- as “conflict” and synonymous terms such as “misunderstanding,”
sive teamwork and decision making, and potentially compromising “dispute,” “communication breakdown,” “miscommunication,” and
patient safety.9 “disagreements” (outlined in Table 1, along with inclusion/exclu-
Conflict itself may hold potential for positive outcomes by fos- sion criteria). There were no time filters specified. Conflicts regard-
tering trust and cohesiveness among team members, enriching ing article selection were resolved through a process of review and

TA B L E 1 Inclusion and exclusion criteria and search terms.

Inclusion and exclusion criteria

Inclusion criteria Full-­text empirical, peer-­reviewed journal articles written in English
Articles related to conflict in the ED context
Studies that address at least one of the research questions (outlined above)
Exclusion criteria Nonempirical articles (e.g., commentaries, opinion pieces, and letters to editors)
Papers lacking a focus on the ED context
Articles that do not address any of the specified research questions
Final search terms
PubMed (“emergency medicine”[Title/Abstract] OR “emergency department*”[Title/Abstract] OR “emergency care”[Title/Abstract]
OR “emergency doctor*”[Title/Abstract] OR “emergency physician*”[Title/Abstract]) AND (“misunderstanding”[Title/
Abstract] OR “dispute”[Title/Abstract] OR “conflict”[Title/Abstract] OR “communication breakdown”[Title/Abstract] OR
“miscommunication”[Title/Abstract] OR “disagreements”[Title/Abstract]))
Web of Science (“emergency medicine” OR “emergency department*” OR “emergency care” OR “emergency doctor*” OR “emergency
physician*”) AND TS = (“misunderstanding” OR “dispute” OR “conflict” OR “communication breakdown” OR
“miscommunication” OR “disagreements”)
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14874 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
540 CONFLICT IN EMERGENCY MEDICINE

discussion between the first two authors to achieve consensus. The New Zealand (n = 4); Canada (n = 4); the United Kingdom (n = 2); and
review was not registered beforehand due to its limited and specific Egypt, Hong Kong, India, Iran, Japan, the Netherlands, Portugal, and
scope. Thailand (n = 1). A prior review of ED consultations (not specifically
A total of 1254 papers were obtained from the initial search conflict) focused mainly on the North American context,16 making
(Pubmed n = 585 and Web of Science n = 669). A total of 446 du- our inclusion of studies from diverse countries a novel approach.
plicates were removed (endnote n = 340 and manually n = 106), and Study participants also varied, inclusive of ED physicians (n = 19),
808 papers were screened for inclusion. Abstract and title screening ED nurses (n = 11), internal medicine (IM) physicians (n = 10) and sur-
excluded 762 papers. Of the remaining papers, six were not retriev- geons (n = 4) as well as ED health care technicians, managers, and
able, resulting in 40 studies assessed for full-­text review and an ad- primary care providers.
ditional 19 papers excluded. Eight additional studies were manually
added through personal sources, resulting in a total of 29 studies
(refer to Figure 1). Given the review involved both quantitative and R E S U LT S
qualitative studies, the Mixed Methods Appraisal Tool (MMAT)15
was used to appraise study quality. The MMAT is used to appraise We summarize the various dimensions of conflict in the ED and
the quality of empirical studies for systematic mixed studies reviews, provide a narrative overview of the underlying individual, inter-
across five specific methodological domains (e.g., for qualitative personal (i.e., team-­level), and organizational (i.e., system-­level)
studies—appropriateness, adequate data collection, adequate der- factors contributing to the conflict, along with identified strat-
ivation of findings, sufficient interpretation of results, and overall egies. (Due to their interconnected nature, and given that indi-
coherence). Most studies rated highly, fulfilling four of five or five of viduals operate within team structures, it is often challenging to
five criteria for methodological quality (refer to Table S1). distinguish between individual and interpersonal-­
/team-­
level
The analyzed papers exhibit diversity in their contextual origins conflict. As a result, there may be a degree of overlap between
across various countries: the United States (n = 11); Australia and/or the factors contributing to conflict at the various levels.) While

Identification of studies
Identification

446 duplicate records removed

1254 records identified from
before screening. Endnote
databases. Pubmed (n = 585),
removal (n = 340), manual
Web of Science (n = 669)
removal (n = 106)

808 records screened for 762 non-relevant records

inclusion excluded

6 reports not retrieved

46 reports sought for retrieval Full-text not available (n = 1)
Article not in English (n = 3)
Abstract only (n = 2)
Screening

19 reports excluded:
Not conflict-specific (n = 10)
40 reports assessed for eligibility
Not specific to ED (n = 2)
Not empirical (n = 7)

8 reports added externally

Included

29 studies included in review F I G U R E 1 Preferred Reporting Items

for Systemic Reviews and Meta-­Analyses
(PRISMA) diagram.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14874 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TJAN et al. 541

conflict itself tends to center around task issues such as hando- Incompatibilities in personal attitudes and motivations
vers, referrals for admission, and making diagnoses, contributing Poor engagement, arrogance, rudeness, and avoidance are exam-
factors tended to revolve around relationship (socioemotional) is- ples of personal behaviors of doctors that erode trust, 25,35 along
sues like distrust, biases, and miscommunication, compounded by with poor listening skills among different parties38 and antagonistic
working in a challenging, strenuous system. A summary of all 29 conflict management strategies. 21 Individual biases and discrimina-
articles is reported in Table S1. tion also play a role, such as when female consultants with more
youthful and softer voices are perceived as having lower seniority.38
Misaligned interests can lead to doubts around underlying mo-
What constitutes conflict in the ED? tives. 21,35 For example, Lawrence et al.38 showed that the likelihood
of receiving specialist teams accepting the ED referral increases if
The primary conflict type in the reviewed papers (n = 12) is conflict the patient has a condition of particular interest to them. In other
emerging between clinicians during handovers, referrals, or admis- words, effective two-­way communication occurs more frequently
sions between ED and admitting/consulting staff such as IM, gen- when both teams perceive tangible benefits.
eral medicine, and surgical teams. These conflicts revolve around
adverse events/near misses; disagreements regarding patient dis- Unfamiliarity
position, priorities, and perspectives; and miscommunication. For A lack of familiarity may breed suspicion and skepticism between
example, Kanjee et al.17 noted that IM physicians express apprehen- health care professionals. 20,22 This stems from limited collabora-
sion about unwarranted hospital admissions, whereas their ED coun- tive interactions over time, compounded by large and varying team
terparts emphasize reasons for admission, such as comprehensive sizes, time constraints, and changing junior medical staff. 20,23,35
outpatient care guidance. Conflict and miscommunication also occur Unfamiliarity with colleagues' workflows and varying preferences
within ED staff and/or between the ED and other services, such as for communication timing can create misunderstandings.17,20
outpatient, general practitioner, and prehospital services.18–24 For instance, regarding the optimal time for contacting inpatient
Other studies ascribed inefficiencies within the handover process teams, some accepting doctors favor early contact, whereas oth-
itself as the main driving force for conflict. 25–27 The focus of three ers believe contact should occur only after investigations have been
other articles was conflict arising from ED crowding (i.e., excessive completed38; such perplexing variability increases the perceived
patient volume overwhelming ED capacity) and high workload. 28–30 workload for ED staff.17,38
Conflict is also portrayed as a contributing factor to occupational
burnout and stress.31,32 Avelino-­Silva and Steinman33 presented Inexperience and lack of self-­confidence
conflict as diagnostic discrepancy, i.e., disparities between initial ED Lastly, inexperience, lack of self-­confidence, and a fear of mak-
diagnoses on admission and final diagnoses on discharge. Lastly, Pun ing mistakes hinder optimal interprofessional communication and
et al.34 explored communication challenges in a multilingual setting. increases the difficulty for making inpatient referrals, especially
among junior ED doctors.6,20,37

What factors contribute to conflict in the ED?

Interpersonal (team-­level) factors
Individual-­level factors
In-­group/out-­group bias
Lack of trust: poor reputation, personality clashes, and unfamiliarity Group identification significantly shapes the dynamics within a
Eleven of the included studies highlight the substantial role of dis- health care environment, especially in the ED with its multiple inter-
trust in driving conflict at the individual (personal) level, stemming actions across and between teams. Individuals naturally favor their
from unfavorable reputation, incongruent personal attributes/atti- own group (in-­group) while exhibiting less favorable attitudes or
tudes/motivations, and a lack of familiarity. perceptions toward other groups (out-­group).39,40 The influence of
specialty identity primes conflicts through a twofold process; first,
Poor reputation in-­group/out-­group biases lead to a divisive “us versus them” mental-
Several studies emphasized the detrimental influence of poor or ity.40 This phenomenon of “tribalism” hinders positive relationships
unknown reputations of ED clinicians impacting on their perceived from being fostered within the ED as well as between the ED and
level of trust. These include being perceived as a nonexpert, hav- other departments.19
ing an unfavorable reputation (e.g., being unhelpful or unreliable), Second, generalizations and stereotypes about approaches to
or even having an unknown reputation.35,36 Additionally, historical patient care erode trust and cooperation. For instance, ED physi-
experiences of poor relationships between ED doctors and inpatient cians perceive IM doctors as less skilled at navigating uncertainty,
doctors, along with a reputation for “aggressive” or “unapproach- whereas IM doctors view ED practitioners as less thorough and
able” behavior during high-­stress situations, are also factors that thoughtful in their assessments.40 Horwitz et al.39 explored the
19,28,36,37
heighten distrust. skepticism of IM physicians toward the competence, judgment, or
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14874 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
542 CONFLICT IN EMERGENCY MEDICINE

professionalism of ED physicians. Unsurprisingly, ED doctors report the workup standards of the ED to be subpar, thus introducing de-
37,41
a sense of condescension and distrust from specialty doctors. lays to the admission process.6,17,18,41,43 This disparity is exemplified
20
Olde Bekkink et al. also noted perceived disparities in experience, in the study from Horwitz et al.39 where a patient initially diagnosed
expertise, and qualifications between ED doctors and nonphysician with pneumonia in the ED was later found to have a pulmonary em-
staff. bolism. The IM physician categorized this as a diagnostic error that
led to improper care, while ED physicians view their role as stabili-
Patient complexity and disposition disagreements zation and disposition and considered the patient suitable for admis-
Eight studies noted patient complexity and disposition disagree- sion with a provisional diagnosis.
ments as team-­level contributors to conflict, especially during the ED and IM doctors also differ in opinion on the barriers to effi-
17
referral and admission process. Patient complexity itself is mul- cient admission and the “right way” of managing common presenta-
tifactorial, because of preexisting comorbidities and acute illness,6 tions.40 IM specialists view patient instability, inadequate workup,
41
uncertain clinical trajectory, and incomplete test results. These and suboptimal initial treatment in ED as more prominent barriers
factors disrupt a shared understanding of the patient, impeding effi- to the admission process, while ED physicians highlight excessive
cient and collaborative decision-­making regarding disposition or fur- IM registrar workload, slow admission process, overly detailed pa-
ther management. 23,25 For example, in Iwata et al.,42 conflict arose tient work-­up, and delays due to timing of referrals near the end of
between emergency physicians advocating for discharge with close shifts.38,43
follow-­up versus inpatient teams recommending hospital admission.
Regarding patient factors, older patients with diagnoses such as uri-
nary tract infections, pneumonia, and congestive heart failure are Organizational (system-­level) factors
more frequently implicated in team-­level conflict.33
High volume, workload, and time pressures
Communication errors Excessive workload and time constraints are important sources
Closely related to patient complexity is miscommunication, where of stress, pressure, and subsequent conflict at the systemic
conflict is generated from omission of crucial information such as level. 28,36,37,43 These may lead to rushed handovers, delayed evalu-
medication history, examination and investigation outcomes, and ations, inadequate medical notes, and reduced opportunities to
39
the ED course. Some articles point to the variability in ED pro- develop relationships.19,34,39 Compounded by staff shortages, the
viders’ training and expertise as reasons for miscommunication, high patient volume necessitates multitasking and passive listening,
leading to missing information during handovers or a lack of closed-­ potentially increasing errors and miscommunication. 24,41 Moreover,
25,26
loop communication around ED outpatient referrals. Sequential this makes reevaluating admission decisions challenging, even when
handovers, where there is no direct communication between the patients improve in the ED.17 Additionally, patient flow issues such
41
first and final caregivers, also contributes to miscommunication. as crowding and unnecessary utilization of ED beds propagate con-
One-­way communication during handovers—simply transferring flict.30,39 High patient volumes also skew perceptions of reasonable
information instead of co-­constructing a mental model of the patient requests, as both ED and admitting teams seek to offload work.17
and their clinical course—adds to conflict. 25 Conflict may also re- Intergroup conflict arises as staff tend to discard accommodating at-
volve around inadequacies in communicating diagnostic uncertainty. titudes in times of resource scarcity, 21 further fueling burnout and
For example, ED physicians prefer provisional diagnoses over defin- occupational stress.31,32
itive diagnoses, but IM doctors perceive that ambiguous diagnostic
labels potentiate diagnostic error and inappropriate care.17,39 Lastly, Ambiguous responsibility after handover
a multilingual environment potentiates information loss through Smith et al. and Horwitz et al. documented scenarios of staff un-
translation errors.34 certainty regarding clinical responsibility after handover, as in the
case of investigations initiated by ED physicians but at the request of
Differences in priorities, perspectives, and expectations admitting physicians. This practice creates ambiguity about follow-
Nine papers highlighted competing priorities, perspectives, and ­up39,41,44 especially in situations where leadership roles and respon-
expectations between ED physicians and other specialists as con- sibilities are unclear. 23
tributors to conflict. These are intimately linked to interdisciplinary
culture differences, as highlighted by Rixon et al.,19 Smith et al.,41 Lack of feedback and training
and Gifford et al. 23 The absence of posthandover feedback between the receiving phy-
ED physicians tend to focus on rapid, uncertain diagnoses, dispo- sician and ED provider prevents quality improvement practices39
sition decisions, stabilization, and referrals, while having to navigate and reduces the effectiveness of a learning environment that relies
17,39
through a high patient load, requiring swift transfers. Conversely, on trial and error. 20 This may result in ED staff lacking knowledge on
admitting physicians emphasize comprehensive assessments, defin- the receiving unit's scope of practice as well as wrongly attributed/
itive diagnoses, and detailed treatment planning.17,19 IM doctors are misplaced admissions (where patients need to be rebooked under a
often concerned about unnecessary admissions and may perceive different unit). 22
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14874 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TJAN et al. 543

Negative hierarchy/power imbalances What strategies are employed to address and resolve
As identified earlier, the dynamics of hierarchy and power imbal- conflict in the ED?
ances are prominent, with ED doctors often being perceived as
“nonspecialists” and lacking discretionary power when making ad- Twenty-­four papers discussed potential strategies to mitigate con-
mission decisions. 23 More commonly, specialty teams hold author- flict, which tend to center around systems-­
level improvements.
ity to accept or reject referrals, although some centers institute an Suggested strategies include training and standardization of hando-
“ED unit transfer privilege” (allowing patient transfer to the ward vers, developing clearer admission guidelines, improving communi-
without consultation). However, both approaches create a sense cation systems, and nurturing interprofessional or interdepartmental
of power imbalance that hinders two-­way communication and col- relationships.
22,38
laborative decision making and, instead, produces discontent
in admitting clinicians on the receiving end of this process. 22 These
hierarchical structures may exacerbate misunderstandings, such as Communication training and handover standardization
when junior staff perceive that their seniors are too busy to assist
them. 28 Such power imbalances are also seen in physician–nurse, in- Proposed solutions to decrease conflict in handovers and refer-
experienced intern–experienced nurse, and nurse–medical student rals include using standardized elements around communicating
interactions. 20,31 patient information, such as content guidelines39 and protocols17
and frameworks like SBAR38 and I-­PASS. 26,27 Standardized com-
Shift culture and physical separation munication protocols increase verbal handover quality and reduce
Two other systemic-­level factors contributing to conflict are the ambiguity concerning patient responsibility. 27,44 Other recom-
workplace culture surrounding night shifts and end-­of-­shift refer- mendations include formal teaching and training around commu-
rals38 as well as the spatial separation of services in the hospital. 23 nication, 20,28,34,37,38 especially around interdisciplinary roles and
Limited senior staff availability during night shifts can lead to prioriti- boundaries.17,24,41,44 In addition, a shared mental model facilitates
zation of managing current patients over accepting new admissions. efficient and effective collaboration by bridging gaps in under-
Additionally, referrals made near the end of a shift are often less standing, which contributes to a more comprehensive and nuanced
likely to be accepted due to impending shift changes.38 Moreover, insight of the patient. 25
the physical separation and limited coordination between specialists
and ED can lead to delays in communication and an underapprecia-
tion for ED operations. Furthermore, sequential consultations lead Improved admission guidelines
to repeated examinations and diagnostics, further delaying patient
care. 23 These contributing factors to conflict within ED are summa- Some studies recommend enhancing referral guidelines and estab-
rized in Table 2 below. lishing protocols to aid disposition decisions.6,17 In addition, granting
ED doctors expanded admitting rights, especially for undifferenti-
TA B L E 2 Summary of contributing factors to conflict within ED. ated patients, may improve patient flow and bed availability. 23 As

Individual factors
shown in Charbonneau et al. 29 where investigation requests and
disposition disagreements remained constant despite end of shift
• Lack of trust
and surge events in the ED, patients need to receive a universal high
◦ Poor reputation
standard of care that is not reliant on ED clinicians adapting well
◦ Personality clashes
to time and surge constraints but on strong, robust, and consistent
◦ Unfamiliarity
admission policies and systems.
• Inexperience and lack of self-­confidence
Interpersonal (team-­level) factors
• In-­group/out-­group bias Communication system changes
• Patient complexity and disposition disagreements
• Communication errors Various articles suggest improvements to how providers involved
• Differences in priorities, perspectives, expectations in patient care communicate, such as real-­time updates for the re-
Organizational (systems-­level) factors sponsible physician and means of contact to facilitate closed-­loop

• High workload and time pressures communication, role-­based pager forwarding, automated alerts to
admitting teams for new patients, and centralized notes to minimize
• Ambiguous responsibility after handover
sequential handovers. 26,39,41 Other recommendations include using
• Lack of feedback and training
technology (e.g., walkie-­t alkies) to improve communication,30 desig-
• Power imbalances
nated notice boards to signify staff whereabouts, 28 and digital trans-
• Shift culture and physical separation
lation for medical documentation.34
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14874 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
544 CONFLICT IN EMERGENCY MEDICINE

Improving interpersonal and interdepartmental Although it is unsurprising that time pressure and resource lack
relationships feature as systematic contributors to conflict, the unstable power
dynamic between the ED and the rest of the system is noteworthy.
Building interpersonal relationships across different units helps In a Canadian paper examining patient flow, Kreindler50 problema-
22
facilitate transfers and can be developed through shared aca- tized the giving of ED more power, as it stressed other parts of the
demic programming, multidisciplinary conferences, and inter- health service. This hints at the notion that power alone is not the
departmental social events.17,35,40 Other recommendations to simple solution, but, as uncovered by Nugus et al.,51 practicing col-
improve team dynamics include increasing face-­to-­face discus- laborative power (i.e., interdependence, collegiality) over competitive
sions, simultaneous bedside evaluation between specialties, power (i.e., dominance) may be more helpful at reducing friction be-
providing positive feedback, motivating junior staff, and involv- tween ED and other specialties.
ing nurses in updates.17,20,28,40 Conflict is also reduced through In light of this, while it is reassuring that the identified strategies
increasing engagement, understanding other teams’ interests, to mitigating conflict in the ED mainly revolve around systems-­level
and focusing on patient-­centred goals. 20,36,43 Finally, colocation improvements, the biggest challenge is presented by perceptions and
of specialists and ED doctors may improve teamwork by reducing prejudices inherent in cultural and professional practice. Improving
structural boundaries. 23 interpersonal and interdepartmental relationships may remain elu-
sive due to complexities and differences in professional identity
and the very nature of ED work.52 Importantly, this could occur be-
DISCUSSION tween emergency medicine clinicians and nonemergency clinicians,
between clinicians themselves within a large emergency medicine
In this first review of conflict specific to emergency medicine, we team or even between leaders and followers. Attempts to foster
identified that for emergency medicine clinicians, conflict gener- unity between these units may paradoxically increase conflict within
ally comprises mainly routine, common task issues of referring pa- another,13,53 such as in the scenario when an ED doctor sides with
tients for admission, handing over, communicating, and diagnosing. an inpatient specialist over an ED nurse or vice versa. Hence, we
Essentially, the underlying reasons for these task-­related conflicts contend that any attempt to improve ED patient care processes (task
are inherently and intrinsically embedded within the complexity of issues) must firstly, and genuinely, foster collaboration between all
4
emergency medicine work : uncertainties with patient diagnosis and parties involved (relationship/socioemotional issues) and consider
disposition, necessity for discussion and negotiation with multiple the various personal, professional, relational, and organizational ten-
other health care providers, and fragmentation of the patient care sions impacting on emergency clinicians.
hospital journey. In alignment with previously identified research gaps around
Examining the contributing factors that stoke the flames of conflict in health care,1 this systematic review highlights particu-
conflict reveals a rich array of relationship (socioemotional) issues. lar opportunities and needs, given the dynamic, demanding, and
These relate more strongly to matters of identity, especially so- data-­r ich ED environment, namely: (i) involving interprofessional
cial identity conflicts, with review findings mirroring the exist- team members; (ii) utilizing conceptual or theoretical bases such
13
ing literature in organizational behavior. On an individual level, as self-­evaluation (e.g., identity) or psychological capital (e.g., re-
emergency clinicians’ competency appears to be undermined by source depletion) to better understand the nature of conflict; (iii)
perceptions of mistrust. As is commonplace in emergency medi- examining the role of power and status implicit in ED conflict;
cine,45 the team dynamism, emotional demands, and psychological and (iv) harnessing organizational metrics to objectify the costs
stresses may not only hinder trust development but potentially of conflict.
generate a negative spiral between mistrust and conflict, when all There is a lack of research on specific conflict management in-
these elements are present.46 terventions and their effectiveness in emergency medicine. Further
At a team level, conflict is exacerbated by entrenched refractory research is needed to identify effective strategies for preventing
professional values (i.e., different interpretations of professional and managing conflicts in emergency medicine. Finally, with the
values; see Wright et al.47) between emergency clinicians and inpa- prevalence and importance of conflict management and resolutions
tient specialties. This is manifested through differing priorities, per- skills being part of clinicians’ capability, it is interesting to note the
spectives and expectations being placed on roles, diagnostic labels, lack of conflict management and resolution as a particular leader-
communication, or even work rate, for instance. This dichotomy is ship competency for health leaders within national and international
epitomized in a recent Agency for Healthcare Research and Quality frameworks54,55—this points the way toward future opportunities
report which perpetuates the perception that misdiagnosis in the for further research on leadership competencies and the role that
ED is mainly attributable to clinicians’ cognitive error,48 although health leaders can play in conflict.56 In this regard, scholars are en-
emergency clinicians’ rebuttal49 was that “the diagnostic role of the couraged to build on recent work that contends that, for leaders, the
emergency physician is critically important, but it is frequently pre- ED work environment itself not only fuels conflict but is threatening
liminary, which does not make the diagnosis incorrect per se.” to their very identity as leaders.57
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14874 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TJAN et al. 545

6. Zhang W, Wong LY, Liu J, Sarkar S. MONitoring knockbacks in

LI M ITATI O N S A N D S TR E N G TH S EmergencY (MONKEY)—an audit of disposition outcomes in emer-
gency patients with rejected admission requests. Open Access
Most of the included articles use qualitative methods such as small-­ Emerg Med. 2022;14:481-490.
7. De Dreu CK, Van Dierendonck D, Dijkstra MT. Conflict at work and
scale interviews and were often conducted as single-­center studies.
individual well-­being. Int J Confl Manag. 2004;15:6-26.
While they still offer valuable insights, they inherently have limited
8. De Dreu CKW, Weingart LR. Task versus relationship conflict, team
generalizability and may introduce subjectivity and bias. Therefore, performance, and team member satisfaction: a meta-­analysis. J
the overall quality of the results should be interpreted with consid- Appl Psychol. 2003;88:741-749.
eration of the study designs and methodologies. However, these lim- 9. Almost J, Wolff AC, Stewart-­P yne A, McCormick LG, Strachan D,
D'Souza C. Managing and mitigating conflict in healthcare teams:
itations are outweighed by the strengths of this review's systematic
an integrative review. J Adv Nurs. 2016;72:1490-1505.
approach, multidisciplinary perspective, and incorporation of find- 10. Skjørshammer M, Hofoss D. Physician in conflict: a survey study
ings from various settings. of individual and work-­related characteristics. Scand J Caring Sci.
1999;13:211-216.
11. Bochatay N, Bajwa NM, Cullati S, et al. A multilevel analysis of pro-
fessional conflicts in health care teams: insight for future training.
CO N C LU S I O N S Acad Med. 2017;92:S84-S92.
12. Dossett LA, Vitous CA, Lindquist K, Jagsi R, Telem DA. Women sur-
In conclusion, this narrative review unveils the multileveled nature of geons’ experiences of interprofessional workplace conflict. JAMA
Netw Open. 2020;3:e2019843-­e.
conflict that health care providers face in ED, where relationship (so-
13. Horton KE, Bayerl PS, Jacobs G. Identity conflicts at work: an inte-
cial–emotional) issues impact on task issues, in a demanding, com- grative framework. J Organ Behav. 2014;35:S6-S22.
plex work environment. Potential solutions center around systemic 14. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 state-
improvements and culture change, of which the latter would be ment: an updated guideline for reporting systematic reviews. BMJ.
2021;372:n71.
harder to implement especially in a system with many moving parts.
15. Hong Q, Pluye P, Fàbregues S, et al. Mixed Methods Appraisal Tool
Further research is needed to develop and assess strategies for cre- (MMAT), Version 2018, User Guide. Canadian Intellectual Property
ating systemic and culture change within the complex ED system. Office, Industry Canada; 2018 Report No.: Registration of
Copyright (#1148552).
16. Lee RS, Woods R, Bullard M, Holroyd BR, Rowe BH. Consultations
AU T H O R C O N T R I B U T I O N S
in the emergency department: a systematic review of the literature.
Timothy Edward Tjan performed the review, analyzed and inter-
Emerg Med J. 2008;25:4-9.
preted results, and drafted and revised the manuscript. Lee Yung 17. Kanjee Z, Beltran C, Smith CC, et al. “Friction by definition”: con-
Wong conceived of and performed the review, analyzed and inter- flict at patient handover between emergency and internal medi-
preted results, and revised the manuscript. Andrew Rixon analyzed cine physicians at an Academic Medical Center. West J Emerg Med.
2021;22:1227-1239.
and interpreted results and revised the manuscript. All authors ap-
18. Maleki M, Mousavi SM, Anjomshoa M, Shaarbafchizadeh N, Naimi
proved the final manuscript. TZ. Does Intradisciplinary conflict influence to outcomes of emer-
gency medicine residency program? A mixed methods study. Bull
C O N F L I C T O F I N T E R E S T S TAT E M E N T Emerg Trauma. 2017;5:292-298.
19. Rixon A, Skinner C, Wilson S. What factors help and hinder ef-
The authors declare no conflicts of interest.
forts to address incivility in Australasian emergency departments?
A modified Delphi study of FACEM perspectives. Emerg Med
ORCID Australas. 2023;35:420-426.
Lee Yung Wong https://orcid.org/0000-0001-8918-9898 20. Olde Bekkink M, Farrell SE, Takayesu JK. Interprofessional commu-
nication in the emergency department: residents' perceptions and
implications for medical education. Int J Med Educ. 2018;9:262-270.
REFERENCES 21. Calha A, Ferreira M, Alminhas S, Pequito T. Interpersonal con-
1. Kim S, Bochatay N, Relyea-­Chew A, et al. Individual, interpersonal, flict management strategies in nursing teams. Rev Rol Enferm.
and organisational factors of healthcare conflict: a scoping review. 2018;41:32-37.
J Interprof Care. 2017;31:282-290. 22. Germack HD, Fekieta R, Britton MC, Feder SL, Rosenberg A,
2. Maximin S, Moshiri M, Bhargava P. Understanding the cost of con- Chaudhry SI. Cooperation and conflict in intra-­hospital transfers: a
flict and an approach to conflict-­management design. J Am Coll qualitative analysis. Nurs Open. 2020;7:634-641.
Radiol. 2015;12:507-509. 23. Gifford R, van der Vaart T, Molleman E, van der Linden MC.
3. Scott C, Gerardi D. A strategic approach for managing conflict in Working together in emergency care? How professional boundar-
hospitals: responding to the joint commission leadership standard, ies influence integration efforts and operational performance. Int J
part 1. Jt Comm J Qual Patient Saf. 2011;37:59-69. Oper Prod Manag. 2022;42:54-78.
4. Nugus P, Holdgate A, Fry M, Forero R, McCarthy S, Braithwaite 24. Janagama SR, Strehlow M, Gimkala A, et al. Critical communication:
J. Work pressure and patient flow management in the emergency a cross-­sectional study of Signout at the prehospital and hospital
department: findings from an ethnographic study. Acad Emerg Med. Interface. Cureus. 2020;12:14.
2011;18:1045-1052. 25. Cohen MD, Hilligoss B, Amaral A. A handoff is not a telegram: an un-
5. Johnston A, Abraham L, Greenslade J, et al. Staff perception of the derstanding of the patient is co-­constructed. Crit Care. 2012;16:6.
emergency department working environment: integrative review 26. Huth K, Stack AM, Chi G, et al. Developing standardized "receiver-­
of the literature. Emerg Med Australas. 2016;28:7-26. driven" handoffs between referring providers and the emergency
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14874 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
546 CONFLICT IN EMERGENCY MEDICINE

department: results of a multidisciplinary needs assessment. Jt medicine and emergency medicine trainees. Emerg Med Australas.
Comm J Qual Patient Saf. 2018;44:719-730. 2016;28:391-398.
27. Huth K, Stack AM, Hatoun J, et al. Implementing receiver-­driven 44. Smith CJ, Buzalko RJ, Anderson N, et al. Evaluation of a novel hand-
handoffs to the emergency department to reduce miscommunica- off communication strategy for patients admitted from the emer-
tion. BMJ Qual Saf. 2021;30:208-215. gency department. West J Emerg Med. 2018;19:372-379.
28. Flowerdew L, Brown R, Russ S, Vincent C, Woloshynowych M. 45. Isbell LM, Boudreaux ED, Chimowitz H, Liu G, Cyr E, Kimball E.
Teams under pressure in the emergency department: an interview What do emergency department physicians and nurses feel? A
study. Emerg Med J. 2012;29:e2. qualitative study of emotions, triggers, regulation strategies, and
29. Charbonneau V, Kwok E, Boyle L, Stiell IG. Impact of emergency de- effects on patient care. BMJ Qual Saf. 2020;29:1-2.
partment surge and end of shift on patient workup and treatment 46. Wildman JL, Shuffler ML, Lazzara EH, et al. Trust development
prior to referral to internal medicine: a health records review. Emerg in swift starting action teams: a multilevel framework. Group Org
Med J. 2018;35:309-315. Manag. 2012;37:137-170.
30. Chartier LB, Simoes L, Kuipers M, McGovern B. Improving emer- 47. Wright AL, Zammuto RF, Liesch PW. Maintaining the values of a
gency department flow through optimized bed utilization. BMJ profession: institutional work and moral emotions in the emer-
Qual Improv Rep. 2016;5:u206156.w2532. gency department. Acad Manage J. 2017;60:200-237.
31. Yuwanich N, Sandmark H, Akhavan S. Emergency department 48. Newman-­Toker DE, Peterson SM, Badihian S, et al. Diagnostic
nurses' experiences of occupational stress: a qualitative study from Errors in the Emergency Department: A Systematic Review. Agency for
a public hospital in Bangkok, Thailand. Work. 2016;53:885-897. Healthcare Research and Quality; 2022.
32. Elshaer NSM, Moustafa MSA, Aiad MW, Ramadan MIE. Job stress 49. Multi-­Organizational Letter Regarding AHRQ Report on Diagnostic
and burnout syndrome among critical care healthcare workers. Alex Errors in the Emergency Department. American College of
J Med. 2018;54:273-277. Emergency Physicians. 2022.
33. Avelino-­Silva TJ, Steinman M. Diagnostic disagreements between 50. Kreindler SA. The three paradoxes of patient flow: an explanatory
emergency department admissions and hospital discharges among case study. BMC Health Serv Res. 2017;17:481.
older adults. J Am Geriatr Soc. 2015;63:S105. 51. Nugus P, Greenfield D, Travaglia J, Westbrook J, Braithwaite J. How
34. Pun JKH, Chan EA, Murray KA, Slade D, Matthiessen C. and where clinicians exercise power: interprofessional relations in
Complexities of emergency communication: clinicians' perceptions health care. Soc Sci Med. 2010;71:898-909.
of communication challenges in a trilingual emergency department. 52. Wong LY, Wilson S, Rixon A, Sendjaya S. Reframing leadership:
J Clin Nurs. 2017;26:3396-3407. leader identity challenges of the emergency physician. Emerg Med
35. Chan T, Sabir K, Sanhan S, Sherbino J. Understanding the impact Australas. 2022;34:127-129.
of residents' interpersonal relationships during emergency depart- 53. Fiol CM, Pratt MG, O'Connor EJ. Managing intractable identity
ment referrals and consultations. J Grad Med Educ. 2013;5:576-581. conflicts. Acad Manage Rev. 2009;34:32-55.
36. Chan T, Bakewell F, Orlich D, Sherbino J. Conflict prevention, con- 54. Australasian College of Health Service Management. Master Health
flict mitigation, and manifestations of conflict during emergency Service Management Competency Framework. ACHSM; 2016.
department consultations. Acad Emerg Med. 2014;21:308-313. 55. Health Leadership Competency Model 3.0. National Center for
37. Reid C, Moorthy C, Forshaw K. Referral patterns: an audit into re- Healthcare Leadership. 2018.
ferral practice among doctors in emergency medicine. Emerg Med J. 56. Zhao E, Thatcher SMB, Jehn KA. Instigating, engaging in, and
2005;22:355-358. managing group conflict: a review of the literature addressing
38. Lawrence S, Spencer LM, Sinnott M, Eley R. It takes two to tango: the critical role of the leader in group conflict. Acad Manag Ann.
improving patient referrals from the emergency department to in- 2019;13:112-147.
patient clinicians. Ochsner J. 2015;15:149-153. 57. Rixon A, Wong L, Wilson S. Are emergency departments leader
39. Horwitz LI, Meredith T, Schuur JD, Shah NR, Kulkarni RG, Jenq GY. identity workspaces? A qualitative study of emergency physicians.
Dropping the baton: a qualitative analysis of failures during the Emerg Med Australas. 2023; Forthcoming. https://​doi.​org/​10.​1111/​
transition from emergency department to inpatient care. Ann Emerg 1742-­​6723.​14346​
Med. 2009;53:701-­10.e4.
40. Schrepel C, Amick AE, Bann M, et al. Who's on your team? Specialty
identity and inter-­physician conflict during admissions. Med Educ. S U P P O R T I N G I N FO R M AT I O N
2022;56:625-633.
Additional supporting information can be found online in the
41. Smith CJ, Britigan DH, Lyden E, Anderson N, Welniak TJ, Wadman
MC. Interunit handoffs from emergency department to inpatient Supporting Information section at the end of this article.
care: a cross-­sectional survey of physicians at a university medical
center. J Hosp Med. 2015;10:711-717.
42. Iwata M, Yamanaka K, Kitagawa Y. The level of agreement regard-
ing patient disposition between emergency physicians and consul- How to cite this article: Tjan TE, Wong LY, Rixon A. Conflict
tants in the emergency department. Int J Emerg Med. 2013;6:22.
in emergency medicine: A systematic review. Acad Emerg
43. Lawrence S, Sullivan C, Patel N, Spencer L, Sinnott M, Eley R.
Admission of medical patients from the emergency department: an Med. 2024;31:538-546. doi:10.1111/acem.14874
assessment of the attitudes, perspectives and practices of internal
Received: 8 September 2023 | Revised: 5 January 2024 | Accepted: 20 January 2024

DOI: 10.1111/acem.14879

ORIGINAL ARTICLE

Clinically significant differences in self-­reported pain scores in

children with headaches

Daniel S. Tsze MD, MPH1 | Christian Thiele PhD2 | Gerrit Hirschfeld PhD2 |
Peter S. Dayan MD, MSc1

1
Department of Emergency Medicine,
Division of Pediatric Emergency Medicine, Abstract
Columbia University College of Physicians
Objectives: Changes in pain scores that represent clinically significant differences in
and Surgeons, New York, New York, USA
2
Faculty of Business and Health,
children with headaches are necessary for study design and interpretation of findings
University of Applied Sciences Bielefeld, reported in studies. We aimed to determine changes in pain scores associated with a
Bielefeld, Germany
minimum clinically significant difference (MCSD), ideal clinically significant difference
Correspondence (ICSD), and patient-­perceived adequate analgesia (PPAA) in this population.
Daniel S. Tsze, Department of Emergency
Medicine, Division of Pediatric Emergency
Methods: We performed a secondary analysis of two prospective studies of children
Medicine, 530 W166th Street, First Floor, with headaches presenting to an emergency department. Two serial assessments were
New York, NY 10032 USA.
Email: [email protected]
performed in children aged 6–17 and 4–17 years who self-­reported their pain intensity
using the Verbal Numerical Rating Scale (VNRS) and Faces Pain Scale–Revised (FPS-­
Funding information
National Center for Advancing
R), respectively. Children qualitatively described any endorsed change in pain score;
Translational Sciences, Grant/Award those who received an analgesic were asked if they wanted additional analgesics to
Number: UL1TR000040; Migraine
Research Foundation
decrease their pain intensity. We used receiver operating characteristic curve–based
methodology to identify changes in pain scores associated with “a little less” (MCSD)
and “much less” (ICSD) pain and patients declining additional analgesics because they
experienced adequate analgesia after treatment (PPAA).
Results: We analyzed 105 children: 63.8% were female and the median (IQR) age was
13 (10–15) years. Ninety-­eight children were analyzed for the VNRS and 101 were
analyzed for the FPS-­R . For the VNRS, raw change and percent reductions in pain
scores associated with MCSD, ICSD, and PPAA were 2/10 and 25%, 4/10 and 56%,
and 3/10 and 50%, respectively, and for the FPS-­R , 2/10 and 25%, 4/10 and 67%, and
4/10 and 60%, respectively. The area under the curve (AUC) associated with a MCSD
for both scales ranged from 94% to 98%; the AUC associated with an ICSD or PPAA
for both scales ranged from 76% to 83%.
Conclusions: We identified changes in pain score associated with patient-­centered
outcomes in children with headaches suitable for designing trials and assigning clinical
significance to changes in pain scores reported in studies.

This study was funded by Columbia University's CTSA grant No. UL1TR000040 from NCATS/NIH and the Migraine Research Foundation (New York, NY).

Supervising Editor: Robert Cloutier

Acad Emerg Med. 2024;31:547–554. wileyonlinelibrary.com/journal/acem © 2024 Society for Academic Emergency Medicine. | 547
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14879 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
548 CLINICALLY SIGNIFICANT DIFFERENCES IN PAIN IN CHILDREN WITH HEADACHES

I NTRO D U C TI O N Selection of participants

The change in pain score that represents a clinically significant dif- The observational study enrolled children aged 4–17 years present-
ference is a fundamental component of designing and interpreting ing to the ED with painful and nonpainful conditions between April
clinical trials that evaluate treatments for patients with headaches. 2014 and March 2016.8 Children who self-­identified as having a
Recommended outcomes for trials have included pain freedom at chief complaint of headache or migraine and endorsed having pain
2 h or headache relief based on an ordinal 4-­point scale representing were included in this current study. The RCT enrolled children aged
no headache, mild, moderate, or severe headache.1,2 A reduction in 8 to 17 years presenting with a migraine headache as defined by the
pain intensity of at least 50% has also been used as an outcome in modified Irma's ED criteria (Table S1) with a self-­reported pain score
3–5
headache trials. of ≥4/10 during one of three recruitment periods between June
These outcomes have been commonly used as desired endpoints 2015 and March 2021, all of whom were included in this current
for patients with headaches, but it is unclear if the changes in pain study.3 Patients presenting during the overlapping period of time
scores associated with these outcomes align with what a child or were enrolled in only one of the two studies. Exclusion criteria com-
adolescent may consider to be clinically meaningful. There are dif- mon to both studies included inability to complete self-­report meas-
ferences in how children and adolescents perceive and report pain ures of pain (e.g., developmental delay, neurological impairment,
compared to adults, and these differences must be considered when altered mental status, intoxication), chronic disease associated with
deciding what change in pain score should be considered a clinically pain other than migraine headaches, underlying medical condition
significant difference in the pediatric population. Changes in pain necessitating multiple painful procedures (e.g., malignancy), or not
score representing clinically meaningful differences have been de- being able to speak English or Spanish. Exclusion criteria specific to
termined in populations of children with a variety of conditions, but patients enrolled in the RCT included any contraindication to receiv-
it is unclear if these estimates are generalizable specifically to chil- ing ketorolac; receipt of any nonsteroidal anti-­inflammatory drug
6,7
dren with headaches. within the previous 6 h; presence of an intranasal obstruction that
To address these gaps in knowledge, we aimed to determine the could not be readily cleared; history of intracranial surgery, struc-
changes in pain scores that represent clinically meaningful outcomes tural abnormalities, or risk factors for intracranial abnormality (e.g.,
in pain in children and adolescents with headaches. Specifically, we coagulopathy, pregnancy); known liver or kidney problems; critical
aimed to determine the minimum clinically significant difference illness; or use of any medication for headaches on more than 10 days
(MCSD), ideal clinically significant difference (ICSD), and patient-­ per month.
perceived adequate analgesia (PPAA) in this population.6,7 These
data are important for both study design and assigning clinical sig-
nificance to changes in pain scores reported in studies of children Procedures
with headaches.
Children from both studies received two serial assessments of pain
intensity using both the VNRS and the Faces Pain Scale–Revised
M E TH O D S (FPS-­R). The first assessment was performed to obtain the patient's
baseline level of pain intensity. For patients in the observational
Study design and setting study, an analgesic was then administered at the treating physician's
discretion. The second assessment was performed 30–60 min after
We conducted a secondary analysis of data combined from two an analgesic was administered or, if an analgesic was not adminis-
separate studies to aggregate a comprehensive cohort that included tered, 30–60 min after the first assessment was performed. For pa-
children with both undifferentiated and migraine headaches to be tients in the RCT, intranasal or intravenous ketorolac and a normal
representative of the spectrum of children with headaches who typ- saline bolus was administered after their baseline pain intensity as-
ically present to and are treated in the emergency department (ED). sessment. The pain scores used for the current study were obtained
The first was an observational cross-­sectional study that aimed to 60 min after analgesic administration.
determine the validity and reliability of the Verbal Numerical Rating To determine the MCSD and ICSD, children who received an
Scale (VNRS) in children; this cohort represented children with un- analgesic were asked: “Is your pain much less, a little less, about
differentiated headaches.8 The second was a randomized clinical the same, a little worse, or much worse compared to before you
trial (RCT) comparing intranasal ketorolac to intravenous ketorolac got your medicine?” Children who did not receive an analgesic were
for reducing pain in children with migraine headaches; this cohort asked: “Is your pain much less, a little less, about the same, a little
represented children with migraine headaches.3 Both studies were worse, or much worse compared to how you felt [number of min-
conducted in the same urban pediatric ED, with an annual census utes since last assessment] ago?” or “… compared with how you
of approximately 55,000 visits. Our institutional review board ap- felt the last time I asked you how much pain or hurt you had?” de-
proved the observational study with verbal informed consent and pending on what was most understandable for each child. The word
the RCT with written informed consent. “hurt” was used interchangeably with “pain,” and the word “smaller”
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14879 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TSZE et al. 549

used interchangeably with “less,” depending on what seemed most Data analysis
understandable for each child.
To determine the PPAA, children from both studies who received We used descriptive statistics to summarize patient characteristics
an analgesic were asked, “Do you want more medicine to make your and the changes in pain scores endorsed by children who reported
pain less?” after their second pain assessment. A standardized pro- a MCSD, ICSD, and PPAA. Outcomes measured using the VNRS
cedure was conducted to ensure that a negative response (i.e., not analyzed data from children 6–17 years old; those measured using
wanting more medicine to make his/her pain less) was due to the the FPS-­R analyzed data from children 4–17 years old. For both
patient having experienced adequate analgesia and not because of pain scales, patients were removed if they provided a response that
reasons unrelated to pain relief (e.g., medication tasted bad, did not did not match a valid option (e.g., when using the FPS-­R , selecting
7
want a needle, afraid of side effects). This procedure was conducted the space in between two faces, instead of selecting a face itself).
in such a way that elicited the preferences of the child rather than Patients who did not understand or did not respond to questions
the patient's caregiver. Children from the observational study who regarding the qualitative description of their change in pain (i.e., “a
did not receive an analgesic were not asked if they wanted additional little less,” “much less”) or regarding whether they wanted additional
analgesia to make their pain less. analgesia were removed from analysis.
We used receiver operating characteristic (ROC)-­based methods
to determine the changes in pain scores representing MCSD, ICSD,
Pain measurements and PPAA.6,7 These methods used dichotomous reference criteria to
determine cut points for continuous measures that optimally differen-
The VNRS is a self-­report measure of pain intensity with strong con- tiate between the two classes of the criteria. This was done by deter-
vergent validity, known-­groups validity, responsivity, and test–retest mining the sensitivity and specificity for all possible cut points of the
reliability in children 6–17 years old.8 It is a recommended measure FPS-­R and VNRS. The cut point with the lowest absolute difference
of pain intensity in patients with headache and may provide a higher between sensitivity and specificity (to achieve the best balance of the
discriminatory capability than a categorical scale.1,9 The VNRS was test characteristics) was considered to be optimal. A balance between
administered by asking, “On a scale from zero to ten, where zero sensitivity and specificity was chosen so that equal importance could
means no pain and ten means the most or worst pain, how much be assigned to correctly identifying patients who achieved a desired
pain do you have right now?” No materials or equipment are required improvement (i.e., sensitivity) those who did not achieve a desired im-
to administer the VNRS. The FPS-­R is a self-­report measure of pain provement in pain (i.e., specificity). The area under the curve (AUC)
intensity with strong convergent validity known-­groups validity, re- was reported as a measure of overall discriminatory ability. Three ref-
sponsivity, and test–test reliability in children aged 4–17 years old. erence criteria were used to determine each outcome. For the MCSD,
The FPS-­R was administered by showing each child a picture of six children who reported that their pain was “a little less” or “much less”
faces that each represent an escalating degree of pain intensity and were compared to those whose pain was “about the same,” “a little
then reading standardized instructions to elicit the child's response worse,” or “much worse.” For the ICSD, children who reported that
(www.​iasp-­​pain.​org/​FPSR).10 For both pain scales, the word “hurt” their pain was “much less” were compared to those whose pain was
was used interchangeably with “pain” depending on what was most “a little less,” “about the same,” “a little worse,” or “much worse.” For
understandable for the child. Both the VNRS and the FPS-­R are PPAA, children who answered “no” when asked if they wanted addi-
strongly recommended for self-­report of acute pain in children.11 tional analgesia to make their pain less were compared to those who
answered “yes.” To guard against finding spurious differences, we used
bootstrapping to determine the magnitude of variability of each cut
Outcomes point.12,13 We reported the change in pain score for the MCSD, ICSD,
and PPAA as both a raw change and a percent reduction in pain score.
We determined three types of clinically meaningful changes in pain The sample size was based on the availability of eligible patients
intensity: the MCSD, ICSD, and PPAA. The change in pain score as- from the two parent studies, for which this study was a secondary
sociated with a child reporting that his or her pain intensity is “a little analysis.3,8 Statistical analyses were conducted using SPSS (Version
less” represents the MCSD; that associated with a report of “much 26), R (Version 4.0.3), and the R cutpointr package.14
less” represents the ICSD. PPAA is the change in pain score associ-
ated with a child declining additional analgesia to make his or her
pain intensity less after an initial treatment is administered because R E S U LT S
he or she feels their pain has been adequately treated (as opposed to
declining additional analgesia for other reasons such as fear or side Patient characteristics
effects). A child is determined to have experienced PPAA if they re-
spond “no” when asked, “Do you want more medicine to make your There were 105 children with headaches that met criteria for this
pain less?” using the aforementioned procedure. PPAA can only be analysis: 56 from the RCT and 49 from the observational study
determined in children who received an analgesic.7 (Figure 1). For the total sample, 63.8% of participants were female
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14879 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
550 CLINICALLY SIGNIFICANT DIFFERENCES IN PAIN IN CHILDREN WITH HEADACHES

Assessed for eligibility Assessed for eligibility

(RCT) (Observational Study)
n=525 n=760

Excluded, n=466 Excluded, n=711

- 458 Ineligible - Non-headache/migraine
- 8 Declined chief complaint

Included in study Included in study

n=56 n=49

Included for analysis

n=105

Excluded, n=7 Excluded, n=4

- Age < 6 years old - Invalid FPS-R response

Analyzed for MCSD, ICSD Analyzed for MCSD, ICSD

using VNRS using FPS-R
n=98 n=101

Excluded, n=8 Excluded, n=10

- Did not receive analgesic - Did not receive analgesic

Analyzed for PPAA Analyzed for PPAA

using VNRS using FPS-R
n=90 n=91

F I G U R E 1 Enrollment flow diagram. FPS-­R , Faces Pain Scale–Revised; ICSD, ideal clinically significant difference; MCSD, minimum
clinically significant difference; PPAA, patient-­perceived adequate analgesia; RCT, randomized clinical trial; VNRS, Verbal Numerical Rating
Scale.

and the median (IQR) age was 13 (10–15) years. Table 1 shows the scales, with estimates having either similar or overlapping 95% con-
patient characteristics of patients analyzed for MCSD, ICSD, and fidence intervals (CIs).
PPAA when using the VNRS and FPS-­R . The largest proportion of
patients reported initial pain scores that represented severe pain
(44.9%). Similarly, the majority of patients reported a change in pain DISCUSSION
that was “much less” (53.1%). The patient characteristics and initial
headache pain intensity were similar between patients from the RCT We identified raw change and percent reduction in pain scores as-
and the observational study (Table S2). sociated with clinically meaningful outcomes in children presenting
acutely with headaches to the ED, which has not been previously
reported in this specific population. Our study provides estimates
Main results based on commonly used pain scales with strong validity and reli-
ability in children and adolescents and can be used both for study
Tables 2 and 3 show the changes in pain scores associated with design and for assigning clinical meaning to changes in pain scores
MCSD, ICSD, and PPAA when using the VNRS and FPS-­R . The AUC reported in studies.
ranged from 76% to 98% for all outcomes when expressed as raw The estimates we identified best representing MCSD, ICSD,
change score or percent reduction for both the VNRS and the FPS-­ and PPAA in children with headaches are comparable to those
R, with the AUC associated with MCSD for both scales ranging from previously identified in cohorts of children with a more heteroge-
94% to 98%. The raw change scores and percent reductions best neous collection of conditions associated with acute pain. 6,7 Our
representing MCSD, ICSD, and PPAA were similar across both pain findings also affirm the clinical significance of existing outcomes
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14879 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TSZE et al. 551

TA B L E 1 Patient characteristics.
Analyzed for VNRS Analyzed for FPS-­R

MCSD/ICSD PPAA MCSD/ICSD PPAA

Characteristics (n = 98) (n = 90) (n = 101) (n = 91)

Sex
Female 63 (64.3) 58 (64.4) 63 (62.4) 58 (63.7)
Male 35 (35.7) 32 (35.6) 38 (37.6) 33 (36.3)
Age (years) 14 (11–16) 14 (11–16) 13 (10–15) 14 (11–16)
Age group (years)
4–7 5 (5.1) 3 (3.3) 9 (8.9) 6 (6.6)
8–12 30 (30.6) 27 (30) 32 (31.7) 28 (30.8)
13–17 63 (64.3) 60 (66.7) 60 (59.4) 57 (62.6)
Race/ethnicity
Hispanic 83 (84.7) 76 (84.4) 85 (84.2) 76 (83.5)
Black 6 (6.1) 5 (5.6) 7 (6.9) 6 (6.6)
White 6 (6.1) 6 (6.7) 6 (5.9) 6 (6.6)
Other 3 (3) 3 (3.3) 3 (3) 3 (3.3)
Primary language
English 92 (93.9) 86 (95.6) 94 (93.1) 86 (94.5)
Spanish 6 (6.1) 4 (4.4) 7 (6.9) 5 (5.5)
Initial pain score reporteda
0–1 3 (3.1) 1 (1.1) 1 (1) 0
2–5 22 (22.4) 19 (21.1) 26 (25.7) 21 (23.1)
6–7 29 (29.6) 27 (30) 43 (42.6) 39 (42.8)
8–10 44 (44.9) 43 (47.8) 31 (30.7) 31 (34.1)
Reported change in pain
Much less 52 (53.1) 51 (56.7) 52 (51.5) 51 (56)
A little less 38 (38.8) 34 (37.8) 37 (36.6) 32 (35.2)
About the same 6 (6.1) 4 (4.4) 8 (7.9) 5 (5.5)
A little worse 1 (1) 1 (1.1) 2 (2) 2 (2.2)
Much worse 1 (1) 0 2 (2) 1 (1.1)
Analgesic administered
Ketorolac only 60 (61.2) 60 (66.7) 60 (59.4) 60 (65.9)
Acetaminophen 13 (13.3) 13 (14.4) 14 (13.9) 14 (15.4)
Ibuprofen 11 (11.2) 11 (12.2) 13 (12.9) 13 (14.3)
Migraine regimen 4 (4.1) 4 (4.4) 3 (3) 3 (3.3)
Morphine 2 (2) 2 (2.2) 1 (1) 1 (1.1)
No analgesic administered 8 (8.2) 0 10 (9.9) 0
Headache type
Migraine 56 (57.1) 56 (62.2) 56 (55.4) 56 (61.5)
NOS 42 (42.9) 34 (37.8) 45 (44.6) 35 (38.5)

Note: Data are reported as n (%) or median (IQR). Migraine regimen = ketorolac + dopamine
antagonist ± diphenhydramine; other = American Indian/Alaska Native, Asian, and “don't know”.
Abbreviations: FPS-­R , Faces Pain Scale–Revised; ICSD, ideal clinically significant difference;
MCSD, minimum clinically significant difference; PPAA, patient-­perceived adequate analgesia;
NOS, not otherwise specified; VNRS, Verbal Numerical Rating Scale.
a
Measured using the VNRS.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14879 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
552 CLINICALLY SIGNIFICANT DIFFERENCES IN PAIN IN CHILDREN WITH HEADACHES

TA B L E 2 Changes in VNRS scores representing MCSD, ICSD, and PPAA.

Change in pain score (95% CI) Sensitivity (95% CI) Specificity (95% CI) AUC (95% CI)

MCSD
Raw change score 2 (1–2) 0.86 (0.79–1) 1 (0.83–1) 0.98 (0.94–1)
Percent reduction 0.25 (0.11–0.50) 0.84 (0.62–1) 0.88 (0.60–1) 0.94 (0.82–1)
ICSD
Raw change score 4 (3–4) 0.65 (0.58–0.95) 0.80 (0.68–0.87) 0.83 (0.74–0.91)
Percent reduction 0.56 (0.43–0.60) 0.79 (0.66–0.87) 0.78 (0.67–0.86) 0.83 (0.74–0.91)
PPAA
Raw change score 3 (3–4) 0.78 (0.58–0.86) 0.65 (0.54–0.87) 0.77 (0.65–0.88)
Percent reduction 0.50 (0.38–0.60) 0.72 (0.60–0.83) 0.73 (0.60–0.84) 0.79 (0.68–0.89)

Abbreviations: AUC, area under the curve; ICSD, ideal clinically significant difference; MCSD, minimum clinically significant difference; PPAA,
patient-­perceived adequate analgesia; VNRS, Verbal Numerical Rating Scale.

TA B L E 3 Changes in FPS-­R scores representing MCSD, ICSD, and PPAA.

Change in pain score (95% CI) Sensitivity (95% CI) Specificity (95% CI) AUC (95% CI)

MCSD
Raw change score 2 (2–2) 0.94 (0.89–0.99) 1 (1–1) 0.98 (0.95–1)
Percent reduction 0.25 (0.25–0.25) 0.94 (0.89–0.99) 1 (1–1) 0.98 (0.95–1)
ICSD
Raw change score 4 (4–4) 0.83 (0.71–0.92) 0.69 (0.56–0.82) 0.79 (0.70–0.87)
Percent reduction 0.67 (0.60–0.67) 0.73 (0.62–0.85) 0.73 (0.61–0.85) 0.79 (0.70–0.88)
PPAA
Raw change score 4 (4–4) 0.73 (0.62–0.84) 0.61 (0.42–0.79) 0.76 (0.65–0.85)
Percent reduction 0.60 (0.50–0.67) 0.67 (0.56–0.82) 0.71 (0.54–0.84) 0.77 (0.66–0.87)

Abbreviations: AUC, area under the curve; FPS-­R , Faces Pain Scale–Revised; ICSD, ideal clinically significant difference; MCSD, minimum clinically
significant difference; PPAA, patient-­perceived adequate analgesia.

being used in the study of headache treatments. For example, changes in pain scores that are attached to more explicit definitions
we demonstrated that the previously utilized outcome of 50% of clinical significance (such as the need for additional analgesia, i.e.,
reduction in pain intensity corresponds to what a pediatric pa- PPAA) can delineate degrees of effectiveness that may not neces-
tient would consider to be both an ideal decrease in pain intensity sarily amount to pain freedom but are still clinically meaningful.
(ICSD) and adequate pain relief after receiving an analgesic (PPAA) We chose cut points representing changes in pain score repre-
3–5
when using the VNRS for assessment. These estimates of ICSD senting MCSD, ICSD, and PPAA that had the best balance between
and PPAA are slightly higher when using the FPS-­R (67% and 60%, sensitivity and specificity. In certain circumstances, one test char-
respectively), which is likely a function of the difference in small- acteristic may be prioritized over the other; for example, a high sen-
est unit of measurement between scales (i.e., smallest unit of mea- sitivity is desirable for screening tests for disease. However, in the
surement of the VNRS is 1/10 compared to 2/10 for the FPS-­R ). context of assessing for clinically meaningful changes in pain, we
The difference noted is a reminder that scale-­specific estimates decided it was equally important to identify both patients who did
should be used during study design due to inherent differences or who did not achieve a desired improvement in pain. For instance,
between scales. having a high sensitivity may correctly identify a large proportion
Our estimates of clinically meaningful changes in pain scores can of children who experienced a desired improvement in pain and ac-
be used in conjunction with other recommended outcomes for as- curately represent the analgesic effect of a treatment. However, it
sessing headache treatments, such as pain freedom or the 4-­point could also incorrectly identify others as having achieved a desired
scale where 0 = no headache, 1 = mild, 2 = moderate, and 3 = severe improvement in pain and lead to withholding needed analgesia or
1,2
headache. For example, achieving pain freedom will always be the the overestimation of the analgesic effect of a treatment. Similarly,
ideal goal for children with headaches. However, trials of analgesics in having a high specificity may correctly identify children who re-
children presenting acutely with migraine headaches show that most quire additional analgesia because a desired improvement in pain
patients do not achieve pain freedom (60%–85%).3–5 Evaluating for was not achieved. However, it could incorrectly identify others as
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14879 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TSZE et al. 553

not having achieved a desired improvement in pain and lead to ad- clinical meaning to changes in pain scores reported in studies of
ministering unnecessary analgesics (and subsequent exposure to this population.
associated side effects) or the underestimation of the analgesic ef-
fect of a treatment. We recognized that both scenarios are equally AU T H O R C O N T R I B U T I O N S
important to balance when seeking to optimize the management of Daniel S. Tsze and Peter S. Dayan conceptualized and designed the
pain in children. study and obtained research funding. Daniel S. Tsze supervised the
conduct of the study and data collection. Christian Thiele and Gerrit
Hirschfeld conducted the statistical analyses for the study. Daniel
LI M ITATI O N S S. Tsze drafted the manuscript, and all authors contributed substan-
tially to its revision. Daniel S. Tsze takes responsibility for the paper
Our study had limitations. We did not use ICHD criteria to iden- as a whole.
tify our patients considered to have migraine headaches. However,
we did use the modified Irma criteria, which has a high sensitivity C O N F L I C T O F I N T E R E S T S TAT E M E N T
(95%) for identifying children with migraine headaches and uses The authors declare no conflicts of interest.
symptoms that constitute what clinicians would typically consider
treating as a migraine in the clinical setting.15 The patients from ORCID
the observational study were a convenience sample rather than Daniel S. Tsze https://orcid.org/0000-0002-7567-8743
consecutive patients. However, the sample included a diverse
distribution of initial pain intensity and representation of anal- REFERENCES
gesics administered. Our patients were limited to those with an 1. Diener H-­C , Tassorelli C, Dodick DW, et al. Guidelines of the in-
acute presentation of headache presenting to the ED. It is unclear ternational headache society for controlled trials of acute treat-
ment of migraine attacks in adults: fourth edition. Cephalalgia.
whether these changes would be generalizable to the context of
2019;39(6):687-­710.
children with chronic daily headaches in the outpatient setting. The 2. Tfelt-­Hansen P, Pascual J, Ramadan N, et al. Guidelines for con-
95% CIs for the test characteristics for ICSD and PPPA are rela- trolled trials of drugs in migraine: third edition. A guide for investi-
tively wide. Future study with a larger cohort may be warranted gators. Cephalalgia. 2012;32(1):6-­38.
3. Tsze DS, Lubell TR, Carter RC, et al. Intranasal ketorolac versus in-
to confirm that the estimates reported are those with the best
travenous ketorolac for treatment of migraine headaches in chil-
balance of sensitivity and specificity. The number of patients who dren: a randomized clinical trial. Acad Emerg Med. 2021;29:465-­475.
comprised the comparison group (i.e., patients who reported that 4. Brousseau DC, Duffy SJ, Anderson AC, Linakis JG. Treatment
their pain was “about the same,” “a little worse,” or “much worse”) of pediatric migraine headaches: a randomized, double-­blind
trial of prochlorperazine versus ketorolac. Ann Emerg Med.
was notably smaller than the patients who reported that their pain
2004;43(2):256-­262.
was “a little less” or “much less,” which could affect the precision 5. Richer LP, Ali S, Johnson DW, Rosychuk RJ, Newton AS, Rowe BH.
of the cut point estimates determined. The precision achieved by A randomized trial of ketorolac and metoclopramide for migraine in
our nonparametric bootstrapping methodology, however, is com- the emergency department. Headache. 2022;62(6):681-­689.
parable to that achieved by other more conservative methods (data 6. Tsze DS, Hirschfeld G, von Baeyer CL, Bulloch B, Dayan PS. Clinically
significant differences in acute pain measured on self-­report pain
available upon request). This was a single-­center study and the
scales in children. Acad Emerg Med. 2015;22(4):415-­422.
generalizability of our results to other settings may require further 7. Tsze DS, Hirschfeld G, von Baeyer CL, Suarez LE, Dayan PS. Changes
study. The majority of patients in our study were 8 years or older, in pain score associated with clinically meaningful outcomes in chil-
so it is unclear if our findings are applicable to children 7 years and dren with acute pain. Acad Emerg Med. 2019;26(9):1002-­1013.
8. Tsze DS, von Baeyer CL, Pahalyants V, Dayan PS. Validity and re-
younger. However, prior studies of children presenting to EDs with
liability of the verbal numerical rating scale for children aged 4 to
acute pain did not demonstrate any differences between clinically 17 years with acute pain. Ann Emerg Med. 2018;71(6):691-­702.e3.
significant changes in pain based on age.6,7 9. Ripamonti CI, Brunelli C. Comparison between numerical rating
scale and six-­level verbal rating scale in cancer patients with pain:
a preliminary report. Support Care Cancer. 2009;17(11):1433-­1434.
10. Tsze DS, von Baeyer CL, Bulloch B, Dayan PS. Validation of self-­
CO N C LU S I O N S report pain scales in children. Pediatrics. 2013;132(4):e971-­e979.
11. Birnie KA, Hundert AS, Lalloo C, Nguyen C, Stinson JN.
In children presenting to an ED with headache pain, we determined Recommendations for selection of self-­report pain intensity mea-
sures in children and adolescents: a systematic review and quality
raw changes and percent reductions in pain scores associated with
assessment of measurement properties. Pain. 2019;160(1):5-­18.
a minimum clinically significant difference, ideal clinically signifi- 12. Carpenter J, Bithell J. Bootstrap confidence intervals: when,
cant difference, and patient-­perceived adequate analgesia when which, what? A practical guide for medical statisticians. Stat Med.
using two highly recommended self-­reported measures of pain that 2000;19(9):1141-­1164.
13. Thiele C, Hirschfeld G. Confidence intervals and sample size plan-
both have strong validity and reliability in children. These estimates
ning for optimal cutpoints. PLoS One. 2023;18(1):e0279693.
can be used when selecting a desired effect size for studies inves- 14. Thiele C, Hirschfeld G. Cutpointr: improved estimation and valida-
tigating treatments for children with headaches as well as assigning tion of optimal Cutpoints in R. J Stat Softw. 2021;98(11):1-­27.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14879 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
554 CLINICALLY SIGNIFICANT DIFFERENCES IN PAIN IN CHILDREN WITH HEADACHES

15. Trottier ED, Bailey B, Lucas N, Lortie A. Diagnosis of migraine in the

pediatric emergency department. Pediatr Neurol. 2013;49(1):40-­45. How to cite this article: Tsze DS, Thiele C, Hirschfeld G,
Dayan PS. Clinically significant differences in self-­reported
pain scores in children with headaches. Acad Emerg Med.
S U P P O R T I N G I N FO R M AT I O N
2024;31:547-554. doi:10.1111/acem.14879
Additional supporting information can be found online in the
Supporting Information section at the end of this article.
Received: 11 September 2023 | Revised: 5 December 2023 | Accepted: 31 December 2023

DOI: 10.1111/acem.14875

ORIGINAL ARTICLE

Timing and causes of death to 1 year among children

presenting to emergency departments

Stefanie G. Ames MD, MS1 | Apoorva Salvi MS2 | Amber Lin MS2 | Susan Malveau MS2 |
N. Clay Mann PhD, MS1 | Peter C. Jenkins MD, MSc3 | Matthew Hansen MD, MCR2 |
4 2 2
Linda Papa MD, MSc | Sabrina Schmitz MD | Cesar Sabogal MD |
2
Craig D. Newgard MD, MPH | for the Pediatric Readiness Study Group

1
Department of Pediatrics, University of
Utah School of Medicine, Salt Lake City, Abstract
Utah, USA
Background and objectives: A better characterization of deaths in children follow-
2
Center for Policy and Research in
Emergency Medicine, Department of
ing emergency care is needed to inform timely interventions. This study aimed to
Emergency Medicine, Oregon Health & describe the timing, location, and causes of death to 1 year among a cohort of injured
Science University, Portland, Oregon, USA
3
and medically ill children.
Department of Surgery, Indiana
University School of Medicine, Methods: We conducted a retrospective cohort study of children <18 years re-
Indianapolis, Indiana, USA quiring emergency care in six states from January 1, 2012, through December 31,
4
Department of Emergency Medicine,
2017, with follow-­up through December 31, 2018, for patients who were not dis-
Orlando Regional Medical Center,
Orlando, Florida, USA charged from the emergency department (ED). In this cohort, 1-­year mortality, time
to death within 1 year, and causes of death were assessed from ED, inpatient, and
Correspondence
Stefanie G. Ames, Department of vital status records.
Pediatrics, University of Utah School of
Results: There were 546,044 children during the 6-­year period. The 1-­year mortal-
Medicine, 295 Chipeta Way, Salt Lake
City, UT 84158, USA. ity rate was 2.2% (n = 1356) for injured children and 1.4% (n = 6687) for medically ill
Email: [email protected]
children. Matched death certificates were available for 861 (63.5%) of 1356 deaths in
Funding information the injury cohort and for 4712 (70.5%) of 6687 deaths in the medical cohort. Among
Eunice Kennedy Shriver National
deaths in the injury cohort, 1274 (94.0%) occurred in the ED or hospital. The most
Institute of Child Health and Human
Development, Grant/Award Number: R24 common causes of death were motor vehicle collisions, firearm injuries, and pedes-
HD085927; U.S. Department of Health
trian injuries. Among the 6687 deaths in the medical cohort, 5081 (76.0%) children
and Human Services Health Resources
and Services Administration (Emergency died in the ED or hospital (primarily in the ED) and 1606 (24.0%) occurred after hospi-
Medical Services for Children Targeted
tal discharge. The most common causes of death were sudden infant death syndrome,
Issue Grant), Grant/Award Number:
#H34MC33243-01-01 suffocation and drowning, and congenital conditions.
Conclusions: The 1-­year mortality of children presenting to an ED is 2.2% for injured
children and 1.4% for medically ill children with most deaths occurring in the ED.
Future interventional trials, quality improvement efforts, and health policy focused in
the ED could have the potential to improve outcomes of pediatric patients.

KEYWORDS
mortality, pediatric emergency care, trauma

Supervising Editor: Jody Vogel

Acad Emerg Med. 2024;31:555–563. wileyonlinelibrary.com/journal/acem © 2024 Society for Academic Emergency Medicine. | 555
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14875 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
556 PEDIATRIC MORTALITY IN THE ED

I NTRO D U C TI O N Study setting

Children with injury and acute medical illnesses account for 30 We included children presenting to 596 EDs in six states over a
million emergency department (ED) visits each year in the United 6-­year period. To be included, EDs had to have a matched 2013
States.1 Prior work has highlighted gaps and variability in the qual- NPRP assessment, as the parent study focused on an outcome-­
ity of emergency care for children, which have spurred national ef- based evaluation of ED pediatric readiness using the 2013 NPRP
forts to improve the quality and consistency of pediatric emergency assessment.12 Matching was completed by linking the NPRP as-
2–8
care. The National Pediatric Readiness Project (NPRP) is a national sessment, a national assessment of adherence to national ED
quality improvement initiative to improve ED pediatric readiness,9 pediatric guidelines, 2 to the index ED record using hospital name,
which has been associated with improved survival among children address, and zip code. Next, we included children residing in six
with critical illness, trauma, and other clinical conditions.10–13 Two states, as determined by having a home zip code listed in one of
recent studies of ED readiness have demonstrated that children the study states, which granted approval to match pediatric ED
who die tend to do so early in the clinical course,11,12 placing extra and inpatient records to state death records. The states included
emphasis on the quality and effectiveness of care at the initial ED. Arizona, California, Florida, Iowa, Maryland, and New Jersey. We
However, the causes and changing distribution of causes by time of selected states based on broad geographic representation, avail-
pediatric deaths remain incompletely characterized. Research per- ability of the necessary hospital and patient identifiers, and ap-
taining to the timing and etiologies of death in children after presen- proval to link vital statistics death records.
tation to an ED is sparse because death is an uncommon outcome,
few large-­scale longitudinal pediatric studies exist, and discerning
the causes of death generally requires matched death certificates or Patient population
autopsy records.
A better understanding of the causes and timing of deaths in We created a patient-­level, chronological data set for consecutive
children following presentation for emergency care may facilitate children <18 years receiving care in 596 EDs, with follow-­up to
improvements in interventions, provider training, ED prepared- 365 days. We identified the first ED visit for each child (marking
ness, and prevention efforts. Furthermore, this research may in- Time 0) from January 1, 2012, through December 31, 2017, with
form the NPRP and other efforts to raise ED pediatric readiness follow-­up through December 31, 2018. Inclusion criteria at the
across the United States. Research on this topic may also inform time of the index ED visit were admission from the ED, transfer
the NPRP and efforts to raise ED pediatric readiness across the to another hospital (to another ED or to an inpatient settings), or
United States. Understanding the most common clinical condi- death in the ED. We also included matched records for inpatient
tions causing early deaths in children may facilitate targeted transfers originating in the ED to ensure the full acute care epi-
strategies to address these conditions, particularly for reversible sode was represented. We tracked mortality for every child to
causes. 365 days from the date of the index ED visit. For children trans-
In this study, we describe the timing, location, and causes of ferred to another hospital, we matched available records from the
death to 1 year among a cohort of injured and medically ill children second hospital to capture complete episodes of acute care. While
presenting to 596 EDs for care in six states, including deaths oc- we included all children meeting the above criteria, the primary
curring in the ED, inpatient, and hospital postdischarge settings. We focus was on children dying within 365 days of the index ED visit.
also characterized the types of children presenting for emergency We excluded children who were treated in EDs without a matched
services who subsequently died in the ED, during inpatient admis- NPRP assessment (based on the parent study),12 missing hospital
sion or after hospital discharge compared to children who survived disposition, transferred from the initial ED without a record from
to 1 year. the second hospital, or missing diagnosis codes or other key data
(Figure S1). To evaluate for potential differences in time to death
and causes of death among children with different clinical con-
M E TH O D S ditions, we divided the sample into children with injuries versus
medical illnesses using ED and hospital discharge International
Study design Classification of Diseases Clinical Modification (ICD) diagnosis
codes from the index ED visit and hospitalization.
This study was a secondary, descriptive analysis of a retrospective
cohort study that was reviewed and approved by institutional review
boards at Oregon Health and Science University and the University Variables
of Utah School of Medicine, which waived the requirement for in-
formed consent. We followed the Strengthening the Reporting of We included variables available from the initial ED visit and accom-
Observational Studies in Epidemiology (STROBE) cohort study panying hospitalization. These variables included age; sex; complex
guidelines.14 chronic conditions15; severity classification system (1–5 acuity scale
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14875 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
AMES et al. 557

for children presenting for emergency care, with higher numbers records). Among children dying after discharge, we quantified the
16
denoting higher clinical severity) ; ED/hospital diagnoses grouped number of postdischarge ED visits and readmissions and used death
by standardized ICD categories; hospital procedures; injury sever- certificates to describe the location of death (ED/hospital, skilled
ity and mechanism of injury (for injured children); and interhospital nursing/long-­term care, hospice, or home). We performed all anal-
transfer. For hospital procedures, we used the Agency for Healthcare yses separately for the injury and medical cohorts. Based on our
Research and Quality Clinical Classification System (CCS)17 and data use agreements, cells with sparse data (n < 30) are reported as
mapped CCS categories to standardized operative domains. For in- “n < 30,” with the accompanying percentage rounded to the nearest
jured children, we used the Abbreviated Injury Scale (AIS) score18 whole number. All analyses were performed in SAS 9.4.
and Injury Severity Score (ISS)19 to measure injury severity. Because
AIS and ISS are not included in administrative data, we used ICD ISS
Map v2.0 (AAAM, Chicago, IL) to convert ICD-­9-­CM and ICD-­10-­CM R E S U LT S
diagnosis codes into standardized injury severity measures, which
we have validated against hand-­abstracted values. 20 This study included 546,044 children who presented to an ED dur-
ing the 6-­year period, including 62,710 (11.5%) injured children and
483,293 (88.5%) children with acute medical illness (Figure S1). The
Outcomes 1-­year mortality rate was 2.2% (n = 1356) for injured children and
1.4% (n = 6687) for children with medical illness. Matched death cer-
The primary outcome was mortality, measured from the date of tificates were available for 861 (63.5%) of 1356 deaths in the injury
initial ED presentation to 365 days. To identify deaths, we used ED, cohort and for 4712 (70.5%) of 6687 deaths in the medical cohort.
inpatient, and vital statistics death records. To link vital statistics re- The median (IQR) time to death was 0 (0–2) days for injured children
cords to the index ED visit, we used probabilistic linkage21 (LinkSolv and 0 (0–8) days for medical children. In Table 1, we compare chil-
22,23
v9, Strategic Matching) and established linkage routines. Linkage dren who survived versus died within 1 year of the initial ED presen-
variables included date of birth, home zip code, date of service, sex, tation. Among the injury cohort, deaths were more common among
race, and ethnicity. We validated the sensitivity and specificity of adolescents, certain injury mechanisms (firearms, motor vehicle,
linkage in the parent study, estimating that these processes identi- and pedestrian events), high injury severity, and serious traumatic
fied 92.5% of all deaths among injured children and 85.6% of deaths brain injury. For the medical cohort, deaths occurred predominantly
among medically ill children within 1 year.12 among young children (<1 year), those with chronic medical condi-
To describe the causes of death, we categorized the ICD-­10 di- tions, children with circulatory and nervous system diseases, and
agnosis codes listed on death certificates into standardized disease-­ higher clinical acuity.
24
based ICD-­10 categories. Because the category of “symptoms, Characteristics of children who died are detailed in Table 2.
signs, and abnormal clinical and laboratory findings, not elsewhere Among the 1356 deaths in the injury cohort, 1274 (94.0%) died
classified” was a common and nonspecific category for deaths in the in the ED or hospital, with relatively few deaths after discharge.
medical cohort, we created subcategories based on the frequency For the injury cohort, the most common causes of deaths occur-
of specific diagnosis codes (e.g., sudden infant death syndrome). ring during the index ED/hospital visit were due to motor vehi-
Similarly, for deaths in the category “injury, poisoning and certain cle crashes, pedestrian events, and firearms, while the majority
other consequences of external causes,” we created subcategories of deaths after discharge were from medical causes. Among in-
for different mechanisms of injury, poisoning, and drowning/suffo- jured children dying after discharge, most deaths occurred weeks
cation. To describe and characterize causes of death, we only used after discharge (median [IQR] time from discharge to death 45
information from death certificates. Among children who survived [5–160] days) and 60% occurred during a subsequent ED visit or
their initial hospitalization, we tracked subsequent ED visits and hos- readmission. Among the 6687 deaths in the medical cohort, 5081
pital readmissions to 365 days from the date of initial ED presenta- (76.0%) children died during their index hospitalization (primarily
tion as a secondary outcome. in the ED) and 1606 (24.0%) occurred after hospital discharge.
Early (i.e., ED) deaths in the medical cohort were most commonly
due to sudden unexplained infant death (SUID), drowning/suffoca-
Statistical analysis tion, and perinatal/congenital conditions. The causes of inpatient
deaths differed from those in the ED, with perinatal/congenital
We used descriptive statistics to compare children who survived conditions being most common. Among deaths in the medical
versus died within 1 year of their initial ED visit. Among children who cohort occurring after discharge, neoplasms/blood/immune dis-
died within 1 year, we described the timing, causes, and location of eases and perinatal/congenital conditions were most common and
death. Causes of death were limited to children with matched vital the majority occurred weeks after discharge (median time from
statistics death certificates. We measured time to death in days, discharge to death = 48 days [IQR 13–145]). Repeat ED visits and
based on the interval between the date of the initial ED visit and the readmissions were common (54.1%), and 60% of postdischarge
date of death (as extracted from the ED, inpatient, and vital statistics deaths occurred during a subsequent ED visit or readmission.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14875 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
558 PEDIATRIC MORTALITY IN THE ED

TA B L E 1 Comparison of children who survived versus died TA B L E 1 (Continued)

within 1 year of ED presentation, separated by injury versus
Survival to Death within
medical illness presentations.
1 year, 1 year,
Survival to Death within
n = 61,215 n = 1356
1 year, 1 year,
(97.8%) (2.2%)
n = 61,215 n = 1356
Top five diagnoses at ED presentation
(97.8%) (2.2%)
Circulatory diseases 36,651 (7.7) 4475 (66.9)
Injury cohort (n = 62,571)
Respiratory diseases 195,606 (41.0) 2167 (32.4)
Age (years), median (IQR) 10 (4–15) 11 (3–16)
Perinatal and congenital 105,618 (22.2) 1924 (28.8)
Age group (years) conditions
<1 5326 (8.7) 134 (9.9) Endocrine, nutritional, 119,681 (25.1) 1590 (23.8)
1–4years 17,686 (28.9) 385 (28.4) digestive, and
metabolic diseases
5–12 15,054 (24.6) 221 (16.3)
Nervous system diseases 55,532 (11.7) 1567 (23.4)
13–17 23,149 (37.8) 616 (45.4)
Severe clinical severity 247,935 (52.0) 6273 (93.8)
Female 22,427 (36.6) 451 (33.3)
(score 4–5)
Chronic complex conditions
Major surgery 64,648 (13.6) 1184 (17.7)
1 2924 (4.8) 204 (15.0)
Interhospital transfer 24,818 (5.2) 241 (3.6)
≥2 831 (1.4) 163 (12.0)
Note: Data are reported as n (%) unless otherwise specified.
Mechanism of injury
Abbreviations: AIS, Abbreviated Injury Scale; ISS, Injury Severity Score.
Fall 23,742 (38.8) 98 (7.2)
Firearm 1336 (2.2) 289 (21.3)
Stab/penetrating 3104 (5.1) 37 (2.7) Figure 1 demonstrates the number of deaths in the injury cohort
Assault 6013 (9.8) 70 (5.2) by day, beginning with the date of the index ED visit. The majority
Motor vehicle 7634 (12.5) 327 (24.1) of deaths in the injured cohort were on Days 0–1 and half of early

Pedestrian or bicycle 6096 (10.0) 276 (20.4) deaths were due motor vehicle collision (26%) or pedestrian/bicycle-­
related injury (24%). Following the initial ED presentation, injury-­
Other 13,290 (21.7) 259 (19.1)
related causes of death continued to predominate until 30 days after
Injury severity
the initial ED presentation, with medical causes increasing to 56% of
ISS 0–8 43,738 (71.4) 211 (15.6)
all postdischarge deaths.
ISS 9–15 13,962 (22.8) 503 (37.1)
In Figure 2, we show the number of deaths by day for the medical
ISS ≥ 16 3515 (5.7) 642 (47.3)
cohort. The majority of deaths occurred within 0–1 days of the index
AIS head ≥ 3 7752 (12.7) 1102 (81.3) ED presentation, with the most common causes of death being SUID
AIS chest ≥ 3 2509 (4.1) 793 (58.5) (20%), drowning or suffocation (17%), and congenital (14%). For chil-
AIS abdomen ≥ 3 689 (1.1) 739 (54.5) dren dying from 2 to 365 days, congenital conditions and neoplasms
AIS extremity ≥ 3 5809 (9.5) 708 (52.2) were the most common causes. There was a small percentage (7%)
Severe clinical severity 29,518 (48.2) 1272 (93.8) of children in the medical cohort who died of injury causes during
(score 4–5) the 0-­ to 1-­day and 30-­ to 365-­day intervals. Particularly among
Major surgery 4324 (7.1) 652 (48.1) children dying early, these were children coded as having a medical
Interhospital transfer 3776 (6.2) 47 (3.5) diagnosis in the ED/hospital (including nonspecific signs and symp-
Medical cohort (n = 483,293) (n = 476,606, (n = 6687, toms), who ultimately had an injury cause of death. Because we con-
98.6%) 1.4%) sidered injuries, poisoning, self-­harm, and suffocation/drowning as
Age (years), mean (range) 3 (0–11) 0 (0–7) separate subcategories (all of which are included under “injury” in
Age group (years) the ICD-­10 framework), injury causes were not related to the other
<1 167,031 (35.0) 3563 (53.3) subcategories.

1–4 126,964 (26.6) 1447 (21.6)

5–12 81,762 (17.2) 781 (11.7)
DISCUSSION
13–17 100,849 (21.2) 896 (13.4)
Female 221,255 (46.4) 2867 (42.9)
In this large, multistate cohort of 546,044 children presenting to an
Chronic complex conditions
ED between 2012 and 2017, we found an overall 1-­year mortality
1 42,534 (8.9) 910 (13.6)
rate of 2.2% for injured patients and 1.4% for medical patients. The
≥2 20,300 (4.3) 1269 (19.0) most common mechanisms associated with death among injured
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14875 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
AMES et al. 559

TA B L E 2 Timing and etiology of death for children who died within 365 days of the ED visit.
ED deaths Inpatient deaths Deaths after discharge
Injury cohort deaths (n = 1356) 693 (51.1) 581 (42.9) 82 (6.1)
Days to death from ED presentation, median (IQR)a 0 (0–0) 2 (1–4) 53 (10–172)
Cause of deathb (n = 858) n = 395 n = 381 n = 82
MVC and other transport 107 (27.1) 102 (26.8) <30 (11)
Pedestrian/bicycle 103 (26.1) 76 (19.9) <30 (5)
Firearm 63 (15.9) 42 (11.0) <30 (7)
Assault <30 (7) 43 (11.3) <30 (6)
Falls/mechanical forces <30 (7) <30 (7) <30 (2)
Self-­harm <30 (7) 34 (8.9) <30 (5)
Neoplasms, blood and immune diseases 0 (0) <30 (<1) <30 (18)
Other injury causes <30 (2) <30 (2) <30 (2)
Medical causes <30 (8) 45 (11.8) 35 (42.7)
Median days from discharge to death (IQR)c — — 45 (5–160)
Repeat ED visits or readmissions after discharge (any) — — <30 (31.7)
Location of postdischarge death
ED — — <30 (20)
Hospital Inpatient — — 33 (40.2)
Hospice — — <30 (9)
Home — — <30 (21)
Other — — <30 (11)
Medical cohort deaths (n = 6687) 4150 (62.1) 931 (13.9) 1606 (24)
Days to death from ED presentation, median (IQR)a 0 (0–0) 3 (1–10) 54 (18–147)
b
Cause of death (n = 4712) n = 2474 n = 632 n = 1606
Sudden infant death syndrome 532 (21.5) 11 (1.7) 21 (1.3)
Suffocation/drowning 442 (17.9) 64 (10.1) 55 (3.4)
Perinatal and congenital conditions 322 (13.0) 133 (21.0) 304 (18.9)
Respiratory diseases 145 (5.9) 62 (9.8) 86 (5.4)
Circulatory diseases 121 (4.9) 61 (9.7) 76 (4.7)
Nervous system diseases 78 (3.2) 59 (9.3) 159 (9.9)
Infectious diseases 53 (2.1) 53 (8.4) 70 (4.4)
Neoplasms, blood and immune diseases 34 (1.4) 62 (9.8) 517 (32.3)
Endocrine, nutritional, digestive, and metabolic diseases 71 (2.9) 49 (7.8) 122 (7.6)
Other medical causes <30 (<1) <30 (<1) <30 (2)
Injury causes, including poisoning and self-­harm 294 (11.9) 47 (7.4) 130 (8.1)
Unknown 366 (14.8) <30 (3.9) 39 (2.4)
Days from discharge to death, median (IQR)c — — 48 (13–145)
Repeat ED visits or readmissions after discharge (any) — — 868 (54.1)
Location of postdischarge death
ED — — 765 (47.6)
In-­hospital — — 191 (11.9)
Hospice — — 80 (5.0)
Home — — 490 (30.5)
Other — — 72 (4.5)
Unknown <30 (<1)

Note: Data are reported as n (%) unless otherwise specified.

Abbreviation: MVC, motor vehicle collision.
a
Days to death was based on patients with known dates for the initial ED visit and day of death (1351 of 1356 deaths in the injury cohort and 6676 of
6687 deaths in the medical cohort).
b
Cause of death was only available for pediatric deaths with a linked death certificate, representing 858 (63.3%) of 1356 deaths in the injury cohort
and 4712 (70.5%) of 6687 deaths in the medical cohort. For cells with less than 30 patients, absolute numbers are not presented and percentages are
rounded to whole numbers based on our data use agreements.
c
Among patients who died after discharge, day -­to death was based on patients with observed values for date of discharge and date of death 73 of 82
deaths in the injury cohort and 1527 of 1696 deaths in the medical cohort.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14875 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
560 PEDIATRIC MORTALITY IN THE ED

(A) 0 - 1 days: causes of death

1000
MVC - 26%
Pedestrian/bicycle - 24%
900
Firearm - 16%
781 Other injury mechanism- 9%
800
Assault - 8%

700

2-7 days: causes of death

Number of deaths

600 MVC - 25%

Pedestrian/bicycle - 16%
500 Medical causes - 12%
Assault - 11%
400 Self- harm - 8%

300

193
200

100 73 60 45 39
17 20
0
0 1 2 3 4 5 6 7
Days since ini al ED visit

(B) 50

45

8 -29 days: causes of death

40
MVC - 31%
Pedestrian/bicycle - 22%
35 Medical causes - 20%
Assault - 12%
30
Drowning/suffocaon - 4%
Number of deaths

30 to 365 days: causes of death

25 Medical causes - 56%
Firearm -12%
20 MVC - 8%
Assault - 8%
Pedestrian/bicycle - 6%
15

10

0
8 10 12 14 17 19 21 24 28 31 34 37 42 46 51 60 72 76 96 111 122 134 146 150 166 172 180 196 234 253 269 293 325 339 354 365
Days since ini al ED visit

F I G U R E 1 Timing and causes of death* among children presenting to the ED with injury (n = 1351 deaths). (A) Deaths within 0–7 days
of ED presentation (n = 1228 deaths). (B) Deaths within 8–365 days of ED presentation (n = 123 deaths). *Deaths from the injury cohort
included in the figures are limited to those where time to death was known (1351 of 1356 deaths). The causes of death are based on patients
with matched death certificates (n = 858 of 1356 deaths). Note change in scale of y-­axis between A and B for clarity.

patients were motor vehicle collisions, firearms, and pedestrian in- is the first, to our knowledge, to describe the etiology and timing
juries. Over 80% of deaths after injury involved a serious traumatic of death in a large multistate cohort of children. The results of this
brain injury. For children in the medical cohort, the most common study have the potential to inform future pediatric interventional
causes of early death were sudden infant death syndrome (SIDS), trials (e.g., the type and timing of interventions to avert death),
drowning or suffocation, and congenital conditions. Importantly, investments in education and quality initiatives, and health policy
most deaths in both cohorts occurred in the ED (which was most focused on children.
pronounced in the medical cohort), illustrating the importance of The timing of deaths among children presenting for emer-
early care and a short time window for interventions. Our study gency care highlights a key window of opportunity for lifesaving
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14875 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
AMES et al. 561

(A)
5000 0 - 1 days: causes of death
SIDS -20%
4500 Drowning/suffocaon - 17%
4129 Congenital - 14%
4000 Unknown – 14%
Injury - 7%
3500

2-7 days: causes of death

Number of deaths

3000
Congenital - 17%
Neoplasms - 16%
2500
Drowning/suffocaon - 12%
Respiratory - 11%
2000
Nervous system - 10%
1500

1000

500 364
132 108 87 65 70 48
0
0 1 2 3 4 5 6 7
Days since ini al ED visit
(B)
50
8 -29 days: causes of death
Neoplasms - 28%
45 Congenital - 23%
Nervous system - 11%
Endocrine - 9%
30-365 days: causes of death
40 Circulatory - 8%
Neoplasms - 32%
Congenital - 18%
35 Nervous system- 10%
Endocrine - 7%
Injury – 7%
30
Number of deaths

25

20

15

10

0
8
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
100
104
108
113
117
122
126
130
134
139
143
147
151
155
159
163
167
173
178
182
186
190
194
199
203
207
211
217
221
227
235
239
244
248
253
258
263
269
273
277
285
290
294
298
303
309
314
319
323
328
332
339
345
351
355
360
364
Days since ini al ED visit

F I G U R E 2 Timing and causes of death* among children presenting to the ED with acute medical illness (n = 6676 deaths). (A) Deaths
within 0–7 days of ED presentation (n = 5003 deaths). (B) Deaths within 8–365 days of ED presentation (n = 1673 deaths). *Patient deaths
included in the figures are limited to those where time-­to-­death was known (6676 of 6687 deaths). The causes of death are based on
patients with matched death certificates (n = 4712 of 6687 deaths). Note change in scale of y axis between panels A and B for clarity.
*Causes of death are provided for deaths with a matched death record (X of X deaths, X%). Note change in scale of y-­axis between panels (A)
and (B) for clarity.

interventions in pediatric health care. Although pediatric deaths required to recruit and consent patients, and loss to follow-­up. 26
are rare compared to adult populations, the majority of deaths oc- Addressing these concerns through deferred consent, waivers of
curred in the ED, rather than other health care settings. 25 These consent, and improved staffing for 24/7 recruitment may be nec-
findings suggest that the opportunity for rescue is early in presen- essary to conduct high-­quality emergency care research aimed at
tation and interventions aimed at reduction of mortality should reducing pediatric mortality. 27
begin in the prehospital and ED settings. Unfortunately, research Based on our results, quality improvement and educational ef-
(e.g., interventional trials) in the prehospital environment and forts to improve early pediatric resuscitation have the potential
ED settings for life-­t hreatening causes involve many challenges, to substantially impact pediatric mortality, given the high burden
including identification and enrollment of eligible patients, time of death in the ED. One potential mechanism for improvement is
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14875 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
562 PEDIATRIC MORTALITY IN THE ED

through the NPRP initiative to increase ED pediatric readiness. This codes between ED/hospital versus death certificate data sources for
national quality improvement effort focuses on ensuring all EDs these patients, there were limitations with the detail of diagnosis
have the necessary equipment, staffing, and policies to provide codes from the different sources and relatively small numbers of pa-
high-­quality emergency care for children. 2 Our data support these tients, which prevented the opportunity to draw conclusions.
essential efforts to raise all EDs to high pediatric readiness to reduce
mortality in children. Future health policy aimed at ensuring high pe-
diatric readiness of EDs may lead to improved pediatric outcomes CO N C LU S I O N S
and reduced mortality in children requiring ED care.10–13 Educational
efforts, including simulation training for pediatric resuscitation in The 1-­year mortality of children presenting to an ED was 2.2% for
28
the prehospital and ED settings, may also be of benefit. injured children and 1.4% for children with medical illness, with most
In addition to timing of death, our study characterized the most deaths occurring early in the child's clinical course. Among children pre-
common etiologies of mortality by the time from ED presentation. senting with injury conditions, the most common causes of death were
Among the injury cohort of children, motor vehicle collisions, firearm motor vehicle collisions, firearms, and pedestrian/bicycle events. For
injuries, and pedestrian/bicycle injuries were the leading mechanisms children presenting with acute medical illness, the most common causes
associated with death. Recent data generally support this finding29; of death were sudden unexplained infant death, drowning/suffocation,
however, in recent years firearms have overtaken motor vehicle and congenital/perinatal conditions, with cancer being a common cause
crashes as the most common cause of death in children.30 Among of death following hospital discharge. Future interventional trials, qual-
types of injury, serious traumatic brain injury occurred in the major- ity improvement efforts, and health policy should consider the timing
ity of children who died within 1 year, highlighting the importance of and etiology of pediatric mortality to target preventable mortality.
traumatic brain injury prevention and early treatment among children.
In the medical cohort, SIDS and drowning/suffocation were the lead- AU T H O R C O N T R I B U T I O N S
31,32
ing causes of death, representing potentially preventable deaths. Stefanie G. Ames conceptualized and designed the study, drafted
Knowledge of the most common etiologies of death and high-­risk the initial manuscript, and critically reviewed and revised the manu-
children can be utilized by emergency medical services agencies, EDs, script. Craig D. Newgard, Amber Lin, and Apoorva Salvi conceptu-
and public health organizations to inform future research, targeted in- alized and designed the study, conducted the statistical analysis,
terventions, quality improvement efforts, and prevention strategies in and critically reviewed and revised the manuscript. N. Clay Mann,
specific patient populations. In addition, this information can shape fu- Peter C. Jenkins, Matthew Hansen, Linda Papa, Sabrina Schmitz, and
ture policy, NPRP efforts, and other national efforts focused on child Cesar Sabogal conceptualized the study and critically reviewed and
advocacy related to addressing the leading causes of death in children. revised the manuscript. All authors approved the final manuscript as
submitted and agree to be accountable for all aspects of the work.

LI M ITATI O N S F U N D I N G I N FO R M AT I O N
This project was supported by the Eunice Kennedy Shriver National
There were limitations in the study. While the data were gener- Institute of Child Health and Human Development (grant R24
ally complete for time to death, the causes of death were missing HD085927) and the U.S. Department of Health and Human Services
for about a third of pediatric deaths due to reliance on matched Health Resources and Services Administration (Emergency Medical
death certificates. Among early deaths in the medical cohort with Services for Children Targeted Issue Grant, grant H34MC33243-­
a matched death certificate, unknown causes were common—these 01-­01). The content is solely the responsibility of the authors. The
cases may reflect SIDS cases that did not have an autopsy com- funding organizations had no role in any of the following: design and
pleted or where there were other reasons to list “unknown” as the conduct of the study; collection, management, analysis, and inter-
cause of death. Also, death certificates do not always provide the pretation of the data; preparation, review, or approval of the manu-
desired granularity for cause of death, particularly if an autopsy is script; or decision to submit the manuscript for publication.
not performed. Therefore, the causes of death were limited by the
presence of a matched death certificate and the available informa- C O N F L I C T O F I N T E R E S T S TAT E M E N T
tion included on the document. Also, we used three data sources to The authors declare no conflicts of interest.
identify deaths in the ED and hospital (i.e., statewide ED databases,
statewide inpatient databases, and statewide vital statistics records), ORCID
which provided a comprehensive surveillance strategy. However, to Stefanie G. Ames https://orcid.org/0000-0002-4972-4777
identify deaths after discharge, we primarily relied on matched vital Peter C. Jenkins https://orcid.org/0000-0002-8527-8268
statistics records, which may have undercounted such deaths. Matthew Hansen https://orcid.org/0000-0003-2958-6423
We were unable to assess whether ED and hospital diagnoses Linda Papa https://orcid.org/0000-0002-1175-3839
from the index visit were related to causes of death for children Sabrina Schmitz https://orcid.org/0000-0001-8810-3757
who died after discharge. While we attempted to align the diagnosis Craig D. Newgard https://orcid.org/0000-0003-1083-3455
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14875 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
AMES et al. 563

18. Abbreviated Injury Scale (AIS). Manual. Association for the

REFERENCES Advancement of Automotive Medicine; 2005:2005.
1. McDermott KW, Stocks C, Freeman WJ. Statistical Brief #242. 19. Baker SP, O'Neill B, Haddon W Jr, Long WB. The injury severity
Overview of Pediatric Emergency Department Visits, 2015. August score: a method for describing patients with multiple injuries and
2018. Accessed August 1, 2023. https://​w ww.​hcup-­​us.​ahrq.​gov/​ evaluating emergency care. J Trauma. 1974;14(3):187-196. https://​
repor​t s/​statb​riefs/​​sb242​-­​Pedia​tric-­​ED-­​Visit​s-­​2015.​pdf www.​ncbi.​nlm.​nih.​gov/​pubmed/​4814394
2. Gausche-­Hill M, Ely M, Schmuhl P, et al. A national assessment 20. Fleischman RJ, Mann NC, Dai M, et al. Validating the use of ICD-­9
of pediatric readiness of emergency departments. JAMA Pediatr. code mapping to generate injury severity scores. J Trauma Nurs.
2015;169(6):527-534. doi:10.1001/jamapediatrics.2015.138 2017;24(1):4-14. doi:10.1097/JTN.0000000000000255
3. Chime NO, Katznelson J, Gangadharan S, et al. Comparing practice 21. Jaro MA. Probabilistic linkage of large public health data files. Stat
patterns between pediatric and general emergency medicine phy- Med. 1995;14(5–7):491-498. (In eng). http://​w ww.​ncbi.​nlm.​nih.​gov/​
sicians: a scoping review. Pediatr Emerg Care. 2017;33(4):278-286. pubmed/​7792443
doi:10.1097/PEC.0000000000000557 22. Newgard CD. Validation of probabilistic linkage to match de-­
4. Hampers LC, Faries SG. Practice variation in the emergency man- identified ambulance records to a state trauma registry. Acad Emerg
agement of croup. Pediatrics. 2002;109(3):505-508. doi:10.1542/ Med. 2006;13(1):69-75. doi:10.1197/j.aem.2005.07.029
peds.109.3.505 23. Newgard CD, Malveau S, Zive D, Lupton J, Lin A. Building a lon-
5. Bekmezian A, Hersh AL, Maselli JH, Cabana MD. Pediatric emer- gitudinal cohort from 91–-­1 to 1-­year using existing data sources,
gency departments are more likely than general emergency probabilistic linkage, and multiple imputation: a validation study.
departments to treat asthma exacerbation with systemic cortico- Acad Emerg Med. 2018;25:1268-1283. doi:10.1111/acem.13512
steroids. J Asthma. 2011;48(1):69-74. doi:10.3109/02770903.20 24. International Statistical Classification of Diseases and Related
10.535884 Health Problems (ICD), ICD-­10 Version:2019. World Health
6. Abulebda K, Whitfill T, Mustafa M, et al. Improving pediatric read- Organization. Accessed July 15, 2023. https://​icd.​who.​int/​brows​
iness and clinical care in general emergency departments: a multi- e10/​2019/​en
center retrospective cohort study. J Pediatr. 2022;240:241-248.e1. 25. O'Malley P, Barata I, Snow S, American Academy of Pediatrics
doi:10.1016/j.jpeds.2021.08.084 Committee on Pediatric Emergency Medicine; American College of
7. Shibata S, Khemani RG, Markovitz B. Patient origin is associ- Emergency Physicians Pediatric Emergency Medicine Committee;
ated with duration of endotracheal intubation and PICU length Emergency Nurses Association Pediatric Committee. Death of a
of stay for children with status asthmaticus. J Intensive Care Med. child in the emergency department. J Emerg Nurs. 2014;40(4):e83-
2014;29(3):154-159. doi:10.1177/0885066613476446 e101. doi:10.1016/j.jen.2014.05.003
8. Brown KM, Ackerman AD, Ruttan TK, et al. Access to optimal 26. Brown J, Lane A, Cooper C, Vassar M. The results of random-
emergency care for children. Ann Emerg Med. 2021;77(5):523-531. ized controlled trials in emergency medicine are frequently
doi:10.1016/j.annemergmed.2021.03.034 fragile. Ann Emerg Med. 2019;73(6):565-576. doi:10.1016/j.
9. National Pediatric Readiness Project. Emergency Medical Services annemergmed.2018.10.037
for Children (EMSC) Innovation and Improvement Center. Accessed 27. Harron K, Woolfall K, Dwan K, et al. Deferred consent for ran-
August 1, 2023. https://​emsci​mprov​ement.​center/​domai​ns/​pedia​ domized controlled trials in emergency care settings. Pediatrics.
tric-­​readi​ness-­​proje​c t/​ 2015;136(5):e1316-e1322. doi:10.1542/peds.2015-0512
10. Ames SG, Davis BS, Marin JR, et al. Emergency department pe- 28. Whitfill T, Gawel M, Auerbach M. A simulation-­based quality im-
diatric readiness and mortality in critically ill children. Pediatrics. provement initiative improves pediatric readiness in community
2019;144(3):1-9. doi:10.1542/peds.2019-0568 hospitals. Pediatr Emerg Care. 2018;34(6):431-435. doi:10.1097/
11. Newgard CD, Lin A, Goldhaber-­Fiebert JD, et al. Association of PEC.0000000000001233
emergency department pediatric readiness with mortality to 1 29. Cunningham RM, Walton MA, Carter PM. The major causes of
year among injured children treated at trauma centers. JAMA Surg. death in children and adolescents in the United States. N Engl J Med.
2022;157(4):e217419. doi:10.1001/jamasurg.2021.7419 2018;379(25):2468-2475. doi:10.1056/NEJMsr1804754
12. Newgard CD, Lin A, Malveau S, et al. Emergency department 30. Goldstick JE, Cunningham RM, Carter PM. Current causes of death
pediatric readiness and short-­term and long-­term mortality in children and adolescents in the United States. N Engl J Med.
among children receiving emergency care. JAMA Netw Open. 2022;386(20):1955-1956. doi:10.1056/NEJMc2201761
2023;6(1):e2250941. doi:10.1001/jamanetworkopen.2022.50941 31. Jullien S. Sudden infant death syndrome prevention. BMC Pediatr.
13. Newgard CD, Lin A, Olson LM, et al. Evaluation of emer- 2021;21(Suppl 1):320. doi:10.1186/s12887-021-02536-z
gency department pediatric readiness and outcomes among 32. Denny SA, Quan L, Gilchrist J, et al. Prevention of drowning.
US trauma centers. JAMA Pediatr. 2021;175:947. doi:10.1001/ Pediatrics. 2021;148:2. doi:10.1542/peds.2021-052227
jamapediatrics.2021.1319
14. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC,
Vandenbroucke JP. The Strengthening the Reporting of
Observational Studies in Epidemiology (STROBE) statement: S U P P O R T I N G I N FO R M AT I O N
guidelines for reporting observational studies. J Clin Epidemiol. Additional supporting information can be found online in the
2008;61(4):344-349. doi:10.1016/j.jclinepi.2007.11.008 Supporting Information section at the end of this article.
15. Feudtner C, Feinstein JA, Zhong W, Hall M, Dai D. Pediatric complex
chronic conditions classification system version 2: updated for ICD-­10
and complex medical technology dependence and transplantation. BMC
Pediatr. 2014;14:199. https://doi.org/10.1186/1471-2431-14-199
16. Alessandrini EA, Alpern ER, Chamberlain JM, et al. Developing a How to cite this article: Ames SG, Salvi A, Lin A, et al. Timing
diagnosis-­based severity classification system for use in emergency and causes of death to 1 year among children presenting to
medical services for children. Acad Emerg Med. 2012;19(1):70-78. emergency departments. Acad Emerg Med. 2024;31:555-563.
doi:10.1111/j.1553-­2712.2011.01250.x
doi:10.1111/acem.14875
17. Clinical Classifications Software (CCS) for ICD-­9-­CM. Agency for
Healthcare Research and Quality. Accessed July 15, 2023. https://​
www.​hcup-­​us.​ahrq.​gov/​tools​softw​are/​ccs/​ccs.​jsp
Received: 10 August 2023 | Revised: 11 December 2023 | Accepted: 27 December 2023

DOI: 10.1111/acem.14873

ORIGINAL ARTICLE

The emergency department trigger tool: Multicenter trigger

query validation

Richard T. Griffey MD, MPH1 | Ryan M. Schneider ACNP-­BC, CPPS1 |

Keith E. Kocher MD, MPH2 | Edmund S. H. Kwok MD, MSc3 | Ellen Salmo PA-­C, MSHS1 |
Nora Malone BSN, RN1 | Carrie Smith MSN2 | Catie Guarnacia MSN, RN2 |
April Rick BScN, RN3 | Tamara Clavet RN3 | Phil Asaro MD1 | Rich Medlin MD2 |
Alexandre A. Todorov PhD4

1
Department of Emergency Medicine,
Washington University School of Abstract
Medicine, St. Louis, Missouri, USA
Objectives: We previously described derivation and validation of the emergency de-
2
Department of Emergency Medicine and
Learning Health Sciences, University of
partment trigger tool (EDTT) for adverse event (AE) detection. As the first step in
Michigan, Ann Arbor, Michigan, USA our multicenter study of the tool, we validated our computerized screen for triggers
3
Department of Emergency Medicine, The against manual review, establishing our use of this automated process for selecting
Ottawa Hospital, Ottawa, Ontario, Canada
4 records to review for AEs.
Department of Psychiatry, Washington
University School of Medicine, St. Louis, Methods: This is a retrospective observational study of visits to three urban, aca-
Missouri, USA
demic EDs over 18 months by patients ≥ 18 years old. We reviewed 912 records: 852
Correspondence with at least one of 34 triggers found by the query and 60 records with none. Two
Richard T. Griffey, MD, MPH, Washington
first-­level reviewers per site each manually screened for triggers. After completion,
University School of Medicine, Campus
Box 8072, 660 S. Euclid Ave., Barnes-­ computerized query results were revealed, and reviewers could revise their findings.
Jewish Hospital, St. Louis, MO 63117,
Second-­level reviewers arbitrated discrepancies. We compare automated versus
USA.
Email: [email protected] manual screening by positive and negative predictive values (PPVs, NPVs), present
population trigger frequencies, proportions of records triggered, and how often man-
Funding information
Agency for Healthcare Research and ual ratings were changed to conform with the query.
Quality, Grant/Award Number: R01
Results: Trigger frequencies ranged from common (>25%) to rare (1/1000) were com-
HS027811#x2010;01
parable at U.S. sites and slightly lower at the Canadian site. Proportions of triggered
records ranged from 31% to 49.4%. Overall query PPV was 95.4%; NPV was 99.2%.
PPVs for individual trigger queries exceeded 90% for 28–31 triggers/site and NPVs
were >90% for all but three triggers at one site. Inter-­rater reliability was excellent,
with disagreement on manual screening results less than 5% of the time. Overall, re-
viewers amended their findings 1.5% of the time when discordant with query find-
ings, more often when the query was positive than when negative (47% vs. 23%).
Conclusions: The EDTT trigger query performed very well compared to manual re-
view. With some expected variability, trigger frequencies were similar across sites and
proportions of triggered records ranged 31%–49%. This demonstrates the feasibility
and generalizability of implementing the EDTT query, providing a solid foundation for
testing the triggers’ utility in detecting AEs.

Supervising Editor: Kristin Laura Rising

564 | © 2024 Society for Academic Emergency Medicine. wileyonlinelibrary.com/journal/acem Acad Emerg Med. 2024;31:564–575.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFEY et al. 565

I NTRO D U C TI O N of overall cases or the denominator of at-­risk cases. Knowing a trig-

ger's distribution in a sample and the population allows estimates
Traditional approaches for emergency department (ED) surveillance for proportions and denominators for associated AEs. For example,
of adverse events (AEs), defined as “unintended physical injury re- if we know (via the trigger “M19: propofol”) the number of visits in
sulting from or contributed to by medical care that requires addi- which this medication was given in our sample and in the general
tional monitoring, treatment or hospitalization, or that results in population, and we know how often AEs were identified in cases in
death,”1 typically consist of reviews of all records meeting common our sample in which propofol was given, then we have a much better
screening criteria (e.g., deaths, 72-­h return visits with admission, up- sense for the scope and magnitude of this issue. In our single-­center
grades in care from an inpatient floor to an intensive care unit [ICU]). study we estimated that propofol-­related AEs occurred in 33% of
Along with event reporting systems, these methods are generally cases in which this medication was used, prompting an ongoing qual-
low yield for identifying harm, 2–7 missing many AEs,8 and resulting ity improvement initiative. 27 Third, knowing the trigger distribution
9
in reviews of records that do not involve AEs. This is inefficient and in our overall population and trigger associations with harm, we can
likely underestimates patient harm, 5 increasing the likelihood that project anticipated levels of harm related to a trigger on a population
5,10,11
AEs go undiscovered and unaddressed. New, efficient methods basis. For example, we estimate that 9% of our annual ED visits are
for AE detection and measurement are needed. for AEs that are present on arrival, for harm that occurred elsewhere
Trigger tools, such as the Global Trigger Tool (GTT) pioneered in the health care system. 28
by the Institute for Healthcare Improvement, offer improved and Most trigger tool applications include a first-­level reviewer
more efficient surveillance that outperforms traditional methods screening a random sample of records for the presence of triggers
for detecting and characterizing all-­cause harm.8,12,13 Trigger tools using a paper list, organized in modules to facilitate an efficient
have been developed for a range of clinical settings across the stepwise approach to manual screening (e.g., laboratory values, vital
globe.1,14–21 These tools detect a broader scope of events, establish signs, patient care elements, medications, procedures), discarding re-
a baseline to assess the effectiveness of improvement efforts, and cords without triggers and reviewing only those with triggers. With
identify priority areas for directing resources to improve quality and an automated trigger tool, triggers are mapped to structured fields in
patient safety. Trigger tools consist of an initial screening of a sample the electronic medical record and the computer query performs this
of records by first-­level reviewers (L1, typically a nurse) for the pres- screening step on all records in the data extract provided (e.g., prior
ence of pre-­defined “triggers”—events that increase the likelihood month). This identifies all records with triggers, from which one can
that an AE is present. For example, the trigger “international normal- select records for review. This can be done using random sampling
ized ratio (INR) > 5” increases the likelihood that a bleeding event or for a higher yield one can rank order records using trigger weights
occurred. The finding of any trigger prompts an in-­depth review for (strength of associations between a trigger and an AE) or the count
evidence of AE(s). Putative AEs undergo confirmatory second level of triggers per record to create a risk distribution and select records
(L2) physician review. that are high risk or based on the presence of specific triggers, etc.
We previously created a trigger tool specific to the ED (EDTT), To use the computer query in this fashion, it must first be validated
describing its development, 22 derivation, 23 use, 24 validation, 25 and for reliability against the existing standard of manual review.
26 27
details of near misses and harm detected in the ED, present on A computerized approach to detection of harm, augmented with
arrival to the ED, 28 and among postacute care and long-­term care natural language processing (NLP) and machine learning, is the fu-
patients. 29 The EDTT features: (1) a computerized query of the ture of safety and quality surveillance. In this study, we take the first
electronic medical record (EMR) for triggers, (2) use of only trig- step in a multicenter validation of the computerized EDTT: the query
gers demonstrated to be associated with AEs (not based on expert validation phase. This was completed at the outset of our ongoing
consensus alone), and (3) the use of trigger weights (the relative multicenter study to validate performance of the EDTT in detect-
strengths of trigger associations with AEs) to allow a deliberate ing harm. In this article we present data on validation of the EDTT
sampling strategy to enhance AE detection. Using this approach, we trigger query against manual review at each site, documenting site-­
achieved high yields and efficiency, while maintaining a broad scope related differences in trigger frequency and query performance.
of detected AE types. 25
Beyond obvious improvements in efficiency, a computerized
query with the ability to screen an entire population of ED visits for M E TH O D S
triggers has powerful applications, particularly when combined with
a risk model relating trigger patterns to AE occurrence. First, rank- Study setting and participants
ing visits in order of increasing AE risk allows for selection of the
higher risk visits for review, significantly increasing the yield and ef- This study was conducted at three urban, academic EDs using data
ficiency of the review process. Second, while traditional approaches from all adult visits between September 1, 2019, and March 28,
to AE detection typically result in a simple tally of cases of some 2021 (Ottawa, 230,246 records; Michigan, 101,915; Washington
type, they provide no sense of magnitude or proportion—whether University School of Medicine [WUSM], 118,691). Patients aged
the specific events detected represent a small or large proportion ≥ 18 were eligible for inclusion. Visits in which patients left without
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
566 THE EMERGENCY DEPARTMENT TRIGGER TOOL: MULTICENTER TRIGGER QUERY VALIDATION

being seen were excluded. All sites used unique instances of the Epic result in inappropriately triggering a record). We thus identified a
EMR. Two sites are located in the United States (WUSM in St. Louis, pool of patient visits with triggers from which we could select sam-
MO, and University of Michigan in Ann Arbor, MI) and one in Canada ples for record review and a set of records without triggers for com-
(University of Ottawa in Ottawa, ON). One site has an integrated ED parison purposes.
ICU. This study was approved by our university institutional review We then established secure, cloud-­based, shared folders. Each
board (IRB) using a single IRB for the two U.S. sites and by the re- site hosted their own instance of a custom web-­based software ap-
search ethics board at the Ottawa Hospital. plication for record review and abstraction (Figure 1). To prevent
transmission of protected health Information, data associated with
pseudo-­IDs (including visit week and triggers identified) were sent
Study design to our coordinating site, where records were selected for review
in batches. Records selected for review in a batch were sent back
This is a retrospective, multicenter observational study to evalu- to participating sites where pseudo-­IDs were matched with actual
ate 34 triggers previously demonstrated in our single-­center study medical record IDs that were loaded into the site's local application
25
to be associated with AEs. The objective of this initial step of for review and abstraction. Deidentified data from the application
the overall multicenter validation of the EDTT is to validate the were shared centrally for analysis.
performance of the computerized trigger query, comparing it to
manual screening (the typical process used in most trigger tools)
for triggers at each site. In this phase of the study we are not eval- Record selection
uating the performance of individual triggers or the tool as a whole
in detecting AEs. For this query validation phase of the study, we deliberately selected
trigger-­rich records to minimize the overall number of records that
would require manual review while also ensuring that reviewers
Trigger query specification and normalization would see each trigger a sufficient (>25) number of times to vali-
across sites date its detection by the query. Random sampling would have been
prohibitive: over 3000 records/site would be required to have a rea-
We harmonized the specifications of each trigger definition across sonable chance (80%) of observing 25 instances of a trigger whose
all three sites. After sharing the core query code for the EDTT population frequency is 1%. An initial set of 912 trigger-­rich records
across sites, over a series of meetings, we discussed each trigger were selected (Michigan, 302; Ottawa, 309; WUSM, 301), including
in succession, identifying any needed modifications to the query 20 records/site without triggers, for comparison purposes. A sample
code to accommodate expected variability across sites. This in- size of 25 is too small to obtain highly precise estimates of predictive
volved minor changes to trigger definitions, decisions about scop- values but provides 80% power to detect poorly performing trig-
ing of trigger specifications, (e.g., should “push dose vasopressors” gers (positive predictive value/negative predictive value [PPV/NPV]
[Trigger M5] include a “bolus from the bag” where a bolus of va- < 0.80, in one-­sided inferiority tests). As detailed below, most trig-
sopressors from an existing infusion is given for hypotension be gers were observed > 25 times and some > 100 times. Therefore,
included?), accommodations for differences in assays and testing in each record there is the opportunity for agreement or disagree-
used at respective sites (e.g., troponin I vs. troponin T), differences ment with the presence of each of 34 different triggers, for a total of
in reference ranges and units, and differences in data architecture 30,974 opportunities (L2 data were missing for one record).
that dictate where the query should search within the EMR for a
given data element.
Query validation and record reviews

Data extraction and data management As described in our prior studies, 23 the review process consists of
two rigorously trained first-­level (L1) reviewers who are assigned ED
We extracted data for all ED visits between September 1, 2019, visits from a queue for dual independent review. The web-­based ab-
and March 28, 2021 (Ottawa, 230,246 records; Michigan, 101,915; straction application displays patient demographics and a complete
WUSM, 118,691). We applied the trigger query to these data ex- list of triggers (Figure 1). Reviewers indicate whether each trigger
tracts and identified 175,416 records with one or more triggers was present (yes/no) on their manual review of a record (e.g., the
(Ottawa, 71,389 records; Michigan, 50,318; WUSM, 53,709). We query says heart rate [HR] > 130. The manual reviewer confirms
cleaned data for duplicates, excluding ineligible visits (e.g., patients whether there was a HR documented as >130 in the record or not).
under 18 years old, those who left against medical advice), and ad- After this manual screening is completed, the interface displays the
dressed various data extraction issues (e.g., data from same-­day query results side by side with the completed manual review results
affiliate community ED referrals or from inpatient portions of a hos- and L1s are then able to recheck the record and modify their re-
pital visit being included in an ED encounter extract, which could sponses if desired. Reviewers then proceed to complete the record
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFEY et al. 567

F I G U R E 1 Data entry platform for trigger validation and AE review. AE, adverse event; BiPAP/CPAP, bilevel positive airway pressure/
continuous positive airway pressure; SIRS, systemic inflammatory response syndrome.

review for the presence of AEs and characterize these with several Outcome measures
qualifiers (outside the scope of the data presented here). During
this query validation phase of the study, Level 2 (L2) reviewers also There were five outcomes measured in this phase of the study:
completed manual screening for triggers with both L1 reviewers’ and
computer query results visible and arbitrated L1 review disagree- 1. Trigger frequencies across sites (number and proportion of
ments with the computerized query and/or with one another. records in which each respective trigger was present).
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
568 THE EMERGENCY DEPARTMENT TRIGGER TOOL: MULTICENTER TRIGGER QUERY VALIDATION

2. Overall proportion of triggered (one or more) records, compared n = 10; C38—diagnosis of aortic dissection, n = 18; M6—push-­dose
across sites. pressors, n = 19). Some common triggers were observed over 100
3. Trigger query performance against manual review. This is defined times per site (e.g., C19—respiratory rate > 24 or < 10, n > 134/site;
as: for each trigger, the proportion of records in which the trigger and C53—systemic inflammatory response syndrome [SIRS] criteria
was identified by the query as being present in which the trigger present, n > 170/site).
was also identified as being present on manual review (PPV) and Trigger frequencies demonstrated some expected site-­to-­site
the proportion of records in which the trigger was not found to be variation, but most were highly consistent across sites and fol-
present by the query in which the trigger was also not identified lowed the same general trends, particularly at the two U.S. sites,
on manual review (NPV). The criterion standard consensus deter- and were comparable to our prior single-­center results (Table 3).
mination was defined as the final L2 rating. Some triggers are relatively rare, resulting in low median overall
4. Inter-­rater reliability of initial L1 reviews prior to review of query trigger frequencies (Michigan, 1.5% [IQR 0.6%–4.1%]; Ottawa,
results and after query results were revealed (and so after L1 re- 0.8% [IQR 0.2%–1.6%]; WUSM, 1.4% [IQR 0.7%–2.8%]). Rare
viewers had made any modifications to their ratings). triggers tend to be rare at all sites, with one exception: ultra-
5. Among cases where initial L1 reviews were discordant with the sound guided IV (P8) is relatively common in U.S. samples (6.2%
query, the proportion of times when L1 reviewers modified their at Michigan, 4.2% at WUSM) but virtually never seen at Ottawa
initial findings to align with the query results once the query re- (one in 10,000 records). The number of rare triggers (frequencies
sults were revealed and the proportion of L2 arbitrations between < 1%) in our sample varied from site to site: 12 at Michigan, nine
final L1 reviews that were discordant with the query in which the at WUSM, and 20 at Ottawa. In the population, 3.9% of records
L2 reviewer agreed with the L1 reviewer versus the query. at Michigan carried one or more rare trigger, 3.8% at WUSM, and
5.8% at Ottawa. Other triggers are relatively common, with fre-
quency ≥ 5% across sites: ED boarding > 6 h (C30), SIRS criteria
Data analyses present (C53), respiratory rate > 24 or < 10 (C19), and SBP > 180
or DBP > 120 (C27). On average, in this enriched sample, triggered
This is a descriptive study and reports outcomes as frequencies with records carried 8.9 and 9.6 triggers at Michigan and WUSM and
exact confidence intervals (CIs). Analyses were conducted using 4.7 triggers at Ottawa. Population frequency averages across sites
30,31
Python (SciPy 1.10 and Sci-­kit learn 0.23). for records with one or more triggers would be 2.4, 2.2, and 2.7,
respectively. This is slightly misleading, however, since the num-
bers and proportions of the more “trigger-­r ich” records to choose
R E S U LT S from differs across sites (Table 4).

Sociodemographics
Proportions of triggered records
The makeup of our sample was driven by the triggers being evalu-
ated, which tends to select for higher acuity patients. Across sites, Collectively, the proportions of triggered records across sites were
the trigger validation samples were comparable in terms of gender 31% (Ottawa), 45.3% (WUSM), and 49.4% (Michigan). Triggers are
and age (Table 1). The Michigan sample is primarily White/Caucasian not independent from one another, such that records containing
(76%), whereas WUSM is majority Black/African American (54%). common triggers may include multiple triggers. For example, the
Information on race/ethnicity is not collected at Ottawa. Visit acuity frequency of SIRS criteria present (C53) appearing is 55% when the
(Emergency Severity Index32; Canadian Triage Acuity Scale33) was trigger ED boarding > 6 h (C30) is present compared to 19% in the
higher at WUSM (p < 0.001). Admission rates in the sample were absence of boarding (consistent finding across all sites). It is not unu-
comparable at WUSM and Michigan (82% and 79%) and lower at sual for records to carry at least three of the more common triggers
Ottawa (67%). Deaths in the sample were higher at WUSM (8% vs. (from 2.9% of records at Ottawa to 6.6% at WUSM). Further, a higher
2% and 4%, respectively). number of common triggers increases the likelihood of the presence
of other triggers: 60% of records (overall) carry two or more ad-
ditional triggers when three or more common triggers are present
Trigger frequency compared to 7% when there are not. The chances of carrying a rare
trigger are vastly increased as the number of common triggers in-
Our list of triggers and brief description is provided in Table 2. All creases. For example, if three or more common triggers are present,
triggers were seen a minimum of 25 times at each site with the ex- the chances of presence of a rare trigger range from 17% at Ottawa
ception of two triggers at Ottawa (P8—US guided IV, n = 5 occur- and 19% at WUSM to 43% at Michigan. Consequently, the propor-
rences in 230K records; C38—diagnosis of aortic dissection, n = 15), tion of records with relatively high numbers of triggers is significant
and three triggers at Michigan (M2—3+ doses of hydromorphone, (Table 4).
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFEY et al. 569

TA B L E 1 Validation sample
WUSM
demographics and clinical characteristics.
Michigan (n = 302) Ottawaa (n = 309) (n = 301)

Gender
Male 106 (35.1) 139 (45.0) 121 (40.2)
Female 196 (64.9) 170 (55.0) 180 (59.8)
Age (years) 57.4 (±18.6) 61.4 (±19.6) 53.7 (±18.1)
Race
Black 46 (15.2) — 162 (53.8)
White 229 (75.8) — 120 (39.9)
Mixed 1 (0.3) — 2 (0.7)
Other 19 (6.3) — 3 (1.0)
Declined 1 (0.3) — 1 (0.3)
Unknown 6 (2.0) – 13 (4.3)
Ethnicity
Non-­Hispanic 284 (96.3) — 276 (91.7)
Hispanic 7 (2.4) — 3 (1.0)
Declined 1 (0.3) — 4 (1.3)
Unknown 3 (1.0) — 18 (6.0)
Visit acuity (ESI)
1 86 (28.5) 43 (13.9) 113 (37.5)
2 188 (62.3) 179 (57.9) 162 (53.8)
3 28 (9.3) 87 (28.2) 26 (8.6)
Visit disposition
Discharge 31 (10.3) 78 (25.2) 15 (5.0)
Admit 248 (82.1) 206 (66.7) 239 (79.4)
Expired 6 (2.0) 13 (4.2) 24 (8.0)
DOA 0 (0.0) 0 (0.0) 2 (0.7)
Send to other department 15 (5.0) 0 (0.0) 18 (6.0)
Transfer 1 (0.3) 11 (3.6) 3 (1.0)
AMA 1 (0.3) 1 (0.3) 0 (0.0)

Note: Data are reported as n (%) or mean (±SD).

Abbreviations: AMA, against medical advice; DOA, dead on arrival; ESI, Emergency Severity Index;
WUSM, Washington University School of Medicine.
a
Ottawa does not collect race/ethnicity information.

Query validation doses of hydromorphone [M2], and opioids and benzodiazepines

administered [M18]), and three trigger queries underperformed
Data on query performance by trigger and site are presented in (procedural sedation [P4; 28%], the very rare ultrasound-­g uided IV
Table 5. Out of approximately a combined 31,000 final ratings of [P8; 60%], and bilevel positive airway pressure/continuous posi-
presence or absence of a trigger, six were excluded for missing val- tive airway pressure [BiPAP/CPAP; C7; 73%]). All other triggers
ues. NPVs were generally very high, consistently greater than 96% had PPVs greater than 90%.
at Ottawa and always greater than 91% at Michigan. At WUSM, We identified five triggers (C7, C27, M6, P4, and P8) that had
three triggers (C53, P2, C39) had NPVs ranging between 76.0% significantly different PPVs between sites. One of those (P4, pro-
and 85.4%, but with relatively wide CIs. PPVs were also gener- cedural sedation) was very rare at one site (only five occurrences)
ally very high. The PPVs at WUSM were over 90% for all but five and PPV was significantly lower at Michigan than at WUSM (86%
triggers, which had PPVs between 83% (M6) and 89% (C27). At vs. 98%). Ultrasound-­guided IV (P8) performed poorly at Ottawa
Michigan, all PPVs were over 90% except for opioids and benzodi- (60%) compared to Michigan (100%) and WUSM (96%). Push-­dose
azepines administered (M18; 85%), procedural sedation (P4; 86%), pressors (M6) and SBP > 180 or DBP > 120 (C27) underperformed at
and ED boarding > 6 h (C30; 89%). At Ottawa, three triggers had WUSM (PPVs 83.5% and 89.9%). BiPAP/CPAP (C7) performed well
PPVs between 80% and 90% (restraint use [C3], three or more at Michigan (97%) and less so at Ottawa (73%) and WUSM (88%).
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
570 THE EMERGENCY DEPARTMENT TRIGGER TOOL: MULTICENTER TRIGGER QUERY VALIDATION

TA B L E 2 Brief trigger descriptions, grouped by module.

1. Patient care
C3 Restraint use Use of physical restraints in the ED.
C4 SBP < 90 on 2+ readings at least Two or more systolic BP readings of less than 90 mm Hg occurring at least 15 min
15 min apart apart.
C7 BiPAP/CPAP Use of BiPAP or CPAP.
C8 HR > 130 Documented HR > 130 at any time.
C13 O2 < 90% Documented O2 saturation of <90%. This can be on room air or oxygen.
C14 Pneumothorax Documentation of pneumothorax of any etiology or listed as ED diagnosis.
C19 RR > 24 or < 10 Documented RR < 10 or > 24 at any time during the ED visit.
C20 Temperature < 35°C or > 38°C Documented temperature of <35 or >38°C at any time during the ED visit.
C27 SBP > 180 or DBP > 120 Any documented SBP > 180 or a DBP > 120.
C28 Chest CT for pulmonary embolus Chest CT PE protocol obtained in the ED.
C30 ED boarding > 6 h Departure from the ED > 6 h after an admit (bed request) order was placed.
C38 Diagnosis of aortic dissection Diagnosis of aortic dissection (includes patients transferred from another ED,
diagnosis in the ED, and/or diagnosed within the first 24 h of hospital admission).
C39 Aspiration Evidence of aspiration event in the diagnosis list, MD notes, RN notes, radiology
reports. Includes aspiration that is present on arrival and as an ED diagnosis.
C48 IV antihypertensives Use of IV administration of the following medications: clevidipine, nicardipine,
nitroprusside, fenoldopam, nitroglycerin, enalapril, enalaprilat, hydralazine,
labetalol, esmolol, metoprolol, nesiritide, and phentolamine.
C53 SIRS criteria present Presence of two or more of the following: max temp > 38°C, min temp < 36°C, max
pulse > 90, max RR > 20, WBC > 12K or < 4K, and PaCO2 < 32.
2. Medication
M2 3+ doses of hydromorphone Administration of three or more doses of hydromorphone in the ED regardless of
route.
M3 Heparin Administration of enoxaparin (lovenox), bivalrudin, fondaparinux, or heparin (drip
and/or bolus).
M4 Diphenhydramine Administration of diphenhydramine, regardless of route.
M5 Dextrose administration Administration of D50, D25, D10, or oral D40.
M6 Push-­dose pressors Administration of neosynephrine or epinephrine IV push.
M9 Nitroglycerin/nicardipine/ Administration of nitroglycerin, nicardipine, or nitroprusside.
nitroprusside
M17 IV calcium Administration of IV calcium chloride or calcium gluconate.
M18 Opiates + benzodiazepines Administration of both an opioid and a benzodiazepine regardless of timing or route.
administered in the ED
M19 Propofol Administration of propofol, including IV push doses and IV propofol drips.
M7 Alteplase (tPA) administration Administration of alteplase in the ED.
3. Laboratory
L2 Lactate > 4 Lactate > 4.
L6 pCO2 > 60 mm Hg pCO2 > 60 mm Hg
L7 Positive troponin Positive (abnormal) troponin, based on thresholds and indices set by your institution.
L11 Elevated BNP BNP or pro-­BNP > 300 or based on thresholds and indices set by your institution.
L14 HCO3 < 18 HCO3 < 18
4. Procedural
P2 Endotracheal intubation Intubation or reintubation that occurs in the ED including patients who arrive to the
ED via EMS or from outside facilities who are intubated prior to arrival.
P4 Procedural sedation Any procedural sedation performed in the ED
P6 Central line insertion Documentation of a central venous catheter placed in the ED.
P8 Ultrasound-­guided IV Placement of a peripheral IV under ultrasound guidance.

Abbreviations: BiPAP, bilevel positive airway pressure; BNP, brain natriuretic peptide; CPAP, continuous positive airway pressure; HR, heart rate; PE,
pulmonary embolism; RR, respiratory rate; SIRS, systemic inflammatory response syndrome; tPA, tissue plasminogen activator.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFEY et al. 571

TA B L E 3 Trigger frequency (%) by module and sitea.

Prior single-­center
Michigan Ottawa WUSM study

Patient care
C53 SIRS criteria present 24.69 18.99 25.34 24.30
C19 RR > 24 or <10 10.36 7.17 17.97 10.60
C27 SBP > 180 or DBP > 120 11.80 6.12 11.79 9.60
C30 Boarding > 6 h 9.37 7.38 10.43 7.08
C56 Two or more consultsa 5.79 4.00 5.31 N/A
C13 O2 < 90% 4.93 2.71 5.28 3.45
C8 HR > 130 4.12 2.70 3.72 2.70
C20 Temperature < 35°C or > 38°C 4.26 3.77 2.84 2.56
C48 IV antihypertensives 2.07 0.53 2.78 2.95
C28 PECT 6.12 1.57 2.56 1.37
C57 Language barriera 3.01 3.29 2.29 N/A
C3 Restraint use 0.29 0.27 1.90 1.12
C4 SBP < 90 ×2 @>15 min 2.69 1.01 1.78 1.51
C7 BiPAP/CPAP 1.27 0.24 1.41 0.72
C55 Upgrade in care within EDa 0.88 0.93 0.72 N/A
a
C58 Upgrade in care of admitted patient 0.42 0.32 0.43 N/A
C14 Pneumothorax 0.19 0.10 0.30 0.16
C54 High-­acuity return within 72 ha 0.16 0.07 0.14 N/A
C39 Aspiration 0.16 0.07 0.14 0.09
C38 Aortic dissection 0.18 0.02 0.09 0.08
Laboratory
L11 Elevated BNP 6.19 0.87 6.29 2.66
L2 Lactate > 4 2.08 1.01 1.82 1.06
L7 Positive troponin 0.94 2.47 1.30 4.91
L6 pCO2 > 60 mm Hg 1.48 0.83 1.14 0.27
L14 HCO3 < 18 1.71 0.60 0.66 8.23
L15 Neutropeniaa 0.26 0.22 0.21 N/A
Medication
M2 3+ doses hydromorphone 1.46 1.59 1.82 1.38
M18 Opiates + benzodiazepiness 2.01 1.34 1.61 3.45
M4 Diphenhydramine 3.72 0.92 1.55 3.49
M17 IV calcium 1.83 0.32 1.47 0.97
M3 Heparin 2.41 0.64 1.19 2.17
M19 Propofol 0.83 0.94 1.17 0.81
M9 IV nitrates 0.74 0.08 1.16 1.23
M5 Dextrose administration 1.30 0.49 0.98 0.86
M6 Push-­dose pressors 0.13 0.19 0.58 0.29
M7 tPA 0.08 0.12 0.14 0.13
M21 Acute anticoagulant reversala 0.10 0.09 0.12 N/A
Procedure
P8 Ultrasound-­guided IV 6.28 0.01 4.18 1.01
P2 Intubation 0.65 0.22 1.02 0.69
P6 Central line insertion 0.66 0.12 0.93 0.53
P4 Procedural sedation 0.68 1.12 0.80 0.63

Abbreviations: BiPAP, bilevel positive airway pressure; BNP, brain natriuretic peptide; CPAP, continuous positive airway pressure; HR, heart rate;
PECT, pulmonary embolism computed tomography; RR, respiratory rate; SIRS, systemic inflammatory response syndrome; tPA, tissue plasminogen
activator; WUSM, Washington University School of Medicine.
a
Indicates new experimental trigger not available in our earlier single-­site study (R18). Frequencies based on all visits: Michigan, 101,915 records;
Ottawa, 230,246; WUSM, 118,691; R18, 92,859.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
572 THE EMERGENCY DEPARTMENT TRIGGER TOOL: MULTICENTER TRIGGER QUERY VALIDATION

TA B L E 4 Distribution of number of
Michigan Ottawa WUSM
triggers per record.
Number of triggers (n = 101,915) (n = 230,246) (n = 118,691)

0 51,597 (50.6) 158,857 (69.0) 64,982 (54.7)

1 22,521 (22.1) 36,569 (15.9) 21,022 (17.7)
2 11,561 (11.3) 14,957 (6.5) 12,280 (10.3)
3 6444 (6.3) 8086 (3.5) 7641 (6.4)
4 3701 (3.6) 4877 (2.1) 4790 (4.0)
5 2421 (2.4) 2983 (1.3) 3030 (2.6)
6 1370 (1.3) 1740 (0.8) 1942 (1.6)
7 884 (0.9) 1033 (0.4) 1165 (1.0)
8 550 (0.5) 542 (0.2) 703 (0.6)
9 325 (0.3) 281 (0.1) 427 (0.4)
10 or more 541 (0.5) 321 (0.1) 709 (0.6)

Note: Data are reported as n (%).

Abbreviation: WUSM, Washington University School of Medicine.

There was variation in PPV for other triggers, but, for these, PPV was (which are infrequently but sometimes discrepant with one another
good (>90%) across all sites. and/or with the query results), it is difficult to make summary com-
parisons of dual manual review performance versus the query. If we
take a very narrow view that favors manual review performance,
Inter-­rater agreement and modifications of screening only considering trigger evaluations where both L1 reviews agree
results after query review on either presence or absence of a trigger (and treat disagreements
as missing values), then the dual L1 approach slightly outperforms
Among three pairs of independent L1 reviewers, there were 1551 the query, with rates of disagreement with the ultimate consensus
disagreements across all triggers (5%, out of 30,974 final calls over- finding of 1.5% for all triggers, except C53 (3.9%, 95% CI 2.6%–5.6%)
all) prior to seeing the query results (Cohen's kappa = 84.9%) and 985 and L7 (3.0%, 95% CI 1.9%–4.3%). However, this ignores all trigger
disagreements after reviewing the query results and modifications evaluations where L1 reviews disagree.
were made (3.1%, kappa = 90.6%). Initial disagreements were most
pronounced (>10%) for five triggers, including ED boarding > 6 h
(C30), aspiration (C39), push-­dose pressors (M6), procedural sedation DISCUSSION
(P4), and especially SIRS criteria (C53; 18.6%, 95% CI 16.1%–21.2%).
Overall, out of 62,118 trigger determinations made independently Trigger tools are used both for measuring and for monitoring the
by all L1 reviewers at all sites, there were 919 (1.5%; 1.4%–1.6%) mod- underlying level of harm in an institution, allowing health care teams
ifications made by reviewers to their initial findings in either direction to assess the impact of interventions and identify areas for quality
after query results were revealed. The proportion of modifications improvement. Particularly in a setting like the ED, with high patient
made overall was consistent across sites, ranging from 1.2% to 1.8%. volume, brief patient encounters, and a low AE rate, tools for detec-
L1 reviewers modified their initial screening result to reflect that a tion of harm need to be efficient (i.e., low proportion of negative
trigger was present 46.8% of the time (95% CI 44.1%–49.5%) if the reviews) and high yield for detecting AEs. Compared to traditional
query found it was present and modified their initial result to reflect approaches that use just a few screening criteria that are reviewed
that the trigger was absent 22.6% of the time (95% CI 20.3%–25.0%) exhaustively, trigger tools consist of broad sets of triggers that ex-
when the query found it was absent. When arbitrating L1 ratings that tend the array of AE types captured. A computerized query makes
were in agreement with each other but discordant with the query, this screening even more efficient. Because very few trigger tools
L2 reviewers agreed with the query 64% of the time (867/1345) and to date have been computerized, the process of validating a query
with the reviewers 36% of the time, more often agreeing with a posi- against manual review is uncommon.
tive query (73%) than with a negative one (61%). To scale the EDTT for multicenter use, we conducted a series
Overall, the query outperformed single L1 review (in terms of of calls to harmonize triggers across sites, addressing expected dif-
lower error rate) for all but two triggers: C39 aspiration, 6.4% (95% ferences related to workflows, patient care practice, medications,
CI 4.9%–8.2%) versus 4.3% (95% CI 3.1%–5.8%), and endotracheal laboratory values, and data architecture. Though all sites were on
intubation (P2) 7.1% (95% CI 5.6%–9.0%) versus 2.9% (95% CI 1.9%– the Epic EMR platform, there were nevertheless differences re-
4.2%). Because our criterion standard L2 determination is made lated to where data elements would be found. We routinely review
following review of both the query results and the final L1 reviews the first 24 h of an inpatient admission to see whether there are
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFEY et al. 573

TA B L E 5 Query performance by trigger and site (%).

PPV NPV

Trigger Michigan Ottawa WUSM Michigan Ottawa WUSM

Patient care
C3 Restraint use 96 80 97 100 99 100
C4 SBP < 90 ×2 @ >15 min 95 100 100 99 99 98
C7 BiPAP/CPAP 97 73 88 100 97 98
C8 HR > 130 100 100 100 100 98 100
C13 O2 < 90 100 98 99 100 99 98
C14 Pneumothorax 95 100 100 100 100 100
C19 RR > 24 or < 10 100 100 100 100 100 97
C20 Temperature < 35°C or > 38°C 100 100 100 100 98 99
C27 SBP > 180 or DBP > 120 97 100 90 100 100 100
C28 PECT 97 100 100 94 99 85
C30 Boarding > 6 h 89 98 86 99 98 96
C38 Aortic dissection 100 100 100 100 100 100
C39 Aspiration 100 100 100 98 97 76
C48 IV antihypertensives 95 100 100 97 97 99
C53 SIRS criteria present 100 100 99 100 100 100
Laboratory
L2 Lactate >4 99 100 100 100 100 98
L6 pCO2 > 60 mm Hg 100 100 99 100 100 98
L7 Positive troponin 100 100 100 100 100 98
L11 Elevated BNP 99 100 100 100 100 100
L14 HCO3 < 18 100 96 100 100 99 93
Medication
M2 3+ doses hydromorphone 100 89 88 100 100 99
M3 Heparin 98 100 98 100 100 100
M4 Diphenhydramine 100 100 100 100 100 100
M5 Dextrose administration 100 100 100 100 99 100
M6 Push-­dose pressors 100 100 84 100 99 99
M7 tPA 100 100 100 100 100 100
M9 IV nitrates 100 100 100 99 100 100
M17 IV calcium 100 100 100 95 99 99
M18 Opiates + benzodiazepines 85 85 97 100 100 100
M19 Propofol 100 100 99 100 100 100
Procedure
P2 Intubation 96 93 99 92 96 83
P4 Procedural sedation 86 28 98 100 100 100
P6 Central line insertion 91 93 94 100 99 100
P8 Ultrasound-­guided IV 100 60 96 100 100 100

Abbreviations: HR, heart rate; NPV, negative predictive value; PECT, pulmonary embolism computed tomography; PPV, positive predictive value;
RR, respiratory rate; SIRS, systemic inflammatory response syndrome; tPA, tissue plasminogen activator; WUSM, Washington University School of
Medicine.

AEs related to ED care and one site had an ED ICU whose data we did make some early minor adjustments, it is possible that such
were capture in the ED record. We took steps to isolate triggers triggers may bear further adjustments and/or that some elements
to the ED encounter itself as at some sites, initially other aspects are simply very difficult to capture with high reliability using a
of post-­ED care were sometimes pulled into the record. Though computerized query, at least in the absence of more complex
early interim assessments of capture looked promising and though search tools like NLP.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
574 THE EMERGENCY DEPARTMENT TRIGGER TOOL: MULTICENTER TRIGGER QUERY VALIDATION

In this analysis, we compare the computerized EDTT to manual providing a more complete trigger evaluation (no data loss due to
screening for trigger detection as the first step in demonstrating its disagreements).
generalizability. Among 34 triggers evaluated, 30 were present in
sufficient frequency at all sites for an adequate evaluation. Overall,
individual trigger frequencies were highly consistent across sites, LI M ITATI O N S
ranging from the common to the rare. Across sites, the overall pro-
portion of triggered records ranged from 31% to 49.4%. Many re- Our study had a few limitations. Our approach to scaling our single-­
cords contained a single trigger, but others contained 10 or more. center trigger tool to multiple sites was effective but imperfect.
Dual independent manual screening resulted in a high level of Though our overall results were quite good, the definitions, specifi-
agreement (kappa = 85%) before query results were revealed. This cations, and applicability of selected triggers appear to have limited
agreement increased further after review of query results, with re- their detection. There is an unavoidable degree of arbitrariness when
viewers changing their findings in cases of discordance to align with using cutoffs for labeling a trigger as “validated” or “not validated”
the query in 65.2% of these cases. The query performed well against and though we present data based on a threshold of 90% for PPV and
manual review, with NPVs greater than 90% for all but three triggers NPV we also present actual numbers in the tables. Our L2 reviews,
at one site and PPVs greater than 90% for all but nine triggers, three while they included a primary review of the record, were not blind
of which were present in insufficient numbers to fully evaluate. to either the L1 reviews or the query. However, this was necessary
We expected variability across sites in terms of which triggers in part to make final determinations. Though this study helps estab-
might be present, in what frequencies, and how the query might per- lish elements of generalizability at new/different sites in the United
form in detecting them. We also expect we will later observe vari- States and Canada, these were all academic rather than community
ability in the associations of triggers with AEs when data collection sites, the inclusion of which should be a goal in future studies.
for that portion of the study is complete, anticipating that a core of
triggers will remain high yield in detecting AEs across sites. In light
of this, we were pleased with the results we observed. It could cer- CO N C LU S I O N S
tainly have been the case that many of the triggers might not have
been present in sufficient numbers, that the query might not capture These results represent the first step in multicenter validation of the
triggers well, or that the proportions from site to site might be highly ED trigger tool. The ED trigger tool query performed very well with
different. The consistency in frequencies bodes well for use of these high negative predictive values and positive predictive values when
trigger in a multicenter evaluation of the tool. compared to manual screening. With a few exceptions, frequen-
The underperformance of the query in detecting a handful of cies for the large majority of triggers were highly conserved across
triggers may be attributed to how those triggers were specified in sites, and overall proportions of triggered records ranged from 31%
the query, differences in documentation and data architecture, dif- to 49.4%. This demonstrates the feasibility and generalizability of
ferent practice patterns, and local variability in the kinds of quality implementing the ED trigger tool query at new sites, providing a
and safety challenges present and in other cases is simply unclear. sound foundation for testing the triggers utility in predicting adverse
For example, ultrasound-­guided IV placement (P8) is an uncommon events, which is the currently ongoing next phase of this study.
practice at Ottawa, and when performed, there is no reliable elec-
tronic capture. Diagnosis of aortic dissection (C38) is rare at all sites. AU T H O R C O N T R I B U T I O N S
Aspiration events (C39) and, somewhat surprisingly, endotracheal All authors contributed to study design and acquisition of funding.
Intubations (P2) may be better captured in narrative text than in a Richard T. Griffey, Ryan M. Schneider, Keith E. Kocher, and Edmund
structured field. Administration of push-­dose pressors (M6) often oc- S. H. Kwok led staff education, training, and data collection. Ellen
curs during critical situations, entered as verbal orders and is some- Salmo, Nora Malone, Carrie Smith, Catie Guarnacia, April Rick, and
times not discoverable in the medication administration record in the Tamara Clavet performed record review and participated in group/
usual fashion. Approaches such as NLP to identify such triggers may data reviews. Phil Asaro and Rich Medlin oversaw programming and
be needed when triggers are not found in structured fields. This is data procurement and Alexandre A. Todorov led data management
something we have attempted in a limited way previously to good and analysis. Richard T. Griffey primarily drafted the manuscript.
effect and are also testing in the late phases of our current study. Richard T. Griffey, Ryan M. Schneider, Edmund S. H. Kwok, Keith
While comparison against manual review was considered the E. Kocher, Carrie Smith, Ellen Salmo, Rich Medlin, Phil Asaro, and
criterion standard, by revealing the computerized query results Alexandre A. Todorov contributed to critical review and editing.
after completion of manual review and allowing for reviewers to
then modify their initial screen for triggers we were able to eval- F U N D I N G I N FO R M AT I O N
uate manual screening with a bit more nuance. Based on changes This work was supported by grant R01 HS027811-­01 (Griffey PI)
made by L1 reviewers to their initial findings after the query results from the Agency for Healthcare Research and Quality. The contents
were revealed, overall, the query outperformed single manual re- of this work are solely the responsibility of the authors and do not
view and was at least equivalent to dual independent review and necessarily represent the official view of the AHRQ or the BJHF.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14873 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFEY et al. 575

17. Mattsson TO, Knudsen JL, Brixen K, Herrstedt J. Does adding an

C O N F L I C T O F I N T E R E S T S TAT E M E N T appended oncology module to the global trigger tool increase its
All authors have no conflicts of interest to report. value? Int J Qual Health Care. 2014;26(5):553-­560.
18. Kalenderian E, Walji MF, Tavares A, Ramoni RB. An adverse event
trigger tool in dentistry: a new methodology for measuring harm in
ORCID
the dental office. J Am Dent Assoc. 2013;144(7):808-­814.
Richard T. Griffey https://orcid.org/0000-0002-6279-7380
19. Carnevali L, Krug B, Amant F, et al. Performance of the adverse drug
Phil Asaro https://orcid.org/0000-0002-6301-7673 event trigger tool and the global trigger tool for identifying adverse
drug events: experience in a Belgian hospital. Ann Pharmacother.
REFERENCES 2013;47(11):1414-­1419.
20. Szekendi MK, Sullivan C, Bobb A, et al. Active surveillance using
1. Griffin FA, Resar RK. IHI Global Trigger Tool for Measuring Adverse
electronic triggers to detect adverse events in hospitalized pa-
Events (Second Edition). IHI Innovation series white paper. Institute
tients. Qual Saf Health Care. 2006;15(3):184-­190.
for Healthcare Improvement; 2009. Accessed June 22, 2020.
21. Resar RK, Rozich JD, Simmonds T, Haraden CR. A trigger tool to
https://​w ww.​i hi.​o rg/​r esou​r ces/​w hite​-­​papers/​i hi-­​globa​l-­​trigg​e r-­​
identify adverse events in the intensive care unit. Jt Comm J Qual
tool-­​measu​ring-­​adver​se-­​events
Patient Saf. 2006;32(10):585-­590.
2. Rising KL, Victor TW, Hollander JE, Carr BG. Patient returns to the
22. Griffey RT, Schneider RM, Adler LM, et al. Development of an emer-
emergency department: the time-­to-­return curve. Acad Emerg Med.
gency department trigger tool using a systematic search and modi-
2014;21(8):864-­871.
fied Delphi process. J Patient Saf. 2020;16(1):e11-­e17. doi:10.1097/
3. Pham JC, Kirsch TD, Hill PM, DeRuggerio K, Hoffmann B.
PTS.0000000000000243
Seventy-­t wo-­hour returns may not be a good indicator of safety
23. Griffey RT, Schneider RM, Todorov AA. The emergency depart-
in the emergency department: a national study. Acad Emerg Med.
ment trigger tool: a novel approach to screening for quality and
2011;18(4):390-­397.
safety events. Ann Emerg Med. 2020;76(2):230-­240. 10.1016/j.
4. Griffey RT, Bohan JS. Healthcare provider complaints to the emer-
annemergmed.2019.07.032.
gency department: a preliminary report on a new quality improve-
24. Griffey RT, Schneider RM, Sharp BR, Vrablik MC, Adler L. Practical
ment instrument. Qual Saf Health Care. 2006;15(5):344-­3 46.
considerations in use of trigger tool methodology in the emergency
5. Griffey RT, Schneider RM, Sharp BR, et al. Description and yield of
department. J Patient Saf. 2021;17(8):e837-­e842. doi:10.1097/
current quality and safety review in selected US academic emer-
PTS.0000000000000448
gency departments. J Patient Saf. 2017;16:e245-­e249. doi:10.1097/
25. Griffey RT, Schneider RM, Todorov AA. The emergency depart-
PTS.0000000000000379
ment trigger tool: validation and testing to optimize yield. Acad
6. Cheng J, Shroff A, Khan N, Jain S. Emergency department return
Emerg Med. 2020;27:1279-­1290.
visits resulting in admission: do they reflect quality of care? Am J
26. Griffey RT, Schneider RM, Todorov AA. Near-­miss events de-
Med Qual. 2016;31(6):541-­551.
tected using the emergency department trigger tool. J Patient Saf.
7. Sabbatini AK, Kocher KE, Basu A, Hsia RY. In-­hospital outcomes
2023;19(2):59-­66.
and costs among patients hospitalized during a return visit to the
27. Griffey RT, Schneider RM, Todorov AA. Emergency department
emergency department. JAMA. 2016;315(7):663-­671.
adverse events detected using the emergency department trigger
8. Classen DC, Resar R, Griffin F, et al. ‘Global trigger tool’ shows that
tool. Ann Emerg Med. 2022;80(6):528-­538.
adverse events in hospitals may be ten times greater than previ-
28. Griffey RT, Schneider RM, Todorov AA. Adverse events present on
ously measured. Health Aff (Millwood). 2011;30(4):581-­589.
arrival to the emergency department: the ED as a dual safety net. Jt
9. IHI Global Trigger Tool for Measuring Adverse Events. Institute for
Comm J Qual Patient Saf. 2020;46(4):192-­198.
Healthcare Improvement; 2014 Accessed June 2020. http://​w ww.​
29. Griffey RT, Schneider RM, Adler L, Todorov A. Post-­acute and long-­
ihi.​o rg/​r esou​r ces/​P ages/​​Tools/​​I HIGl​o balT​r igge​rTool​f orMe​a suri​
term care patients account for a disproportionately high number of
ngAEs.​aspx
adverse events in the emergency department. J Am Med Dir Assoc.
10. Aaronson EL, Wittels KA, Nadel ES, Schuur JD. Morbidity and mor-
2021;22(4):907-­912.e1.
tality conference in emergency medicine residencies and the cul-
30. Pedregosa F. Scikit-­learn: machine learning in python. J Mach Learn.
ture of safety. West J Emerg Med. 2015;16(6):810-­817.
2011;12:2825-­2830.
11. Seigel TA, McGillicuddy DC, Barkin AZ, Rosen CL. Morbidity
31. Virtanen PGR, Oliphant TE, et al. SciPy 1.0: fundamental al-
and mortality conference in emergency medicine. J Emerg Med.
gorithms for scientific computing in python. Nat Methods.
2010;38(4):507-­511.
2020;17(3):261-­272.
12. Landrigan CP, Parry GJ, Bones CB, Hackbarth AD, Goldmann DA,
32. Gilboy N, Tanabe P, Travers D, Rosenau AM. Emergency Severity
Sharek PJ. Temporal trends in rates of patient harm resulting from
Index (ESI): A triage tool for emergency department care (Version 4).
medical care. N Engl J Med. 2010;363(22):2124-­2134.
Schaumburg, IL: Emergency Nurses Association; 2020.
13. Levinson DR. Adverse Events in Hospitals: Methods for Identifying
33. Bullard MJ, Musgrave E, Warren D, et al. Revisions to the Canadian
Events. OEI-­06-­0 8-­0 0221. Department of Health and Human
emergency department triage and acuity scale (CTAS) guidelines
Services, Office of Inspector General; 2010.
2016. CJEM. 2017;19(S2):S18-­S27.
14. Suarez C, Menendez MD, Alonso J, Castano N, Alonso M, Vazquez
F. Detection of adverse events in an acute geriatric hospital over
a 6-­year period using the global trigger tool. J Am Geriatr Soc.
2014;62(5):896-­900.
15. Sharek PJ, Horbar JD, Mason W, et al. Adverse events in the neo- How to cite this article: Griffey RT, Schneider RM, Kocher
natal intensive care unit: development, testing, and findings of KE, et al. The emergency department trigger tool:
an NICU-­focused trigger tool to identify harm in north American
Multicenter trigger query validation. Acad Emerg Med.
NICUs. Pediatrics. 2006;118(4):1332-­1340.
2024;31:564-575. doi:10.1111/acem.14873
16. Najjar S, Hamdan M, Euwema MC, et al. The global trigger tool
shows that one out of seven patients suffers harm in Palestinian
hospitals: challenges for launching a strategic safety plan. Int J Qual
Health Care. 2013;25(6):640-­6 47.
Received: 2 October 2023 | Revised: 16 January 2024 | Accepted: 22 January 2024

DOI: 10.1111/acem.14880

ORIGINAL ARTICLE

A qualitative assessment of emergency physicians’ experiences

with robust emergency department buprenorphine bridge
programs

Jessica Marie Heil MS1 | Jonathan M. Lassiter PhD2 | Matthew S. Salzman MD1,3,4 |
Andrew Herring MD5 | Jason Hoppe MD6 | Michael Lynch MD7 |
Scott G. Weiner MD8 | Brian Roberts MD3,4 | Rachel Haroz MD1,3,4

1
Center for Healing, Division of Addiction
Medicine, Cooper University Health Care, Abstract
Camden, New Jersey, USA
Objectives: Emergency departments (EDs) are a critical point of entry into treatment
2
Department of Psychology, Rowan
University, Camden, New Jersey, USA
for patients struggling with opioid use disorder (OUD). When initiated in the ED, bu-
3
Department of Emergency Medicine, prenorphine is associated with increased addiction treatment engagement at 30 days
Division of Addiction Medicine and when initiated. Despite this association, it has had slow adoption. The barriers to ED
Medical Toxicology, Cooper University
Health Care, Camden, New Jersey, USA buprenorphine utilization are well documented; however, the benefits of prescribing
4
Cooper Medical School of Rowan buprenorphine for emergency physicians (EPs) have not been explored. This study
University, Cooper University Healthcare,
utilized semistructured interviews to explore and understand how EPs perceive their
Camden, New Jersey, USA
5
Department of Emergency Medicine, experiences working in EDs that have successfully implemented ED bridge programs
Highland Hospital—Alameda Health (EDBPs) for patients with OUD.
System, University of California, San
Francisco, California, USA Methods: Semistructured interviews were conducted with EPs from four geographi-
6
Department of Emergency Medicine, cally diverse academic hospitals with established EDBPs. Interviews were recorded
University of Colorado School of
and transcribed, and emergent themes were identified using codebook thematic anal-
Medicine, Colorado, Aurora, USA
7
University of Pittsburgh School of
ysis. Analysis credibility and transparency were confirmed with peer debriefing.
Medicine Pittsburgh, Pittsburgh, Results: Twenty-­t wo interviews were conducted across the four sites. Three key
Pennsylvania, USA
8
themes were constructed during the analyses: (1) provided EPs agency; (2) trans-
Brigham and Women's Hospital, Boston,
Massachusetts, USA formed EPs’ emotions, attitudes, and behaviors related to treating patients with OUD;
and (3) improved EPs’ professional quality of life.
Correspondence
Jessica Marie Heil, Center for Healing, Conclusions: Participants in this study reported several common themes related to
Division of Addiction Medicine, Cooper participation in their hospital's BP. Overall our results suggest that physicians who
University Health Care, Camden, NJ
08103, USA. participate in EDBPs may feel a renewed sense of fulfillment and purpose in their
Email: [email protected] personal and professional lives. These positive changes may lead to increased job sat-
Funding information isfaction in hospitals that have successfully launched EDBP.
Bloomberg Philanthropies

This project was funded by Bloomberg Philanthropies as part of the Bloomberg Overdose Prevention Initiative.

Presented at the Society for Academic Emergency Medicine Annual Meeting, Austin, TX, May 2023.

Supervising Editor: Mark B. Mycyk

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2024 The Authors. Academic Emergency Medicine published by Wiley Periodicals LLC on behalf of Society for Academic Emergency Medicine.

576 | 
wileyonlinelibrary.com/journal/acem Acad Emerg Med. 2024;31:576–583.
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14880 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HEIL et al. 577

I NTRO D U C TI O N would report the benefits of these programs, including improved

care for their patients with OUD and decreased moral injury.
Background

The Centers for Disease Control and Prevention estimated more M E TH O D S

than 105,000 overdose deaths in the United States in 2022, with
more than 82,000 deaths involving opioids.1 As the opioid epidemic Eligibility criteria and recruitment
continues, only 13.4% of patients with opioid use disorder (OUD)
receive evidence-­based treatments. 2 There are three FDA-­approved To be eligible for this study, physicians had to be board certified
medications for OUD (MOUD): methadone, buprenorphine, in emergency medicine, possess a DEA waiver to prescribe
and extended-­release naltrexone (XR-­NTX). Methadone and buprenorphine (the “X-­waiver,” which has since been abolished but
buprenorphine are considered criterion standard treatments.3 was in place during this study), have self-­reported to have written ≥10
Accessibility remains one of the biggest barriers for patients to buprenorphine prescriptions after board certification, and not be an
4
receive such care. To alleviate the obstacles to care, the emergency addiction medicine fellow or board certified in addiction medicine.
department (ED) has become a critical point of entry into treatment The physicians must work in the ED while prescribing buprenorphine
for patients struggling with OUD.5 to be invited to participate in the study. The authors of this paper
In the United States, between 2005 and 2017, ED opioid-­related understand that some hospitals have “bridge clinics” within their ED
6
encounters increased by approximately 180%. At the beginning of that serve patients with substance use disorders. However, when we
the COVID-­19 pandemic, in March 2020, the mean weekly rate of refer to “bridge program” for this study, we are specifically referring
all-­cause ED visits decreased, while drug overdose visit rates held to providers prescribing buprenorphine to patients in the ED and
steady.7–10 Mortality rates after ED visits for nonfatal opioid over- referring them to an outside clinic for further treatment. 27–29 The
doses are high; more than 5% of patients discharged after a nonfatal four urban academic EDs were selected because they are known to
overdose die within 1 year.11 Because of this risk and the increased the last author as institutes that have robust ED bridge programs.
number of patients with OUD seeking treatment in the ED, it is a Key stakeholders (RH, AH, JH, ML) recruited participants who were
perfect venue to initiate MOUD. employed to work within the ED at one of four urban academic
Buprenorphine, a strong, partial mu-­opioid receptor agonist, de- hospital systems. The stakeholders were known to the last author
creases mortality risk in patients with OUD and increases treatment to be a person who championed the buprenorphine bridge program
retention.12–15 Emergency physicians (EPs) can use buprenorphine within the hospital's ED. The stakeholders sent out email invitations
to initiate treatment and manage withdrawal. When buprenorphine that included the eligibility criteria to all EPs in their hospitals.
is initiated in the ED, patient engagement in addiction treatment at Interested EPs contacted the research coordinator to be screened
30 days increases.16–18 for eligibility. All EPs (n = 28) that contacted the research coordinator
A much-­studied model is the “emergency department bridge pro- were screened as eligible for the study. Six of the eligible participants
gram” (EDBP), in which MOUD is initiated in the ED, and the patient did not respond to multiple email invitations and were deemed lost
is then followed up urgently in an affiliated longitudinal clinic de- to follow-­up. The research coordinator scheduled and completed
signed to stabilize patients further and support them with long-­term remote video interviews with the eligible remaining participants
care. The EDBP model has been shown to be effective.19 Identified (n = 22). The interviews took place between February 2022 and June
barriers to adopting buprenorphine in the ED include lack of depart- 2022. This study was reviewed by our institutional review board and
ment support, limited follow-­up services, and EPs holding negative found to be exempt. The study is reported in accordance with the
views toward initiating buprenorphine in the ED. 20 Additionally, pro- Standards for Reporting Qualitative Research (Appendix S2).30
vider stigma and negative attitudes conveyed toward patients with
OUD and prescribing MOUD have been described as a major barrier
to initiating ED MOUD. 21–26 Qualitative interview procedures

The interview protocol was constructed by JML with input from

Goals of this investigation the project team members based on the aims of the study and prior
research literature related to EP experiences working with patients
The potential benefits to physicians of a successfully implemented with OUD.4,28 The protocol contained six prompts that explored
EDBP have not been explored. In this cross-­sectional study, we (a) participants’ motivation for participating in an EDBP; (b) the
utilized qualitative methods to explore and understand how EPs professional impact of participating in EDBP, (c) the personal impact
make meaning of their perceptions and experiences treating patients of participating in EDBP; (d) barriers, resilience, and strengths
with OUD within an EDBP model. EPs who work in EDs that have related to EDBP implementation and maintenance; (e) physicians’
successfully implemented buprenorphine programs for patients with attitudes about patients and perceived impact on patients with
OUD (EDBP) were interviewed. A priori, we hypothesized that EPs OUD; and (f) the impact of the EDBP on professional relationships
 A QUALITATIVE ASSESSMENT OF EP'S EXPERIENCES WITH ED BUPRENORPHINE BRIDGE
|

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14880 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
578 PROGRAMS

(Appendix S1). We used the critical incident technique (CIT) to team members used best practices (i.e., memos and reflexive
uncover existing realities or truths so they can be measured and discussions) throughout the project to track subjectivities and
predicted, which involves participants recalling a specific incident distinct perspectives. We worked to bolster the trustworthiness
(or story) in which the phenomenon under study was activated in of our findings by including, throughout this article, logical and
their consciousness and influenced their experiences. 29,30 The clear descriptions of the research processes and rationale for
technique gives participants freedom in describing the experience them.
and assumes that participants will describe incidents (stories) that We avoided claims of saturation, as recommended by the de-
have high priority and that have been most impactful for them. For velopers of thematic analysis, and instead focused on information
example, participants were instructed, “Tell me a story about how power.31 Information power is having enough participants enrolled
participating in the ED bridge program has impacted you as a person.” a study to develop new knowledge from the analysis.32 Information
All participants were presented with all six CIT prompts. Short power is said to be higher when (a) the study aim is narrow, (b) the
follow-­up questions or prompts were utilized to gather more detail study sample is targeted while also possessing between participant
when a participant gave their initial response and the interviewer variation in the phenomena under study, (c) a specified coherent the-
perceived needing additional details. The research coordinator ory is used to guide study design and data analysis/interpretation,
reiterated throughout the interview that we were specifically asking (d) there is strong and clear communication between interviewer
about their work within the ED and not a clinic within the ED that and interviewee, and (e) an in-­depth narrative analysis of a few
serves patients with substance use disorders, sometimes called a participants is conducted. Our study meets these criteria. We are
“bridge clinic.” confident that our information power is sufficiently high given our
The research coordinator conducted the individual interviews development of new knowledge.
through a HIPAA-­compliant video conferencing platform. In addition
to recording audio and video, the research coordinator took notes.
The notes were written up into memos directly following each inter- Rationale for choice of methods
view. The memos contained noteworthy statements from the par-
ticipant and the interviewer's reaction to the interview. The content Qualitative data were analyzed using thematic analysis. Thematic
of these memos was used to adjust the interview guide allowing for analysis is a robust data analytic method for identifying, analyzing,
a more iterative process. Interviews took an average of 38:03 min and describing patterns of meaning across a data set in rich detail.33
to complete (range of 8:55–42:35 min). Participants received a $100 This study utilized a codebook approach to thematic analysis.34,35
gift card for completing the interview. The codebook allowed us to develop codes from our textual data.
The interviewer was a research coordinator with 7 years of The codebook also facilitated consistency among coders, thus
experience in quantitative and qualitative research focusing on bolstering the reliability of findings.
substance use treatment in and outside the ED. Additionally, the
interviewer works with EPs and is married to an EP. This experi-
ence allowed the interviewer to create a better rapport with the Codebook development
participants.
Author JML developed a codebook using a bottom-­up (inductive
process) whereby he adhered to guidelines of the first two phases
Analysis of thematic analysis: familiarization with data (Phase 1) and coding
(Phase 2). First, he began by coding a random sample of five of
Positionality statement the transcripts from the data set using an inductive approach. He
created codes primarily at the semantic level—staying close to the
The qualitative analysis team consisted of five members from participant's language—which entailed reading the transcripts
diverse backgrounds. Author JML, a Black American same-­ and labeling sections that seemed to be relevant to the research
gender–loving licensed clinical psychologist and health equity questions. Next, latent coding was conducted with each transcript
researcher with 10+ years of experience conducting qualitative to uncover implicit meaning in participants’ transcripts. Through this
research, led the analyses. Four research assistants (RAs; two process, 18 codes were developed and comprised the codebook,
graduate students and two undergraduate students) who were which included code names, code abbreviations, and extracted
members of author JML's laboratory comprised the analysis team. quotes from the transcripts that illustrated each code. Four RAs
All RAs were women of color who received training in coding using in JML's laboratory were then trained how to use the codebook.
the codebook by author JML. The analysis team noted several RAs were deemed proficient in codebook utilization when they
similarities in their subjectivities while conducting the analyses. began to achieve at least 90% accuracy in coding the original five
As people of color and nonphysicians, they were cultural outsiders transcripts utilized by JML to make the codebook. Once proficiency
who were not completely familiar with the professional, cultural, was achieved by all RAs, JML randomly assigned the remaining 17
historical, and sociopolitical contexts of the participants. Overall, transcripts to the RAs. Each transcript was coded by two different
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14880 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HEIL et al. 579

RAs. Discrepancies in coding were resolved with discussion until a themes that described the positive effects of participating in the
consensus was reached. EDBP for physicians (Table 2).

Data analysis process Theme 1: Providers gained agency through their

participation in EDBP
NVivo was used to analyze the qualitative data. Each coder followed
the steps outlined by Braun and Clarke in analyzing transcripts.30 EPs expressed that participating in the EDBP helped them develop
Specifically, each coder familiarized themselves with the data by a sense of agency in their work by giving them access to treatment
reading and rereading each transcript before coding. Coding was options they needed to take multilevel action. This sense of agency
informed by the codebook. Nine additional codes were identified allowed them to provide effective treatment to their patients with
and added to the codebook during coding of subsequent transcripts. OUD and address barriers that hindered the provision of treatment.
All transcripts were then reanalyzed to determine the presence or EDBP help physicians effectively respond to the needs of a
absence of those new codes. Once all transcripts were completely population with frequent contact in ED and help EPs to provide an
coded, author JML utilized NVivo to generate initial themes (Phase opportunity for patients to overcome the seemingly insurmountable
3) by clustering together potentially connected codes based on problem of OUD (Table 2).
similarities into candidate themes. This process yielded three One participant (P4, 55, White, male) discussed how EDBP was
themes. After examining the themes in the context of the full data a way not only to help patients get effective medical care from EPs
set to ensure that they clustered around a singular central idea but also to get additional services related to recovering from OUD.
(Phase 4), in Phase 5, the themes were presented to the larger study He shared, “So we were doing more than just, you know, prescribing
team for feedback. After revisions, theme names and definitions medications—that we actually were getting people plugged into a
were finalized. system that could help them, sort of, move from coming in—get help
to, uh, successfully move on to treatment.”

Trustworthiness
Theme 2: Transformation in providers' emotions,
To ensure the trustworthiness of our findings, we have focused on attitudes, and behaviors related to participation
four criteria specified by Nowell and colleagues36 throughout the in EDBP
research process: credibility, transferability, dependability, and
confirmability. We established credibility through peer debriefing. Providers disclosed how they evolved from pathological and
Peer debriefing was completed by a professor at the same institution patient-­blaming attitudes and emotions to developing more
as author JML but who works in a separate department from any of holistic understandings of their patients with OUD. These holistic
the study team members. This person has several years of experience understandings facilitated positive emotions and more engaged
conducting qualitative research. Her assessment of our analysis physician-­patient relationships. One participant (P13, 51, White,
processes was that all codes were applied consistently across the male) revealed, “I don't want to say that I judged them—I judged
data set. We have tried to ensure transferability of our findings by them in the past, but probably to a degree I did, because I didn't
providing clear descriptions of our research process and results. With understand as much [as] now.”
these descriptions, readers will be able to determine transferability Participating in an EDBP seems to help EPs cultivate a greater
of our findings from this study to their sites of research. We have degree of empathy for patients suffering from OUD. This empathy,
worked to ensure dependability of our findings by providing a logical once developed by EPs, is then shared with colleagues through ad-
description of our methods and analyses. Finally, we have attempted vocating for them in conversations with colleagues (Table 2).
to bolster the confirmability of our findings by detailing the rationale
for our analytical choices throughout this article.
Theme 3: EDBP participation improved professional
quality of life
R E S U LT S
Physicians shared stories about how their participation in an EDBP
Twenty-­t wo EPs with a mean (±SD) age of 43 (±10) years par- contributed to a better quality of life for themselves, their patients,
ticipated in the study. The predominately White (95.5%) and male and society as a whole. They reported feeling more helpful, appreci-
(68.2%) sample reported primarily living in the Northeast (59.0%) ated, energized, and supported. One participant (P14, 38, White, male)
region of the United States. Participants had been practicing medi- shared, “… in terms of … getting the X-­waiver and then prescribing it …
cine for a mean (±SD) 13 (±10) years and most (59.0%) reported hav- it is actually rewarding to do so … I find it's actually really important.
ing a DEA X-­waiver for 3 to 5 years (Table 1). We constructed three … I didn't know that at the time when I started it … but I'm glad I did it
 | 580

TA B L E 1 Qualitative results.

Age Geographic Years practicing Years since Themes endorsed Theme 1 endorsed by Theme 2 endorsed by Theme 3 endorsed by
Participant (years) Race Gender location medicine X-­waviered by each participant number of participants number of participants number of participants

1 62 White Female Northeast 30 3–5 years All 22 20 22

2 40 White Female Northeast 7 >5 years All
3 35 White Male Northeast 6 3–5 years All
4 55 White Female West 27 1–3 years All
5 48 White Male West 20 3–5 years All
6 55 White Male West 25 3–5 years All
7 32 White Female West 2 6 months–1 year All
8 36 White Male Northeast 7 3–5 years All
9 45 White Male Northeast 12 1–3 years All
10 32 White Male Northeast 3 3–5 years All
11 32 Asian Male Northeast 4 3–5 years All
12 48 White Female Northeast 19 3–5 years All
13 51 White Male Northeast 23 >5 years All
14 38 White Male Northeast 9 3–5 years All, except 2
15 34 White Male West 3 1–3 years All
16 33 White Male West 3 3–5 years All
17 36 White Male West 4 1–3 years All, except 2
18 60 White Male Northeast 32 3–5 years All
19 46 White Female West 16 1–3 years All
20 54 White Male Northeast 23 3–5 years All
21 34 White Male Northeast 2 1–3 years All
22 37 White Female West 11 3–5 years All

Notes: Theme 1: Providers gained agency through their participation in ED bridge. Theme 2: Transformation in providers' emotions, attitudes, and behaviors related to participation in ED bridge. Theme 3:
ED bridge participation improved professional quality of life.
PROGRAMS
A QUALITATIVE ASSESSMENT OF EP'S EXPERIENCES WITH ED BUPRENORPHINE BRIDGE

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14880 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
 |

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14880 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
HEIL et al. 581

TA B L E 2 Themes and exemplar quotes.

Physicians perceived improving their patients’ quality of life
Theme 1: Providers gained agency through their participation in BP by providing more in-­depth education about OUD treatment.
Exemplar quotes
Participants also endorsed being able to better care for this patient
… all of a sudden, now you have a tool that you can use to kind of
population, resulting in experiencing several professional benefits,
help this population that you see over and over and over again in
the department.—P2, 40, White, female including helping reduce burnout. Overall, physicians who partici-
pated in EDBP reported feeling a renewed sense of fulfillment and
… it made treating what was a very difficult problem a lot easier,
because the challenge was you felt sort of helpless, like you purpose in their personal and professional lives.
couldn't do anything to help them. It was sort of the inevitability
of the train's coming, you're tied to the tracks, and I can't help
you. And [inaudible] now you had a way of getting them off the
DISCUSSION
track a little bit and giving them a chance.—P11, 32, Asian, male
So we were doing more than just, you know, prescribing medications—
that we actually were getting people plugged into a system that In this cross-­sectional study, we identified three major themes that
could help them, sort of, move from coming in—get help to, uh, suggest a positive impact of prescribing buprenorphine from the ED.
su-­successfully move on to treatment.—P6, 55, White, Male Working in an EDBP (1) provided EPs agency; (2) transformed EP
Theme 2: Transformation in providers' emotions, attitudes, and emotions, attitudes, and behaviors related to treating patients with
behaviors related to participation in BP OUD; and (3) improved EP professional quality of life. While there is
Exemplar quotes
ample evidence of the potential benefits of EDBP to patients, to our
I don't want to say that I judged them—I judged them in the past, but
knowledge, this is the first report to demonstrate a potential benefit
probably to a degree I did, because I didn't understand as much
now.—P14, 38, White, male to ED physicians.
We found that EPs gained agency by working in EDs with bridge
… it's changed my, um, my views on people with opioid use disorder.
Um, you know, I–again, my—you know, my—I was talking about programs because they can provide effective treatment to patients
my, my initial views opioid use disorder as a—as a new physician. that were not previously available to them. This sentiment is con-
Um, and I think, you know, kind of the ways that we were sistent with previous literature that found the ability to provide
thinking about opioid disorder have changed societally, um, as
high-­quality care to patient populations is associated with job satis-
well, kind of it around the same time period. Um, you know, I
think—I think now I'm more sympathetic to, you know, people.— faction.37–39 This is also the case when physicians have the capability
P8, 36, White, male to treat complex patients.40
So the thing—the thing that, um—that's interesting about working in Concurrently, the stigma surrounding patients with OUD is a
the bup clinic, I mean, the—and, and—that I, I can take to my, my barrier to those seeking care for their disorder. This stigma is perva-
colleagues is, is sort of like—you don't understand. These folks, sive in the ED, where physicians' negative attitudes toward patients
these are your neighbors, man. This is like your cousin. Um, the
with OUD are a major barrier to initiating MOUD programs. 21–26
guy who's sitting in the waiting room all pissed off, um, isn't
pissed off because he's an angry soul. It's that if he doesn't get Most EPs in this study reported having negative attitudes toward
seen soon, um, he's gonna be late for his shift at the mine, and patients with OUD prior to participating in an EDBP. Participating in
once the train or whatever goes into the mine, if he misses it, he EDBP changed these views, decreasing stigma and increasing em-
misses his whole day, you know? These are folks who just wanna
pathy. In the ED, patients with OUD perceive discrimination and
go to work.—P20, 54, White, male
stigma due to their substance use, so modifying physician behav-
Theme 3: BP participation improved quality of life
Exemplar quotes ior and attitudes may be a key toward engaging more patients into
treatment.41
… in terms of … getting the X waiver and then prescribing it … it is
actually rewarding to do so … I find it's actually really important. Finally, we discovered that EDBP participation increased the
… I didn't know that at the time when I started it … but I-­I'm glad professional quality of life of participating physicians. Physicians are
I did it, because you kind of are the, the doc that can do that for at higher risk for burnout than the general U.S. working population,
people.—P14, 38, White, male
likely because of the high intensity of clinical practice, higher risk of
I just think that we're able to provide a greater value for the patients that
litigation than some other specialties, and chronic fatigue related to
present with opioid use disorder and, uh, it gives us, uh, a chance
to kind of educate more than what we had been doing previously, circadian rhythm disruption.42 EPs reported the highest burnout rate
where again, we were just treating symptoms.—P18, 60, White, male of all specialties in 2022 (60%), a large jump from 43% the year prior
I think that it helped with provider—specifically, my burnout because and likely related to changes in the health system post-­COVID.43
you're able to provide resources and, um, not just say, “I'm so There may also be a component of moral injury or second victim
sorry you're going through this.” It's helpful to have a solution.— syndrome resulting either from feeling helpless when treating pa-
P7, 32, White, female
tients with OUD or from the trauma incurred to the physician when
Abbreviation: BP, bridge program. a patient under their care dies from an opioid overdose. The results
of this study suggest that the first two themes (gained agency and
because you kind of are the, the doc that can do that for people.” The changed attitudes) augment the experience of caring for this patient
participant's comment hints at a sense of personal pride in being able population, making EPs feel helpful and rewarded, potentially de-
to serve his clients in a way that he was unable to do before. creasing at least one of the many reasons for burnout.
 A QUALITATIVE ASSESSMENT OF EP'S EXPERIENCES WITH ED BUPRENORPHINE BRIDGE
|

15532712, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/acem.14880 by Shifa Tameer-e-Millat, Wiley Online Library on [26/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
582 PROGRAMS

Unfortunately, many hospitals have been slow to adopt bu- results can help hospital leadership and policymakers to adopt
44
prenorphine programs because of well-­described barriers. an ED bridge program at their hospitals. Future research should
Inadequate follow-­up remains a key barrier for EPs and contributes include collecting quantitative data (quality-­of-­life scales, burnout
to reluctance to prescribe buprenorphine. A recent Canadian study measures, and stigma) from emergency physicians before and after
cited access to follow-­up resources as one of the top barriers to ini- introducing a bridge program into their hospital or health care
tiating buprenorphine in their EDs, along with inadequate training system. Researchers should also carry out this study in different
on buprenorphine prescribing and time restraints related to patient settings (rural, community, etc.) to generate more generalizable data.
education on safe buprenorphine use. 28 There was a 300% increase
in the number of EPs prescribing buprenorphine between 2016 AU T H O R C O N T R I B U T I O N S
45
and 2021. This increase may be due to increased implementation Rachel Haroz conceived of the study and obtained research funding.
of EDBP. For example, California's CA Bridge program resulted in Rachel Haroz, Jessica Marie Heil, and Jonathan M. Lassiter designed
100% of the 52 participating hospitals routinely treating OUD with the study. Rachel Haroz. Andrew Herring, Jessica Marie Heil, and
buprenorphine in the ED. 27 Michael Lynch undertook recruitment of participating centers. Jessica
Not only do EDBPs improve clinical outcomes for patients, but Marie Heil collected the participant's data. Jonathan M. Lassiter ana-
this study provides evidence that such programs may also be ben- lyzed the data and interpreted the results. Jessica Marie Heil drafted
eficial for EPs as well. Buprenorphine prescribing provides agency the manuscript, and all authors contributed substantially to its revision.
to EP clinicians, potentially decreases stigma toward patients with
OUD and improves professional quality of life. The first-­hand testi- C O N F L I C T O F I N T E R E S T S TAT E M E N T
monies from EPs in this study may help provide leverage to influence The authors declare no conflicts of interest.
EDs across the globe to pursue implementing EDBPs.
ORCID
Jessica Marie Heil https://orcid.org/0000-0002-8876-0922
LIMITATIONS Matthew S. Salzman https://orcid.org/0000-0002-8277-4928
Scott G. Weiner https://orcid.org/0000-0002-4672-5184
This study has several limitations that may impact the interpretation Brian Roberts https://orcid.org/0000-0002-7690-997X
of the findings. The study's cross-­section design limits temporality
and causation conclusions. All interviews took place after the REFERENCES
physician worked in an established EDBP; therefore, there is a 1. Spencer MR, Warner M, Cisewski JA, et al. Estimates of Drug
potential for recall bias. It is possible that this sample of EPs had Overdose Deaths Involving Fentanyl, Methamphetamine, Cocaine,
Heroin, and Oxycodone: United States, 2021. 2023.
lower baseline stigma, which made them more likely to work in an
2. Krawczyk N, Rivera BD, Jent V, Keyes KM, Jones CM, Cerdá M.
ED with a bridge program, as opposed to their participation in an Has the treatment gap for opioid use disorder narrowed in the
established bridge program thereby reducing stigma. However, the US?: A yearly assessment from 2010 to 2019. Int J Drug Policy.
consistency of the themes across EPs makes recall bias less likely. 2022;110:103786.
3. TIP 63: Medications for Opioid Use Disorder -­ Full Document.
Also, these data cannot be generalized to all EPs who prescribe
Substance Abuse and Mental Health Services Administration. 2018.
buprenorphine, as most EPs in this sample were White males. 4. Mackey K, Veazie S, Anderson J, Bourne D, Peterson K. Barriers
Virtual interviews may have hindered rapport building between the and facilitators to the use of medications for opioid use disorder: a
interviewer and the interviewee. Participants may have felt less at rapid review. J Gen Intern Med. 2020;35:954-963.
5. Hawk K, Hoppe J, Ketcham E, et al. Consensus recommendations
ease in divulging certain types of information during the interview
on the treatment of opioid use disorder in the emergency depart-
process. Our study may be limited by selection bias in that physicians ment. Ann Emerg Med. 2021;78(3):434-442.
with more positive experiences may have elected to participate in 6. Emergency Visits for Opioid Use Rose While Treatment for Illicit
our study than those with more negative or neutral attitudes and Drug Use Remained Unchanged. Agency for Healthcare Research
experiences related to bridge programs. Thus, our findings may not and Quality. AHRQ Publication No. 21- 0052. August 2021. https://​
www.​a hrq.​g ov/​s ites/​​d efau​l t/​f iles/​​w ysiw​y g/​r esea​r ch/​f indi​n gs/​
be generalizable to all EPs and EDs.
nhqrdr/​qdr-­​data-­​spotl​ight-­​opioi​ds-­​edvis​its-­​tx.​pdf
7. Salzman M, Jones CW, Rafeq R, Gaughan J, Haroz R. Epidemiology of
opioid-­related visits to US emergency departments, 1999–2013: a ret-
CO N C LU S I O N S rospective study from the NHAMCS (National Hospital Ambulatory
Medical Care Survey). Am J Emerg Med. 2020;38(1):23-27.
8. Venkatesh AK, Janke AT, Kinsman J, et al. Emergency department
The ED remains the safety net and point of entry into the health care utilization for substance use disorders and mental health conditions
system for many patients struggling with substance use disorders. during COVID-­19. PloS One. 2022;17(1):e0262136.
In this study, the presence of an ED bridge program empowered 9. Soares WE III, Melnick ER, Nath B, et al. Emergency depart-
ment visits for nonfatal opioid overdose during the COVID-­19
physicians; led to a shift in their emotions, attitudes, and behaviors
pandemic across six US health care systems. Ann Emerg Med.
when treating patients with opioid use disorder; and ultimately 2022;79(2):158-167.
enhanced their overall professional quality of life. This study's
Other documents randomly have
different content
 143. Lamas Drinking Tea in the Court of Ceremonies in Tashi-lunpo.
Sketch by the Author.

On February 21 I spent nearly the whole day in parts of the monastery I had not
previously seen. We wandered through narrow winding corridors, and lanes in deep
shadow, between tall white-washed stone houses, in which the monks have their
cells. One of the houses was inhabited by student monks from the environs of Leh,
Spittok, and Tikze, and we went into the small dark cubicles, hardly larger than my
tent. Along one of the longer sides stood the bed, a red-covered mattress, a pillow,
and a frieze blanket. The other furniture consisted of some boxes of books,
clothing, and religious articles. Holy writings lay opened. A couple of bags contained
tsamba and salt, a small altar with idols, votive vessels, and burning butter lamps,
and that is all. Here it is dark, cool, damp, and musty—anything but agreeable; very
like a prison. But here the man who has consecrated his life to the Church, and
stands on a higher level than other men, spends his days. Monks of lower rank live
two or three in one cell. Gelongs have cells to themselves, and the chief prelates
have much more elegant and spacious apartments.
 Each monk receives daily three bowls of tsamba, and takes his meals in his own
cell, where tea also is brought to him three times a day. But tea is also handed
round during the services in the temple halls, in the lecture-rooms, and in the great
quadrangle. No religious rite seems to be too holy to be interrupted at a convenient
time by a cup of tea.

One day from the red colonnade (Kabung) I looked down on the court full of
lamas, who were sitting in small groups, leaving only narrow passages free, along
which novices passed to and fro with hot silver and copper pots, and offered the
soup-like beverage stirred up with butter. It had all the appearance of a social “Five
o’clock tea” after some service. But the meeting had a certain touch of religion, for
occasionally a solemn, monotonous hymn was sung, which sounded wonderfully
beautiful and affecting as it reverberated through the enclosed court. On March 4
the quadrangle and other places within the walls of Tashi-lunpo swarmed with
women—it was the last day on which the precincts of the monastery were open to
them; they would not be admitted again till the next Losar festival (Illustration
150).

The young monk who, when accompanying the Tashi Lama in India, had had an
opportunity of learning about photography, had his dark room beside his large
elegant cell. I, too, was able to develop my plates there. He asked me to come
frequently and give him instructions. He had solid tables, comfortable divans, and
heavy handsome hangings in his room, which was lighted with oil lamps at night.
There we sat and talked for hours. All of a sudden he took it into his head to learn
English. We began with the numerals, which he wrote down in Tibetan characters;
after he had learned these by heart he asked for other of the more common words.
However, he certainly made no striking progress during the few lessons I gave him.

Care is necessary in walking through the streets of the cloister town, for the
flags, which have been trod by thousands of monks for hundreds of years, are worn
smooth and are treacherous. Usually there is a good deal of traffic, especially on
feast days. Monks come and go, stand talking in groups at the street-corners and in
the doorways, pass to and from the services, or are on their way to visit their
brethren in their cells; others carry newly-made banners and curtains from the
tailor’s shop into the mystical twilight of the gods; while others bear water-cans to
fill the bowls on the altars, or sacks of meal and rice for the same purpose. Small
trains of mules come to fill the warehouse of the convent, where a brisk business is
going on, for a family of 3800 has to be provided for. And then, again, there are
pilgrims, who loiter about here only to look in on the gods, swing their prayer mills,
and murmur their endless “Om mani padme hum.” Here and there along the walls
beggars are sitting, holding out their wooden bowls for the passer-by to place
something in, if it is only a pinch of tsamba. The same emaciated, ragged beggars
are to be found daily at the same street-corners, where they implore the pity of the
passengers in the same whining, beseeching tone. In the narrow lanes, where large
prayer mills are built in rows into the wall, and are turned by the passers-by, many
poor people are seated, a living reproof of the folly of believing that the turning of a
prayer mill alone is a sufficiently meritorious action on the way to the realms of the
blessed. In one particularly small room stand two colossal cylindrical prayer mills
before which a crowd is always collected—monks, pilgrims, merchants, workmen,
tramps and beggars. Such a praying machine contains miles of thin paper strips
with prayers printed on them, and wound round and round the axis of the cylinder.
There is a handle attached, by which the axle can be turned. A single revolution,
and millions of prayers ascend together to the ears of the gods.

CHAPTER XXIX
WALKS IN TASHI-LUNPO—THE DISPOSAL OF THE DEAD

Immediately below the red colonnade stands the Sokchin-rungkang-chimbo,

the kitchen, with its walled-up stove of colossal dimensions and six huge caldrons
embedded in masonry. The first supplies all the 3800 monks with tea at one boiling.
On the part of the caldron which rises above the masonry are inscriptions and cast
ornaments (Illustration 141). Each caldron has a wooden cover which is put on
when the caldron is not in use. Tea was being prepared in two of these gigantic
pots; probably allowance was made for any chance guests. Glowing, blazing
fireplaces yawn below the caldrons, and faggots of branches and sticks are thrust in
with long iron forks. There is an opening in the roof for the smoke, which rises up
in grey rings and produces a picturesque illumination in the holy kitchen. A
continuous succession of young lamas and workmen ascend the steps leading up
from the street, carrying on their backs water-tubs of different capacity according to
the strength of the bearer; for there are quite small boys among them, who have
recently been consigned by their relations to the care of the monks. One after
another tips his tub over the edge of the caldron, while the stoker thrusts fresh
faggots of wood into the stove. Other serving brothers bring in a quantity of cubes
of brick tea which they throw into the boiling water, whence clouds of steam ascend
and mingle with the smoke. At the side of the caldron stand two cooks, who stir
with huge staves larger than oars, and disappear in the rising steam, becoming
visible again, like shadow figures lighted from above, when a slight draught from
the door clears the air. They sing a slow rhythmical song over their work.

When the tea is ready, it is poured into large bright copper pots with shining
yellow brass mountings, handles, and all kinds of ornamentation. Novices carry the
vessels on their shoulders to all the various halls and cells. A loud signal is given on
a sea-shell from a temple roof that the monks may not miss their tea, but may be
on the look-out. I frequently looked into the kitchen, the scene was so picturesque,
and the cooks were ready for a joke and were not averse to being sketched
(Illustration 148).

Two large and several small chhortens are erected on an open square in front of
the mausoleums, of exactly the same design as those so frequently seen in Ladak.
There are also stone niches filled with idols and other objects. A crowd of people
was collected on the terrace when I was sketching, and it was not easy to get a
clear view. It was a striking picture, with all the red and many-coloured garments
against the background of the white-washed walls of the memorial towers
(Illustration 151).
144. Part of Shigatse.
 145. The Tashi Lama returning to the Labrang after a Ceremony.

One day when I had sat a long time talking in the cell of the photographing lama,
it was dark when I went home. We passed, as we often had before, the entrance
gate to the forecourt of the Namgyal-lhakang, the temple in which the Tashi Lama
had once provided us with refreshments. There the evening service was in full
swing, and of course we entered to look on. The illumination was more dimly
religious than usual, but we could at any rate make out our surroundings after
coming straight out of the outer darkness. The monks sat on long red divans, and
their black profiles were thrown up by the row of forty flames burning in bowls
before the altar. The gilded lotus blossoms of the pedestal were brightly lighted,
and the yellow silken scarves in the hands of Tsong Kapa’s statue and the garlands
draped over the images stood out conspicuously. But the upper parts of the figures
under the roof were plunged in darkness, and Tsong Kapa’s countenance, with
plump rosy cheeks and broad nose, was so curiously lighted up from below that his
smile was not perceptible. The four coloured pillars in the middle of the hall
appeared black against the altar lamps. The monks wore yellow robes, sat bare-
headed, and chanted their melancholy litanies, now and then interrupted by ringing
of bells and the roll of drums. At first the leather head of the drumstick falls slowly
and regularly on the tight skin, then the beats become more and more frequent,
and at last the drum becomes silent in an instant. A monk recites “Om mani padme
hum” in rising and falling tones with the rapidity of an expert, and the others join
in, making some kind of responses. The recitation passes into a continuous hum, in
which often only the words “Om mani” are heard aloud, and the word “Lama”
uttered more slowly. The whole ritual has a singularly soporific effect; only Tsong
Kapa listens attentively, sitting dreamily with wide staring eyes, and ears hanging
down to the shoulders. Here, too, the indispensable tea is handed round; a monk
with an oil lamp attends the server that he may be able to see the cups. The monks
were now quite accustomed to my visits and took no particular notice of me, but
they always greeted me politely and asked what I had been sketching during the
day.

A lama gave me information about a remarkable custom. Certain monks consent

of their own free will to be walled up in dark grottoes or caves for the space of
three, six, or at most twelve years. Near a small monastery, Shalu-gompa, a day’s
journey from Tashi-lunpo, there is a monk who has already spent five years in his
grotto, and is to remain there seven more. In the wall of the grotto is an opening a
span in diameter. When the twelve years are over, and the hermit may return to the
light of day, he crawls out through this opening. I insinuated that this was a
physical impossibility, but the lama replied that the miracle does take place, and,
besides, the enclosed monk has become so emaciated in the twelve years that he
can easily slip through the opening. One of the monks of the monastery goes daily
to the grotto with tea, water, and tsamba, and pushes these provisions through the
opening, but he may not speak to the prisoner or the charm would be broken. Only
sufficient light penetrates through the opening to allow the anchorite to distinguish
between day and night. To read the holy scriptures, which he has taken with him
into the cave, he must use an oil lamp, and a fresh supply of oil is placed from time
to time in the opening. He says his prayers all day long, and divides the night into
three watches, of which two are spent in sleep and one in reading. During the
twelve years he may not once leave his grotto, never look at the sun, and never
kindle a fire. His clothing is not the usual monk’s dress, but a thin cotton shirt, and
a girdle round the body; he wears no trousers, head-covering, or shoes.

Among other abstruse subjects, this penitent must study a composition on some
kind of magic, which renders him insensible to cold and almost independent of the
laws of gravity. He becomes light, and when the hour of release arrives, travels on
winged feet: whereas he used to take ten days to journey from Tashi-lunpo to
Gyangtse, he can now cover the distance in less than a day. Immediately the twelve
years of trial are ended, he must repair to Tashi-lunpo to blow a blast of a horn on
the roof, and then he returns to Shalu-gompa. He is considered a saint as long as
he lives, and has the rank of a Kanpo-Lama. No sooner has he left his grotto than
another is ready to enter the darkness and undergo the same test. This lama was
the only one in this neighbourhood then confined in a grotto, but there are hermits
in abundance, living in open caves or small stone huts, and maintained by the
nomads living near them. We were later on to hear of fanatical lamas who renounce
the world in a much stricter fashion.

In Tashi-lunpo the cloister rule seems to be strictly enforced: there are especial
inspectors, policemen and lictors who control the lives of the monks in their cells
and take care that no one commits a breach of his vows. Recently a monk had
broken the vow of chastity; he was ejected for ever from the Gelugpa confraternity
and banished from the territory of Tashi-lunpo. He has, then, no prospect of finding
an asylum in another monastery, but must embrace some secular profession.

One day we visited the Dena-lhakang, a temple like a half-dark corridor, for it is
lighted only by two quite inadequate windows. In the middle of the corridor there is
a niche which has doors into the hall, for the walls are very thick. Thus between the
doors and the window is formed a small room in which the lama on duty sits as in a
hut. He belongs to the Gelong order, is named Tung Shedar, came from Tanak, and
is now seventy years old, has short white hair, and a skin as dry as an old yellow
crumpled parchment.

On entering, one sees on the right a bookcase with deep square pigeon-holes, in
which holy books are placed. On the outer, longer wall, banners painted with
figures hang between the two windows, in the deepest shadow, most of which are
of venerable age, and are dusty and faded—a Lamaist picture gallery. Pillars are
ranged along the longer wall, of red lacquered wood, and between them is
suspended trellis-work of short iron rods, forming geometrical figures. They are
intended to preserve the valuables from theft. In such a niche we see hundreds of
small idols set round in rows, four to eight inches high, in silken mantles. Before
them are taller statues of gods, and Chinese vases of old valuable porcelain.
Especial reverence is shown to a cabinet with an open door, within which is
preserved a tablet, draped with kadakhs, and inscribed with Chinese characters, in
memory of the great Emperor Kien-Lung who was admitted by the third Tashi Lama
into the confraternity of the yellow monks. Above, covering the capitals of the
pillars, is hung strange, shabbily-fine drapery, of pieces of variously coloured cloth
and paper strips. For the rest, the hall abounds in the usual vessels, brazen
elephants with joss-sticks, large chalices and bowls, small and large flags, and other
things.

Another time I had been drawing in a sepulchral chapel and taken the
opportunity of making a sketch of some female pilgrims who were praying there.
When the work was finished, we crossed a paved court fully 20 yards broad by 90
long, which was situated just under the façade of the Labrang. It was full of people
waiting to see the Tashi Lama, who was to pass by on his way to some ceremony.
He came in a red monk’s frock and the yellow mitre; above his head was held the
yellow sunshade, and he was accompanied by a train of monks. He walked with his
body slightly bent and an air of humility. Many fell down before him full-length and
worshipped him, while others threw grains of rice over him. He did not see me, but
his smile was just as kind and mild as when we last met. So he is evidently affable
to all alike.

I made daily visits to the monastery and so gained a thorough knowledge of the
solitary life of the monks. Gompa signifies “the abode of solitude,” or monastery;
the monks in the convent certainly live isolated from the outer world, its vanities
and temptations. Once, in the Kanjur-lhakang, I purposed to draw the images with
the lamps burning before them on the innermost, darkest wall, but just as I was
about to begin monks filled the hall. Their places on the long divans were made
ready for them, and before each seat a huge volume of the holy scriptures, the
Kanjur, lay on a long continuous desk. The large yellow robes which are put on at
service time, but may not be worn in the open-air, were laid ready. The young,
brown-skinned, short-haired monks entered in red togas, threw the yellow
vestments over their shoulders, and sat down cross-legged before the books. An
older lama, a Kanpo, mounted the pulpit on the shorter wall and intoned the sacred
text in a harsh, solemn, bass voice. The pupils joined in a monotonous rhythm.
Some read from the pages in front of them, while others seemed to know the
words of the chant by heart—at any rate they looked all about. Exemplary order is
not observed. Some young fellows, who certainly were much more at home in the
world than in the Church, talked during the chant, giggled, and buried their faces in
their robes to stifle their laughter. But no one took any notice of them; they caused
no disturbance. Others never raised their eyes from the book. The hall was as dark
as a crypt, being lighted only by a narrow skylight, and through two small doors
(Illustration 154).
 146. The Panchen Rinpoche, or Tashi Lama.

After they had sung awhile there was an interval, and lama boys passed along
the gangways between the rows of benches and poured tea, with wonderful
adroitness and without spilling a drop, into the wooden cups held out to them. But
almost before the pupils have begun to drink, the deep bass of the leader drones
out in the gloom above, and the proceedings recommence. Meanwhile pilgrims pass
along the gangways to the altar and place small heaps of tsamba or meal in the
bowls standing before the images, from the bags and bundles they bring with them.

A tall lama stands erect at the entrance door. A pilgrim says to him: “I will pay 3
tengas for a blessing.” The lama sings out aloud the contribution and the purpose
for which it is given, and then a strophe is sung especially on behalf of the pilgrim,
after which all the monks clap their hands. This is repeated whenever fresh pilgrims
come up. I myself paid 5 rupees for a blessing, and received it together with a
noisy clapping. For ten minutes the lamas stand up, run along the passages outside
the lecture-hall, or take stock of me while I am sketching the schoolroom and the
pupils. Often a handful of rice rains down upon the youths—some pilgrim is passing
by the window opening. At these readings and at the high mass the monks who
have been longest in the monastery occupy the front seats, and the last-comers the
back seats. When the lecture is over the Kanpo-Lama counts the receipts that have
flowed in from the pockets of the pilgrims, wraps the coins in paper, which is sealed
up and conveyed to the treasury, and enters the amount in a large account book.

The images on the altar table of the Kanjur-lhakang are small, and composed of
gilded metal, and most of the other idols in Tashi-lunpo are of the same kind. Some
are of carved wood, and a few, like the great statue of Tsong Kapa, are composed
of powdered spices cemented together by a gum extracted from roots of plants.
The statue of Tsong Kapa is said to have been constructed seventy-two years ago,
and to have cost as much as one of gold. The Tashi Lama has 1500 small gods cast
for the New Year festival, each costing 7 rupees; they are manufactured in Tashi-
lunpo, and are given away or sold. The manufacture of these images is regarded as
a peculiarly blessed work, and the lamas engaged in it may count with certainty on
a long life. Especially is this the case with those who make images of the
Tsepagmed. The oftener they utter his name and produce his likeness from the
rough metal, the longer it will be before their poor souls have to set out on their
travels again. No idol, however, possesses any miraculous power or the slightest
shadow of divine influence unless it is properly consecrated and blessed by an
incarnated lama.

I must by this time have tried the patience of my readers with my personal
recollections of the monastery of Tashi-lunpo. I have unintentionally tarried too long
with the fraternity of the yellow-caps, and quite forgotten events awaiting our
attention elsewhere. I might have remembered that temples and monks’ cells may
not have the same interest for others as they have for myself, but the remembrance
of this period is particularly dear to me, for I was treated with greater friendliness
and hospitality in Tashi-lunpo than in any town of Central Asia. We came from the
wastes of Tibet to the greatest festival of the year, from solitude into the religious
metropolis swarming with thousands of pilgrims, from poverty and want to
abundance of everything we wanted, and the howling of wolves and storms gave
place to hymns and fanfares from temple roofs glittering with gold. The balls in
Simla and the desolate mountains of Tibet were strange contrasts, but still greater
the solitude of the mountain wilderness and the holy town, which we entered in the
garb of far-travelled pilgrims, and where we were hospitably invited to look about
us and take part in all that was going on.

It is now time to say farewell to Tashi-lunpo, its mystic gloom and its far-
sounding trumpet blasts. I do so with the feeling that I have given a very imperfect
and fragmentary description of it. It was not part of my plan to thoroughly
investigate the cloister town, but on the contrary it was my desire to return as early
as possible to the parts of Tibet where I might expect to make great geographical
discoveries. Circumstances, however, which I shall hereafter refer to in a few words,
compelled us to postpone our departure from day to day. As we were always
looking forward to making a start, our visits to the monastery were curtailed.
Moreover, I wished, if possible, to avoid exciting suspicion. Tashi-lunpo had on two
occasions, more than 100 years ago indeed, been pillaged by Gurkhas from Nepal.
The English had quite recently made a military expedition to Lhasa. Many monks
disapproved of my daily visits, and regarded it as unseemly that a European, of
whose exact intentions nothing was known, should go about freely, sketch the
gods, see all the treasures of gold and precious stones, and make an inventory. And
it was known that the dominant race in Tibet, the Chinese, were displeased at my
coming hither, and that I had really no right to sojourn in the forbidden land. If,
then, I wished to accomplish more, I must exercise the greatest caution in all my
proceedings.

A few words on funeral customs before we take leave of Tashi-lunpo.

South-west of Tashi-lunpo lies a small village, Gompa-sarpa or the New

Monastery, where, according to tradition, a temple formerly stood which was
plundered by the Dzungarians. Here is now the cemetery of Shigatse and of the
monastery, the Golgotha where the bodies of monks and laymen are abandoned to
corruption in the same fashion.
 When the soul of a lama grows weary of the earthly frame in which it has spent
its human life, and the lama himself, after living perhaps fifty years in his dark
cloister cell, perceives that the lamp of life is going out for want of oil, some
brethren gather round his sick-bed, recite prayers, or intercede with the gods set up
in his cell, whose prototypes in Nirvana or in the kingdom of the dead have
something to do with death and the transmigration of souls. As soon as life is
extinct, special prayers for the dead are recited to facilitate the severance of the
soul from the body, and console it during its first steps on the dark road beyond the
bounds of this life. The corpse of a lama lies in his cell for three days, that of a
layman as long as five days, that there may be sufficient time for all the funeral
rites and services. Rich people retain the corpse longer in the house, which is
certainly more expensive, but allows more time for prayers which will benefit the
deceased. Monks fix the date of interment and the moment when the soul is
actually freed from its earthly fetters and soars up in search of a new habitation.

The dead lama in a new costume of the ordinary cut and style is wrapped in a
piece of cloth and is carried away by one or two of his colleagues; a layman is
borne on a bier by the corpse-bearers. These are called Lagbas, and form a
despised caste of fifty persons, who live apart in fifteen small miserable cabins in
the village Gompa-sarpa. They are allowed to marry only within the guild of corpse-
bearers, and their children may not engage in any other occupation but that of their
fathers, so that the calling is hereditary. They are obliged to live in wretched huts
without doors or windows; the ventilators and doorways are open to all the winds
of heaven and all kinds of weather. Even if they do their work well they are not
allowed to build more comfortable houses. It is their duty also to remove dead dogs
and carcasses from Tashi-lunpo, but they may not enter within the wall round the
convent. If they have any uneasiness about their souls’ welfare, they pay a lama to
pray for them. When they die, their souls pass into the bodies of animals or wicked
men. But in consequence of the afflictions they have endured they are spared too
hard a lot in the endless succession of transmigrations.

The Lagbas have only to hack in pieces lamas, their own relations, and the bodies
of the homeless poor. Well-to-do laymen have this operation performed for their
own people without calling in professional aid.

When the monks come with a dead brother to the place of dissection they strip
him completely, divide his clothes among them, and have no compunction in
wearing them the very next day. The Lagbas receive 2 to 5 tengas (11d. to 2s. 3d.)
for each body and a part of the old clothing of a lama; in the case of a layman the
Lagba receives all the raiment of the deceased, and the ear-rings and other simpler
ornaments of a woman. The monks who have brought the body hurry off again
with all speed, partly because the smell is very bad and partly that they may not
witness the cutting up of the corpse, at which only the Lagbas need be present;
even when the body is that of a layman, it is divided only in the presence of
Lagbas.

147. Portrait of the Tashi Lama.

(before retouching.)

A cord fastened to a post driven into the ground is passed round the neck of the
corpse, and the legs are pulled as straight as possible—a feat requiring great
exertion in the case of a lama, who has died and become rigid in a sitting posture.
Then the body is skinned, so that all the flesh is exposed; the Lagbas utter a call,
and vultures which roost around come sailing up in heavy flight, pounce down on
the prey, and tear and pluck at it till the ribs are laid bare. There are no dogs here
as in Lhasa, and even if there were, they would get no share in the feast, for the
vultures do their work quickly and thoroughly. We afterwards visited convents
where sacred dogs were fed with the flesh of priests. The Lagba sits by while the
vultures feed, and these are so tame that they hop unconcernedly over the man’s
legs.

The head is usually cut off as soon as the body is skinned. The skeleton is
crushed to powder between stones, and is kneaded with the brains into a paste,
which is thrown to the birds in small lumps. They will not touch the bone-dust
unless it is mixed with brains. The guild of corpse-cutters pursue their task with the
greatest composure: they take out the brains with their hands, knead it into
powder, and pause in the midst of their gruesome employment to drink tea and eat
tsamba. I am exceedingly doubtful if they ever wash themselves. An old Lagba,
whom I summoned to my tent to supplement the information I had received from
the monks, had on that very morning cut up the body of an old lama. Muhamed Isa
held his cap before his face all through the conversation, and had at last to go out,
for he began to feel ill. The man had an unpleasant rough aspect, wore a small
grey soft cap, and was dressed in rags of the coarsest sacking. He had his own
theories of post-mortem examination and anatomy. He told me that when an
effusion of blood was found in the brain it was a sign that the man had been
insane, and that when the substance of the brain was yellow the man had been an
habitual snuff-taker.

In some cases, so a monk assured me, the corpse is not skinned, but the head is
cut off, the trunk is divided in two along the spine with a sharp knife, and each half
is cut into small pieces, and the vultures are not called till this has been done. Small
children and grown-up men are cut up in the same manner. There is not the least
respect shown for the nakedness of dead women. The whole aim of this method of
disposing of the body is that the deceased may have the merit of giving his body to
the birds, which would otherwise be famished. Thus even after his death he
performs a pious deed which will promote the peace of his soul. The vultures here
act the same part as in the Towers of Silence among the Parsees of Bombay and
Persia.

As soon as the demands of religion are fulfilled, the relatives take leave of the
deceased. He is then gone away, and his body is quite worthless; when the soul has
recommenced its wanderings, the body may be consigned to the brutal treatment
of the Lagbas without the least hesitation. No one follows the corpse to the home
of the vultures when it is carried out of the house at night to be cut up before the
sun rises. There is no legal regulation, and when the bodies are numerous, the sun
has generally risen before the work is finished. After that, one, or at most two, of
the corpses are left till evening and are taken in hand after sunset. This is also
because the vultures are satiated with their morning’s feed and must have a rest
before supper. It is seldom that more than two deaths are reported in Gompo-sarpa
in one day. About twelve years ago when an epidemic of smallpox raged in
Shigatse, forty to fifty bodies were removed daily. Then, after the vultures had
gorged themselves, the rest of the bodies were wrapped in thin shrouds and buried.

One would suppose that the dying man would shudder at the thought that, at the
very moment when the gates of death were opened for him, his body, with which
he was so closely connected during his life, which he had cared for so anxiously,
endeavouring to shield it from danger and sickness, nay, from the slightest pain,
would be consigned to such barbarous treatment. But probably he thinks more of
his soul in his last moments, and counts up the good deeds he has performed and
the millions of manis he has recited.

There is, then, not the slightest touch of sentiment in the funeral customs of the
Tibetans and their attitude towards the dead. The children of Islam visit the graves
of their loved ones and weep out their sorrow under the cypresses, but the
Tibetans have no graves and no green-covered mounds where they may devote an
hour to the remembrance of a lost happiness. They weep not, for they mourn not,
and they mourn not, because they have loved not. How can they love a wife whom
they possess in common with others, so that there is no room for the idea of
faithfulness in marriage? The family ties are too loose and uncertain, and the
brother does not follow his brother, the man his wife, and much less his child, to
the grave, for he does not even know if the child is really his own. And, besides, the
corpse in itself is a worthless husk, and even a mother who has tenderly loved her
child feels not a shadow of reverence for its dead body, and has no more horror of
the knife of the corpse executioner than we have of the doctor.
 CHAPTER XXX
OUR LIFE IN SHIGATSE

The time that was not taken up by visits to Tashi-lunpo I occupied in many ways.
We had friends to visit us, and I frequently spent many hours in transferring types
of the people to my sketch-book, and I found good material among the citizens and
vagrants of the town and the monks of the convent.

148. Lamas with Copper Tea-Pots. 149. Female Pilgrim from Nam-tso and Mendicant Lama.
Sketches by the Author.

On one of the first days the Consul of Nepal paid me a visit. He was a lieutenant,
twenty-four years old, was named Nara Bahadur Chetteri, and bore between his
eyes the yellow marks of his caste. He was dressed in a black close-fitting uniform
with bright metal buttons, and a round forage cap on his head without a shade, but
with a gold tassel, and in front the sun of Nepal surrounded by a halo of rays. He
had been four months in Lhasa and two here. He and his young wife had taken two
months to travel from Khatmandu; they had ridden the first week, but had then
sent their horses back, and had tramped through very dangerous, pathless,
mountainous regions for fifteen days; the rest of the journey they had
accomplished on hired Tibetan horses. Here he had to protect the interests of the
150 Nepalese merchants and assist the pilgrims of his country when they were in
difficulties. The merchants have their own serai, called Pere-pala, for which they
pay an annual rent of 500 tengas; they buy wool from the nomads in the north, and
pay for it with corn and flour, which therefore is scarce and dear in Shigatse,
especially during the festival time when so many pilgrims flock in. The Consul
received 200 tengas a month, or rather less than £60 a year, and considered that
the Maharaja paid him very badly. Bhotan has no consul in Shigatse, though many
pilgrims come from that country.

On February 14 I received a very unexpected visit, a lama and an official from

Lhasa. When the Devashung, the Government, had received the letter of Hlaje
Tsering announcing my arrival at the Ngangtse-tso, the Chinese Ambassador and
the Government, after consulting together, had despatched these two gentlemen in
forced marches to the lake, where, however, they arrived several days after my
departure. Singularly enough they had been given quite erroneous information
about the route we had taken, perhaps because our wanderings over the ice across
the lake in all directions had confused the nomads. Therefore they had sought for
us for twenty-two days on the shores of the Ngangtse-tso and the Dangra-yum-tso,
until they had at length discovered that we had gone off southwards a long time
before. Then they had followed our track and had made further inquiries among the
nomads, all of whom said that they had been kindly treated, and well paid for all
they sold us. The gentlemen rode on, and heard in Yeshung that we had passed
through a couple of days before; our camp-fires were scarcely cold. They changed
horses, and spurred them on at a faster pace, for they had been ordered to force
us at any cost to return northwards by the same way we had come. But I had got
the better of them, for they did not reach Shigatse till thirty-six hours after us, and
another party sent from Lhasa to intercept us by a more direct road had quite lost
our trail in the labyrinth of valleys and mountains into which we had plunged.

“We have carried out our mission as well as we have been able,” they said, “and
it only remains for us to ask for your name and all particulars of your journey and
companions.”
 “I have already communicated everything to Ma Daloi and Duan Suen, who have
seen my passport, but if you want a second edition, you are welcome to it.”

“Yes, it is our duty to send a report to the Devashung. In virtue of the treaty of
Lhasa only the market-towns of Yatung, Gyangtse, and Gartok are free to the
Sahibs under certain conditions, but no other routes. You have come by forbidden
roads and must turn back again.”

“Why did you not close the way to me? It is your own fault. You can inform the
Devashung that I shall never be content till I have seen the whole of Tibet. Besides,
the Devashung will not find it worth their while to place obstacles in my way, for I
am on good terms with your gods, and you have seen yourself how friendly the
Tashi Lama has been to me.”

“We know it, and it seems as though you bore the sign of the favour of the gods
on your forehead like caste-markings.”

“How is Hlaje Tsering getting on?”

“He is suspected of receiving a bribe from you; he has been dismissed, and has
lost his rank and all his property.”

“It is very mean of the Devashung to persecute him. But the Government is
composed of the most despicable rogues in all Tibet. You ought to be glad that you
are at length properly under Chinese protection.”

At first they exchanged meaning looks, but gradually they came round to my
opinion and admitted that their Government was a disagreeable association. The
reason they had not shown themselves immediately after their arrival was that they
wished first to spy out our occupations and our associates; for, if they found out
that we had friends, these would of course be denounced. Otherwise they were
decent men, and readily partook of tea and cigarettes. Unfortunately Tsaktserkan
was just then with me, and he must have thought the affair serious, for he made
himself scarce as soon as they entered my tent, but afterwards asked me to tell him
what they had said.

It impressed them most of all that, in spite of all the ambushes and traps in the
form of scouting patrols, who were on the look-out for us, we had after all
succeeded in advancing to Shigatse. Now they would wait for orders from Lhasa.
No heed was paid to the Dalai Lama, who was as good as dead and buried.
 150. The Great Red Gallery of Tashi-lunpo.
Sketch by the Author.

They came frequently during the following days to greet us, and then expressed
their opinions of their superiors more and more frankly. Their remaining on the
scene proved, however, that both the Chinese and the Government had their eyes
on me. I wondered how the affair would terminate.

When I returned from the equestrian performance on the 15th, I found a large
packet of letters from Major O’Connor, and greedily seized letters from home and
from friends in India, Lady Minto, Colonel Dunlop Smith, Younghusband, and
O’Connor himself, who welcomed me most heartily, and expressed a hope that we
should soon meet. He had also kindly given me a great surprise with two boxes
containing preserved meats, cakes, biscuits, whisky, and four bottles of champagne.
Fancy my drinking champagne alone in my tent in Tibet! I drank a glass at dinner
every day to the health of Major O’Connor as long as the supply lasted.
 In the chapter on Leh I mentioned the Hajji Nazer Shah and his son Gulam Razul.
The old Hajji had another son in Shigatse, named Gulam Kadir, who had been ten
years in Tibet and now managed the branch in Shigatse. He sold chiefly gold-
embroidered stuffs from China and Benares, which the lamas bought for state
robes, and he told me that he made a yearly profit of 6000 rupees. A bale of such
material as he showed me was worth 10,000 rupees. Gulam Kadir rendered me
many services at this time, and supplied us with anything we wanted.

There is a fine view from the roof of his house of the Dzong, or fort, the stately
front of which seems to grow out of the rock. The windows, balconies, roof
decorations and streamers have a harmonious and picturesque effect. In the middle
of the structure is a red building; all the rest is white, or rather an undecided
greyish-yellow colour, which the plaster has assumed in the course of time.

At the southern foot of the Dzong hill lies the open market-place, where trade
was carried on two hours a day. There are no tables and stands, but the dealer sits
on the dusty ground and spreads out his wares on cloths or keeps them beside him
in baskets. In one row sit the dealers in implements and utensils, in others boards
and planks are sold, ironware, woven goods, coral, glass beads, shells, sewing
thread, needles, dyes, cheap oleographs, spices and sugar from India, porcelain,
pipes, figs and tea from China, mandarins from Sikkim, dried fruits and turquoise
from Ladak, yak hides and tails from the Chang-tang, pots, metal dishes, covers
and saucers manufactured in the town, religious books and other articles for the
use of pilgrims, etc. Straw and chaff, rice, grain, tsamba and salt are sold by many
traders. Walnuts, raisins, sweets, and radishes are other wares in which there is a
large trade. Horses, cows, asses, pigs, and sheep are also on sale; for the last, 7
rupees a head are asked. In Chang-tang we had paid at most 4 rupees, and a
sheep can be got for 2 rupees. Every kind of ware has its particular place, but the
traders, so far as I could see, were all Tibetans; for the merchants of Ladak, China,
and Nepal have shops in their own houses.
 151. Chhorten in Tashi-lunpo.
Sketch by the Author.

Most of the traders are women, and they sell even hay, firewood, and meat. They
wear huge coils of hair with inferior turquoise, glass beads, and all kinds of
pendants, which contrast strongly with their faces smeared with black salve. If they
would dispense with this finery and give themselves a good washing instead, some
of them would perhaps look quite human. What was the original colour of their
clothes is hard to guess, for they are now caked with dust, soot, and dirt. But these
hucksters are always polite and obliging; they sit in rows parallel to the north wall
of the Chinese town, which is little more than a ruin. Now and then a mule caravan
passes along the pathway between the rows, bringing new goods to market.
Frequently gentlemen partly dressed in Chinese fashion ride past from the Dzong;
and among the swarms of customers are seen all kinds of people—clergy and
pilgrims, children of the country and strangers, white turbans from Ladak and
Kashmir, and black skull-caps from China. In the market all the gossip of Shigatse is
hatched; all sorts of reports more or less probable reach us from there. As soon as
any one comes from Lhasa he is driven almost frantic with questions, for all take a
deep interest in the new Chinese régime. It was current in the bazaar that lamas in
Lhasa were organizing a bloody insurrection against the Chinese, because the latter
had demanded that half the lamas should serve in the army. It was further reported
that I and my companions would soon be compelled to leave the country, and that
before very long the English commercial agency in Gyangtse would be closed. Every
one who has heard anything fresh carries it at once to the market, where the
visitors who come to hear news are as numerous as those who make purchases. In
a word the market is Shigatse’s only newspaper.

Gulam Kadir told me that the two gentlemen from Lhasa employed spies, who
reported daily all that they could find out about us. These men used to come as
hawkers into our tents and sit there by the hour. Ma also encompassed us with
spies. With the help of Gulam Kadir I set two Ladakis as spies to spy upon the spies
of the Lhasa spies. We could now be on our guard, for we knew what was going on
around us.

My own Ladakis enjoyed in Shigatse a very necessary period of rest. I gave them
money for new clothes, which they made up themselves; in a few days they
appeared in all the glory of a new outfit from head to foot. Nor could I refuse them
a jug of chang daily; they very seldom drank too much, after one of them one day
under the influence of beer painted his face black, and in this guise made ridiculous
pirouettes about the court. Muhamed Isa happened to come home from the market
just at the moment, and, catching hold of the dancer, gave him such a thorough
drubbing that he never thought of painting himself again. Both Chinamen and
Tibetans said that the conduct of my men was exemplary and gave no cause for
quarrels. But to hear Tsering’s singing in the evening! It was like the creaking of a
badly oiled wicket-gate to a shed in my own country, and therefore I listened with
pleasure to his rude song. When he had sung for three hours on end, it became a
little too much, but I put up with it—it is so pleasant to have cheerful, contented
men about one.

Under February 19 the following entry stands in my diary: “In spite of the windy,
dusty weather I have all day long been sketching various types, chiefly women,
who sat for me as models in front of my tent.” The first were from Nam-tso (Tengri-
nor) (Illustration 155), wore head-dresses decorated with shells, china beads, and
silver spangles, and in their sheepskins trimmed with red and blue ribands looked
like girls from Dalecarlia. They had large bones, were strongly built, looked fresh
and healthy, and their broad faces were remarkably clean. The women of Shigatse,
on the other hand, had smeared their faces with a brown salve mixed with soot
which looked like tar. This mask makes them hideous, and it is impossible to tell
whether they are pretty or not; the black colour interferes with the lights and
shadows, and confuses the portrait painter. One had painted only her nose and
rubbed it bright as metal. This singular custom is said to date from a time when the
morality of the Lhasa monks was at a low level, and a Dalai Lama issued orders that
no female should show herself out of doors unless painted black, so that the
charms of the women might be less seductive to the men. Since then the black
paint has remained in fashion, but seems now to be going out.

The clothes are always black with age, dirt, and soot. The women pay most
attention to their head decoration, and the higher they are in the social scale the
more profusely they deck their coiffures with bows, pendants, and jewelry. The hair
is frequently so closely entwined with all this finery that it can scarcely be let down
every night, but only when it becomes so entangled that it must be put straight.
Those who are rich wear large heavy ear-rings of solid gold and a few turquoises,
but others simpler and smaller rings. On the neck are worn chains of various
coloured beads and gaos, small silver cases studded with coral and turquoise and
containing amulets. Poor women have to be contented with copper gaos of the
clumsy kind so common among the Tsaidam Mongols.

A woman of forty belonging to Shigatse was named Tashi-Buti; she looked sixty,
for women age very soon here. Above her ordinary clothing she wore a coarse
shawl over the shoulders, fastened in front with brass clasps, plates, and rings.
152. Portal in Tashi-lunpo.
 153. Group of Lamas in Tashi-lunpo.

A nomad woman from Kamba had the right arm and shoulder bare, and was as
powerfully and muscularly built as a man, but was so horribly dirty that it was
impossible to perceive her complexion. She had no head-dress; but the dark hair
was plaited into innumerable thin rat’s tails hanging over the shoulders, and tied
together on the forehead into a mane of cords. She would have been good-looking
if her features had not been so masculine; she sat still and solemn as a statue of
Buddha. A fifteen-year-old girl had a parting in the middle, and her hair frizzed in
two pads down to the ears, which were combed, oiled, and shiny like those of a
Japanese, and she wore a diadem studded with coral. She was dainty and clean
and had rosy cheeks (Illustrations 157, 158, 159).

Burtso was a little Shigatse lady of seventeen summers, and bore the dirt of
those seventeen summers on her face. Like most of the others her features had the
sharply marked characteristics of the Mongolian race—oblique narrow eyes
contracting to a point at the sides, and the lower part of the eyelid telescoped into
the upper so that a slightly curved line is formed and the short lashes are almost
covered; the iris is dark chestnut brown, and appears black within the frame of the
eyelids; the eyebrows are usually only slightly marked, are thin and irregular, and
never form the finely curved Persian and Caucasian arch like a crescent. The cheek-
bones are rather prominent, but not so high as with the Mongolians; the lips are
rather large and thick, but the nose is not so flat as among the Mongols. Faces with
handsome features are seen among the male Tibetans. But the differences between
individual Tibetans are often as great as between Tibetans on the one hand and
Mongols, Chinamen, and Gurkhas on the other. The nomads of the Chang-tang are
apparently a tribe of themselves, and seldom, if ever, intermarry with the others.
Otherwise the Tibetan people is undoubtedly much mixed with neighbouring
elements. Chinamen living in Lhasa and Shigatse marry Tibetan women. In the
Himalayas, south of the Tibetan frontier, live the Bothias, a mixed people, sprung
partly from Indian, partly from Tibetan elements. The people of Ladak have mingled
to a large extent with their Aryan and Turkish neighbours, because they have been
in closer and more active contact with them. The Tibetan people present
remarkable and peculiar problems in anthropological, ethnographical, and linguistic
science, which must be solved by future investigation.

I drew on and on, and one type after another found its way into my sketch-book.
The expression of my models is listless and devoid of animation; they seem absent-
minded and passionless. They take little interest in the proceedings; all they care
about is to pocket the rupees after the sitting. They sit motionless, without laughing
or complaining. They are rather too solemn, and not a smile plays round the
corners of their mouths when their eyes meet mine. I passed the greater part of
the day in this silent, apathetic female society.

Now and then comes a party of inquisitive people to watch me, Tibetans,
Chinamen, or pilgrims who want to have something to tell when they get home
again to their black tents. They stand round me, wondering whether it is dangerous
to be drawn by a European and what is the object of it. Of course there are many
spies among them. There is an endless variety of types and costumes, and as I ride
through the streets and see the inhabitants at their various occupations, I feel
oppressed by the thought that I have not time to draw them all. Here stands a man
splitting wood, there come two young fellows driving before them asses laden with
twigs and branches. There go a couple of women with large water-jugs on their
backs, while small girls collect cattle dung from the street. Here a group of officials
approaches in yellow garments on fine horses, while some lamas stroll slowly
towards the monastery. All is so picturesque, so charming for the pencil; one is
constantly delighted with attractive subjects, genre pictures of unusual character,
strikingly grouped parties of salesmen and customers; one could spend months
here, drawing again and again. I am grieved at the prospect of an early departure.
 154. Lecture in Tashi-lunpo.
Sketch by the Author.

In the afternoon a company of dancers, male and female, frequently appears in

the court and gives no despicable performances, reminding me strongly of the
dances in Leh. They are always introduced by our little old mother Mamu, who has
the management of the garden, and hops about smiling and friendly as a sparrow.
She speaks Urdu, so Robert employs her as interpreter. Then come caravans,
bringing hay, firewood, chaff, and barley for our remaining animals, or provisions for
ourselves, and people are constantly coming to sell all kinds of goods—chickens,
eggs, butter, or fish from the Tsangpo; milkmen run with their clattering metal cans,
and stringed instruments and flutes make music in our groves. A beggar comes up
like a troubadour to my tent with a lute, and sings a melodious air. When I look at
him he stops singing and puts out his tongue. Barefooted boys, who could be no
blacker if they were drawn twice up a chimney, run about laughing loudly, and peep
out from among the trees. Three of them perform on a tight rope, dance like
professional rope-dancers, and beat drums, while they turn summersaults all mixed
up together (Illustration 164).

Pious visitors also frequent my courtyard: two nuns, for instance, with a large
tanka representing a series of complicated episodes from the holy scriptures. While
one chants the explanation, the other points with a stick to the corresponding
picture (Illustration 165). She sings so sweetly and with so much feeling that it is a
pleasure to listen to her. Or a mendicant lama comes with his praying mill in his
hand and two hand-grooves hung by a strap round his neck. In these he pushes his
hands as in a curry-comb, when he prostrates himself on the ground in making a
circuit of the temple. They are much worn, and this moves the hearts of the people
to generosity, so that his alms bowl is filled daily.

These pious men are the parasites of Tibet, living at the expense of the working
population. And yet they are endured and treated by every one with the greatest
consideration and respect. To give them a mite brings a blessing on the giver. The
people are kept by the lamas in spiritual slavery, and the lamas themselves are
docile slaves to those tomes of narrow-minded dogmas which have been
stereotyped for centuries, which may not be interfered with or criticised, for they
are canonical, proclaim the absolute truth, and stand in the way of all free and
independent thought. The clergy form a very considerable percentage of the scanty
population of this poor country. Without the Peter’s pence Tibet could not make
both ends meet. Tashi-lunpo is, then, a huge savings-box, in which the rich man
places his pile of gold, the poor man his mite. And with what object? To propitiate
the monks, for they are the mediators between the gods and the people. Scarcely
any other land is so completely under the thumb of the priests as Tibet. And while
the people toil, the monks gather round their tea-pots and bowls of tsamba at the
summons of the conch.

On three evenings in succession large numbers of wild-geese have flown low over
our garden from north-west to south-east. The ravens are as bold as usual; of other
birds only sparrows roost in our trees. Our camp within the wall is quiet, but we
have posted a night-watch outside, for in a town like Shigatse, full of all sorts of
vagabonds, there are many scoundrels. Two monks, who were with me one evening
to answer my inquiries, durst not return to Tashi-lunpo in the dark, unless I sent
some of my men armed with guns to take them home. Recently a lama was
attacked at night between the town and the monastery and stripped to the skin.
 On February 20, after only 17.6 degrees of frost, it snowed all day long, the wind
howled dismally through the poplars, and the snow fell on my tent. Nothing was to
be seen of the golden temple roofs, and the ground and the mountains were white;
there was no one in the bazaar, and no inquisitive visitors pestered us. It was just
as in the Chang-tang.

155. Female Pilgrims from the Nam-tso.

 156. Tibetans in Shigatse.
Sketches by the Author.

On March 4 Gulam Kadir paid me a farewell visit, for he was going next day to
Lhasa, which, according to his reckoning, was nine days’ journey distant. As he
would pass through Gyangtse, he took a large letter-bag to Major O’Connor. On the
day before, he had sent off a caravan of 201 yaks laden with brick tea to Ladak. A
yak carries 24 bricks, and a brick costs in Shigatse 6 rupees, but in Ladak 9 to 11. It
is only the refuse of the tea, which is despised in China, but is good enough for
Tibetans and Ladakis. Gulam Kadir hires the yaks at a cost of 5 rupees a head to
Gartok—uncommonly cheap, but they follow the mountain paths and their keep
costs nothing. They are five months on the way, for the caravan makes short
marches and stays at places where grass grows luxuriantly. From Gartok, where the
Hajji Nazer Shah has a large warehouse, managed by Gulam Razul, the tea is
transported on other yaks. By a single caravan of this kind the commercial house of
the Hajji makes a very large profit. Musk, coral, Chinese textiles, and other valuable
goods are forwarded on mules along the great highway which runs along the
Tsangpo and the upper Indus.

I had on several occasions met Kung Gushuk, the Duke, in the monastery, and
had thanked him for his kindness in sending my letters to the lakes, but it was not
till March 7 that I paid him a visit in his house. The walls in the entrance hall are
painted with tigers and leopards. In the court, round which the stables and
servants’ quarters are situated, a large black watch-dog, with red eyes and a red
swollen ring round his neck, is chained up, and is so savage that he has to be held
while we pass. After mounting two ladder-like staircases we come to the reception-
room, which is very elegant, and has square red pillars with carved capitals in green
and blue. Along the walls stands a row of shrines of gilded wood with burning
butter-lamps in front of them, and over them hang photographs of the Tashi Lama
which were taken in Calcutta. The rest of the walls are draped with holy banners
(Illustration 167).

The trellised window pasted over with paper, which occupies nearly the whole
length of the wall towards the courtyard, and is draped with white curtains on the
outside, is placed rather high above the floor. Immediately below the window runs a
long divan mattress, on which a square cushion covered with panther skin marks
the seat of honour. Before this cushion stand two small stool-like lacquered tables
on golden feet. Seated here one has on the left hand, against the shorter wall, a
cubical throne with steps leading up to it, and here the Tashi Lama takes his seat
when he visits his younger brother, now twenty-one years of age.

Kung Gushuk is, then, quite young. He is very shy, and is evidently relieved when
his guest talks and he is not obliged to strain his own small, poorly furnished brain.
His recollections of India, whither he had accompanied his illustrious brother, were
very hazy: he did know that Calcutta is a large town, and that the weather was
excessively hot there, but for the rest the journey seemed to be to him only an
unintelligible dream. He did not venture to give an opinion on the journey before
me, but said openly that the lamas did not like to see me so often in Tashi-lunpo.
His wife had sent to ask me if I would take her portrait, and I now begged to be
told what time would suit her. “Any time.” When I went away, Her Highness was
standing with her black court ladies at the other end of the open gallery
surrounding a court (Illust. 168). I saluted her politely, and certainly fascinated the
lady as I passed; there was no danger, as she was quite passée, for she had
belonged in common to Kung Gushuk and an elder brother, who died in Sikkim on
the return from India. It is said that she rules the house and keeps the finances in
order, and with good reason, for Kung Gushuk leads a fast life, is over head and
ears in debt, and plays hazard. This is bad form in a brother of the Tashi Lama.
 157, 158, 159. Tibetan Girl and Women in Shigatse.
Sketches by the Author.

On March 22 the portrait-drawing came off; it was executed in the large saloon
and in pencil. The Duchess is big and bloated, and asserted that she was thirty-
three years old—I should put her down at forty-five. Her complexion is fair and
muddy, the white of her eyes is dull. She had put on for this occasion all the finery
she could find room for; a pearl pendant which hung on the left side of her façade
had cost 1200 rupees. In her hair were thick strings of pearls, bunches of coral and
turquoises. She was friendly and amiable, and said that she did not mind how long
she sat, if only the result were good. Her small carpet-knight of a husband sat by
and looked on, and round us stood the other inmates of the house, including a
small brother of Kung Gushuk and the Tashi Lama. They drank butter tea, but did
not offer me any, which made the visit all the pleasanter (Illustration 170).

Then we were shown the other apartments, which even on sunny days are dark
as dungeons, for the windows are small, the paper thick, and the white curtains
outside help to increase the gloom. A small oratory with red pillars was so dark that
the images of the gods could scarcely be distinguished. In the study of the Duke a
low divan stood at the window, with paper, inkstand, pens, and a religious book on
a table in front of it. The bedroom was adorned with tankas, statues, and cups.
Here and there butter-lamps struggled with the darkness, while braziers of brass on
stands of dark carved wood were used to counteract the chilliness of the air. The
whole house is like a temple, which is quite as it should be when the owner is
brother of the Grand Lama.

Two passages connecting parts of the upper storey are not covered in, so are
exposed to all the winds of heaven. A third staircase leads to the top of the roof,
which is surrounded by a parapet a yard high, and is white-washed. A thicket of
roof decorations and bundles of rods with streamers frightens away evil spirits.
There was a violent wind, and dust and bits from the streets of Shigatse flew up in
the air, so that our eyes received their share. With the portrait-drawing the visit
lasted four good hours, and at the end I had become as intimate with the family as
if I had known them from childhood.

CHAPTER XXXI
POLITICAL COMPLICATIONS

In the first chapters of this book I described very briefly the difficulties placed in
my way by the English, and told how the Liberal Government in London had not
only refused the favours I had asked for, but had even tried to suppress my
expedition altogether. In consequence I had been compelled to make a wide detour
all through the Chang-tang, where more than once our lives hung by a thread, and
we had suffered great losses. Then we met with a weak resistance on the part of
the Tibetans, but, nevertheless, came to Shigatse; it was pure good luck that the
patrols sent out to intercept us had not fallen in with us. On February 14 the
representatives of the Tibetan Government had intimated to me that I had no right
to make a prolonged sojourn in Tibet, and that I must leave the country. As though
I had not enough to do with the English, Indian, and Tibetan Governments, the
Chinese Government also appeared on the scene on February 18. I was now
opposed to a fourfold combination of Governments, and wished all politics and
diplomatists at Jericho.

On this day the young Chinaman Duan Suen appeared on behalf of Gaw Daloi,
the Chinese political agent in Gyangtse. He brought me a letter from him with the
following curt contents:
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebookball.com