Critical Care

accurately predicted by the queuing model. Overall, the

Queuing Theory Accurately Models the mean difference between observed and predicted values was
Need for Critical Care Resources 0.4% with a maximum difference of 13% and a minimum of
0. For the entire 2-year period, the observed overall turn-
Michael L. McManus,* Michael C. Long,y Abbot Cooper,z
away rate was 21%, and that predicted by the model was
and Eugene Litvakz
22%. Turn-away rates increased exponentially when utiliza-
(Anesthesiology, 101:12711276, 2004) tion exceeded 80% to 85%, in practice and in simulation.
The observed rejection rate was best viewed as an expo-
*Department of Anesthesia, Pain and Perioperative Medicine, Childrens nential function of utilization, behavior that is consistent
Hospital; yDepartment of Anesthesia, Massachusetts General Hospital; with predictions from queuing theory. An average admis-
and zHealth Policy Institute, Boston University, Boston, MA. sion rate of 5.7 patients per day and a 3.5-day average dura-
Copyright n 2005 by Lippincott Williams & Wilkins
DOI: 10.1097/01.sa.0000158562.87099.59
tion of stay yielded a predicted utilization rate of 86%.
Sensitivity analysis using the model showed rapid and severe
degradation of system performance with even the small
I ntensive care units (ICUs) are among the most complex and
expensive of all medical resources, and administrators are
challenged to meet the demand for intensive care services
changes in bed availability that might result from sudden
staffing shortages or admission of patients with very long
with an appropriate capacity. Queuing theory is used in ICU stays.
engineering and industry for analysis and modeling of The queuing theory may be used to accurately model
processes that involve waiting lines. It allows calculation of ICU bed utilization in a large unit operating at or near
the optimal supply of fixed resources necessary to meet a capacity. The nature of patient flow may falsely lead health
variable demand. Attempts have been made to apply queu- planners to underestimate resource needs in busy ICUs.
ing analysis to various hospital activities as a means of Although the nature of arrivals for intensive care warrants
directing the allocation of scarce resources. However, most further study, when demand is random, queuing theory
proposed queuing models lack real-world validation and provides an accurate means of determining the appropriate
have not been prospectively evaluated over significant supply of beds.
periods. This study explored the utility and implications of
queuing theory as it relates to the supply and demand COMMENT
for critical care services by attempting to validate a simple The high cost of intensive care necessitates an optimal
queuing model in a busy ICU. use of the available resources, recognizing that delayed
Queuing analysis depends on accurate measurement critical care admission is unacceptable. From an administra-
of arrival rate, service time, and number of servers in the tive point of view, approaching 100% use of these resources
system. The authors prospectively collected 2 years admis- indicates a highly efficient system. However, it is obvious
sion, discharge, and turn-away data in a busy urban ICU. to everyone that 100% occupation of intensive care unit
Data were analyzed for frequency of admission requests, (ICU) beds means that for many other patients, there will
durations of stay, and crowding. With observed monthly be no bed available. This study applies queuing theory to
admission rates, available beds, and stay durations as inputs, describe supply and demand for critical care services in a
monthly utilizations and rejection probabilities were cal- large pediatric medical center. Queuing theory predicted
culated using queuing theory. A mathematical model was accurately the relation between utilization and rejection
constructed for patient flow to compare predictions from the rates of the ICU beds. When utilization increased above
model to observed performance of the ICU and to explore 80% to 85%, rates of rejection increased immediately. This
the sensitivity of the model to changes in unit size. finding supports previous data indicating that the average
During the study period, there were 3786 requests occupancy target of 85% may be optimal; however, the fact
for admission; 3680 were admitted to the hospital, and 106 that the majority of patients in this study had elective surgery
were diverted to other institutions. Monthly average admis- may affect the conclusion if the ICU is faced with more
sion request rates ranged from 4.6 to 6.2 patients per day. emergency admissions.
Individual durations of stay ranged from 1 to 190 days. The solutions for finding and allocating ICU beds
Monthly average durations of stay ranged from 2.4 to 5.5 will probably differ at different institutions. The available
days. The monthly turn-away rates varied from 3% to 47% step-down units, the postanesthesia care unit, or other units
but corresponded closely to calculated utilization and were in the hospital or with different subspecialties may be used

Survey of Anesthesiology  Volume 49, Number 2, April 2005 63

Copyr ight Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Critical Care Survey of Anesthesiology  Volume 49, Number 2, April 2005

as alternatives, because there always will be a shortage of intensive care unit, recently, several therapeutic modalities
ICU beds in the real world. have been shown to improve the outcome of critically ill
patients. In the current article, these studies are reviewed
Comment by Reuven Pizov, MD
very thoroughly.
Sepsis, especially severe sepsis and septic shock, is
associated with high mortality and contributes significantly
Evidence-Based Management of to the cost of health care. The introduction of recombi-
Critically Ill Patients: Analysis and nant activated protein C has been shown to reduce mortality,
Implementation particularly in special subgroups of patients. However, pro-
tein C is very expensive, and its more widespread use will
Michael A. Gropper depend on its cost. Other therapies that may improve the
outcome of septic patients are very simple and do not re-
(Anesth Analg, 99:566572, 2004)
quire the introduction of new drugs or significantly in-
Department of Anesthesia and Perioperative Care and Cardiovascular
creasing intensive care unit workload. The use of low-dose
Research Institute, University of California, San Francisco, CA. steroids has been shown to reduce mortality in patients with
Copyright n 2005 by Lippincott Williams & Wilkins proven functional adrenal insufficiency, which is not
DOI: 10.1097/01.sa.0000158563.25218.52 uncommon in the critically ill patient. The use of 50 mg/d
of hydrocortisone for a limited period of time is not expected

A number of important clinical trials focusing on critically

ill patients have been completed in recent years. These
trials have been among the first critical care clinical trials
to cause significant adverse effects. Most critically ill
patients develop hyperglycemia, regardless of whether they
suffer from diabetes mellitus. Several investigators have
to show mortality reduction in the critically ill. As in any now shown that tight glycemic control with continuous
adaptation of evidence-based medicine, it is important to infusion of insulin improves the outcome of critically ill
examine the trials and ascertain whether the benefits dem- patients. The implementation of tight glycemic control is
onstrated can be translated to the individual patient. In ad- easily utilized.
dition to the primary outcome, usually survival benefit, it During the last 15 years, various studies have eval-
is also important to look at cost-effectiveness. All of the uated the role of lung protective ventilation and concluded
trials examined in this review could demonstrate mortality that ventilation with low tidal volume reduces mortality.
reduction. Most focused on patients with severe sepsis, be- Finally, the use of early goal-directed therapy has been
cause this population has been associated with both frequent shown to reduce mortality in acutely septic patients. How-
mortality and increased hospital costs. Some interventions, ever, the process of implementation of critical care in the
such as a small tidal volume mechanical ventilation in pa- emergency medicine department is fraught with difficulties.
tients with acute lung injury or the administration of low- The current study excellently summarizes all of these
dose corticosteroids for patients with septic shock, are landmark investigations in critical care and is of great
cost-effective and relatively simple to implement. Others, importance, especially to physicians whose practice touches
such as use of activated protein C in patients with severe upon critical care but who do not provide critical care
sepsis or tight glycemic control in patients with hyper- routinely.
glycemia, require either major pharmaceutical expenditures
or, possibly, additional health care personnel. Notwithstand- Comment by Arieh Eden, MD
ing, the trials discussed represent significant advances in
the field of critical care medicine and should at least be
considered for implementation in all intensive care units.
Evaluation of Simple Criteria to Predict
COMMENT Successful Weaning From Mechanical
Until the last few years, there was very little evidence
to support the use of specific therapeutic modalities in the
Ventilation in Intensive Care Patients
intensive care unit. Several large-scale, prospective, random- T. S. Walsh, S. Dodds, and F. McArdle
ized trials failed to show any benefit from modulating the
immune and inflammatory response in sepsis. Although (Br J Anaesth, 92:793799, 2004)
initially promising, no consistent benefit was shown for
Anesthetics, Critical Care and Pain Medicine, New Royal Infirmary of
aggressive hemodynamic monitoring and maximizing oxy- Edinburgh, Little France, Edinburgh, Scotland.
gen delivery in several prospective studies. Contrary to Copyright n 2005 by Lippincott Williams & Wilkins
the previous paucity of evidence to guide therapy in the DOI: 10.1097/01.sa.0000158564.32842.0a

64 n 2005 Lippincott Williams & Wilkins

Copyr ight Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Survey of Anesthesiology  Volume 49, Number 2, April 2005 Critical Care

I ncreasing evidence indicates that weaning protocols can

improve outcome from mechanical ventilation, but imple-
menting such protocols in large intensive care units (ICUs)
The investigators concluded that this simple bedside
weaning checklist can reliably predict patients who achieve
ventilator independence, particularly if the day from ven-
is problematic. This prospective observational cohort study tilation on which these criteria were first met is considered.
in a 12-bed general ICU evaluated a checklist of simple These criteria are a potential method of introducing nurse-
bedside criteria to determine whether it could be used reli- led weaning protocols.
ably to predict successful discontinuation of mechanical
ventilation.
The checklist included metabolic, cardiorespiratory,
COMMENT
and neurological criteria that suggested patients could start
The authors report their experience in implementing
the weaning process. Daily assessments throughout ICU
daily screening of ventilated intensive care patients to
stay were performed, and whether the criteria were met
facilitate weaning. The screening is comprised of simple
was recorded. The reference standard was ultimate ventilator
clinical questions relating to the respiratory, neurological,
independence.
metabolic, and cardiovascular status of the patient. A pa-
The checklist included the following criteria: cooper-
tient is deemed to have passed the screening only when
ative and pain-free, good cough reflex to tracheal suctioning,
PaO2 to FIO2 ratio >24 kPa, positive end-expiratory pres- passing each and every aspect of the evaluation. The au-
thors showed that patients who passed the screening had a
sure 10 cm H2O, hemoglobin >7 g/dL, axillary tempera-
much better chance of achieving independence from me-
ture of 368C to 38.58C, plasma K+ concentration >3.0 and
chanical ventilation than those who did not. Furthermore,
<5.0 mmol/L, plasma Na+ concentration >128 and <150
the authors showed that the screening was most predictive
mmol/L, inotropes reduced or unchanged over the previ-
in patients who passed it on the first day of hospitali-
ous 24 hours, and spontaneous ventilatory frequency >6
zation in the intensive care unit (ICU). All of the 55
breaths per minute. If all fields were met simultaneously,
patients who passed the test on day 1 in the ICU were
then it was considered that acute physiological derangement
had improved sufficiently to consider starting the weaning extubated within 3 days, and most of those during the
next 24 hours after reaching weaning criteria. On the
process.
other hand, patients who reached the weaning criteria only
A total of 325 sequential admissions to the ICU were
after the third day of hospitalization had a much lower
studied. Data were available for 98% of the patients; 97%
success rate in achieving weaning from mechanical venti-
were mechanically ventilated on admission to the ICU. The
lation. Regrettably, information concerning the severity of
patients had a median age of 61 years, mean APACHE II
disease of the patients according to subgroups was not
score of 20.7, ICU mortality rate of 28%, and median ICU
provided.
length of stay of 1.7 days. Overall, 205 (67%) of the 308
ventilated patients achieved ventilator independence during I am compelled to comment that of the 125 patients
who never met the weaning criteria, 35 were successfully
ICU admission; the others died or were transferred venti-
extubated. One wonders whether such a large group of
lated to other ICUs. After these exclusions, during their
patients would not suffer from the implementation of
ICU stay, 180 (83%) admissions met the weaning criteria
weaning criteria.
and 114 admissions did not. Fulfilling the criteria was a
A point of major interest is whether such a protocol
moderately strong predictor of ultimate ventilator indepen-
can actually facilitate the weaning process and thus reduce
dence: sensitivity 84%, specificity 89%, positive and nega-
the ICU length of stay. Because the majority of the patients
tive likelihood ratios 7.6 and 0.18, respectively, and positive
and negative predictive values 94% and 70%, respectively. who would appear to benefit from such a protocol were
extubated 1 day after ICU admission, it seems that the ben-
The test was a strong predictor of subsequent ventilator in-
efit from such a protocol would be minimal, at least in
dependence when criteria were met by day 1 or day 2, but
an institution where most patients are rapidly weaned off
weaker when met by 4 days or more. Those who met the
mechanical ventilation.
criteria after 4 days or more often had prolonged weaning
The challenge of truncating the time to extubation
and a high incidence of reintubation. Nine of 178 patients
and reducing ICU length of stay will not be resolved by
who met weaning criteria required reintubation within 48
the current report, although actively seeking patients who
hours of first extubation; 8 of these patients first met the
criteria in >4 days from admission. Thirty-five patients may be able to achieve respiratory independence is one of
the paths that has to be followed.
achieved ventilator independence but did not fulfill the
weaning criteria; the most frequent reason for failing crite-
ria before independence was a PaO2/FIO2 ratio >24 kPa
(49% of cases). Comment by Arieh Eden, MD

n 2005 Lippincott Williams & Wilkins 65

Copyr ight Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Critical Care Survey of Anesthesiology  Volume 49, Number 2, April 2005

in multivariable analysis. Patients with delirium during their

Delirium as a Predictor of Mortality ICU stay also had a longer post-ICU stay, fewer median days
in Mechanically Ventilated Patients alive and without mechanical ventilation, and a higher inci-
dence of cognitive impairment at hospital discharge. Delir-
in the Intensive Care Unit ium and no-delirium groups did not differ importantly in
E. Wesley Ely,*y Ayumi Shintani,*z Brenda Truman,*y baseline comorbidities, severity-of-illness scores, organ dys-
Theodore Speroff,*z Sharon M. Gordon,*x function scores, or admission diagnoses.
Frank E. Harrell, Jr.,z Sharon K. Inouye,*# The development of delirium in mechanically venti-
Gordon R. Bernard,y and Robert S. Dittus* lated patients was associated with a 3-fold increase in death,
a longer hospital stay, and a greater likelihood of cog-
(JAMA, 291:17531762, 2004) nitive impairment at discharge. Few ICUs monitor for
delirium, a risk factor for adverse outcomes that might be
*Department of Medicine, Division of General Internal Medicine and Center
for Health Services Research and the Veterans Affairs Tennessee Valley prevented or treated.
Geriatric Research, Education and Clinical Center (GRECC); yDivision
of Allergy/Pulmonary/Critical Care Medicine, zDepartment of Bio-
statistics, and xDepartment of Psychiatry, Vanderbilt University School
of Medicine, Nashville, TN and #Department of Medicine, Yale Uni- COMMENT
versity School of Medicine, New Haven, CT. In the current article, the authors conclusively show the
Copyright n 2005 by Lippincott Williams & Wilkins association between the development of delirium and poor
DOI: 10.1097/01.sa.0000158565.09971.bc
outcome in ventilated patients in the intensive care unit
(ICU). This adds to previous knowledge about the delete-
rious effects of delirium on surgical patients. In the past,
A lthough respiratory failure and the need for a mechanical
ventilator account for most cases of intensive care unit
(ICU) admission, only 16% of patients receiving mechani-
although it was known that delirium occurs frequently in
the ICU (and to prove it there are no less than 25 synonyms
cal ventilation die as a direct result of respiratory failure. in the literature), its impact on outcome was not at all clear.
A prospective cohort study examined the contribution of The current study prospectively followed a cohort of ICU
delirium, a common yet underdiagnosed form of organ patients treated with mechanical ventilation and showed that
function, to the outcome of patients receiving mechanical the presence of delirium at any time during the ICU stay
ventilation. was associated with a 3-fold increase in the adjusted risk
Enrolled in the study were 275 consecutive mechan- for death at 6 months.
ically ventilated patients admitted to ICUs at the Vanderbilt This present article is of great importance because
University Medical Center between February 2000 and the tools used to assess delirium are easily applied and
May 2001. The most abnormal values obtained during the are independent of language or cultural background. The
first 24 hours of ICU stay were used to calculate patient Richmond Agitation Sedation Scale is a numerical scale
scores on the Acute Physiology and Chronic Health Eval- ranging from 5, indicating coma, to 0, reflecting an alert
uation II and the Sequential Organ Failure Assessment. and calm patient, to +4, describing combative behavior.
Standard scales (Confusion Assessment Method and the The confusion assessment method for the ICU is a short
Richmond Agitation-Sedation Scale) completed daily by evaluation process that establishes whether the patient can
study nurses provided data on neurological status. Patients answer simple question and perform simple commands. It
with delirium at any time while in the ICU were included was designed specifically for the evaluation of mechani-
in the delirium group. Primary outcome variables were cally ventilated patients. A tutorial for the use of these tools
6-month mortality, overall hospital length of stay, and is available at www.icudelirium.org.
duration of post-ICU stay. Ventilator-free days and cog- These tools can serve as measuring stones for future
nitive impairment at discharge were secondary outcome studies evaluating the incidence and significance of delir-
variables. ium on the one hand and of methods to control it on the
Fifty-one patients who had persistent coma and died in other. As the ICU population becomes older and there-
hospital were excluded from outcome analyses. Severity of fore more prone to delirium, such studies will be of great
illness was greater in these patients than in the remaining importance. One can hope that as a consequence of the
224 patients, 183 (81.7%) of whom had delirium during present study, more ICUs will include delirium evaluation
their ICU stay. Compared with the no-delirium group, scales as 1 of their routine tools.
patients with delirium had higher 6-month mortality rates
(34% vs. 15%) and spent 10 days longer in the hospital.
Delirium was independently associated with both outcomes Comment by Arieh Eden, MD

66 n 2005 Lippincott Williams & Wilkins

Copyr ight Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Survey of Anesthesiology  Volume 49, Number 2, April 2005 Critical Care

advantage in conditions with decreased systemic vascular

Does Arginine Vasopressin Influence resistance (vasodilatory shock and anesthetized patients
the Coagulation System in Advanced treated with angiotensin-converting enzyme inhibitors) and
during resuscitation. Vasopressin results in more pronounced
Vasodilatory Shock With Severe vasoconstriction in skin, skeletal muscles, and adipose tissue
Multiorgan Dysfunction Syndrome? than in the mesenteric, coronary, and cerebral circulation.
Martin W. Dunser,* Dietmar R. Fries,* In contrast with catecholamines, vasopressins vascular
Wolfgang Schobersberger,* Hanno Ulmer,y effect is preserved during hypoxia and acidosis. (For more
Volker Wenzel,* Barbara Friesenecker,* details on this topic, I recommend an excellent recent
Walter R. Hasibeder,* and Andreas J. Mayr* review).1 Notably, it has been shown recently that decreased
platelet count may occur during AVP infusion.
(Anesth Analg, 99:201206, 2004) In this current study, the effect of AVP infusion on the
coagulation system was evaluated in patients in advanced
*Division of General and Surgical Intensive Care Medicine, Department of vasodilatory shock. Although the authors used advanced
Anesthesiology and Critical Care Medicine and the yInstitute of Medical
technologies to evaluate the coagulation changes resulting
Biostatistics, The University of Innsbruck, Austria.
Copyright n 2005 by Lippincott Williams & Wilkins from AVP administration, the study added no new infor-
DOI: 10.1097/01.sa.0000158566.17595.e0 mation beyond the already established association between
AVP therapy and decreased platelet count. The major

A rginine vasopressin (AVP) is a potent supplementary

vasopressor in advanced vasodilatory shock. However,
decreases in platelet count have been reported during AVP
problem of the study is that almost all patients received
different combinations of blood products in an attempt to
correct coagulopathy on one hand, and therapy with heparin
therapy. The effects of AVP infusion on the coagulation on the other. Additionally, the authors did not neutralize
system were evaluated in advanced vasodilatory shock when the heparin effect in laboratory blood tests, and only half
compared with norepinephrine infusion alone. Forty-two of the patients completed the planned 48 hours of observa-
patients with advanced vasodilatory shock (norepinephrine tion. The authors conclusion that AVP-induced thrombocy-
requirements >0.5 mg/kg/min, mean arterial blood pres- topenia does not adversely affect the coagulation system is
sure <70 mm Hg) were prospectively randomized to be given based only on a brief observation of several laboratory tests
an additional AVP infusion (4 U/h) or norepinephrine in- and should be evaluated more thoroughly.
fusion alone. Most patients received coagulation active
Comment by Reuven Pizov, MD
treatment (fresh-frozen plasma, thrombocyte concentrates,
coagulation factors, and continuous venovenous hemofiltra-
REFERENCE
tion with heparin). At baseline and 1, 24, and 48 hours
1. Mutlu GM, Factor P. Role of vasopressin in the management of septic
after randomization, coagulation laboratory variables and a shock. Intensive Care Med. 2004;30:12761291.
modified thromboelastography were measured. No differ-
ences existed between groups in plasmatic coagulation
variables. Although there was no important difference be-
tween groups, platelet count markedly decreased in the
Twenty Months Routine Use of a New
AVP patients. There were no differences in the results of Percutaneous Tracheostomy Set Using
modified thromboelastography analyses between groups. Controlled Rotating Dilation
AVP infusion in advanced vasodilatory shock with severe
multiorgan dysfunction syndrome does not increase plasma Neel Sengupta, Keng Leong Ang, Doraiswamy Prakash,
concentrations of factor VIII, von Willebrand factor anti- Vivien Ng, and Shane J. George
gen, and ristocetin cofactor but may stimulate platelet ag-
(Anesth Analg, 99:188192, 2004)
gregation and induce thrombocytopenia. Global coagulation,
which is assessed by modified thromboelastography, is not Intensive Care Unit, Harefield Hospital, Royal Brompton and Harefield
different from patients receiving norepinephrine infusion Hospitals NHS Trust, Middlesex, United Kingdom.
only. Copyright n 2005 by Lippincott Williams & Wilkins
DOI: 10.1097/01.sa.0000158567.55712.a8
COMMENT
Arginine vasopressin (AVP) is strong vasoconstrictor
acting via V1 receptors located on the vascular endothelium,
which remain sensitive during septic shock. To date, the
A fter a favorable trial period, PercuTwist, a new percu-
taneous tracheostomy set was introduced in February
2002 for routine procedures. During the next 20 months,
vasoconstrictive properties of AVP have been used to 90 procedures were performed with minimal complications.

n 2005 Lippincott Williams & Wilkins 67

Copyr ight Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Critical Care Survey of Anesthesiology  Volume 49, Number 2, April 2005

This experience was prospectively evaluated by collecting COMMENT

information on reasons for unit admission, operators In the current report, the authors describe their
previous experience, the duration of previous tracheal experience with a new percutaneous tracheostomy technique
intubation, the time required for the procedure, the grading (PercuTwist). They report 90 cases of successful use of the
of the difficulty, the amount of bleeding, and the complica- technique. It is interesting that residents, who were not very
tions of the procedure. Twenty-two (24.4%) of 90 procedures experienced, performed most of the procedures. It is unclear
were performed by senior consultants with experience; 68 whether this method of percutaneous tracheostomy is better,
(75.6%) of 90 were performed safely by intensive care or cheaper, than others.
residents under close supervision. The mean time required Because serious problems were associated with other
for the procedure was 13 minutes 7 seconds. During the percutaneous tracheostomy techniques, it is encouraging to
entire study, only 1 procedure presented any difficulty during see new technical developments that appear to prevent or
the insertion process. It occurred because the initial skin remedy some of the previous complications.
incision was too small; no major bleeding or complications
were encountered, however. Comment by Arieh Eden, MD

68 n 2005 Lippincott Williams & Wilkins

Copyr ight Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.