Serhat Gul, PhD

Industrial Engineering
TED University

Surgery Scheduling

Main reference:

Erdogan and Denton. 2010.

Surgery Planning and
Scheduling: A Literature Review
Introduction

◼ The National Academy of Engineering and Institute of Medicine

report identifies the design and implementation of better planning
and scheduling systems as an important area.

◼ Poor surgery scheduling causes:

 Inefficient use of resources
 Decreased return on investment
 Long wait lists for patients

◼ Surgery accounts for more than 40% of a hospital’s total revenues

and expenses

◼ Recent studies indicate resource utilization, overtime, and on-time

start performance could be improved
The challenge of surgery scheduling

◼ Surgery schedules impact the

 Expensive resources including operating rooms (ORs), post anesthesia
care unit (PACU) beds
 Equipment resources such as mobile diagnostic imaging devices
 Human resources including surgeons, nurses, anesthesiologists and
other staff
◼ There is considerable uncertainty on the day of surgery due to the
 Duration of surgery and patient recovery
 Arrival of add-on cases

◼ There are a number of competing criteria

 For example: Utilization of ORs/staff/equipment vs patient waiting time
Background on surgery scheduling

◼ Surgery is performed either on inpatient or outpatient basis.

◼ Outpatient: Patient is not admitted to the hospital bed. Patient enters

and leaves the hospital on the day of surgery.
 Ambulatory service centers serve only outpatients.

◼ Inpatient: Patient is admitted either ahead of the day of surgery or

on the day of surgery (also called as same-day admit). Patient stays
in a hospital bed for at least one day following surgery and recovery.
Stages of perioperative system
Surgical suite configuration

◼ Several ORs share resources such as an equipment storage area,

sterilization resources, preoperative intake and postoperative
recovery rooms.

◼ Technological equipment restrictions

 For example: Some cardiac surgeries require cardiopulmonary bypass
equipment
 Some equipments are dedicated while some are mobile

◼ Intake and recovery rooms, and ORs can also be dedicated

Surgical suite configuration example
Flow of patient through a suite (linear flow)
Flow of patient through a suite (reentrant
flow)
Surgery planning strategies

◼ Open scheduling:
 Surgery time blocks are pooled together and any surgeon can use any
one of the available blocks

◼ Block scheduling:
 Certain blocks are first assigned to surgeons/surgical groups in
advance. The surgeons/surgical groups utilize these allocated blocks to
schedule their surgeries.

◼ Modified block scheduling:

 Some portion of the blocks may be left as non-dedicated and thus be
available for any surgical group.
 Also, all blocks may be reserved up to a certain point before the day of
surgery, at which point any unused blocks are released to other
surgeons/surgical groups.
Surgery block schedule of the VA Medical
Center – Boston

Note that 5 ORs are open each day

MONDAY TUESDAY WEDNESDAY THURSDAY FRIDAY

GU-O/E ENT GU-O/E ORTHO GU-O/E
THOR ORTHO ORTHO THOR ORTHO
CARDIAC CARDIAC VASCULAR CARDIAC CARDIAC
NEURO NEURO GENERAL GENERAL PLAS-AM
OPEN-PM

VASCULAR VASCULAR VASCULAR GYN-1ST&3RD VASCULAR

OPEN 2ND&4TH
Decision making at surgical suites
◼ Strategic decisions
 Whether to use an open-booking or block booking system
 Number of ORs to utilize
 Investment in new technology
 Number and type of staff to employ

◼ Tactical level decisions

 Reserving OR time and capacity for surgical specialties
 Finding an efficient surgery mix
 Determining how much time to plan for each type of surgery

◼ Operational decisions
 Sequencing surgeries in a particular OR
 Setting appointment times for patients to arrive for surgery
 How to manage add-on cases
 Whether to cancel scheduled cases
Factors affecting the quality of the surgery
schedules

◼ Uncertainty in the duration of activities

◼ Uncertainty in the number of patients served on the day of surgery

◼ The potential congestion in the downstream resources such as

PACUs, ICUs, surgical wards.

◼ Conflicting criteria of different stakeholders

Uncertainty in the duration of activities

◼ Pre-operative time can depend on

 Whether the patient requires some prescreening (e.g. blood tests)
 If the patient requires a translator

◼ Surgery time depends on

 The type of procedure
 Physical characteristics of the patient such as body-mass-index, age,
gender
 Surgical environment (e.g. academic vs. non-academic medical center,
or hospital vs. outpatient procedure center)

◼ Postoperative time depends on

 The outcome of surgery
 Patient’s response to anesthesia
An example for the scope of uncertainty
Uncertainty in the number of patients

◼ Surgery at hospitals can be divided into two major categories:

elective and non-elective cases

◼ Elective cases: Surgery may be planned well in advance to be

performed on a future date

◼ Non-elective cases: Surgery is unanticipated. These are urgent add-

on or emergent cases.

◼ Patient punctuality, no-shows, cancellations

Elective Surgery Scheduling Process

◼ The surgeon and patient select potential surgery dates

◼ The surgeon’s assistant checks the available blocks

◼ The scheduler books the case during a phone call with the
surgeon’s assistant

◼ Necessary information about the surgical case, patient and surgeon

are also inputted in the system.
Other Surgery Scheduling Processes
◼ How to handle requests for add-on surgeries?
 If a request comes after, for example, 12 pm one day before the day of
surgery, this surgery is considered as urgent and the appointment time
is set the day after the surgery is requested.

 Emergent cases bumps up all urgent and elective cases

◼ How to handle cancellation requests?

 Surgery can be rescheduled or completely cancelled due to several

different reasons
Some Cancellation Reasons
◼ Anesthesia cancellation
◼ Anesthesia unavailable
◼ Case bumped due to emergency
◼ Doctor underbooked previous case – delay
◼ Equipment unavailable
◼ Incomplete work-up
◼ Insurance related
◼ No beds available
◼ Patient ate or drank
◼ Patient ill
◼ Patient no-show
◼ Patient reschedule
◼ Surgeon rescheduled
◼ Surgeon unavailable
◼ Surgery not necessary
◼ Too late to perform procedure
◼ Weather
Surgical suite criteria

◼ Patient waiting time

 Wait on the day of surgery
 Wait on the future surgery list
◼ OR overtime
◼ OR idle time
◼ Surgeon idle time
◼ Staff utilization
◼ OR utilization
 Some criteria and factor definitions

Surgeon 1 1 2 3 4 5 Pooled set of

identical ORs

Surgeon 2 1 2 3 4 OR 1

OR 2
Surgeon 3 1 2

OR 1 1 2 3 4 4 5 2

OR 2 1 2 3 1 Overtime

OR Turnover Time Surgeon

Idle Time Surgeon Turnover Time
 Conflicting Criteria

Surgeon 1 1 2 OR 1

Surgeon 2 1 2 3 OR 2

OR 1 1 2 Is it possible to decrease
Schedule 1 overtime?
OR 2 1 2 3

OR 1 1 3 Surgeon idle time increases!

Schedule 2

OR 2 1 2 2
From Sakine Batun, Informs
2010.
Multi-criteria Optimization

Thanks to Prof. John Fowler, Arizona State University.

.
A

.
B
.I
.C
.J
.D .E
.F .G
.
H
Multicriteria Optimization

◼ In any multicriteria problem, it is important to point out the nature

of optimization being performed (Notation: T’Kindt and
Billaut(2002))
 Lexicographic optimization
 Linear combination
 Generating the extreme points
 Epsilon constraint
 Generating the set of non-dominated points
.
A

.
B
.I
.C
.J
.D .E
.F .G
.
H
Multicriteria Optimization

◼ In any multicriteria problem, it is important to point out the nature

of optimization being performed (Notation: T’Kindt and
Billaut(2002))
 Lexicographic optimization
 Linear combination
 Generating the extreme points
 Epsilon constraint
 Generating the set of non-dominated points
.
A

.
B
.I
.C
.J
.D .E
.F .G
.
H
Multicriteria Optimization

◼ In any multicriteria problem, it is important to point out the nature

of optimization being performed (Notation: T’Kindt and
Billaut(2002))
 Lexicographic optimization
 Linear combination
 Generating the extreme points
 Epsilon constraint
 Generating the set of non-dominated points
.
A

.
B
.I
.C
.J
.D .E
.F .G
.
H
Surgery Scheduling Literature

◼ Methodological Categorization

 Queuing Models

 Simulation

 Mathematical Programming

 Heuristic Methods
Queuing Models

◼ Many earlier studies used queuing models for surgery

and outpatient clinic appointments
 For example: Welch and Bailey (1952), Welch (1964), Mercer (1960),
Jansson (1966), Soriano (1966), Mercer (1973)

◼ Used for single OR scheduling problems

 Assign start times for a number of surgeries to be scheduled in
that OR on a given day
 Consider the variability in surgery durations
 Minimize waiting, idling and overtime
Single OR scheduling problem
Queuing Models

◼ Advantage: Offer valuable general insights about the effect of

uncertainty on scheduling decisions.

◼ Shortcoming in the context of surgery scheduling : Assumptions

need strong validation effort.

◼ Commonly made assumptions which are not valid:

 The system reaches steady state
 Exponentially distributed surgery durations
Simulation

◼ Found considerable application to surgery planning and scheduling

◼ Comparison to queuing models:

 Advantage: More flexible in terms of assumptions about the probability
distributions of surgery or other activity durations
 Disadvantages:
◼ Do not provide a closed-form solution
◼ More computationally intensive than queuing models and may require
considerable computation time

◼ Generally used to evaluate surgery schedules and assess the

impact of disruptions that arise due to uncertainty in duration and
demand
Simulation application examples from the
literature
◼ OR process and schedule performance analysis
 A single OR with open-booking scheduling (Charnetski 1984)
◼ The relationship between the starting times of surgeries and the
costs of surgeon, staff and OR idle times.
 More complex multi-OR surgical environments with recovery
areas (Lowery 1992, Schmitz and Kwak 1997, Lowery and Davis 1999)
◼ Determine the number of ORs to open on a day given that the
number of surgeries is known

◼ Testing the performance of scheduling heuristics

 Surgery sequencing and patient appointment time setting
heuristics (Huschka et al. 2007, Gul et al. 2011, Hans et al. 2008)
Optimization (Deterministic Models)
 Two-phase approach for scheduling (Guinet and Chaabane 2003,
Jebali et al. 2006)
◼ Phase 1: Assign surgeries to ORs and days (Fei et al. 2010)

◼ Phase 2: Sequence surgeries and set patient appointment times

 Modeling approach:
◼ Mixed Integer Programming (MIP), goal programming
 Objectives:
◼ Phase 1: Minimize the lead time (time between the day surgery is
requested and the day surgery is performed), OR overtime, fixed
cost of opening ORs
◼ Phase 2: Patient waiting time on the day of surgery, surgeon idle
time, OR overtime
 Constraints:
◼ Equipment availability in the ORs
◼ Surgeon time availability
◼ OR resource capacity
Deterministic OR allocation problem
(DORA)

◼ A number of ORs need to be opened (i.e. staffed) on a particular

day

◼ Surgeries to be performed that day must be allocated to the open

ORs

◼ Performance measure:
 Daily operating cost
◼ Fixed cost of opening ORs
◼ OR overtime cost

◼ ORs are open only for a fixed amount of time on that day
Based on “Denton et al. 2010. Optimal Allocation of Surgery Blocks to
Operating Rooms Under Uncertainty. Operations Research”
DORA
Exercise (DORA)

◼
Exercise (DORA)

◼ Now, suppose that you formulated this model to help a surgical suite
operations manager solve her problem for the case where n=13 and
m=10.

◼ In your formulation, surgery blocks having indices i=1,2,3,4,5 denote

cardiac surgeries. The operations manager warns you that the
assignment of these surgeries must be restricted to the ORs that
contain cardiopulmonary bypass equipment. Note that only the ORs
having indices j=1, 2 contain these equipments; and these
equipments are not mobile. Considering this restriction, formulate a
linear constraint to add into the DORA formulation.
Active Learning Exercise 1

◼ In your formulation, each one of the surgery blocks with indices i= 6

and 7 actually represents an individual surgery. Furthermore, these
surgeries are performed by the same surgeon. The surgeon informs
the surgical suite manager that he can benefit from parallel
processing while performing his surgeries, because the incision
durations are relatively low for his surgeries. Therefore, the manager
concludes that the OR for surgery 6 must be different from the OR
for surgery 7. Considering this restriction for allocation, formulate a
linear constraint to add into the DORA formulation.
A simple heuristic based on LPT

LPT(n) sorts all surgery blocks in decreasing order of

duration and allocates each successive item (starting with
the largest) to one of the n ORs with the lowest current
level.

The LPT heuristic solves the

problem with a 13/12
approximation guarantee
Surgery scheduling as a bin packing
problem
◼ Problem: Finding the (near) optimal mix of surgical cases for multi
OR surgical environments

◼ Bin packing problem: Given a list of items and their sizes and a
collection of bins of fixed size, find the smallest number of bins so
that all of the objects are assigned to a bin.

◼ Bin packing problem is an NP-hard combinatorial optimization

problem, so many heuristics have been applied to solve the
problems
Bin packing problem (cont’d)
◼ Surgery scheduling as a bin packing problem
 ORs are available for a fixed time during the day and represent bins
 Surgeries are the items to be packed in the bins
 Duration of the surgeries represent the item sizes

◼ On-line bin packing problem: Surgeries are scheduled sequentially

one at a time without having perfect information related to the future
requests.

◼ Off-line bin packing problem: Surgeries are batched and

simultaneously assigned to ORs
Bin packing problem example

◼ Example: 7 different types of urology surgery requests arrive

sequentially (i.e. prostatic needle biopsy, transurethral greenlight
laser, transurethral bladder biopsy and others) to an ambulatory
surgery center. They need to be scheduled to either one of the 10
identical ORs as each one is requested. Note that the mean duration
for a surgery type is considered when the surgery of that type is
scheduled.
◼ The following surgery type index indicates the surgery arrival
sequence: {4, 8, 5, 1, 7, 6, 1, 4, 2, 2}

OR 1 OR 2 OR 3 OR 4 OR 5 OR 6
Bin packing problem (cont’d)

◼ On-line heuristics:
 Next Fit: Schedule into the current OR (the OR which was used the latest) if it
fits. If it does not, close that OR and open a new OR
 First Fit: Schedule into the lowest numbered OR (the OR which was opened the
earliest) if it fits. If it does not, close that OR and open a new OR
 Best Fit : Schedule into the OR which has the lowest amount of remaining time
which is also sufficient to perform the case
 Worst Fit: Schedule into the emptiest OR which has also enough time to
perform the case
◼ Off-line heuristics:
 First Fit Decreasing Heuristic: Sort the surgeries in decreasing order of
estimated duration. Schedule the next surgery into the lowest numbered OR in
which it fits. If it does not fit into any open OR, open a new OR.
 Best Fit Decreasing Heuristic: Sort the surgeries in decreasing order of
estimated duration. Schedule the next surgery into the OR which has the lowest
amount of remaining time which is also sufficient to perform the case
Bin packing problem (cont’d)
Arrival sequence: {4, 8, 5, 1, 7, 6, 1, 4, 2, 2}

Next Fit result First Fit result

Best Fit result Worst Fit result

Exercise

Consider the below parameter values for an OR allocation problem.

ORs are open for 8 hours during regular time. Assume that no overtime
is ‘planned’ in the ORs (i.e. overtime may be observed only when
surgeries take longer than their expected durations). Solve the problem
using best fit decreasing heuristic and determine the number of ORs
needed, accordingly.
Exercise (Sequencing)
◼ Suppose that there are seven patients waiting for their operations in
a wait list of a surgical oncologist. It is important for the surgeon to
finish each surgery as soon as possible due to some needs or
preferences of patients.
◼ To facilitate her decision-making process, the surgeon assigned an
urgency level to each surgery j (wj). With the help of the scheduling
office, she also came up with the expected surgery durations
(duration amounts are scaled and represented by pj values). As her
consultant, you’ve concluded that she can achieve her goals by
minimizing the total weighted completion time while creating the
schedule. Then, answer the following question based on the
parameter values given on the table below.
◼ Find all optimal surgery sequences for the surgical oncologist.
Total Weighted Completion Time

◼ For single machine problems, WSPT rule is optimal if

 w j C j is the objective function
◼ WSPT is Weighted Shortest Processing Time
 Surgeries are ordered in decreasing order of wj/pj
 wj is the weight, or importance factor, of surgery j
◼ May represent urgency level
Exercise (sequencing with precedence
constraints)
◼
◼ Find at least one optimal sequence.
Exercise (modified)

◼ Now consider the following condition and resolve the problem

accordingly:
 It is not necessary for a surgeon to complete all of her surgeries in her
list before another surgeon is allowed to start working on her surgeries.