Appointment Systems

Modeling

Serhat Gul, Ph.D.

TED University
Case Study at a Campus Health Center
(CHC)

◼ An outpatient clinic system serving mainly the students of a major

university (a population of about 65000 students)

◼ Provides service in 4 different units

 Acute Care
 Primary Care and Women’s Care
 Health and Wellness
 Sports Medicine

Polo Peng
Kelly Hession
Paula Sleiman
Stefanie Schroeder, MD
Serhat Gul, Ph.D.
Case study steps

◼ Study steps:

 Draw a process map for the CHC patient flow (including primary and
acute care area)

 Identify the input and output parameters to consider in the model

 Build a discrete-event simulation model

 Analyze the simulation output values

 Identify the bottlenecks and test different improvement ideas

Check-in process

Teller 1
Confirm and
Update Patient
Information Patient Waits in
Acute Care Unit
Appointment Checks In with Wait Room
Patient arrives to next available Acute
wait room teller Teller 2 Care
Confirm and
Update Patient
Information Patient Waits in
Primary & Physician and
Women’s Specialist
Care Consultation Unit
Teller 3 Wait Room
Confirm and
Update Patient
Information
Wellness
Walk In Patient Checks In with Patient Signs In
arrives to wait next available and Waits for
room teller Teller 4 Clinician
Confirm and
Update Patient
Information
Bumped Out
Patient

Patient Patient exists the

Reschedules Health Center
Acute care flow

Patient Waits for a Patient waits for a Clinician Makes

Triage Nurse (TN) YES clinician in an and Confirms
in Waiting Room examination room Diagnosis

Examination Room
TN checks Ready?
patient’s vitals and
asks questions
about condition

Treatment Needed?
NO
(X-Rays, Tests…)

NO
TN Updates
Tests needed for YES
NO information in
diagnosis?
EMR

Patient Gets
YES Treatment
Patient Waits for a
room

TN Conducts
Tests
Primary care flow
Patient Waits for a
Patient waits for a Clinician Makes
Medical Assistant
YES clinician in an and Confirms
(MA) in Waiting
examination room Diagnosis
Room

Examination Room
MA checks Ready?
patient’s vitals and
asks questions
about condition

Treatment Needed?
NO
(X-Rays, Tests…)

NO
MA Updates
Tests needed for YES
NO information in
diagnosis?
EMR

Patient Gets
YES Treatment
Patient Waits for a
room

MA Conducts
Tests
Discharge procedure

Clinician makes Clinician Inputs Pharmacy

Drug prescription
assessment of YES Prescription in prepares and
needed?
patient’s condition EMR gives prescription

NO

Clinician inputs
assessment in Checkout
EMR

Patient exists the

health center
Tasks

◼ Find out what type of parameter values (such as the resource and
duration information) you should obtain from the CHC managers or
by time studies
 Note: The CHC actually has information on only the time triage nurse
spends with a patient

◼ Identify the stochastic parameters whose values might vary

significantly due to particular factors
 For example, new patients could spend more time than existing patients
for the check-in process
Resources

◼ Patient Service Assistant (PSA)

◼ Triage Nurse
◼ Medical Assistant
◼ Physician
◼ Examination Room
Some necessary inputs

Main parameters:
◼ Patient arrival rate for walk-ins
◼ Pre-determined schedule for patients with appointments
◼ Check-in time
◼ Triage nurse service time
◼ Physician service time (visit time)
◼ Time for lab tests/ radiology exams

Extra parameters:
◼ No-show rates
◼ Average delay in arrival for the appointment
Stochastic parameters having significant
variability
◼ Patient arrival rate
 Patients with appointments
◼ Patient are asked to arrive 20 minutes early
◼ No-show
◼ Punctuality
 Random walk-in arrivals
 Seasonality affect
◼ Factors affecting the check-in time:
 New patient
 Missing records
 Health insurance policy
 Language barriers
◼ Physician service time
 Acuity levels might not be helpful for duration estimation
Tasks (cont’d)

◼ Select some performance measures (outputs) that might be

important for the performance of the CHC
Potential criteria

◼ Patient waiting time

 Waiting for check-in
 Waiting for triage
 Waiting for examination room
 Waiting for physician

◼ Physician AM overtime
 Might cause idle time for a PM physician sharing the same room with
the AM physician

◼ Physician PM overtime
Tasks (cont’d)

◼ What might be the major bottlenecks of the system?

◼ What factors might make the CHC different from other outpatient
clinics?

◼ What are the potential directions for the improvement of the system?
 Potential improvement efforts

◼ Scattered arrivals:
 Current Status: There are 9 physicians in the clinic but only 3 PSAs.
First 9 patients are asked to arrive at the same time.
 Suggestion: New patients could arrive 5 minutes earlier than old
patients

◼ Variability in the check-in wait time

 Current status: Each PSA is responsible for only three physicians’
records, so there are 3 separate queues in the system
 Improvement idea: Pool the queues

Chand et al. (2009). Improving patient flow at an outpatient clinic: study of sources of
variability and improvement factors. Health Care Management Science
Potential improvement efforts (cont’d)

◼ Variability in the check-in time

 Restrict the number of external appointment requests that the PSAs
receive

◼ Variability in the waiting time for triage

 Current status: Patient records are batched and then delivered
 Suggestion: Avoid batching

◼ Fast-track option
 Dedicate certain resources to certain type of patients (e.g. Nurse
practitioners handling simple cases)
 Resource reservation decisions should be carefully given to justify the
benefit
Appointment Scheduling for Ambulatory
Services
Cayirli et al. (2006). Designing appointment scheduling systems for ambulatory care services.
Health Care Management Science.

◼ Patient classification
 New/return
 Pediatric/adolescent/geriatric
 Difficult/easy

◼ Sequencing rules – six rules tested

 Determines the order in which calling patients are assigned to time
blocks based on a particular patient classification scheme

◼ Appointment rules – seven rules tested

 Determines the basic template of the appointment schedule by
specifying the number of patients scheduled to each appointment slot
(i.e. block size) and the length of appointment slot
Sequencing rules
◼ FCFA: Patients receive appointment slots on a first-call first-appointment
basis

◼ ALTER: Orders new and return patients in an alternating pattern

(RNRNRNRN…)

◼ NWBG: Schedules new patients in the beginning and return patients in the
remaining part of the session based on the expected percentage of each
patient class (NNN….RRRR)

◼ RTBG: Schedules return patients in the beginning (RRRR…NNN)

◼ NWBND: New patients in the beginning and in the end (NN..RRR..NN)

◼ RTBND: Schedules return patients in the beginning and in the end

(RR…NNN..RR)
Appointment rules
Appointment rules (cont’d)
Appointment rules (cont’d)
Findings of the study

◼ Sequencing decisions have a more pronounced impact on clinic

performance than the choice of an appointment rule

◼ No-shows, walk-ins, clinic size and patient punctuality are found to

be important factors affecting the performance measures (e.g.
patient waiting time, doctor idle time and overtime)

◼ Placing new (return) patients in the beginning of the session is

preferred when doctor’s (patient’s) time is highly valuable compared
to patient’s (doctor’s) time

◼ Alternating new and return patients performs the best in between

these two extremes
Findings (cont’d)

◼ Fixed intervals of 2BEG, MBFI, IBFI are the best performers

◼ Individual block rules are mostly suited to specialties with short

consultation times; they should be avoided in clinics having long
consultation times, low walk-ins, high no-shows.

◼ The best choice of appointment rule depends on the sequencing

rule selected
Outpatient Clinics Literature

White et al. 2011. The effect of integrated scheduling and capacity policies
on clinical efficiency. Production and Operations Management
 Optimal Appointment Scheduling

◼ A single server system at which patients arrive punctually at

scheduled appointment times and are served in the order of their
arrival (i.e. job sequence is fixed and based on the FCFA rule).

◼ n jobs (appointments) needed to be scheduled on a particular day

◼ Job durations are stochastic

◼ Performance measures:
 Expected patient waiting time
 Expected clinic idle time
 Expected clinic overtime

Denton and Gupta. 2003. A sequential bounding approach for optimal appointment scheduling. IIE
Transactions
 Optimal Appointment Scheduling
𝑛 = number of patients to be scheduled

𝐾 = number of scenarios

𝑖 = index for appointment

𝑘 = index for scenario

𝑑 = regular capacity of the clinic

𝑧𝑖𝑘 = actual duration of the appointment of patient 𝑖 under scenario 𝑘

𝑥𝑖 = allocated time for the appointment of patient 𝑖

𝑤𝑖𝑘 = amount of time patient i waits under scenario 𝑘

𝑠𝑖𝑘 = amount of time the clinic stays idle before the appointment for patient 𝑖 begins

𝑙𝑘 = clinic overtime (tardiness) under scenario 𝑘

𝑔𝑘 = clinic earliness under scenario 𝑘

𝑐𝑖𝑤 = cost of waiting per unit time for patient i

𝑐 𝑠 =cost of having clinic idle per unit time

𝑐 𝑙 =cost of clinic overtime per unit time

◼ Suppose now that the clinic manager would like to make sure the
mean waiting time for each patient does not exceed θ units.
Formulate a linear constraint for this restriction to add into the
current formulation.