Radiation Oncology CT Simulator: Efficiency and Utilization
Final Report
Submitted to:
Dawn Johnson, Director of Clinical Operations, Radiation
Oncology
Chris Alcala, Chief Radiation Therapist, Radiation Oncology
Zac Costello, Performance Improvement Engineer, Performance
Improvement
Kate Sell, Performance Improvement Fellow, Performance
Improvement
Dr. Mark Van Oyen, IOE 481 Professor, Industrial and Operations
Engineering
Submitted by:
IOE 481 Team 2
Julia Clark
Alexander Mize
Maddie Price
Karan Shah
Date Submitted:
April 17, 2018
18W2-final-report
ii
Table of Contents
Executive Summary1
Introduction4
Background4
Current Situation4
Appointment Time Length Scheduling Guidelines5
CT Simulator Patient Flow5
Key Issues5
Goals and Objectives6
Project Scope6
Appointment Scheduling Guidelines Data Collection and Analysis
Methods6
Appointment Time Length Data Collection7
Patient Attribute Data Collection7
Appointment Scheduling Guidelines Findings and Conclusions7
General Linear Model Regression7
Statistical Analysis8
30-Minute Appointments9
60-Minute Appointments10
90-Minute Appointments17
Appointment Scheduling Guidelines Recommendations20
Consider Five Significant Factors in Appointment
Scheduling20
Pilot Reducing Appointment Time Length Increments from 30 to 15
Minutes20
Patient Flow Data Collection and Analysis Methods21
Historical Data Analysis21
Patient Flow Benchmarking21
Observations21
CT Simulator Patient Flow Process Map22
Time Studies22
Value Stream Map22
Patient Flow Findings and Conclusions23
Historical Delays23
Induction Room Benchmarking Research23
Main CT Simulator Patient Flow Tasks23
Patient Flow Bottlenecks24
Patient Flow Design Methods, Requirements, Constraints, and
Standards25
Soft Design Constraints (Design Requirements)25
Hard Design Constraints 25
Design Standards26
Patient Flow Recommendations27
Designate Patient Flow Overflow Room27
Watch Educational Videos Prior to Room 1027
Assign Pre-Simulation Checklist to MAPSA28
Implement Induction Room28
Pugh Matrix Recommendation28
Expected Impact28
New Appointment Scheduling Guidelines29
Improved Patient Flow29
References31
Appendix A: Appointment Time Length Data Collection Sheet32
Appendix B: Appointment Time Length General Linear Model
Regression33
Appendix C: Historical Data CT Simulator Delay Root Cause
Frequency37
Appendix D: CT Simulator Patient Flow Process Map38
Appendix E: Time Study Data Collection Sheet41
Appendix F: Patient Flow Value Stream Map42
Appendix G: Patient Flow Design Constraints and Standards
Matrix44
Appendix H: Future State Patient Flow46
Appendix I: Patient Flow Pugh Matrix48
List of Figures and Tables
Table 1: Mean and Standard Deviation of Actual Appointment Time
Lengths9
Figure 1: 30-Minute Appointment Types Contributing to Unused
Scheduled
Appointment Time9
Table 2: 30-Minute Appointment Type with Highest Unused
Scheduled Time10
Figure 2: 60-Minute Appointment Types Contributing to Unused
Scheduled
Appointment Time11
Table 3: 60-Minute Appointment Types with Highest Unused
Scheduled Time12
Table 4: 60-Minute Appointment Types Unused Scheduled Time per
Week12
Figure 3: Head and Neck Appointment Actual Time Length
Distribution13
Table 5: Head and Neck Appointment Scheduled Time
Comparison13
Figure 4: Prostate Appointment Actual Time Length
Distribution14
Table 6: Prostate Appointment Scheduled Time Comparison14
Figure 5: Lung (No SDX) Appointment Actual Time Length
Distribution15
Table 7: Lung (No SDX) Appointment Scheduled Time
Comparison15
Figure 6: Spine Appointment Actual Time Length
Distribution15
Table 8: Spine Appointment Scheduled Time Comparison16
Figure 7: Pelvis Appointment Actual Time Length
Distribution16
Table 9: Pelvis Appointment Scheduled Time Comparison17
Figure 8: 90-Minute Appointment Types Contributing to Unused
Scheduled
Appointment Time17
Table 10: 90-Minute Appointment Type with Highest Unused
Scheduled Time18
Table 11: 90-Minute Appointment Types Unused Scheduled Time per
Week18
Figure 9: Abdomen Appointment Actual Time Length
Distribution18
Table 12: Abdomen Appointment Scheduled Time Comparison19
Figure 10: Abdomen Appointment Actual Time Length
Distribution19
Table 13: Abdomen Appointment Scheduled Time Comparison20
Table 14: Value Stream Map Process Time, Wait Time, and First
Time Quality Values24
Figure B-1: General Linear Model Regression with ACTV Patient
Attributes34
Figure B-2: General Linear Model Regression with REVC Patient
Attributes35
Figure B-3: General Linear Model Regression without Height,
Weight, and BMI36
Table G-1: Soft Design Constraints (Design Requirements)44
Table G-2: Hard Design Constraints44
Table G-3: Design Standards45
Executive Summary
The following section is an executive summary of the IOE 481
Radiation Oncology CT Simulator: Efficiency and Utilization project
which concluded on April 17, 2018.
Introduction
One of the vital resources in the Michigan Medicine Radiation
Oncology Department is the CT Simulator: a machine that almost
every patient must be scanned with to plan radiation dosages.
Therefore, the optimization of the patient flow process and
appointment schedule for the CT Simulator are important. The
Radiation Oncology Department’s Director of Clinical Operations and
the Chief Radiation Therapist requested the help of an IOE 481
student team from the University of Michigan to analyze and make
recommendations for improving the CT Simulator appointment
scheduling guidelines and CT Simulator patient flow.
Key Issues
The clients identified the following key issues:
· Frequent delays in the CT Simulator patient flow process
· Long patient wait time during CT Simulator appointments
· CT Simulator room idle time when schedule is delayed
· High variation in the difference between scheduled appointment
time length and actual appointment time length
Goals and Objectives
The team’s primary goal was to create new appointment scheduling
guidelines for optimizing appointment time lengths and design a
more efficient CT Simulator patient flow process. The team’s
recommendations aimed to achieve the following objectives:
· Decrease patient wait time during CT Simulator
appointments
· Decrease CT Simulator room idle time
· Decrease the difference between scheduled appointment time
length and actual appointment time length
· Increase patient throughput
Project Scope
The project included the process of scheduling patients for the
CT Simulator and the process surrounding the CT Simulator, starting
from the moment the patient who will be going through the CT
Simulator enters the Radiation Oncology Department and ending when
the patient leaves the department.
Appointment Scheduling Guidelines Data Collection and Analysis
Methods
The team created a data collection sheet, shown in Appendix A,
that the CT Simulator employees used to collect appointment time
length data. The CT Simulator employees collected this data for a
total of 255 patients from February 26, 2018 to March 28, 2018. The
team’s coordinator obtained patient attribute data using the
patient contact serial numbers (CSNs) for the patients included in
the appointment time length analysis.
To analyze the collected data, the team used general linear
model regression in Minitab to identify the significant factors
affecting appointment time lengths to create new appointment
scheduling guidelines. The team hypothesized that patient
characteristics could be statistically significant in determining
appointment times. Therefore, the team used general linear model
regression to predict the length of the appointments using factors
such as age, weight, and inpatient or outpatient status.
Appointment Scheduling Guidelines Findings and Conclusions
The team used 207 patient data records to perform a general
linear model regression in Minitab to determine the factors that
are significant at a significance level of 0.05. Variables that are
statistically significant were: IV Contrast (yes or no), SDX
training (yes/assess or no), number of treatment areas (isos),
vacuum bag (yes or no), and Qfix (yes or no). However, the adjusted
R-squared value was found to be only 34.41%. Therefore, the team
concluded using a linear regression equation to predict appointment
time lengths is infeasible.
The team then split the data by scheduled appointment time
length (30, 60 or 90 minutes) and performed general statistical
analysis. For each of these three groups of appointments, the team
analyzed the appointment time lengths for each scan type. To
identify the most important types of appointments to analyze, the
team used the Pareto Principle which states that only a few
appointment types contribute to a large percentage of the CT
Simulator appointment scheduling optimization issue.
For 30-minute appointments, the mean of the actual time length
was 29.11 minutes. The team concluded that the appointments are
scheduled appropriately because the appointments only contributed
to an additional 10.75 minutes per week in the CT Simulator
schedule.
For 60-minute appointments, the mean of the actual time length
was 35.05 minutes. Head and neck, prostate, lung (no SDX), spine,
and pelvis appointments contributed to 2.38, 2.10, 1.46, 1.12, and
1.02 unused scheduled hours per week, respectively. After analyzing
the distribution of these appointment types, the team concluded
that the scheduled appointment time lengths for these appointments
could be shifted down to 45 minutes.
For 90-minute appointments, the mean of the actual time length
was 62.10 minutes. Liver and abdomen appointments contributed 1.10
and 0.97 unused scheduled hours per week, respectively. After
analyzing the distribution of these appointment types, the team
concluded that the scheduled appointment time lengths for these
appointments could be reduced to 75 minutes.
Appointment Scheduling Guidelines Recommendations
In the long-term, the team recommends that the department
decrease the scheduled appointment time increments from 30 minutes
to 15 minutes. From the statistical analysis performed, the team
recommends that department conduct a pilot that decreases the
scheduled time for head/neck, prostate, lung (no SDX), spine, and
pelvis appointments to 45 minutes and decreases the scheduled time
for liver and abdomen appointments to 75 minutes.
Patient Flow Data Collection and Analysis Methods
The team used historical CT Simulator data for 102 patients to
determine root causes of CT Simulator room delays. Additionally,
the team performed benchmarking research to understand the
organization of other CT Simulator patient flows at hospitals
around the United States. Then, the team observed the CT Simulator
patient flow for 32 patients in order to create a process map in
Microsoft Visio, shown in Appendix D, and the time study data
collection sheet, shown in Appendix E. The team collected time
studies for 90 patients over 102 hours in the department. Finally,
a value stream map, shown in Appendix F, was created using the
process map framework. The team calculated the average process
times, wait times, and first time quality values from the time
study data.
Patient Flow Findings and Conclusions
From the value stream map, the team found that, on average, the
patient flow process time was 62 minutes, the wait time was 48
minutes, the lead-time was 110 minutes, and the total first time
quality was 42%. The time identified three bottlenecks from the
value stream map. First, the team identified patient check-in as an
initial bottleneck for patients due to high wait time. The second
bottleneck is the completion of the pre-simulation tasks due to the
high wait time, high process time, and low first time quality. The
third bottleneck is the pre-scan set up and patient positioning
step due to the high process time. The team concluded that these
three bottlenecks would be the focus of the patient flow
recommendations and improved patient flow design.
Patient Flow Design Methods, Requirements, Constraints, and
Standards
The main design task the team performed was creating an improved
patient flow process. When designing this process, several design
requirements, constraints, and standards guided the team’s
approach. Details on the design considerations of the project are
described in Appendix G.
Patient Flow Recommendations
Overall, four key recommendations were made to improve patient
process flow. The future state patient flow map is shown in
Appendix H. First, adding a second patient preparation room
(similar to Room 10) could decrease customer wait time after
checking in. Second, ensuring that the patients watch CT Simulator
videos prior to their appointment could decrease patient process
time in Room 10. Third, allowing MAPSA employees the role of
highlighting and listing next day missing patient questionnaire
tasks could decrease process and wait times in Room 10. Fourth,
implementing a separate patient positioning room could increase
overall CT Simulator patient flow efficiency. The team used a Pugh
Matrix, shown in Appendix I, to recommend that the department
heavily focus on changing the process for watching the educational
videos in the patient flow.
Expected Impact
The team expects the new appointment time length pilot will
increase patient throughput by 288 patients and decrease CT
Simulator room idle time and the difference between the scheduled
appointment length and actual appointment length by 234 hours.
Additionally, the team expects the four patient flow
recommendations to decrease patient flow process time by 24%, from
62 minutes to 47 minutes, primarily from the decrease in time spent
in Room 10 watching the video. After removing unnecessary patient
wait time in the check-in process, the patient wait time is
expected to decrease by 35%, from 48 minutes to 31 minutes. The
lead time is expected to decrease by 29%, from 110 minutes to 78
minutes. Since patients would experience fewer delays when a MAPSA
employee is assigned to document the next day patient missing
questionnaire, the first time quality is expected to increase from
42% to 57%.
Introduction
The Michigan Medicine Radiation Oncology Department treats
cancer patients through personalized radiation therapy treatment
plans and radiation therapy. Radiation therapy is used to treat
cancer and relieve pain for patients, which ultimately improves the
quality of life for the individual receiving treatment. One of the
vital resources in the Radiation Oncology Department is the CT
Simulator: a machine that almost every patient must be scanned with
prior to radiation therapy. The CT Simulator is used to simulate
the patient’s exact position for treatment. Therefore, the patient
is immobilized on the CT Simulator table using several devices such
as cushions and masks. After the patient is positioned, a CT scan
is taken of the treatment area. The CT scan is then used by a team
of dosimetrists to precisely plan the radiation treatment dosage
that minimizes radiation to the surrounding healthy organs. The
optimization of the appointment schedule and the patient flow for
the CT Simulator are important because the machine is vital in the
treatment planning process.
The Radiation Oncology Department’s Director of Clinical
Operations and the Chief Radiation Therapist have seen issues such
as frequent delays in the CT Simulator patient flow process, long
patient wait times during the CT Simulator appointments, CT
Simulator room idle time when schedule is delayed, and high
variation in the difference between scheduled appointment length
and actual appointment length. Therefore, the clients requested the
help of an IOE 481 student team from the University of Michigan to
analyze and make recommendations for improving the patient flow and
appointment scheduling guidelines. The team collected qualitative
and quantitative data by observing the process to understand the
patient flow and conducting time studies to analyze the appointment
time lengths and bottlenecks in the patient flow. Using this data,
the team performed analysis to create new appointment scheduling
guidelines for optimizing appointment time lengths and design a new
patient flow process. This final report documents the background,
key issues, goals and objectives, scope, methods, findings and
conclusions, recommendations, and expected impact of the
project.
Background
This section of the proposal describes the current appointment
scheduling guidelines and patient flow for the CT Simulator in the
Radiation Oncology Department.
Current Situation
The CT Simulator is open Monday through Friday, from 7:00 AM to
6:00 PM. Most days, the CT Simulator is booked for all 11 hours.
The department sees both inpatients and outpatients; therefore,
several other hospital departments are often involved with the CT
Simulator patient flow such as transportation services and
anesthesiology. Because of the important and urgent nature of the
simulation, the department does not cancel appointments for any
reason. Therefore, if one patient is delayed, the following
scheduled patients for that day are also delayed. During most
delays, the CT Simulator is not being used. Thus, any delays in the
schedule not only extend patient wait times, but also cause CT
Simulator room idle time.
Currently, appointment time lengths are scheduled based on the
anatomical location of the tumor on the body. The appointments are
either 30, 60, or 90 minutes long. The client identified that the
scheduling guidelines may not be accurate. The department does not
consider factors such as age, weight, and inpatient or outpatient
status, which could affect appointment length. Additionally,
patients often finish early or late. This trend could signal that
the appointment time length scheduling guidelines should be
adjusted to provide accurate appointment time lengths. New
appointment time length scheduling guidelines could minimize the
difference between scheduled appointment length and actual
appointment length and increase patient throughput.
Appointment Time Length Scheduling Guidelines
Once a patient has attended a consultation in the Radiation
Oncology Department, a CT Simulator appointment is scheduled
through the Medical Assistants and Patient Services Associates
department (MAPSA). The length of each appointment is either 30
minutes, 60 minutes, or 90 minutes depending on the anatomical
location of the tumor. The physician indicates the length of the
appointment on the patient’s directive. The physician-recommended
appointment time length is used to schedule the appointment.
CT Simulator Patient Flow
The patient’s first step after diagnosis is to have a
consultation with a physician in the Radiation Oncology Department.
If the physician recommends radiation therapy, the patient is
scheduled immediately for a CT Simulator scan.
The patient is asked to arrive one hour prior to the appointment
start time to complete pre-simulation tasks. The tasks that need to
be completed prior to simulation include a physician signature on
the patient directive, verification of consent, verification of an
IV form if necessary, verification of a face photo, and education
through a 20-minute radiation therapy video. After the patient
checks in at the clerk desk, the clerk reviews the pre-simulation
checklist to see which tasks the patient needs to complete. Then,
the patient is taken to a holding room (Room 10) to change into a
gown and complete the uncompleted tasks. The patient waits in the
holding room until the CT Simulator is prepared.
When the CT Simulator room is prepared, the simulation team
retrieves the patient from the holding room. A team of radiation
therapists, radiation therapy students, and other medical staff
then immobilize the patient for the scan. The comfort of the
patients during the CT scan is extremely important because the
patient must be in the same exact position for every radiation
treatment in order for the treatment to work effectively. Once the
scan is complete, the simulation team pages a physician to approve
the scans. After physician approval, the simulation team takes
photos of the reference marks and patient setup for
reproducibility. After the patient has completed the CT Simulator
scan, the patient will go on a tour of the Radiation Oncology
Department. The purpose of the tour is to show the patients where
they will be coming every day for the next two to four weeks to
receive their radiation therapy treatments. The CT Simulator
patient flow is complex because of the varying medical needs of
each patient. Therefore, the team observed the patient flow process
and focused on the most common delays when designing the improved
patient flow.
Key Issues
The key issues causing CT Simulator inefficiency that drove this
project were:
· Frequent delays in the CT Simulator patient flow process
· Long patient wait time during CT Simulator appointments
· CT Simulator room idle time when schedule is delayed
· High variation in the difference between scheduled appointment
time length and actual appointment time length
Goals and Objectives
The primary goal of this project was to create new appointment
scheduling guidelines for optimizing appointment time lengths and
design a more efficient CT Simulator patient flow process. To reach
this goal, the team completed the following 12 tasks:
· Analyzed historical data
· Attended Radiation Oncology Lean meetings
· Observed current patient flow
· Created current patient flow process map
· Conducted preliminary time studies
· Conducted time studies of the pre-simulation process
· Conducted time studies in the CT Simulator room
· Recorded appointment time length and patient characteristic
data
· Interviewed key team members of the CT Simulator
· Created value stream map for current patient flow
· Designed future state patient flow
· Created new appointment time length scheduling guidelines
The team aimed to achieve the following design objectives using
the qualitative and quantitative information gathered:
· Decrease patient wait times during CT Simulator
appointments
· Decrease CT Simulator room idle time
· Decrease the difference between scheduled appointment time
length and actual appointment time length
· Increase patient throughput
Project Scope
The project included the process of scheduling patients for the
CT Simulator and the process surrounding the CT Simulator, starting
from the moment the patient who will be going through the CT
Simulator enters the Radiation Oncology Department and ending when
the patient leaves the department. The team solely analyzed the
pre-simulation and during-simulation patient flow process for the
CT Simulator.
Any process outside the CT Simulator, such as the MRI Simulator
process and treatment room patient flow, was out of scope. Any
factors that are not within the control of the Radiation Oncology
Department were not considered and were out of scope for this
project, such as transportation services, in-patient procedures,
and other departments.
Appointment Scheduling Guidelines Data Collection and Analysis
Methods
This section describes the data collection methodology and
analysis techniques used to formulate findings and conclusions
leading to recommendations for the appointment scheduling
guidelines. The team created a data collection sheet that the CT
Simulator employees used to collect appointment time length data
and obtained patient characteristic data from the IOE 481
coordinator.
Appointment Time Length Data Collection
An obstacle regarding the appointment scheduling guidelines was
gathering enough data to run a general linear model regression
analysis that identified statistically significant factors. To
mitigate this obstacle and obtain a large sample size, the team
asked the employees in the CT Simulator room to collect data. The
team designed an appointment time length data collection sheet
using a format similar to other forms used in the CT Simulator
room. Therefore, the data collection had minimal interference with
normal employee work flow. The data collection sheet is shown in
Appendix A. The CT Simulator employees collected data for a total
of 255 patients from February 26, 2018 to March 28, 2018. Of the
total 255 patients, 207 data records were used for analysis. The
remaining data records were eliminated due to missing data
values.
To analyze the collected data, the team used general linear
model regression in Minitab to identify the significant factors
affecting appointment time lengths to create new appointment
scheduling guidelines. The team hypothesized that patient
characteristics could be statistically significant in determining
appointment times. This hypothesis was supported by a study
performed on the consideration of patient characteristics to
improve medical resource utilization in a hospital radiology
department [1]. Therefore, the team planned to use a regression
equation to predict the length of the appointment using factors
such as age, weight, and inpatient or outpatient status, in
addition to the anatomical location of the tumor on the body.
Patient Attribute Data Collection
The IOE 481 team’s coordinator pulled patient attribute data
using the patient contact serial numbers (CSNs) for the patients
included in the appointment time length analysis. The data was
pulled from the Health System Warehouse’s activity (ACTV) and
revenue (REVC) data tables [2]. The team used the ACTV height, ACTV
weight, ACTV body mass index (BMI), REVC height, REVC weight, and
REVC BMI. The team requested this information as patient
characteristic data to test whether these factors are statistically
significant in determining appointment time length.
Appointment Scheduling Guidelines Findings and Conclusions
Findings from the appointment time length and patient attribute
data analysis helped the team form conclusions that supported
appointment scheduling recommendations. This section describes the
findings and conclusions for the appointment scheduling
guidelines.
General Linear Model Regression
First, the team used general linear model regression in Minitab
to determine which factors listed in Appendix B are significant for
predicting the actual appointment time length with a significance
level of 0.05. The actual appointment time length is the time
between when the patient enters the CT Simulator room and when the
CT Simulator employees finish cleaning the room.
The team did not have height, weight, and BMI data for all
patients. Therefore, to use Minitab general linear model
regression, the team had to subset the data. The team created two
subsets: (1) patients with ACTV height, ACTV weight, and ACTV BMI
records and (2) patients with REVC height, REVC weight, and REVC
BMI records. For the second subset of data, none of the patients
needed a body fix positioning and, therefore, the body fix factor
was excluded from the analysis.
Using the first subset of data, the team found that the
statistically significant variables were ACTV height, ACTV weight,
whether or not the patient needed IV contrast, whether or not the
patient needed or needed to be assessed for SDX training, the
number of scan locations (isolations), and whether or not the
patient needed to be positioned with a vacuum bag. The output from
the general linear model regression with the first subset is shown
in Appendix B, Figure B-1. The adjusted R-squared, the amount of
variation in the response variable described by the factors tested,
was only 31.82%.
Using the second subset of data, the team found that the
statistically significant variables were age, whether or not the
patient needed IV contrast, whether or not the patient needed or
needed to be assessed for SDX training, and whether or not the
patient needed to be positioned with a vacuum bag. The output from
the general linear model regression with the second subset is shown
in Appendix B, Figure B-2. The adjusted R-squared, the amount of
variation in the response variable described by the factors tested,
was only 40.24%.
Since general linear model regression with the first two subsets
of data yielded low adjusted R-squared values, the team chose to
run a general linear model regression without the height, weight,
and BMI factors. Therefore, all 207 patient data records were
included in the analysis. With the actual appointment time length
as the response variable, the team tested to see whether the
factors listed in Appendix B are significant for predicting the
actual appointment time length with a significance level of 0.05.
The Minitab output is shown in Appendix B, Figure B-3. The team
found that the following variables are statistically
significant:
· IV Contrast (yes or no)
· SDX training (yes/assess or no)
· Number of treatment areas (isos)
· Vacuum Bag (yes or no)
· Qfix (yes or no)
Although these three variables are significant in determining
the actual appointment time length, the adjusted R-squared value,
which is the amount of variation in the response variable described
by the factors tested, is only 34.41%. This adjusted R-squared
value is also low. Therefore, the team concluded that using a
linear regression equation to accurately determine appointment time
lengths for patients in the Michigan Medicine Radiation Oncology
Department is infeasible. However, because these variables are
statistically significant, the department should consider the five
factors above when scheduling appointments.
Statistical Analysis
Due to the infeasibility of a linear regression equation to
predict appointment time lengths, the team used general statistical
analysis to make recommendations for appointment time lengths. The
appointment time length data was analyzed using Microsoft Excel.
First, the team split the data by scheduled appointment time length
and identified the mean and standard deviation of actual
appointment time lengths, shown in Table 1 below. The actual
appointment time length is the time between when the patient enters
the CT Simulator room and when the CT Simulator employees finish
cleaning the room.
Table 1: Mean and Standard Deviation of Actual Appointment Time
Lengths
Scheduled Length (minutes)
Mean of Actual Length (minutes)
Standard Deviation of Actual Length (minutes)
% of Appointments Shorter Than Scheduled Time
% of Appointments Longer Than Scheduled Time
30
29.11
11.21
62%
38%
60
35.05
18.35
85%
15%
90
62.10
20.33
83%
17%
Sample size: 207; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 1, on average, the actual appointment time
lengths are lower than the schedule appointment time lengths.
Hence, the CT Simulator room is experiencing idle time when time is
scheduled for appointments but not used. However, the standard
deviations of the actual appointment lengths from the scheduled
lengths are wide which indicates that the actual time lengths vary.
Therefore, the team decided to break down the analysis further by
looking at appointments grouped by scheduled time length.
30-Minute Appointments
First, the team created a Pareto chart, shown in Figure 1 below,
of the total positive difference between the scheduled appointment
time and actual appointment time by scan type for the appointments
scheduled in 30-minute time slots. Therefore, the y-axis in the
graph displays the amount of extra time available for additional
appointments during the data collection period by 30-minute
appointment type. The type of scan is determined by the area of the
body that needs radiation therapy treatment.
Figure 1: 30-Minute Appointment Types Contributing to Unused
Scheduled Appointment Time
Sample size: 47; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
In order to identify the types of appointments that were most
important to the CT Simulator schedule, the team used the Pareto
Principle which states that only a few appointment types contribute
to a large percentage of the CT Simulator appointment scheduling
optimization issue. Therefore, the team focused on the spine
appointments, which contribute to a large amount of the extra time
available, as shown in the graph in Figure 1. From this analysis,
the team calculated the mean and standard deviation of the spine
appointments, displayed in Table 2 below.
Table 2: 30-Minute Appointment Type with Highest Unused
Scheduled Time
Appointment Type
Mean of Actual Time Length (minutes)
Standard Deviation of Actual Time Length (minutes)
Spine
21.40
5.08
Sample size: 5; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 2, the mean of the actual appointment time
lengths is 21.40 minutes and the standard deviation is 5.08 minutes
for spine appointments. The team concluded that the spine
appointments could, on average, provide an additional 10.75 minutes
in the CT Simulator room each week. Since this time is minimal, the
team concluded that the 30 minutes currently scheduled for spine
appointments is appropriate.
60-Minute Appointments
For the appointments scheduled for 60 minutes, approximately 50%
of appointments lasted 30 minutes or less and 77% of appointments
lasted 45 minutes or less. The team created a Pareto chart, shown
in Figure 2, of the total positive difference between the scheduled
appointment time and actual appointment time by scan type for the
appointments scheduled in 60-minute time slots. Therefore, the
y-axis in Figure 2 displays the amount of extra time available for
additional appointments during the data collection period by
60-minute appointment type.
Figure 2: 60-Minute Appointment Types Contributing to Unused
Scheduled Appointment Time
Sample size: 131; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
Similar to the analysis with the 30-minute appointments, in
order to identify the types of appointments that were most
important to the CT Simulator schedule, the team used the Pareto
Principle. This principle states that only a few appointment types
contribute to a large percentage of the CT Simulator appointment
scheduling optimization issue. Therefore, the team decided to
investigate the distributions of the five appointment types
contributing to the most extra time: head/neck, prostate, lung (no
SDX), spine, and pelvis. The team calculated the mean and standard
deviation of these five appointment types, displayed in Table 3
below.
Table 3: 60-Minute Appointment Types with Highest Unused
Scheduled Time
Appointment Type
Mean of Actual Time Length (minutes)
Standard Deviation of Actual Time Length (minutes)
Head/Neck
36.17
14.43
Prostate
30.35
22.92
Lung (No SDX)
28.09
11.41
Spine
26.50
10.60
Pelvis
31.67
16.83
Sample size: 69; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 3 above, the mean appointment times are
significantly lower than the 60-minute scheduled time. Each week,
these five appointment types contribute to over eight hours of
extra time collectively, as shown in Table 4 below.
Table 4: 60-Minute Appointment Types Unused Scheduled Time per
Week
Appointment Type
Unused Scheduled Appointment Time per Week (hours)
Head/Neck
2.38
Prostate
2.10
Lung (No SDX)
1.46
Spine
1.12
Pelvis
1.02
Sample size: 69; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
Given the large amount of opportunity available, the team
analyzed the individual actual time length distributions for each
appointment type. First, the team created a histogram of actual
appointment time lengths for the head and neck appointments to
determine the appropriate time length to schedule for the
appointments, shown in Figure 3 below.
Figure 3: Head and Neck Appointment Actual Time Length
Distribution
Sample size: 24; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in the histogram above, most of the appointments are
shorter than the scheduled 60-minute appointment time. Table 5
below shows the percentage of head and neck appointments above and
below the scheduled time currently and the effect of changing the
appointment time length to 45 minutes.
Table 5: Head and Neck Appointment Scheduled Time Comparison
Scheduled Time
% of Appointments Shorter Than Scheduled Time
% of Appointments Longer Than Scheduled Time
60 Minutes (Current)
96%
4%
45 Minutes (New)
75%
25%
Sample size: 24; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 5 above, 75% of appointments would still
finish earlier than expected if the scheduled appointment time were
45 minutes instead of 60 minutes. Although 25% of the appointments
would be finishing after the scheduled time, the CT Simulator room
idle time would be decreased and the department would be able to
schedule more head and neck patients.
Second, the team analyzed the distribution of actual appointment
time length for the prostate appointments. A histogram of the
actual appointment times observed is shown in Figure 4 below.
Figure 4: Prostate Appointment Actual Time Length
Distribution
Sample size: 17; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
After finding that most of the prostate appointments finish
before 60 minutes, as shown in Figure 4 above, the team
investigated the expected effect of reducing the scheduled
appointment time from 60 minutes to 45 minutes, as shown in Table 6
below.
Table 6: Prostate Appointment Scheduled Time Comparison
Scheduled Time
% of Appointments Shorter Than Scheduled Time
% of Appointments Longer Than Scheduled Time
60 Minutes (Current)
88%
12%
45 Minutes (New)
71%
29%
Sample size: 17; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 6 above, the team expects that 71% of the
prostate appointments would finish early if the scheduled
appointment time length was 45 minutes. Although 29% of the
appointments would finish late, the extra time would allow patient
throughput to increase. However, since the prostate patients
require a full bladder for the CT Simulator scan, the department
would need to closely communicate with MAPSA and the patients prior
to the appointment to ensure the patient drinks at the correct time
to avoid delays.
Third, the team investigated the distribution of the actual
appointment time lengths for lung (no SDX) appointments, as shown
below in Figure 5.
Figure 5: Lung (No SDX) Appointment Actual Time Length
Distribution
Sample size: 11; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
After observing that all of the lung (no SDX) appointments
finished before the 60 minutes scheduled, the team analyzed the
anticipated effect of changing the scheduled appointment time from
60 minutes to 45 minutes, shown in Table 7 below.
Table 7: Lung (No SDX) Appointment Scheduled Time Comparison
Scheduled Time
% of Appointments Shorter Than Scheduled Time
% of Appointments Longer Than Scheduled Time
60 Minutes (Current)
100%
0%
45 Minutes (New)
100%
0%
Sample size: 11; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown above in Table 7, 100% of lung (no SDX) appointments
would be expected to finish early if the appointment time length
was scheduled as 45 minutes. Therefore, the team expects that
decreasing the scheduled time would allow the department to
schedule more lung (no SDX) appointments and increase patient
throughput.
Fourth, a histogram for the actual appointment time lengths for
spine appointments was analyzed, as shown in Figure 6 below.
Figure 6: Spine Appointment Actual Time Length Distribution
Sample size: 8; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in the histogram above, all of the appointments were
shorter than the scheduled 60-minute appointment time. Table 8
below shows the percentage of spine appointments above and below
the scheduled time currently and the effect of changing the
appointment time length to 45 minutes.
Table 8: Spine Appointment Scheduled Time Comparison
Scheduled Time
% of Appointments Shorter Than Scheduled Time
% of Appointments Longer Than Scheduled Time
60 Minutes (Current)
100%
0%
45 Minutes (New)
100%
0%
Sample size: 8; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 8 above, 100% of appointments would still be
expected to finish earlier than the scheduled appointment time if
the scheduled appointment time was changed to 45 minutes.
Therefore, the CT Simulator room idle time would be decreased and
the department would be able to schedule more spine appointments
and increase patient throughput.
Fifth, the team analyzed the distribution of actual appointment
time lengths for the pelvis appointments. A histogram of the actual
appointment times observed is shown in Figure 7 below.
Figure 7: Pelvis Appointment Actual Time Length Distribution
Sample size: 9; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in the histogram above, most of the appointments were
shorter than the scheduled 60-minute appointment time. Table 9
below shows the the effect of changing the appointment time length
to 45 minutes.
Table 9: Pelvis Appointment Scheduled Time Comparison
Scheduled Time
% of Appointments Shorter Than Scheduled Time
% of Appointments Longer Than Scheduled Time
60 Minutes (Current)
100%
0%
45 Minutes (New)
78%
22%
Sample size: 9; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 9 above, the team expects that 78% of the
pelvis appointments would finish early if the scheduled appointment
time length was 45 minutes. Although 22% of the appointments would
finish late, the extra time would allow patient throughput to
increase through additional pelvis appointments.
90-Minute Appointments
Following the same analysis methodology used for the 30-minute
and 60-minute appointments, the team created a Pareto chart, shown
in Figure 8 below, of the total positive difference between the
scheduled appointment time and actual appointment time by scan type
for the appointments scheduled in 90-minute time slots. The y-axis
in Figure 8 displays the total positive amount of extra time
available for additional appointments during the data collection
period by 90-minute appointment type.
Figure 8: 90-Minute Appointment Types Contributing to Unused
Scheduled Appointment Time
Sample size: 29; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
The team used the Pareto Principle which states that only a few
appointment types contribute to a large percentage of the CT
Simulator appointment scheduling optimization issue. The team found
that liver and abdomen appointments contribute to a large percent
of the extra time available, as shown in Figure 8. Following this
finding, the team calculated the mean and standard deviation of the
liver and abdomen appointments, displayed in Table 10 below.
Table 10: 90-Minute Appointment Type with Highest Unused
Scheduled Time
Appointment Type
Mean of Actual Time Length (minutes)
Standard Deviation of Actual Time Length (minutes)
Liver
66.09
19.56
Abdomen
43.40
13.94
Sample size: 17; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 10 above, the mean appointment times were
significantly lower than the 90-minute scheduled time. Each week,
these appointments contribute to almost two hours of extra time
collectively, as shown in Table 11 below.
Table 11: 90-Minute Appointment Types Unused Scheduled Time per
Week
Appointment Type
Unused Scheduled Appointment Time per Week (hours)
Liver
1.10
Abdomen
0.97
Sample size: 17; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
Given the opportunity for increased patient throughput with the
extra time available shown in Table 11 above, the team analyzed the
individual actual time length distributions for each appointment
type.
First, the team created a histogram of actual appointment time
lengths for the liver appointments to determine the appropriate
time length to schedule for the appointments, shown in Figure 9
below.
Figure 9: Abdomen Appointment Actual Time Length
Distribution
Sample size: 12; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Figure 9, the majority of the appointments observed
during data collection were shorter than the scheduled 90-minute
appointment time. Table 12 below shows the percentage of liver
appointments above and below the scheduled time currently and the
effect of changing the appointment time length to 75 minutes.
Table 12: Abdomen Appointment Scheduled Time Comparison
Scheduled Time
% of Appointments Shorter Than Scheduled Time
% of Appointments Longer Than Scheduled Time
90 Minutes (Current)
83%
17%
75 Minutes (New)
58%
42%
Sample size: 12; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 12, the team expects 58% of the appointments
to end early if the appointment time length is changed to 75
minutes. Although 42% would still finish late, the change would
decrease overall CT Simulator room idle time and the department
would be able to schedule more liver appointments annually.
Second, the team analyzed the distribution of actual appointment
time length for the abdomen appointments. A histogram of the actual
appointment times observed is shown in Figure 10 below.
Figure 10: Abdomen Appointment Actual Time Length
Distribution
Sample size: 5; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
After finding that all abdomen appointments finished before 90
minutes, as shown in Figure 10 above, the team investigated the
expected effect of reducing the scheduled appointment time from 90
minutes to 75 minutes, as shown in Table 13 below.
Table 13: Abdomen Appointment Scheduled Time Comparison
Scheduled Time
% of Appointments Shorter Than Scheduled Time
% of Appointments Longer Than Scheduled Time
90 Minutes (Current)
100%
0%
75 Minutes (New)
100%
0%
Sample size: 5; Source: University of Michigan Radiation
Oncology Department; Time Period: February 26, 2018 to March 28,
2018
As shown in Table 13 above, the team expects that 100% of the
abdomen appointments would finish early if the scheduled
appointment time length was 75 minutes. Therefore, scheduling
abdomen appointments for 75 minutes would reduce CT Simulator room
idle time and would allow patient throughput to increase.
Appointment Scheduling Guidelines Recommendations
This section describes the recommendations the team has for the
Radiation Oncology Department’s appointment scheduling guidelines.
First, the recommendations resulting from the general linear model
regression are described. Then, the section details the
recommendations for reducing scheduled appointment time lengths for
specific appointment types.
Consider Five Significant Factors in Appointment Scheduling
The general linear model regression showed that the following
factors were statistically significant in determining the
appointment time length:
· IV Contrast (yes or no)
· SDX training (yes/assess or no)
· Number of treatment areas (isos)
· Vacuum Bag (yes or no)
· Qfix (yes or no)
Although a linear function cannot be used to accurately predict
the actual appointment times, the Michigan Medicine Radiation
Oncology Department should consider these five factors when
scheduling. Currently, the IV contrast, SDX training, vacuum bag,
and Qfix are considered. The team recommends that the department
require multiple appointments for patients with more than one
treatment area (iso). This process would also require physicians to
complete a simulation directive for each treatment area. By
considering these factors, the department will be able to collect
data for using regression to accurately predict appointment time
lengths in the future.
Pilot Reducing Appointment Time Length Increments from 30 to 15
Minutes
After performing statistical analysis on the appointment time
length data collected from February 26, 2018 to March 28, 2018, the
team recommends that the Michigan Medicine Radiation Oncology
Department reduce specific appointment time lengths to minimize the
amount of CT Simulator room idle time, decrease the difference
between scheduled appointment length and actual appointment length,
and increase patient throughput. The team found that few
appointment types contribute heavily to the unused scheduled time
in the department. Therefore, the team recommends that the
department take measures to reduce the appointment time increments
from 30 minutes to 15 minutes.
In the short-term, the team recommends that the department
perform a pilot test for reducing the time increments from 30
minutes to 15 minutes. For the pilot, the team suggests the
department reduce the scheduled time for head and neck, prostate,
lung (no SDX), spine, and pelvis appointments from 60 minutes to 45
minutes. Additionally, the team recommends that the department
reduce the scheduled time for liver and abdomen appointments from
90 minutes to 75 minutes.
After initiating a pilot of these new appointment time lengths,
the team recommends that the Radiation Oncology Department continue
to collect appointment time length data for all appointment types.
The team predicts an additional three to four months of data will
be needed for full analysis of all appointment types and for full
integration of 15-minute appointment time length scheduling
increments.
Patient Flow Data Collection and Analysis Methods
This section describes the data collection methodology and
analysis techniques used to formulate findings and conclusions
leading to recommendations for the CT Simulator patient flow. The
team analyzed historical data, performed benchmarking research and
observations, created a patient flow process map, collected time
study data, and designed a value stream map.
Historical Data Analysis
The Radiation Oncology Department collected data in the CT
Simulator room from January 5, 2018 to January 18, 2018. The team
was provided with this historical data for 102 patients. The CT
Simulator employees kept handwritten notes regarding the delays
that occurred during the patient’s appointment, the time the
patient entered the CT Simulator room, and the time the patient
exited the CT Simulator room. Of the 102 patients in the CT
Simulator room during the data collection period, 58 patients
experienced some type of delay.
The team used this data to determine root causes of CT Simulator
room delays during the period that the data was collected. The team
analyzed this information by categorizing the delays into delay
types. Using Microsoft Excel, the graph shown in Appendix C was
created from this analysis to understand the relative frequency of
each delay type.
Patient Flow Benchmarking
The team performed benchmarking research to understand how CT
Simulator patient flow is organized at other hospitals in the
United States. The team found a research study from the Ohio State
University Medical Center’s James Department of Radiation Oncology.
The study was performed in collaboration with the Academy for
Excellence in Healthcare in August 2014 [3]. The team used the
findings from the benchmarking research to inform recommendations
for alternative patient flows in the Michigan Medicine Radiation
Oncology Department.
Observations
To understand the current CT Simulator patient flow process and
detect possible bottlenecks and areas for improvement, the team
observed the CT Simulator patient flow process for 32 patients from
January 26, 2018 to February 23, 2018. During the observations, two
team members were present. One team member sat at the front desk
and took notes when the patient checked into the department to
track the pre-simulation checklist tasks. The second team member
sat in the CT Simulator control room to observe and take notes on
any delays or variations during the scan. The notes were taken by
hand and the team members switched places halfway through each
observation period. At the end of the observation period, notes
were compared to understand the full patient flow process. The
observations were used to create the patient flow process map,
shown in Appendix D, and the time study data collection sheet,
shown in Appendix E.
CT Simulator Patient Flow Process Map
One team member attended the Radiation Oncology Department’s
Lean meetings on Tuesday, January 30, 2018 and Tuesday, February 6,
2018 from 9:30 AM to 10:30 AM in the Radiation Oncology Department
at the University of Michigan Hospital. The clients, the Director
of Clinical Operations and the Chief Radiation Therapist,
participated in the meeting along with staff across various groups
in the department such as Physics, Dosimetry, and Medical
Assistants and Patient Services Associates. The employees at the
meeting mapped out the current patient flow process for the CT
Simulator using post-it notes on the walls of the conference
room.
The IOE 481 project team utilized the employee-perspective
patient flow process in addition to their observations to create a
process map for the CT Simulator patient flow process in Microsoft
Visio. The patient flow process map is shown in Appendix D. The
process map begins when the patient is ready to be taken into the
CT Simulator room. The process map was used as a framework for the
value stream map.
Time Studies
To record task times for use in the value stream map, the team
collected time study data in the Radiation Oncology Department from
March 5, 2018 to March 30, 2018. The team used the data collection
sheet shown in Appendix E. Time study data was collected for 90
patients during a total of 102 hours in the department.
An obstacle regarding time studies was that the team members had
to observe for long time periods to collect data because the CT
Simulator appointments are either 30, 60, or 90 minutes. This
obstacle was mitigated by staggering the team members’ observation
times so the team could record data for a maximum number of
patients. Each team member recorded time study data independently
each week. The team member stood in the CT Simulator control room
and recorded data on the time study data collection sheet. Every
hour, the team member went to the clerk desk and requested time
data from the MAPSA employees regarding when the patient checked
in, when the patient was moved to the holding room, and what
pre-simulation tasks needed to be completed.
The team entered the time study data in Microsoft Excel and
utilized general statistical analysis to obtain the metrics needed
to create a value stream map for the patient flow.
Value Stream Map
The team created a value stream map to identify the current
bottlenecks in the patient flow process, which led to
recommendations for the future state patient flow process. The team
used a 2010 IOE 481 team’s patient flow value stream map as a
reference for the CT Simulator value stream map [4]. The value
stream map is shown in Appendix F. Ultimately, the value stream map
was used to identify key bottlenecks in the current process so the
team could make useful and specific recommendations when designing
the improved patient flow process.
Patient Flow Findings and Conclusions
Findings from the historical data analysis, benchmarking,
process map, and value stream map helped the team form conclusions
that informed the patient flow recommendations. This section
describes the findings and conclusions for the CT Simulator patient
flow.
Historical Delays
After analyzing the historical data using the bar graph in
Appendix C, the team found that the patient was the cause of the
greatest number of delays, including showing up late to the
appointment time, asking many questions before the simulation,
using the bathroom multiple times, drinking contrast slowly, and
having pain or discomfort. This analysis, in addition to a review
of the Radiation Oncology delays identified in an IOE 481 report
from 2007 [5] and IOE 481 report from 2012 [6], helped the team
conclude that the patient flow process varies by patient and is
difficult to standardize. Therefore, the team needed to make
general recommendations that would still allow all patient needs to
be satisfied.
Induction Room Benchmarking Research
After benchmarking research, the team found that the James
Department of Radiation Oncology at the Ohio State University
Medical Center recently changed its CT Simulator patient flow
process after a patient flow improvement study with the Academy for
Excellence in Healthcare [3]. The result of the project was a
patient flow change to a two-step immobilization process for CT
Simulator appointments. The first step is to bring the patient to
an induction room where the patient is immobilized with several
devices. The second step is to bring the patient to the CT
Simulator room for the scan. The researchers expect that this
two-step process will optimize the CT Simulator room so patients
can be scanned while complicated immobilizations are prepared in
the induction room. The hospital predicts that this improved
patient flow will allow 14 appointments in an 8-hour day.
Therefore, the IOE 481 team concluded that an induction room could
be an option for increasing patient throughput in the Michigan
Medicine Radiation Oncology Department.
Main CT Simulator Patient Flow Tasks
From the patient flow process map shown in Appendix D, the team
found that the patient flow process is complicated by various
conditional tasks. For example, the physician, physics, or
dosimetrists can be called at any point in the patient positioning
and scanning process. Therefore, the team concluded that the value
stream map must only include the main tasks for effective analysis
purposes. Therefore, the team concluded that the process has 11
main tasks, which are listed below.
1. Patient check-in
2. Completion of pre-simulation tasks in Room 10
3. Team notified patient ready
4. Patient brought to CT Simulator room
5. Patient moved to CT Simulator table
6. Pre-scan set up and patient positioning
7. Scout
8. Scan
9. Physician scan approval
10. Patient exits CT Simulator room
11. CT Simulator room cleaned
These 11 tasks were used to create the value stream map which is
discussed in the following subsection.
Patient Flow Bottlenecks
The team used the time study data for 90 patients to create a
value stream map for the CT Simulator patient flow process, shown
in Appendix F. The team calculated the average process time, wait
time, and first time quality, for each step in the patient flow.
Overall, on average, the patient flow process time was 62 minutes,
the wait time was 48 minutes, the lead time was 110 minutes, and
the total quality was 42%. Table 14 below lists the wait time
(W/T), process time (P/T), and first-time quality (FTQ) for the 11
main tasks in the patient flow process.
Table 14: Value Stream Map Process Time, Wait Time, and First
Time Quality Values
Task
P/T (minutes)
W/T (minutes)
FTQ (%)
Patient check-in
1.00
22.04
100.00%
Completion of pre-sim tasks in Room 10
25.00
13.94
66.30%
Team notified patient ready
1.00
8.74
100.00%
Patient brought to CT Simulator room
2.00
0.00
93.48%
Patient moved to CT Simulator table
1.27
0.00
98.81%
Pre-scan set up and patient positioning
11.68
0.00
94.57%
Scout
3.98
0.00
84.78%
Scan
3.24
0.00
93.48%
Physician scan approval
2.13
3.24
91.30%
Patient exits CT Simulator room
4.87
0.00
100.00%
CT Simulator room cleaned
5.41
0.00
100.00%
From the information in Table 14 above, the team identified
three bottlenecks based on high process time, high wait time, or
low first time quality. The three bottlenecks were later used to
formulate patient flow recommendations.
The team identified patient check-in as the first bottleneck for
patients. On average, patients wait approximately 22.04 minutes
before they are brought to Room 10 for completion of pre-simulation
tasks. Therefore, the team decided to focus on this bottleneck when
making patient flow improvement recommendations for the department
to decrease patient wait time.
The completion of the pre-simulation tasks in Room 10 is the
second bottleneck, where the average process time is 25.00 minutes,
wait time is 13.94 minutes, and first time quality is rated at
66.30%. The first time quality for this task is the percent of
patients without any possible source of delay present including
consent, face photo, or other pre-simulation tasks. Therefore, the
team concluded that patient throughput and patient wait time could
be reduced by addressing this bottleneck.
The value stream map informed the team that the pre-scan set up
and patient positioning was the third bottleneck because of the
high process time of 11.68 minutes on average. Therefore, the team
concluded that this step in the patient flow process could be
improved to increase patient throughput in the department.
Patient Flow Design Methods, Requirements, Constraints, and
Standards
The main design task is to create an improved patient flow
process for the CT Simulator. This section describes the design
requirements, constraints, and standards for the CT Simulator
patient flow that were used to create the alternative patient
flows. The design requirements, constraints, and standards were
identified through discussions with department stakeholders,
observations, and online research. The alternative patient flows
were analyzed using a Pugh Matrix. The three tables in Appendix G
form the project’s constraints and standards matrix for the new
patient flow process.
Soft Design Constraints (Design Requirements)The main design
task was the improved patient flow process. When designing this
process, several design requirements guided the team’s approach.
The design requirements are listed below and were used to evaluate
the alternative patient flows. The labels following the
descriptions refer to an entry in the constraints and standards
matrix in (Appendix G, Table G-1).
1. The new patient flow process should incur minimal costs to
the Radiation Oncology Department. (R-D-1)
2. The new patient flow process should be easily implemented in
the department and should not require intensive training.
(R-E-1)
3. The new patient flow process should minimize interaction with
new departments or new staff that are not already included in the
CT Simulator patient flow. (R-E-2)
4. The CT Simulator staff should accept the new process in order
for it to be successfully implemented and sustained. (R-F-1)
5. The new CT Simulator patient flow process should be
acceptable to patients because the patient flow directly affects
the experience of the patients in the Radiation Oncology
Department. (R-G-1)
6. The average patient wait time should decrease as a result of
the new patient flow process. (R-H-1)
7. The CT Simulator room idle time should decrease as a result
of the new patient flow process. (R-H-2)
8. The CT Simulator throughput should increase as a result of
the new patient flow process. (R-H-3)
Hard Design ConstraintsThe team identified several hard
constraints that were adhered to when creating the new patient flow
process. The hard constraints are listed below. The labels
following the descriptions of the hard constraints refer to an
entry in the constraints and standards matrix (Appendix G, Table
G-2).
1. The new patient flow process must follow specific
organizational policies. These policies require the patient to stay
on the CT Simulator table until the physician approves the scan and
require signed consent and a complete physician directive before
the patient enters the simulation room. (C-A-1)
2. The new patient flow process must not violate the health or
safety of patients. Therefore, the patient lift team, the
anesthesiology team, the IV team, and other hospital departments
will not be removed from the patient flow process. (C-C-1)
3. An additional CT Simulator cannot be purchased due to budget
and space limitations. (C-D-1)
4. The new patient flow process must not disrupt any of the
other processes in the Radiation Oncology Department. (C-E-1)
5. All process maps must be designed in Microsoft Visio so the
Radiation Oncology Department can update the maps after the IOE 481
project is complete. (C-F-1)
6. Observations can only occur Monday through Friday from 7:00
AM to 6:00 PM, the days and times that the CT Simulator is open.
(C-H-1)
7. The new patient flow process must be applicable to the CT
Simulator Monday through Friday from 7:00 AM to 6:00 PM, the days
and times that the CT Simulator is open. (C-H-2)
8. All observations and data collection must be completed within
the University of Michigan semester time frame, allowing a final
report including the new patient flow process to be delivered to
the client by April 25, 2018. (C-H-3)
Design StandardsThe standards the team followed when designing
the new patient flow process include HIPPA, Radiation Oncology
Department standards, Michigan Medicine best practices, IEEE, and
MLearning. The implications of the standards on the project are
listed below. The labels following the descriptions of the
standards refer to an entry in the constraints and standards matrix
(Appendix G, Table G-3).
1. When collecting and analyzing the Radiation Oncology
simulation data, the team must follow the Health Insurance
Portability and Accountability Act of 1996 (HIPPA) standard. HIPPA
includes a Privacy Rule “to assure that individuals’ health
information is properly protected while allowing the flow of health
information needed to provide and promote high quality health care
and to protect the public's health and well being” [7]. Following
this standard involves de-identifying patient data to refrain from
potentially leaking protected health information (PHI). To protect
PHI, the team and clients have access to a University of Michigan
MBox folder to allow safe data and file sharing. When collecting
time study data for the new patient flow process, the team has been
labeling each patient with a unique random identifier and does not
record any data such as gender, age, or weight. (S-1-1)
2. The Radiation Oncology Department has several standards of
work surrounding the CT Simulator process that must be followed
during the new patient flow process design. These departmental
standards include processes surrounding patient education prior to
the simulation, physician scan approval, and patient positioning
techniques. These organizational standards were communicated by the
team’s clients during the initial client meeting on January 19,
2018. (S-2-1)
3. The team’s coordinator, Zac Costello, provided the team with
best practice templates for process maps and value stream maps. The
team used these resources to construct the current patient flow
process map and plans to utilize them again when designing the
value stream map. The resources were provided to the team in an
email following a coordinator meeting on February 15, 2018 [8].
(S-3-1)
4. The Radiation Oncology Department uses Microsoft Visio to
construct process maps within the department. Therefore, the
clients requested that the team use Microsoft Visio as a best
practice to enable the department to update the maps after the IOE
481 project ends. (S-3-2)
5. When documenting the final report with the new patient flow
process design, the team will use the Institute of Electrical and
Electronics Engineers (IEEE) standards to cite literature used for
analysis and developing recommendations [9]. (S-4-1)
6. The team has been certified through MLearning, the Michigan
Medicine training portal on compliance, patient safety, and other
important policies within the hospital. The team must adhere to
these standards when observing, collecting data, and creating the
new patient flow process [10]. (S-5-1)
The team searched the websites of OSHA, MiOSHA, NIOSH, ASTM,
ANSI, SAE, and Mil-Specs on March 12, 2018 from 6:00 PM to 6:30 PM
using the key words “patient,” “patient flow,” “data,” and “data
collection” but did not find these to be relevant standards.
Patient Flow RecommendationsThe team used the findings from the
value stream map in addition to the design requirements,
constraints, and standards to formulate the CT Simulator patient
flow recommendations. First, the team identified four patient flow
recommendations associated with the three bottlenecks identified. A
future state map with kaizen bursts is shown in Appendix H. Then,
the team used a Pugh Matrix with the design requirements as
criteria to determine the preferred patient flow alternative for
the Michigan Medicine Radiation Oncology Department CT Simulator.
This section describes the patient flow alternatives and the
decision process leading to the final recommendation.
Designate Patient Flow Overflow Room
The first recommendation addresses the wait time after the
patient checks in, prior to being brought to Room 10. This
alternative patient flow includes designating a second room as a CT
Simulator patient overflow room. This room could still be utilized
by other patients, but it would be set as a higher priority for CT
Simulator patients. Therefore, this room would only be used by
non-CT Simulator patients if there were no other rooms available.
With this patient flow, two CT Simulator patients could be
completing pre-simulation tasks that were not completed prior to
arrival at the same time. Therefore, the overall patient wait time
would be decreased.
Watch Educational Videos Prior to Room 10
The second recommendation for the patient flow process is to
ensure that patients watch the necessary educational videos prior
to going into Room 10 to decrease patient wait time and delays. The
team recommends that patients watch the videos with a nurse,
doctor, or physician before leaving their initial consultation
appointment, instead of in Room 10 before their CT Simulator
appointment. Another option could be to have the patient watch the
educational videos in the waiting room after check-in on a portable
tablet such as an iPad. The current average wait time after
check-in before the patient is brought to Room 10 is 22.04 minutes.
Given that the main educational video is 20 minutes long, ensuring
that patients watch the video in the waiting room would be an
efficient utilization of space and patient wait time.
Assign Pre-Simulation Checklist to MAPSA
The third recommendation is for MAPSA employees to be assigned
the task of preparing a “next day patient missing questionnaire”
list that highlights pre-simulation tasks needed for all upcoming
CT Simulator patients. Currently, this task is led by a CT
Simulator staff member. Since MAPSA employees are responsible for
ensuring the pre-simulation tasks are complete prior to the patient
entering the CT Simulator room, the team believes assigning this
task to MAPSA employees would better prepare the department for
delay mitigation.
Implement Induction Room
The final alternative patient flow recommendation is for the
department to implement a second patient positioning room similar
to the two-step CT Simulator patient flow process at the Ohio State
University Medical Center [3]. This process was identified through
departmental benchmarking. In the two-step process, an additional
room is available for complex immobilizations. After the patient is
immobilized properly in the positioning room, the patient would be
transported to the CT Simulator scan room. This process would
decrease CT Simulator room idle time and increase patient
throughput.
Pugh Matrix Recommendation
After creating alternatives for a new patient flow process, the
team used a Pugh Selection Matrix, shown in Appendix I, to weigh
the advantages of the new patient flow process alternatives when
compared to the datum, the current process. The team identified the
requirements of cost, training required, number of outside
resources required, CT Simulator employee acceptability, patient
acceptability, patient wait time, CT Simulator room idle time, and
patient throughput as the design requirements for analysis. The
weights for the requirements are shown in the Overall Importance
column in the Pugh Selection Matrix. The weights represent
percentages out of 100%. Each alternative was given a score of +1,
0, or -1 for each criterion compared to the datum.
As shown in Appendix I, the highest weighted alternative was the
pre-watch video alternative. Therefore, the team highly recommends
that the Radiation Oncology Department design the patient flow for
patients to watch the educational video prior to entering Room 10.
This change would reduce the amount of time the patient is in Room
10 and reduce the number of delays in the department. Additionally,
the team recommends that the department have the patients watch the
SDX video prior to the general video because the SDX video is a
higher priority. The second highest weighted alternative was the
MAPSA questionnaire preparation alternative. Therefore, MAPSA
should take over the role of contacting patients for the next day
and tracking incomplete pre-simulation tasks to reduce patient flow
delays.
Expected Impact
This section describes the expected impact of the team’s
recommendations for the Michigan Medicine Radiation Oncology
Department regarding the CT Simulator appointment scheduling
guidelines and the CT Simulator patient flow.
New Appointment Scheduling Guidelines
If the Michigan Medicine Radiation Oncology Department reduces
the scheduled appointment time length increments from 30 minutes to
15 minutes as recommended, the team expects that the patient
throughput will be maximized and the CT Simulator room idle time
and the difference between the scheduled time length and actual
time length will be minimized. Therefore, the team expects that
these changes will directly address the current key issues
associated with the appointment scheduling guidelines.
From February 26, 2018 to March 28, 2018, the appointment time
length observation period, the team calculated the expected number
of additional appointments that could be scheduled as a result of
the pilot if the scheduled appointment time lengths were reduced as
recommended. The team aggregated the accumulated unused scheduled
appointment time for the appointments recommended to be included in
the pilot and found that the pilot is expected increase patient
throughput by 288 patients annually. The appointment type
distribution of the expected 288 extra appointments is listed
below.
· 96 Head/Neck
· 60 Prostate
· 36 Lung (No SDX)
· 24 Spine
· 36 Pelvis
· 24 Liver
· 12 Abdomen
Furthermore, the 288 additional appointments would decrease CT
Simulator room idle time and the difference between the scheduled
appointment length and actual appointment length by 234 hours.
Improved Patient Flow
The CT Simulator future state patient flow map, shown in
Appendix H, was designed based on the four patient flow
recommendations for eliminating the patient flow bottlenecks.
The recommendation of opening a second room for CT Simulator
patients would decrease the wait time for patients between checking
in and moving to Room 10 because an additional room would be
available for the patient to complete the pre-simulation tasks if
someone was occupying Room 10 at the time of check-in. The expected
impact of having this overflow room for CT Simulator patients is a
decrease in wait time after check-in to five minutes. The five
minutes is estimated as the amount of time needed for the MAPSA
employee to prepare the second room and bring the patient in.
The recommendation for the patient to complete any
pre-simulation educational videos after their consultation
appointment or on a tablet in the waiting room prior to entering
Room 10 would reduce the process time of the patient in Room 10 to
ten minutes because the patient would no longer need to watch the
20-minute educational video in Room 10. The only tasks necessary to
complete in Room 10 would be talking with a nurse, changing
clothes, and, if necessary, other tasks such as contacting the IV
team or drinking contrast. Due to factors outside of Room 10, the
wait time decrease cannot be estimated.
Next, the recommendation for MAPSA employees to be assigned the
task of preparing a “next day patient missing questionnaire” list
that highlights pre-simulation tasks needed for all upcoming CT
Simulator patients would increase the visibility of patients with
outstanding pre-simulation tasks. Since MAPSA employees are
responsible for bringing the patients to Room 10 and ensuring
completion of pre-simulation tasks, the team expects the first time
quality percentage in Room 10 to increase to 90%. The team does not
expect 100% first time quality because some delay sources are
outside the control of MAPSA and the CT Simulator team, such as a
myelogram recovery delay.
For the two-step immobilization patient flow process
recommendation based on the Ohio State University Medical Center
study, the team was unable to predict the expected changes in
process time and wait time because of the scale of the change.
However, since the CT Simulator room would only be needed for the
scan, which takes 3.24 minutes on average, the team expects that
patient throughput would increase.
Overall, the expected impact of the improved patient flow is
that the process time will decrease by 24%, from 62 minutes to 47
minutes, primarily from the decrease in time spent in Room 10
watching the educational videos. The patient wait time is expected
to decrease by 35%, from 48 minutes to 31 minutes, after removing
unnecessary patient wait time in the check-in process. The lead
time is expected to decrease by 29%, from 110 minutes to 78
minutes. The first time quality is expected to increase from 42% to
57%, since patients would experience fewer delays when a MAPSA
employee is assigned to document the next day patient missing
questionnaire.
The average patient flow lead time is expected to decrease by 32
minutes. The data collected is the average of all appointment types
and lengths. Therefore, these recommendations and expected time
decreases will affect different types of appointments and scheduled
lengths with variation in the actual change in time. However, on
average, the lead time is expected to decrease regardless of
appointment type or scheduled length.
References
[1] Y. L. Huang and J. Marcak. (2008, February). “Radiology
Scheduling with Consideration
of Patient Characteristics to Improve Patient Access to Care and
Medical Resource Utilization.” Health Systems. [Online]. vol. 2
(no. 2), pp. 93-102. Available:
https://www.tandfonline.com/doi/full/10.1057/hs.2013.1?scroll=top&needAccess=true
[2] Z. Costello. “RE: IOE 481 - Coordinator Meeting Wednesday
3/28” Personal email (March 29, 2018).
[3]A. McCabe and M. Pennington. “Improved CT Simulation and
Patient Access at the James.” The Ohio State University: Columbus,
Ohio. 11, Mar. 2015.
[4] A. R. Hershberger, D. Y. Lee, D. J. Jansen, and P. Raswono,
“Improving Patient Flow
within the Physical Medicine and Rehabilitation and
Anesthesiology Clinics,” IOE 481,Univ. Michigan, Ann Arbor, Final
Report. Apr. 21, 2010. [5] S. Mowlavi, Z. Shoup, and A. Wang,
“Utilization Study of Linear Accelerators in theRadiation Oncology
Department,” IOE 481, Univ. Michigan, Ann Arbor, Final Report.
Apr. 16, 2007.
[6] J. Card, A. Conlin, H. Frey, and K. Ogrodzinski, “Improving
the Scheduling Process ofReferred Radiation Oncology Patients,” IOE
481, Univ. Michigan, Ann Arbor, FinalReport. Apr. 17, 2012.
[7] United States Department of Health and Human Services,
Health Information Privacy,
Health Information Privacy, 2018. [Online]. Available:
https://www.hhs.gov/hipaa/index.html. [Accessed: 12 March,
2018].
[8] Z. Costello. “RE: IOE 481 - Coordinator Meeting Tomorrow”
Personal email (February 15, 2018).
[9] IEEE, IEEE Standards, IEEE Standards Association, 2018.
[Online]. Available: http://standards.ieee.org. [Accessed: 12
March, 2018].
[10] University of Michigan Health System, MLearning, Training
Portal, 2018. [Online].
Available: https://trainingportal.med.umich.edu/Saba/Web/Main.
[Accessed: 12 March,
2018].
Appendix A: Appointment Time Length Data Collection Sheet
Patient Sticker
Time sim room set up begins
Time sim room clean up ends
IV contrast on directive?
4D sim on directive?
SDX on directive?
Appendix B: Appointment Time Length General Linear Model
Regression
Response variable: Actual appointment time length
Categorical variables:
· Gender
· IV contrast (yes or no)
· 4DX scan (yes or no)
· SDX training (yes/assess or no)
· Inpatient or outpatient
· Number of isos
· 3-point mask (yes or no)
· 5-point mask (yes or no)
· Dental guards (yes or no)
· Bite block (yes or no)
· Bolus (yes or no)
· Cradle (yes or no)
· Vacuum Bag (yes or no)
· Accuform (yes or no)
· Qfix (yes or no)
· Body fix (yes or no)
Covariates:
· Age
· ACTV height
· ACTV weight
· ACTV BMI
· REVC height
· REVC weight
· REVC BMI
Minitab Output:
Figure B-1: General Linear Model Regression with ACTV Patient
Attributes
Figure B-2: General Linear Model Regression with REVC Patient
Attributes
Figure B-3: General Linear Model Regression without Height,
Weight, and BMI
Appendix C: Historical Data CT Simulator Delay Root Cause
Frequency
Sample size: 58 patients
Source: University of Michigan Radiation Oncology Department
Data collection period: January 5, 2018 to January 18, 2018
Appendix D: CT Simulator Patient Flow Process Map
Note: The team has provided a Microsoft Visio version of the CT
Simulator Patient Flow process map to the clients. The process map
below may be difficult to read due to space constraints.
Appendix E: Time Study Data Collection Sheet
Process
Start Time
End Time
Reasons for delays
Patient check in
_ _ : _ _
N/A
Late
At another appointment
Patient taken to Room 10
_ _ : _ _
N/A
Room 10 not available
Taken to a different room
Patient Questionnaire Filled?
N/A
N/A
Face Photo
Consent Paged: _ _ : _ _ Arrival: _ _ : _ _
IV Team Needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Anesthesiology Needed Paged: _ _ : _ _ Arrival: _ _ : _ _
NPO Requested Wait Time: _ _ : _ _
Oral Contrast
SDX Training
Myelogram Needle out time: _ _ : _ _
Simulation team notified patient ready
_ _ : _ _
N/A
Patient enters simulation room
_ _ : _ _
N/A
Myelogram recovery
Not finished drinking oral contrast
Transportation team
Patient moved to simulation table
_ _ : _ _
_ _ : _ _
Lift team Paged: _ _ : _ _ Arrival: _ _ : _ _
Pre-scan set up / patient positioning
_ _ : _ _
_ _ : _ _
IV Contrast Start: _ _ : _ _ End: _ _ : _ _Power port? Y/N
SDX Training Start: _ _ : _ _ End: _ _ : _
Physician needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Physics needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Dosimetry needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Patient needed to go to bathroom
Complicated bolus
Could not complete SDX training
> 1 iso?
Inaccurate or incomplete simulation directive
Scout
_ _ : _ _
_ _ : _ _
> 1 repositioning needed
Physician needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Physics needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Dosimetry needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Patient scan
_ _ : _ _
_ _ : _ _
Input error
Physician needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Physics needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Dosimetry needed Paged: _ _ : _ _ Arrival: _ _ : _ _
Physician scan approval
_ _ : _ _
_ _ : _ _
Physician Paged: _ _ : _ _ Arrival: _ _ : _ _
Physics Paged: _ _ : _ _ Arrival: _ _ : _ _
Dosimetry Paged: _ _ : _ _ Arrival: _ _ : _ _
Patient exit
_ _ : _ _
N/A
Room cleaned
_ _ : _ _
N/A
Appendix F: Patient Flow Value Stream Map
Appendix G: Patient Flow Design Constraints and Standards
Matrix
This appendix contains three tables that map the soft design
constraints, hard design constraints, and design standards for the
new patient flow process. Together, these three tables form the
project’s constraints and standards matrix for the new patient flow
process.
Table G-1: Soft Design Constraints (Design Requirements)
Entry #
1
2
3
R-A. Organizational Policy
N/A
R-B. Ethical
N/A
R-C. Health & Safety
N/A
R-D. Economic
(R-D-1)
R-E. Implementability
(R-E-1)
(R-E-2)
R-F. User Acceptance
(R-F-1)
R-G. Patient Acceptance
(R-G-1)
R-H. Task Duration
(R-H-1)
(R-H-2)
(R-H-3)
Table G-2: Hard Design Constraints
Entry #
1
2
3
C-A. Organizational Policy
(C-A