7/24/2014
Experience with Lean/Six Sigma
Todd Pawlicki
What is Six Sigma?
• A methodology
• Focuses on process variation and defects
• Tools and methods to improve process quality
and process costs
What is Lean?
• A methodology that focuses on process speed,
flow, and agility
Journal of Oncology Practice 3(4):189-193, 2007.
1
7/24/2014
What is Lean/Six-Sigma?
• An integrated quality, speed, and cost
methodology and toolset
• Use more lean tools if trying to improve process
speed or reduce process costs,
• Use more Six-Sigma tools if trying to improve
process quality
Lean/Six-Sigma Demystified
• The scientific process applied to quality
improvement
Not quite...also includes many new tools, change
management, and sustaining change strategies.
Six-Sigma Excellence
• Defective parts per million (ppm) opportunities
Limits in
Probability of having a
sigma around product outside the limits
the mean
(Centered distribution)
Probability of having a
product outside the limits
(distribution shifted by 1.5𝝈)
3 sigma
2700 ppm
66,810 ppm
4 sigma
63.4 ppm
6,210 ppm
5 sigma
0.34 ppm
233 ppm
6 sigma
2 ppb
3.4 ppm
2
7/24/2014
Creator of Six-Sigma
• Bill Smith
– TQM spinoff; a better mousetrap
• 1952 Naval Academy graduate
– 35 years engineering and QA
• Joined Motorola in 1987
– Using 6𝜎, Motorola was the 1988 Baldrige winner
• Died at work of a heart attack in 1994
DMAIC
Define
What problem to solve?
Measure
What is the process capability?
Analyze
When & where do defects occur?
Improve
Go after root causes.
Control
Control process to sustain gains.
UCSD Six-Sigma Experience
• Support obtained ($20k)
– May 2010
• June – December 2010
• 5 members / 5 projects
– Clin Ops Manager, Physics, IT, Dosimetry, Therapy
– Example
• Reduce the time for patients to start SRS treatment
3
7/24/2014
Six Sigma – Project Scorecard
Define
Define
Measure
Measure
Devel op a Project Cha rter wi th
the Project Focus , Key Metri cs ,
a nd Project Scope
Crea te a Proces s Ma p of the key
proces s i nvol vi ng key pers onnel
i nvol ved i n the proces s .
Form a n i mprovement tea m
i ncl udi ng key s takehol ders
Analyze
Analyze
Improve
Improve
Control
Control
Percent
Complete
Process map
Ana l yze proces s fl ow a nd
i dentify wa s te
Pri ori tize potentia l s ol utions
i ncl udi ng cos t benefi ts .
Crea te a Control Pl a n for
s ol ution
10%
Crea te a pl a n for col l ecting da ta
Data plan
Determi ne s ources of va ri a tion
a cros s proces s
Identify, eva l ua te, a nd s el ect,
bes t s ol ution
Continue to moni tor a nd
s tabi l ze proces s us i ng control
cha rts
20%
Va l i da te probl em s tatement
a nd goa l s wi th s takehol ders
Determi ne proces s performa nce
/ ca pa bi l i ty
Ana l yze da ta col l ected for
trends , pa tterns , a nd
rel a tions hi ps .
Devel op, optimi ze a nd Impl ement
pi l ot s ol ution
Devel op SOP's a nd proces s
ma ps for i mpl emented s ol ution
30%
Crea te a communi ca tion pl a n
wi th a ction i tems
Va l i da te the mea s urement
s ys tems
Perform root ca us e a na l ys i s a nd
pri ori tize ca us es .
Devel op "To Be" va l ue s trea m
ma p
Tra ns i tion project to proces s
owner
40%
Crea te a Va l ue Strea m Ma p of
the s el ected proces s i nvol vi ng
key pers onnel i nvol ved i n the
proces s .
Col l ect da ta for "As -Is " proces s
Ana l yze two s a mpl es us i ng
Hypothes i s Tes ts
Va l i da te pi l ot s ol ution for
portentia l i mprovements wi th
feedba ck from key s takehol ders
Communi ca te project s ucces s &
cha l l enges to crea te
opportuni ties for s ys tem wi de
a doption.
Communicate
50%
Devel op a hi gh l evel proces s
ma p (SIPOC)
Ana l yze three or more s a mpl es
us i ng ANOVA
FMEA of potentia l fa i l ures
Fa ci l i tate cha nge ma na gement
60%
Col l ect ba s el i ne da ta i f exi s ts
Unders tand rel a tions hi ps i n two
va ri a bl es Correl a tion
Des i gn of Experi ments
Charter
Validate prob.
Communicate
“as is” VSM
Collect data
Find variation
RCA
ANOVA
Verify VOC
Prioritize
Control
charts
SOPs
“to be” VSM
FMEA
DOE
70%
Determine relationships in
variables using Regression
Determi ne "Voi ce of Cus tomer"
a s i t rel a tes to the project
Regression
80%
90%
Review with Sponsor
Review with Sponsor
Review with Sponsor
Review with Sponsor
Review with Sponsor
100%
Project Charter
Project Charter - Six Sigma Performance Improvement Projects
Project Name: Reducing the time for identifying and starting cranial SRS patients on treatment
Start Date: 6/29/2010
End Date: 9/30/2010
Project Team Members
Responsibility
Team Leader
Project Sponsor
Executive Champion
Process Owner
Stakeholder
Team Member
Team Member
Team Member
Team Member
Name
Todd Pawlicki & Greg White
Josh Lawson, MD
AJ Mundt, MD
Mary Collins
John Alksne, MD
Grace Kim & Jia-Zhu Wang
Matt Taylor
Rich Fletcher
Polly Nobiensky
Department
Title
Physicist & Dosimetrist
Physician (Rad Onc)
Department Chair (Rad Onc)
Clinical Operations Manager
Physician (Neurosurgen)
Physicists
Chief Therapist
Chief IT
Chief Nurse
Phone
Problem Statement and Scope
Problem Statement
Currently, it takes 16.3 work days to get a cranial SRS patient
on treatment. This results in stress for the patients in waiting
for their treatment as well as deviations from high-quality care,
in some cases due to medical conflicts with chemotherapy, for
example.
Scope
Starts after consult is completed and ends when the patient's
first fraction begins.
4
7/24/2014
Objective and Metrics
Objective
Decrease the time from consult to first fraction treated by 70%.
The goal is to have all SRS patients treated within 5 work days
of their consult.
Primary Metric(s) Work day hours (7am - 5pm): Consult to first fraction treated.
Secondary Metric(s) Work day hours: Consult to Sim
Work day hours: Sim to PDOS started
Work day hours: PDOS to first fraction treated
Process Map
Consult/Scheduling
Simulation/Planning
QA/First Tx
CT Sim is
completed
SRS Referral
comes to Tanya
(usually fax but
can be direct
communication
with MD
Sub CMD
for Staff
Tanya schedules
with appropriate
MD
PDOS is
run by the
RTT at the
Linac
Cone or
Fiducial Plan?
No
CMD picks up QA
data and finalizes
plan for treatment
Yes
Staff identifies
SRS case on their
Task Pad
MD meets with the
patient
Sub PhD
for Staff
Physics checks
the plan
Staff imports CT
data into Eclipse
Treat?
End
Yes
MD fills out DISPO
form (MRI order
and Rx intent)
MD or RN email
Lon (scheduler)
that an SRS case
is coming
Is plan
acceptable?
Staff finds MRI,
imports into
Eclipse, & fuses
with CT data
Yes
Staff emails Rad
Onc MD to see
who will draw
turmor
The patient arrives
for first treatment
on scheduled day/
time
Staff emails
appropriate MD to
draw tumor
End
MD draws tumor
and enters intent
Lon get the
authorization for
MRI and Tx
Staff draws normal
tissues
Lon schedules
MRI and CT Sim
(tries to be on the
same day)
Staff finished
treatment plan per
MD Rx
Lon calls patient
with MRI and CT
Sim date(s)
Staff emails Rad
Onc MD to reviewapprove plan
RTT’s schedule
patient’s first
treatment
No
Staff puts PDOS
delivery on RTT
Task Pad
Plan
acceptable?
Data Collection Plan
Staff puts 2nd
check on Physics
Task Pad
Yes
Staff finished
approved plan
Process Owner: Todd Pawlicki
Process: Reducing the time for identifying and
starting cranial SRS patients on treatment
Staff schedules
Staff creates
PDOS on the
Data
PDOS
planCollection Plan
appropriate Linac
Data
What?
Measure Units
& Data Type
Operational Definition and Procedures
Sample Size
Stratification
Factors
Who will collect
data
How
Measured?
How will data be Where will data
collected
be collected
Consult
date & time
By case type
(cranial SRS)
Schedulers
Entered in Aria
CT Sim
date & time
Taken from
consult case
Schedulers
Entered in Aria
PDOS
date & time
Taken from CT
Sim case
Dosimetrist
Entered in Aria
First treatment
date & time
Taken from
PDOS case
Therapist
Entered in Aria
Entered by staff
(part of routine
procedures)
Entered by staff
(part of routine
procedures)
Entered by staff
(part of routine
procedures)
Entered by staff
(part of routine
procedures)
Consult to first
treatment
duration (7am5pm workdays)
By case type
(cranial SRS)
n/a
calculated from
above data
n/a
41
Aria database
Aria database
Aria database
Aria database
Aria database
5
7/24/2014
Baseline Data
Consult to First Treatment (workdays, 7am - 5pm)
Raw Data
Outlier
Removed
Mean
19.6
16.3
Standard
deviation
11.9
6.5
Minimum
8.3
8.3
Maximum
48.7
28.1
Fishbone Diagram
Correlation Analysis
Step 1
Step 2
Consult
CT Sim
Step 3
1st Treatment
QA/PDOS
Consult to First Treatment (workdays)
45.0
40.0
35.0
30.0
R2 = 0.1951
25.0
20.0
15.0
10.0
5.0
0.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
PDOS to First Tx (workdays)
6
7/24/2014
Correlation Analysis – Conclusions
Step 1
Consult
Step 2
Step 3
CT Sim
1st Treatment
QA/PDOS
• Strongest correlation for primary metric
– Consult to CT Sim (Step 1)
• The steps are only weakly correlated
– Improving one step won’t have an effect on the other steps
Before Change
Process Map: “To-Be” Process
Classification
YES
C
Input Spec /
Settings
INPUTS (x's)
Start
Decision
Block
(For all "C" & "K"
Classifications)
NO
- Pre-Heat temp
- Cycle time
- Travel tank speed
- Cure Temp
- Dip time
600-700
65-80
20-35
400-500
12-17
PROCESS STEP
Process Block
OUTPUTS (y's)
Rework
Preparation
or Data
Delay
Output Specs /
Measurment Levels
(If applicable)
Inventory
Off Page
Connector
Data
Entry
Warm up the oven before
starts shift for 1 hour
Tune all process parameters
at the oven
Align Wire form
Wires in Good
shape
C
C
N
C
Replace wires
- Silicone concentration 2.5-3.5 ratio
- Silicone recirculation
Oven starts with wire form
dipping into silicone tank
- Silicone Temperature
- Silicone remove
xx psi
Open/Close
- Pre-Heat Temperature 600-700
C
C
N
C
C
K
C
K
N
N
- Thermocouples
- Vents
Calibrated
Pre-heat wire form
+ Implementation Plan
- Top Zone heaters
- Dip time
- Plastisol Temperat
- Travelup tank speed
- Traveldown tank speed
- Buil up surface
- Insulation from oven
400-500
Open/Close
12-17
< 80 F
8-15 sec
25-35 sec
Wire form dipping into
plastisol tank
7
7/24/2014
After Change
Cost Benefit Analysis
Cost Benefit Analysis
Process:
Project Team:
Team Leader:
Expected Costs
Description
Unit
Cost
Expected Costs $
-
Benefits
Description
Unit
Savings
Estimated Procurement Savings $
-
Projected Project Savings $
-
VSM IERC 2008
Using Value Stream Map on The
Issue of Activity Capture and Billing
at an Academic Radiation Oncology
Department
Claribel Bonilla, PhD, CSSBB
Brigitte Wesselink, (Student)
Ashlee Enriquez, (Student)
Todd Pawlicki, PhD
University of California, San Diego
Department of Radiation Oncology
University of San Diego
Department of Industrial & Systems
Engineering
8
7/24/2014
Area of focus
CAT Scan
+
Immobilization
Treatment
Planning
Position
Verification
Plan
Verification
Treatment
Delivery
• Staff was asked to decrease capture errors
• Staff suggested a focus on treatment delivery
Value Stream Mapping (VSM)
• Part of Lean Thinking
– Used to identify and remove waste
• Value stream
– All the actions required to bring a product or service to a
customer
• Steps
–
–
–
–
Establish the process scope
Construct a current state map
Construct a future state map
Develop a plan to implement changes
Our VSM Application
• Step 1
– Map process by someone not familiar with the process (USD Student)
– Collect data on accuracy of activity capture
• Step 2
– Analysis of data (“hard evidence”)
• Step 3
– Create future state
• Engineering expert + Domain expert = Final recommendations
9
7/24/2014
Trilogy
21EX
Trilogy
2.4 errors/day
21EX
2.4 errors/day
10
7/24/2014
Future
State
Map
Results – Current State
• Data collected over a two month period
21EX
# % tx's
20 1.43
17 1.21
12 0.86
8 0.57
Captured incorrectly
Forgot to capture
Billed when shouldn't
Double billed
Trilogy
# % tx's
11 0.79
10 0.71
20 1.43
3 0.21
Captured incorrectly
Forgot to capture
Billed when shouldn't
Double billed
• Total error: ~$240,500 / month
11
7/24/2014
Results – Future State
• Data collected over a two month period
21EX
# % tx's
13 0.93
7 0.50
15 1.07
0 0.00
Captured incorrectly
Forgot to capture
Billed when shouldn't
Double billed
Trilogy
# % tx's
6 0.43
5 0.36
2 0.14
4 0.29
Captured incorrectly
Forgot to capture
Billed when shouldn't
Double billed
• Total error: ~$123,800 / month
• 48.5% improvement
Lean/Six-Sigma Lessons Learned
• Requirements
– Direct line of accountability to “senior” management
• Key to success
– Need protected time for participants
– Make use the guided problem solving
• Most difficult part
– Data collection and analysis
12