Upload
afia
View
63
Download
0
Embed Size (px)
DESCRIPTION
Healthcare process optimization research University Twente. Prof. Wim H. van Harten Prof. Quality management & healthcare technology.(UT) Exec. Board Netherlands Cancer Institute Prof. Erwin W. Hans, MSc, PhD Associate prof. Operations Management and Process Optimization in Healthcare - PowerPoint PPT Presentation
Citation preview
Healthcare process optimization research University TwenteProf. Wim H. van Harten
Prof. Quality management & healthcare technology.(UT)Exec. Board Netherlands Cancer Institute Prof. Erwin W. Hans, MSc, PhD
Associate prof. Operations Management and Process Optimization in HealthcareCenter for Healthcare Operations Improvement & Research
04/22/[email protected] 3
Strategic
Operational offline
Tactical
Resource capacity planning
Material planning
Medical planning
Financial planning
Operational online
managerial areas
hierarchical decom
position
Case mix planning, layout planning,
capacity dimensioning
Allocation of time and resources to
specialties, rostering
Elective patient schedulingworkforce planning
Supply chain and warehouse design
Supplier selection, tendering, forming
purchasing consortia
Purchasing, determining order
sizes
Care pathwayplanning
Diagnosis and planning of an
individual treatment
Research planning, introduction of new treatment methods
Agreements with insurance companies,
capital investments
Budget and costallocation
DRG billing, cash flow analysis
Monitoring, emergency rescheduling
Rush ordering, inventory replenishing
Triage, diagnosing complications
Expenditure monitoring, handling billing complications
A modern framework for health care planning & control(Hans, Houdenhoven, Hulshof, 2010)
Society
04/22/[email protected] 4
Research development 2003 - 2007: Focus on single departments
• Operating rooms (planning, scheduling, etc.)
• Radiology (CT, MRI)
2008 - 2012: Focus on care pathways within hospitals• STW funded project “LogiDOC”
• 12 hospitals, 7 PhD students
• PhD students are at hospitals 2-3 days per week
2010 - 2016: • Optimization of complex clinical pathways
• Optimization of the “transmural” care pathway
• Optimization of rehabilitation processes
04/22/[email protected] 6
Tactical resource allocation in hospitals
Minimize waiting for resources Minimize access time Minimize care pathway lead time Maximize utilization of resources Equitable distribution amongst patient types Prototype “dashboard” / tactical planner based on MILP
Peter Hulshof
04/22/[email protected] 7
Tactical resource allocation in hospitals
Static vs. dynamic problem MILP:
• Decision variables• Weights
• Queues, patient types, resources, time waiting
Peter Hulshof
Details: [email protected]
04/22/[email protected] 8
Analytical Comparison of the Patient-to-Doctor Policy and the Doctor-to-Patient Policy in Outpatient Clinics
Peter Hulshof
04/22/[email protected] 9
Analytical Comparison of the Patient-to-Doctor Policy and the Doctor-to-Patient Policy in Outpatient Clinics
Analytical stochastic model, based on the recursion of the moment that the doctor finishes a consultation
Discrete-event simulation to observe behavior when general assumptions are relaxed
doctor travel time
patient preparation time
Peter Hulshof
04/22/[email protected] 10
Access to CT scannersPatient preferences (2009) Ranking of patient preferences
(Analytic Hierarchy Process method):1. One-stop shop 43%2. Access time 22%3. Waiting time 19%4. Autonomy in selection of appointment time 16%
Majority of patients prefer regular hours for appointments “One-stop shop” can be obtained by:
• Planning combination appointments• Walk-in
Patients accept a higher waiting time in case of walk-in
MaartjeZonderland
NikkyKortbeek
04/22/[email protected] 11
Combine appointments with walk-in
Walk-in: Extra service Zero access time contributes to lead-time optimization Not always possible
• Medical reasons
• Demand fluctuates
Combine walk-in with appointments
MaartjeZonderland
NikkyKortbeek
04/22/[email protected] 12
Combine appointments with walk-in
Given walk-in demand, when should appointments be planned?
When on 2 levels:
On which day?
When on the day?
Trade-off: access time (app) & acceptance rate (walk-in)
MaartjeZonderland
NikkyKortbeek
04/22/[email protected] 13
Algorithm: Balance
Result:
Appointment plan: number of appointments per
day in the cycle
(satisfy access time norm)
Day plan with appointment slots
(maximize walk-in acceptance rate)
Tuesday’s plan
MaartjeZonderland
Model I:Access TimeAppointments
Model II:Acceptance RateWalk-in Patients
Algorithm:Balance!
Details: [email protected]
04/22/[email protected] 14
Emergency OR, or not?
Concept: “emergency
ORs”
Concept: “No
emergency ORs”
04/22/[email protected] 15
Simulation results operational problem
Waiting time less than:
First emergency procedure
Second emergency procedure
Third emergency procedure
No BII opt.
BII opt.
No BII opt.
BII opt.No BII opt.
BII opt.
10 minutes
28.8% 48.6% 34.9% 44.9% 40.4% 46.2%
20 minutes
53.0% 75.8% 56.9% 73.6% 63.0% 69.8%
30 minutes
70.5% 90.9% 71.8% 87.2% 76.3% 86.7% Case mix Academic Hospital
04/22/[email protected] 16
Results after simulation
“Emergency surgery in elective program” instead of “emergency ORs” yields:
Improved OR utilization (3.1%) Less overtime (21%)
Break-in-moment optimization yields: Reduced waiting time for emergency surgery, especially
for the first arrival(patients helped within 10 minutes: from 28.8% 48.6%)
Evidence based management (HBR 2006)
Weak evidence Many self proclaimed experts Wide array of sources Variation in company configuration
Evidence based management (HBR 2006)
Trust of own experience above research
Capitalise on own strenghts Hype and marketing Dogma and belief Casual benchmarking
Evidence based management (HBR 2006)
Examine the logic Always Pilot your Programs The art of implementation cf EBM
Survey Operations management in Dutch Hospitals
To explore the use of business improvement approaches to improve patient logistics in Dutch hospitals • Used business improvement approaches• Tools that hospitals used• Results achieved with these approaches
Respons 46 – 6 UMC, 13 STZ, 27 Alg.
Method used
Total Method used
Used in combination with
CP BM BPR
LM TOC CI TQM OR FF LSS SS
CP 42 CP 32 20 20 18 15 10 12 10 8 6
BM 36 BM 32 20 18 18 13 11 11 7 6 4
BPR 22 BPR 20 20 14 10 6 6 10 6 5 4
LM 22 LM 20 18 14 13 7 4 8 7 2 5
TOC 20 TOC 18 18 10 13 8 6 7 6 2 4
CI 15 CI 15 13 6 7 8 4 1 5 2 1
TQM 13 TQM 10 11 6 4 6 4 5 2 2 1
OR 13 OR 12 11 10 8 7 1 5 4 3 3
FF 10 FF 10 7 6 7 6 5 2 4 1 1
LSS 8 LSS 8 6 5 2 2 2 2 3 1 3
SS 6 SS 6 4 4 5 4 1 1 3 1 3
Frequency of methods used and combinations of methods used (n=46)
Combination of type of hospital and priority (N=46)
Teaching General Top clinical Total
CP 4 9 7 20
BM 1 2 2 5
BPR 1 4 1 6
LM 0 5 0 5
TOC 0 4 0 4
CI 1 1 1 3
TQM 0 5 0 5
OR 0 1 0 1
FF 1 0 0 1
LSS 1 3 2 6
SS 0 0 1 1
Diverse ziekenhuizen hebben meerdere prioriteiten
Results per performance aspect
UMC General Top clinical
Efficiency(n= 31)
+ 1 12 4
- 3 6 5
Timely(n=29)
+ 0 10 4
- 4 6 5
Financial(n=25)
+ 2 10 2
_ 0 5 6
+ = objectives achieved or results above objectives
- = objectives were not achieved
35 hospitals answered this question on different aspects General hospitals report more successes Top clinical hospitals report least successful No relation with method; with consultants less successful
Ontwikkeling ziekenhuismanagement
1980 Professional Quality 1985 Continuous improvement (audit ) 1990 Quality systems(EFQM/TQM) (2005 “seperate track” Safety) 2000 Operations management/logistics 2015 Evidence and formats OM/OR?
Antoni van Leeuwenhoek Hospital- Medical & Surgical Oncology, Radiotherapy
Budget (2011): 135 million euro
Number of employees ca. 1400 (1600 pers.)
First contacts 29.000
Admissions 6600
Chemotherapy Day care 15.000
180 beds; 30 Chemo day care ; 10 linacs; 6 Operating rooms120 medical specialists (employed)
Framework voor planning en control in ziekenhuizen
Alignment of CDU and Pharmacy processes
Queue: Make in advance
Queue: Make on demand
Phar
mac
y
CDULeftovers
No wait
Risk of Wastage
Patient will wait
No Wastage
MedicineOrders
DecisionCriteria?
Lab GreenLight
Modelled with stochastic programming
Alignment of CDU and Pharmacy processes
Prepare nothing in advance:• There is no medicine wasted• Patients must wait while medicine is prepared
Prepare everything in advance• Patient do not have to wait for their medicine
to be prepared• There is a risk that expensive medicine will
be wasted
Prepare some in advance
Results
Preparenothing
in advance
Preparesome
in advance
Cost of wasted medicines per day
Wai
ting
Tim
e pe
r pati
ent
Results mathematical analysis
The outcome from this study resulted in a new policy at the cancer centre which is expected to decrease the waiting time by half, while only increasing pharmacy’s costs by 1-2%
Lit. review successful simulation
Defined by Robinson and Pidd 19981. The study achieves its objectives
and/or shows a benefit.2. The results of the study are
accepted3. The results are implemented4. Implementation proved the results
of the study to be correct
Figure 2 Overview selected papers for literature review
161 abstracts from Pubmed 125 abstracts from BSE
277 different abstracts
70 relevant papers
21 cross references of
papers reporting
implementation
2 papers inaccessible
89 papers included in literature review
Simulation review
Results Section II: implementation
Implementation phases
Yes Partially Intention is mentioned
No Not stated
Did the study achieve the clients objectives
89 0 N/A 0 0
Show the study results direct benefits to the client?
71 2 N/A 11 5
The study results are accepted by the client
21 5 N/A 0 63
The study results are executed
10 6 3 1 69
Results of the electronic survey of the authors
Question Yes Partially No Do not know
Missing Not relevant for study
Total
The study results are accepted by the hospital
11 10 1 4 3 12 41
The study results are executed
7 11 5 6 0 12 41
Implementation proved the study to be correct?
9 5 1 3 11 12 41
Conclusions review & survey• Implementation rate of literature review
and the survey was 18% and 44% actual implementation occurred more often than reported in literature
• Quality of the research methods used for evaluation is higher in reality (17%) than is reported in the literature (3%)
• Availability and Technical quality of the data
• Culture, knowledge and “Contingency” (what to apply when)
physicianphysician
Process analysis
throughput timeradiologist’s
report
radiologist
access timeCT scanner
access timeoutpatient
consult
throughput timediagnostic track
Data collection: Variation in access time
Variation is influencedby capacity outage
1. Variation2. High average
Design of experimentsSpecial slots?
max. access time CT
(working days)
max. throughput time scan – consult 2
(working days)
max. throughput time
(working days)
IP OPU OPS IP OPU OPS IP OPU OPS
CurrentNone, based on urgency
1 5 - 0 5 5 1 10 -
Approach 1Only for long term outpatients
1 5 15 0 5 5 1 10 20
Approach 2Separate slots for all groups
1 2 7 0 3 3 1 5 10
Approach 3Shared slots forurgent and short term, rest separate slots
1 2 2 0 3 3 1 5 5
Modeled results
Number of slots Expected idle time Expected overtime
IPUPU
OPST
OP LT
totalMean St
devCI mean stdev CI
Current37
37 21,1 53,1 -83,1-125,3
38,4 35,7 -31,6-108,3
Appr 124
15 39 14,1 29,0 -42.8-70.9 45,4 32,1 -17,6-108,4
Appr 2 4 8 16 15 43 19,2 23,1 -26,1-64,4 10,3 19,8 -28,5-49,0
Appr 3 4 24 15 43 20,1 29,7 -38,2-78,3 11,4 23,8 -35,3-58,1
Implemented changes
1. Set service levels: not formally done. Growth causes problems
2. Change allocation of capacity• Separate slots for long and short term patients.
• Plan short term max 7 working days ahead • Slots based on model: 19 long term, 19 short term, 3
inpatients, rest is urgency. • But after short term group contained ‘long term’
patients 22, 17,2 3. Set service levels for throughput time radiologist report 4. Improve maintenance management
• Labor regulations
Results
Old situation New situation
Urgent Short term Total Total
Patients included 199 613 812 962
Access time CT (A) Average 2.1 12.3 9.8 7.2 -2.6
St dev 1.2 3.6 5.4 5.0 -0.4
Throughput time radiologist’s report (B)
Average 1.0 1.1 1.0 1.1 +0.1
St dev 1.2 0.9 1.0 2.5 +1.5
Access time second outpatient consult (C)
Average 4.9 4.6 4.7 4.3 -0.2
St dev 3.5 2.7 2.9 3.5 +0.6
Total throughput time (D= A+C)
Average 7.0 16.9 14.5 12.7 -1.8
St dev 3.6 4.6 6.1 6.3 +0.2
Conclusions CT track simulation
OR shows that significant improvements are possible
The best solution is not always implemented in reality, other factors play a role
Unexpected differences between model and reality complicate the use of a strict before and after design. iterative approach preferred
Continuous measurement of data is needed to maintain the results
• A case from : Peter T. Vanberkel et al• Published in Anasthesia &
Analgesia
• Netherlands Cancer Institute- Antoni van Leeuwenhoek Hospital
(NKI-AVL)• Expand OR capacity from 5 to 6
with limited additional bed use• Design new OR schedule in
combination with optimal ward occupancy
ACCOUNTING FOR INPATIENTS WARDS WHEN DEVELOPING A MASTER SURGICAL SCHEDULE
Project Description Develop prognostic model that relates OR
production to ward occupancy Calculate effects of various OK scheme’s
Verification of alternative schedules with observed limitations in practice
Scenario’s were limited to changing sessions on tactical level.
Implementation of modelled OR schedule . Pre modelled data were compared with observed data in practice
Model: Ward workload as a function of the MSS
Conceptual Model Scheme
Assumptions• No cancelations due to lack of ward space (extra
nurses will be called in)• Acceptable Risk of “calling in a nurse” is ~10%
• Time scales is days. • Count patients on the day of admission, not on the
day of discharge
WardBatches of patients arrive daily according
to the MSS Recovery
Discharge
Model: Ward workload as a function of the MSS
Conceptual Model Scheme
Metrics
1) Recovering Patients in the Hospital
2) Ward occupancy
3) Rates of admissions and discharges
4) Patients in recovery day n
WardBatches of patients arrive daily according
to the MSS Recovery
Discharge
Model: Ward workload as a function of the MSS
Conceptual Model Scheme
Data
For each surgical specialty
Empirical Distributions of Cases/Block (batch size)
Empirical Distribution of Length of Stay (LOS)
WardBatches of patients arrive daily according
to the MSS Recovery
Discharge
Baseline situation• 2 Surgical wards, total capacity 100 beds• Objective: sufficient beds to serve upto 90th percentile
of patient load on every weekday
90th percentile of demand on each day of the MSS
30
35
40
45
50
Mon Tue Wed Thu Fri Sat Sun
Sta
ffed
Bed
s N
eed
ed
Ward A Ward B
Predicted results
90th percentile of demand on each day of the MSS
30
35
40
45
50
Mon Tue Wed Thu Fri Sat Sun
Sta
ffed
Bed
s N
eed
ed
Ward A Ward B
Tactical change in OR session schedule.
Observed results
Better occupancy over the week
The “old appraoch”calculating mean occupancy rates would have lead to bed shortage in 51% of the days.
90th percentile of demand on each day of the MSS (Ward B)
30
35
40
45
Mon Tue Wed Thu Fri Sat Sun
Sta
ffed
Bed
s N
eed
ed
Observed Projected
Master Surgical Schedule
Simulation algorithm relating OR planning and ward occupancy
Periodical review to optimise tactical planning.
Use of Real Time data Enrich tactical scheduling options &
Deduct operational scheduling instructions
JORS 2010, Anesthesia & Analgesia 2011
Improving Cancer Care NKI-AVL One stop shop outpatient clinics 40% Efficiency chemotherapy day care: 25-40%
(EJC 2009) Benchmark Efficiency Radiotherapy Improved efficiency Operation Rooms 15%
(Simulation project) Algorithm based OR/ward planning Open access (simulation based) projects
Radiology Mathematic defining trade off: batch based
chemo prep vs patient wait times
Figure 2 Number of scientific staff (faculty and non-faculty) plotted against the total overhead costs per scientific staff. The costs are adjusted for purchasing power parity.
A
B
C
D
E
F
0
5
10
15
20
25
30
35
0 200 400 600 800 1000 1200 1400
# of scientific staff
overh
ead
costs
per
scie
nti
fic s
taff
(in
kE)
Evidence based health service management
Ask for published evidence: large series and quasi experimental designs.
Verify logics and feasible results Pilot on limited scale Exchange (long term) results
AVL: Contingency and Evidence… Lean tools and OR-based-simulation..
Future developments
Use real time data Various related hospital “funtions” Research into effectiveness and
implementatiuon in at least 4 hospitals in NL, (CHOIR/UT & NKI-AVL), IBI and US ??
Thanking:
Erwin Hans, Peter van Berkel, Richard Boucherie, Wineke van Lent, Joost Deetman, Rene Brouwers, Marloes Sanders, and many others.
Wim H. van Harten