Discreteevent simulation applied to health services research: analysis of demand and waiting time fo

1
Discrete-event simulation applied to health
services research analysis of demand and waiting
time for knee arthroplastyComas M1, Castells
X1, Román R1, Cots F1, Hoffmeister L1, Mar J2,
Quintana JM3, Barceló J4.1 Hospital del
Mar-IMIM, Barcelona, Spain. 2 Hospital Alto
Deba, Mondragón, Spain. 3 Hospital de Galdakao,
Galdakao, Spain. 4 Department of Statistics and
Operational Research, Universitat Politècnica de
Catalunya (UPC), Barcelona, Spain.International
Society for Clinical Biostatistics,
Alexandroupolis, Greece, 29 July-2 August 2007
Funding Catalan Agency for Technology Assessment
and Research (AATRM) Fondo de Investigación
Sanitaria (FIS) and
2
Background

• Healthcare systems are complex
• Usefulness of simulation in decision-making
• Alternative to analytical modeling
• Discrete-event simulation
• Operations research technique
• Widely used in other fields (military research,
  industry)
• Health services research
• Decision trees, Markov models
• Cost-efectiveness analysis of new treatments
• Scarcity of resources!!!

3
Background

• Unmet needs in spite of the increase in knee
  arthroplasty utilization
• Aging
• Widening of indication criteria
• Long waiting lists and waiting times
• Supply does not meet demand
• Prioritization of patients in waiting lists
• Priority ? Need

4
Background

• The events of interest are discrete and occurr at
  discrete times.
• Transition probabilities are difficult to
  determine
• They come from complex relationships among the
  parameters and the entities attributes.
• Non feasible analytical solution ? Simulation
• Event Scheduling approach.
• Waiting lists are queues.

5
Objective

• To define and implement a simulation model to
  analyze demand and waiting time for knee
  arthroplasty.
• To assess the impact of introducing a waiting
  list prioritization system compared with the
  usual first-in, first-out (FIFO) system.

6
Setting

• Subjects
• general population
• aged 50 years or more
• at risk of need of knee replacement
• Demand in the public health system of Spain
• 4 regions (Andalusia, Aragon, Canary Islands,
  Catalonia)
• 16.6 Million inhabitants (about 40 of the
  population of Spain)
• Indication criteria
• Radiological osteoarthritis symptoms (pain)
• Simulation horizon
• 60 months (5 years)
• Time units months

7
Methodology

• Stages
• 1.- Specification of the conceptual model.
• 2.- Estimation of the model parameters.
• 3.- Simulation.

8
Stage 1 Conceptual Model
No need
Incidence
Non Expressed Need
Demand
Waiting List
Surgery
Operated
1
9
Stage 1 Conceptual Model
No need
Incidence
Non Expressed Need 1st Surg.
Demand
Waiting List
2nd surgery
Operated bilateral
1
10
Conceptual Model Assumptions

• Surgery indication is always appropriate.
• Prevalent/incident cases have bilateral knee
  osteoarthritis.
• Cases accessing the private sector do not go back
  to the public sector waiting list.
• Need is not a dichotomous variable
• Prioritization System
• Linear scoring system ranging from 0 (lowest
  priority) to 100 (highest priority) according to
  clinical, functional and social criteria.

Developed by the Catalan Agency for Health
Technology Assessment and Research (CAHTA)
11
Prioritization system
Score
Criteria and levels
Disease severity
Score
Criteria and levels
0
Moderate
Limitation on ability to work
18
Severe
0
No or does not work
Pain
10
Yes
0
Mild
Has someone looking after him/her
17
Moderate
33
Severe
0
Yes
Recovery probability
9
No
0
Moderate
Has someone to look after
4
High
0
No
Limitation in doing everyday activities
6
Yes
0
Some difficulty
10
Great difficulty
20
Unable to do most of the everyday activities
12
Stage 2 Parameter Estimation
No need
Non Expressed Need 1st Surg.
Non Expressed Need 2nd Surg.
Death
1
Postoperative State
Waiting List
Private Sector
1st surgery
2nd surgery
Operated bilateral
1
1
13
Stage 2 Parameter estimation

• Sources of information
• INE (Instituto Nacional de Estadística)
  Population and number of deceased (2001).
• Regional Health Service
• Register of waiting lists and
• Minimum Data Set.
• Vizcaya Study Population-based study on
  prevalence of knee osteoarthritis in the North of
  Spain.
• CATHAs pilot test of the introduction of the
  prioritization system in the clinical practice
  (in every region).
• Field work on patients of Hospital de lEsperança
  (Barcelona).

14
Stage 2 Parameter Estimation
No need

Incidence
Non Expressed Need 1st Surg.
Demand
Waiting List
2nd surgery

Operated bilateral
1
15
Stage 3 Simulation

• Discrete Event Simulation
• Software SIMUL8

16
Stage 3 Simulation
17
Stage 3 Simulation
18
Stage 3 Simulation

• Analysis of results
• n20 runs
• Sensitivity analysis
• FIFO vs Prioritization System
• Comparisons were paired by stream of random
  numbers
• Outcome variable
• Waiting time weighted by priority score
• For all cases that have entered the waiting list

19
Results Waiting times
20
Results Relationship between priority score and
waiting time
21
Conclusions

• Valid and credible model to analyze need and
  demand of knee arthroplasty
• Introducing a waiting list prioritization system
  increased benefits
• But patients entering the waiting list with low
  priority scores had excessive waiting times
• This tool will allow
• assessing different scenarios, like changes in
  clinical practice or definition of need.
• applying it to other elective procedures.