Hospital Waiting List Dynamics Dave Worthington Management Science Dept. Lancaster University d.worthington@lancaster.ac.uk
2 Outline The Issue Problems with previous approaches Modelling potential problems with current approaches Conclusions
3 The Issue Long waiting lists and long waiting times have been a major problem for the UK NHS since its inception in 1948. Many attempts to solve the problem, all of which have failed (though some local successes). Will current attempts to manage waiting lists and meet 18 week targets fare any better?
4 Problems with previous approaches - 1 Consultant X has a long inpatient waiting list So give consultant X some more beds - FAIL! because: Beds not the bottleneck Consultant X likes to have a long waiting list
5 Problems with previous approaches - 2 Consultant X has a long inpatient waiting list and insufficient theatre time So give consultant X more theatre time - FAIL! because: X transfers more patients from outpatient WL And allows more patients onto OP WL; Current GPs refer more patients to X; Extra GPs start to refer to X
6 Problems with previous approaches - 3 Classic Queue behaviour Observed Queue behaviour, explained by: (a) Unnecessary systematic delays, and/or (b) Waiting time acting as rationing mechanism.25.2.15.1.5 5 1 15 2 25
Potential Problems with Current Approaches 7 Concentrating on (a) unnecessary systematic delays: NHS Institute for Innovation and Improvement believe: No Evidence that Demand exceeds Capacity; Experienced Trust Manager argues: where clinical gatekeepers/triage mechanisms or tangible clinical thresholds exist there is no reason why referral rates should rise when waits drop.
Potential Problems with Current Approaches 8 and crossing fingers about (b) rationing mechanism: Evidence of history; Experienced Trust Manager: caveat is diagnostics where our intervention rates are well below European rates; Having said all that I admit to having a nervousness about referral rates Choose and Book
9 Simple System Dynamics model Designed to demonstrate risks to which current measures are susceptible. Model incorporates: Throughput changes Possible natural growth in underlying demand Rationing effect of waiting times Deterrence effect of distance Model (deliberately) ignores: queue discipline, OP/IP shunting, random variation.
1 Components of Model Simple Bathtub principle, with: Where: arrivalrate(areaa tohospitalh) = D ( W A d AH ) A e γ + D A = underlying demand in area A, γ indicates strength of rationing/deterrence effects, W A = anticipated waiting time; d AH = deterrence effect of distance from area A to hospital H, in equivalent weeks
Scenario 1: Throughput at Hospital 1 increased for 1 months to reduce backlog 11 No other factors Waiting List Sizes Waiting time (week Waiting Times 3 25 2 15 1 5 5 1 15 2 25 3 time (months) Patients on Lis 4 3 2 1 5 1 15 2 25 3 % of Patients Referr Access to Care 1 8 6 4 2 5 1 15 2 25 3 Ar ea_a Ar ea_b Ar ea_c
Scenario 2: Throughput at Hospital 1 increased for 1 months to reduce backlog 12 Some natural growth in demand Waiting List Sizes Waiting time (week Waiting Times 3 25 2 15 1 5 5 1 15 2 25 3 time (months) Patients on Lis 4 3 2 1 5 1 15 2 25 3 % of Patients Referr Access to Care 1 8 6 4 2 5 1 15 2 25 3 Ar ea_a Ar ea_b Ar ea_c
Scenario 3: Throughput at Hospital 1 increased for 1 months to reduce backlog 13 Patients on Lis 6 4 2 Some feedback, no Choose and Book Waiting List Sizes 5 1 15 2 25 3 Waiting time (week % of Patients Referr Waiting Times 3 25 2 15 1 5 5 1 15 2 25 3 time (months) Access to Care 1 8 6 4 2 5 1 15 2 25 3 Ar ea_a Ar ea_b Ar ea_c
Scenario 4: Throughput at Hospital 1 increased for 1 months to reduce backlog 14 Some feedback, some Choose & Book From Area B to Waiting time (week Waiting Times 5 4 3 2 1 5 1 15 2 25 3 time (months) Waiting List Sizes Patients on Lis 6 4 2 5 1 15 2 25 3 % of Patients Referr Access to Care 1 8 6 4 2 5 1 15 2 25 3 Ar ea_a Ar ea_b Ar ea_c
Scenario 5: Throughput at Hospital 1 increased for 1 months to reduce backlog 15 Some feedback, some Choose and Book from Areas B & C to Waiting List Sizes Waiting time (week Waiting Times 7 6 5 4 3 2 1 5 1 15 2 25 3 time (months) Patients on Lis 1 8 6 4 2 5 1 15 2 25 3 % of Patients Referr Access to Care 1 8 6 4 2 5 1 15 2 25 3 Ar ea_a Ar ea_b Ar ea_c
16 In Summary Current measures to solve waiting lists may well not have the predicted effects of achieving 18 week targets; They may also have some undesirable implications for equity of access; Careful choice of model may be key to understanding issues, and SD seems appropriate here.