DECODE: Modelling the impact of a dementia triage system on Exeter memory clinic waiting times
A large proportion of people who are referred to memory clinics are cognitively healthy and do not receive a diagnosis of dementia or another cognitive deficit. This places unnecessary strain on stretched services locally, nationally and internationally.
Dr David Llewellyn and his team from the University of Exeter have developed a clinical decision support system called DECODE to triage patients being referred for a dementia assessment. David and his team are working with Devon Partnership Trust and the Devon Memory Clinic service to develop a trial of the DECODE system.
If used as a triage system, DECODE has the potential to reduce waiting lists, reduce the cost per diagnosis, and prevent unnecessary distress in cognitively healthy patients who can be reassured at a much earlier stage.
This PenCHORD project modelled the anticipated impact that using DECODE would have on patient waiting times for assessment.
Using modelling techniques, the project aimed to provide evidence of the impact the DECODE system would have on waiting times at Exeter memory clinic, to help determine whether it would be beneficial to implement DECODE.
To achieve this objective, the project aimed to:
- Map the current assessment pathway for patients at Exeter memory clinic
- Determine the impact the DECODE tool would have on waiting times for assessments
- Determine the impact that reducing the number of assessments would have on the neuropsychological assessment pathway.
The first stage of the modelling process was to create a process map of the Exeter memory clinic assessment pathway. From this process map a conceptual model was created and the memory clinic provided data to populate the actual model. Individual patient episode data from December 2014 to November 2016 inclusive was used. The model structure was a simplified version of the assessment pathway: referrals for assessment arrive at the rate ascribed by the data, patients then wait for an assessment, and then based on the prevalence of dementia seen at the clinic patients have one of five outcomes and exit the system. The model was then used to simulate the queue for assessment over two years to capture trends in the data.
The graph on the left-hand side in Figure 1 shows that in the baseline scenario where there is no DECODE triage, the waiting time for assessment continues to steadily increase over time. The right-hand graph of the experimental scenario using the DECODE triage system, shows a different trend. Waiting times continue to increase for about 50 days following the implementation of DECODE. At 50 days the trend reverses and waiting times begin to decrease until at about 160 days where it levels out at a wait time of less than one day.
The simulation found that using the system could potentially increase the number of people with dementia who could be diagnosed within a two-year period by 21%. At the same time, a large proportion of cognitively healthy people could be picked up by the DECODE triage system; this has the potential to reduce the number of unnecessary assessments by up to 400%.
The app was found to triage a small proportion of patients incorrectly; however this could be accounted for by changing the point in the assessment pathway at which DECODE is used.
Outputs and impact
The findings from this project have been used to inform Devon Partnership Trust of the expected impact of implementing the DECODE triage system at the Exeter memory clinic. The model showed that DECODE could reduce the waiting times at the Exeter memory clinic and additional capacity would be released from the system.
From the point when DECODE is implemented it could take approximately 50 days for there to be a measurable reduction in waiting times and it could then take between 150 and 380 working days until the minimum waiting time is reached.
Following the results of the project, the collaborators would like to extend the modelling to the two other memory clinics in Devon: North Devon and South Devon & Torbay. A second phase of the project will now assess the impact of the DECODE dementia triage app across all three memory clinics, extending the modelling work already undertaken.
The results from Phase 1 of the project demonstrated that the recovery of the triage system to a zero waiting time state was non-linear: variation in the use of the tool produced different rates of system recovery. If DECODE is tested as a multi-site trial it will be important to understand how the systems at the different sites are likely to recover.
Phase 2 will also explore the three way risk-benefit trade-off between:
- the number of incorrectly triaged patients produced by the DECODE process,
- the length of time it takes to reduce the queue size,
- and the amount of capacity released in the system.