Cunningham, Clare. 2008. Development of an electronic treatment decision aid for Parkinson's disease using mutli-criteria decision analysis. PhD Thesis, Cardiff University. |
![]() |
PDF
- Accepted Post-Print Version
Download (11MB) |
Abstract
Clinicians constantly weigh the relative importance of multiple attributes when they make decisions about how to treat patients. The literature shows that this is generally done in a relatively informal manner using intuition rather than evidence-based medicine. Decision analysis methods and computer decision support systems (CDSS) have been developed to help implement evidence-based medicine and to aid clinicians in their decision making. Multi-criteria decision analysis (MCDA) is a methodology used to break complex problems into manageable pieces, allow data and judgement to bear on them and then reassemble them to present an overall picture of the problem. The aim of the study was to use MCDA to develop a model to aid practitioners to choose the most effective drug treatments for Parkinson's disease (PD). A CDSS was developed from this model. Two surveys were sent to 304 neurologists, 88 geriatricians as well as Parkinson's disease nurse specialists across the UK to determine the criteria for the model. The seven steps of developing a MCDA model were carried out. A value tree was created from the criteria established from the surveys. The drugs were scored for their performance against the criteria using data from clinical trials and the weights were determined by the clinician for each individual patient. Software was developed using Excel and Visual Basic for Applications (VBA) to implement the functions of the model. A sensitivity analysis was carried out to determine whether the model was suitable for use with individual PD patients and whether the software was quick and easy to use. A total of 68 criteria were generated from the surveys, which was reduced to 11. This showed that clinicians were perhaps using personal experience more than evidence-based medicine. Scoring the data on the drugs showed that some drugs performed either better or worse than expected. The weights were phrased so that users could use swing-weighting to weight the criteria for their importance to each patient. The combined scores and weights were calculated by Excel and the result returned on the screen to the user by VBA. An expert panel carried out the sensitivity analysis and showed that there were some issues with the scores developed, such as potential bias from the trials data and that not all the expected criteria were included in the model, for example bradykinesia and tremor were not included. However, the expert panel felt that the software was quick and easy to use and overall the principle of the model was approved, subject to some modifications. Therefore, a model was successfully developed for Parkinson's disease using MCDA and a CDSS developed to implement the model's functions. The model needs further refinement but has the potential to be successfully used in a clinical setting. MCDA could additionally be used to develop models for other diseases.
Item Type: | Thesis (PhD) |
---|---|
Status: | Unpublished |
Schools: | Pharmacy |
Subjects: | R Medicine > RC Internal medicine > RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry |
ISBN: | 9781303196904 |
Date of First Compliant Deposit: | 30 March 2016 |
Last Modified: | 12 Feb 2016 23:12 |
URI: | https://orca.cardiff.ac.uk/id/eprint/54501 |
Actions (repository staff only)
![]() |
Edit Item |