Neurological Conditions: estimating the prevalence in Scotland of selected conditions using GP and Hospital Admissions datasets

4 Comments including Data Improvement Plans

The data commitment in the Neurological Framework is ambitious. It requires improvements in gathering and extracting reliable data on neurological conditions. Developing reliable estimates of the prevalence of neurological conditions in the Scottish population is an important first step to improve data. Consequently, the Scottish Government's Neurological Care and Support - Framework for Action 2020-2025: Midpoint Progress Report has prioritised the need to understand how we can collect more reliable data on prevalence. This experimental statistics report supports this aim to develop more reliable data by comparing estimates from alternative data sources.

The incompleteness of the general practice dataset is a significant limitation in this report, with the data extract covering only 72.7% of registered patients. The accuracy of future data extracts would be improved by the participation of all 14 NHS Boards, and by improving consistency in recording of general practice patient lists to avoid exclusion of data following quality assurance processes.

Public Health Scotland are undertaking a wider programme of work to expand the available analysis relating to disease prevalence as recorded in general practice, as well as providing support to improve consistency of coding across Scotland. This report will provide additional context to those discussions.

Detailed diagnostic data are not currently available from hospital outpatient clinics, where a considerable proportion of people with neurological conditions are diagnosed. Improved availability of outpatient diagnostic data would enhance our understanding of the prevalence of specific neurological conditions. Timely feedback of these diagnoses to patients' general practices would improve the accuracy of general practice data.

Work is required to address the other challenges around data set out in the Scottish Government's Neurological Framework's Midpoint Progress Report, such as data linkage, or combining two or more sets of administrative data. This work could improve the quality and accuracy of the original datasets, clarify the extent to which they overlap and improve understanding of re-diagnosis. For example, the NHS Greater Glasgow & Clyde: Scottish Epilepsy Register healthcare improvement project, funded through the Framework, is working to improve local data collection and patient safety by linking prescribing data and the use of unscheduled care and hospital admissions data. The learning from this project will be relevant for scaling up and replicating data improvement at a national level.

There would be value in a wider group of analysts, researchers and clinicians considering how data of this kind could be improved. This could include consideration of patient pathways for specific neurological conditions and in which part of the health system those patients are most likely to be seen, diagnosed and treated. Depending on the nature of disease presentation, most cases will be defined based on general practice data, whereas for other conditions some cases will be defined based on hospital admissions data. It will also be important to consider how the overlap between cases defined using both methods differs between conditions. This is potentially important in terms of comparing these prevalence estimates with those from other sources.