Automated differentiation of incident and prevalent cases in primary care computerised medical records (CMR)

Conference item

Identifying incident (first or new) episodes of illness is critical in sentinel networks to inform about the seasonal onset of diseases and to give early warning of epidemics, as well as differentiating change in health service utilization from change in pattern of disease. The most reliable way of differentiating incident from prevalent cases is through the clinician assigning episode type to the patient's computerized medical record (CMR). However, episode type assignment is often made inconsistently. The objective of this collaborative study between the Royal College of General Practitioners Research and Surveillance Centre (RCGP RSC), University of Surrey and the National Physical Laboratory (NPL) is to develop a methodology to reconstruct missing or miscoded episode types. The data, gathered from the RCGP RSC network of over 230 practices, are analyzed and poor episode typing reconstructed by disease type. The methodology is tested in practices with good episode type data quality. This method could be used to improve prediction of epidemics, and to improve the quality of historical rates retrospectively.

Authors: Smith, N; Livina, V; Byford, R; Ferreira, F; Yonova, I

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: IOS Press
• Host title: Studies in Health Technology and Informatics
• Journal: Studies in Health Technology and Informatics
• Volume: 247
• Pages: 151-155
• Publication date: 2019-04-24
• Acceptance date: 2019-04-24
• Event location: Netherlands
• DOI: 10.3233/978-1-61499-852-5-151
• ISSN: 0926-9630
• Pmid: 29677941
• ISBN: 9781614998518
• Language: English
• Keywords: FFR