Measuring follow-up time in routinely-collected health datasets: Challenges and solutions

• DOI: 10.1371/journal.pone.0228545
• Publication Date: 2020-02-11T13:33:14Z
• Authors: Daniel Thayer; Arfon Rees; Jon Kennedy; Huw Collins; Dan Harris; Julian Halcox; Luca Ruschetti; Richard Noyce; Caroline Brooks
• Keywords: Data set

A key requirement for longitudinal studies using routinely-collected health data is to be able measure what individuals are present in the datasets used, and over time period. Individuals can enter leave covered population of administrative a variety reasons, including both life events characteristics themselves. An automated, customizable method determining individuals' presence was developed primary care dataset Swansea University's SAIL Databank. The covers only portion Wales, with 76% practices participating. start end date varies by practice. Additionally, change or Wales. To address these issues, two step process developed. First, period which each practice had available calculated measuring changes rate recorded time. Second, registration records individual were simplified. Anomalies such as short gaps overlaps resolved applying set rules. result analyses cleaned indicating dates individual. Analysis GP showed that 91.0% occurred within periods having algorithm. 98.4% those observed at same computed standardized solving this common problem has enabled faster development set. Using rigorous, tested, verifying study will also positively influence quality research.