Estimation of the prevalence of diagnosed diabetes from primary care and secondary care source data: comparison of record linkage with capture-recapture analysis

Study objective: To compare multiple source linkage and capture-recapture analysis in determining the current age and gender specific prevalence of type 1 and type 2 diabetes in a UK white population. To assess whole population trends in diabetes prevalence and treatment by comparison with previous studies. Design: Data were obtained from hospital sources and all 74 general practices in the study population. Analyses were carried out both by record linkage and by use of a two source capture-recapture model to correct for incomplete ascertainment. Setting: County of Clwyd, North Wales: total population 418 200. Main results: By record linkage the age adjusted prevalence of all diabetes was 2.04 (95% confidence intervals 2.00 to 2.09)% . Using the capture-recapture method it was 2.29 (2.24 to 2.33)%. From capture-recapture data the age adjusted prevalence of type 1 diabetes was 0.40 (0.37 to 0.43)% in men and 0.28 (0.25 to 0.30)% in women; the prevalence of type 2 was 2.03 (1.97 to 2.09)% in men and 1.67 (1.62 to 1.72)% in women. These figures represent an increase compared with previous surveys. The age specific prevalence of type 2 diabetes was greater in men in a ratio of approximately 1.5:1 and there were more patients treated by diet alone. Conclusions: Record linkage using multiple sources underestimates the prevalence of diabetes compared with capture-recapture estimates. The results suggest the prevalence of known diabetes in the UK has approximately doubled in less than 20 years. There is an increasing preponderance of male patients and of patients treated currently with diet alone.