Estimating the prevalence of diagnosed diabetes in a health district of Wales: the importance of using primary and secondary care sources of ascertainment with adjustment for death and migration.

SummaryAims To determine diagnosed diabetic prevalence within our district (population 434 398) in 1996 using data from two sources. Methods A general practice audit comprising data on patients with diabetes from 61 (82%) of 74 general practices was linked to a record linkage-derived patient index in which data from secondary care and other sources underwent a process of probability matching to identify records relating to the same patient and to flag those with diabetes. By linking this dataset to a mortality dataset, patients known to have died before 1996 could be excluded. Age and sex-stratified emigration rates were applied to those identified by the hospital dataset for each year from 1991 onwards. Results A total of 386 988 residents (89.1%) were listed with a general practitioner participating in the audit, of whom 6050 patients were identified as having diabetes in 1996; a prevalence rate of 1.56%. From the hospital-based source, 7639 patients were identified who were alive in 1996, a period prevalence of 1.76%. By combining the two sources, and extrapolating the general practice audit to the population as a whole, a total of 10 530 patients were identified of whom 8735 were confirmed as still resident within South Glamorgan during 1996. This represented a period prevalence of between 2.01% to 2.42%. By applying age and sex-stratified migration rates to the diabetic population identified by hospital sources, a diagnosed diabetic population of 10,004 was identified, a prevalence of 2.3%. Conclusions This study demonstrates that to calculate the true prevalence of diagnosed diabetes from health sources, it is necessary to use both primary and secondary care sources. [ABSTRACT FROM AUTHOR]