Diabetes Res Clin Pract. 2025 Apr;222:112113. doi: 10.1016/j.diabres.2025.112113. Epub 2025 Mar 18.
ABSTRACT
AIMS: To compare an algorithm for identifying individuals with diabetes using linked administrative health data with an Australian diabetes registry (National Diabetes Services Scheme, NDSS).
METHODS: This prospective cohort study linked baseline survey data for 266,414 individuals aged ≥ 45 years from the 45 and Up Study, Australia, to administrative health data sets. An algorithm for identifying individuals with diabetes was developed based on a combination of claims for dispensed insulin and glucose lowering medicines, diabetes-related hospital admissions, and diabetes-specific Medicare claims. Using the algorithm, participants were classified as 'certain', 'uncertain' or 'no' diabetes. The algorithm was compared to NDSS registrations as the reference standard.
RESULTS: Amongst the 45 and Up Study cohort, there were 53,669 individuals with certain diabetes identified by the algorithm, and 35,900 NDSS registrants. Compared with the NDSS, the sensitivity of the algorithm was 96.9% (95%CI 96.7-97.1) and specificity 91.8% (95%CI 91.7-91.9). Of the 53,699 individuals with diabetes identified by the algorithm, 34,864 were registered to the NDSS (PPV = 64.9%, 95%CI: 64.6-65.2).
CONCLUSIONS: This study demonstrates the value in using linked administrative data for diabetes monitoring and surveillance. National estimates using the NDSS alone may underestimate the diabetes burden by up to 35%.
PMID:40113176 | DOI:10.1016/j.diabres.2025.112113