Background: The objectives of this study were to describe in detail the ascertainment and verification of prevalent and incident diabetes in the Dutch contributor to the European Prospective Investigation into Cancer and Nutrition (EPIC-NL cohort) and to examine to what extent ascertained diabetes agreed with general practitioner (GP) and pharmacy records.

Methods: In total, 40,011 adults, aged 21 to 70 years at baseline, were included. Diabetes was ascertained via self-report, linkage to registers of hospital discharge diagnoses (HDD) and a urinary glucose strip test. Ascertained diabetes cases were verified against GP or pharmacist information using mailed questionnaires.

Results: At baseline, 795 (2.0%) diabetes cases were ascertained, and 1494 (3.7%) during a mean follow-up of ten years. The majority was ascertained via self-report only (56.7%), or self-report in combination with HDD (18.0%). After verification of ascertained diabetes cases, 1532 (66.9%) [corrected] were defined as having diabetes , 495 (21.6%) as non-diabetic individuals, and 262 (11.5%) as uncertain. Of the 1538 cases ascertained by self-report, 1350 (positive predictive value: 87.8%) were confirmed by GP or pharmacist. Cases ascertained via self-report in combination with HDD were most often confirmed (334 (positive predictive value: 96.0%)).

Conclusions: Two out of three ascertained diabetes cases were confirmed to have been diagnosed with diabetes by their GP or pharmacist. Diabetes cases ascertained via self-report in combination with HDD had the highest confirmation.