Development of risk prediction equations for 5-year diabetes incidence using Japanese health check-up data: a retrospective cohort study.

Objectives: This study aimed to develop risk prediction equations for the 5-year incidence of diabetes among the Japanese population using health check-up data. We hypothesised that demographic and laboratory data from health check-ups could predict diabetes onset with high accuracy.

Design: Retrospective cohort study.

Setting: Data from a health examination in Japan between 2008 and 2016.

Participants: Data were analysed from 31 084 participants aged 30-69 years. The presence of baseline diabetes and endocrine disease was included in the exclusion criteria, as were participants with missing data for the analysis. The study population was randomly divided into derivation and validation cohorts in a 1:1 ratio.

Primary outcome measures: The primary outcome was the incidence of diabetes at the 5-year follow-up, defined as a fasting blood glucose level ≥126 mg/dL, glycosylated haemoglobin A1c (National Glycohemoglobin Standardization Program (NGSP)) ≥6.5%, or initiation of diabetes treatment. Predictor variables included age, sex, body mass index, blood pressure, underlying diseases, lifestyle factors and laboratory measurements. The primary measure was the area under the receiver operating characteristic curve (AUC) for the predictive equations.

Results: In the derivation cohort, diabetes incidence was 5.0%. The prediction equation incorporating age, sex, body mass index, fasting blood glucose and glycosylated haemoglobin A1c showed good discriminatory ability with an AUC of 0.89, sensitivity of 0.81 and specificity of 0.81 in the validation cohort.

Conclusions: The equation with laboratory measures effectively predicted the 5-year diabetes risk in the general Japanese population. It has potential clinical utility for identifying individuals at high risk of diabetes and guiding preventive interventions.