Development of a future prediabetes risk assessment model for individuals with normal glucose levels using data envelopment analysis

Abstract Identifying healthy individuals at risk of prediabetes for primary prevention is crucial because current tools often focus on secondary prevention. We investigated whether efficiency scores derived from data envelopment analysis (DEA) could predict the development of prediabetes in a healthy population. This historical cohort study analyzed annual health checkup data. Cox proportional hazards analysis was used to assess the relationship between the efficiency scores and incident prediabetes. A classification tree analysis was also performed, incorporating efficiency scores, hemoglobin A1c (HbA1c), and other diabetes-related variables. The cohort comprised 923 individuals (49.7% female) with a mean efficiency score of 0.72 (0.07). During follow-up, 175 participants developed prediabetes (79.3 per 1,000 person-years). A 0.1-point increase in efficiency score was associated with an adjusted hazard ratio of 0.51 (95% CI 0.39-0.68, P 0.0001) for prediabetes, whereas a 0.1% increase in HbA1c yielded an adjusted hazard ratio of 2.26 (95% CI 1.88-2.71, P 0.0001). The classification tree identified a high-risk group of 31 individuals (3.4%) with a sensitivity of 12.1% and specificity of 98.7%. Efficiency scores were linked to the 3-year risk of prediabetes in healthy subjects. The combined use of DEA and classification tree analysis is a potentially valuable approach for developing primary prevention strategies in clinical practice.