Systematic literature review of the health economic implications of early detection by screening populations at risk for type 2 diabetes.

Background: Undetected/uncontrolled diabetes is associated with substantial morbidity and mortality and consequent costs. Early detection through screening identifies patients at risk, allowing for earlier treatment initiation.

Objectives: To determine the economic impact of screening for type 2 diabetes (T2DM).

Data Sources: We systematically reviewed health economic analyses of screening programs for T2DM/pre-diabetes.

Study Eligibility Criteria: Published between 2000 and 2015 in any language. Articles must have reported costs of screening, test/patient outcomes and cost-effectiveness.

Participants And Interventions: Any type of screening (universal, targeted, opportunistic) was accepted.

Methods: Data were extracted from Scopus/Medline/Embase, then tabulated.

Results: There were 137 studies identified, 108 rejected; 29 were analyzed. Screening types included 18 universal, 8 targeted and 8 opportunistic. One study screened for pre-diabetes, 16 for T2DM and 12 examined both. Fourteen (48%) reported costs of screening only, 9 (31%) costs of screening combined with interventions and 6 (21%) presented all costs separately. Screening was compared to no screening in 13 studies (45%); screening was cost-effective in 8 (62%), not cost-effective in 4 (31%) and neither in 1 (8%). When comparing different screening methods, 6 found targeted screening was cost-effective compared with universal screening (none found the opposite), 2 found opportunistic superior to universal. Sensitivity analyses generally confirmed primary findings. Cost drivers included prevalence of T2DM/pre-diabetes, type of blood test used and uptake of testing. For optimal cost-effectiveness, screening for both T2DM and pre-diabetes should be initiated around age 45-50, with repeated testing every 5 years.

Conclusions/implications: Targeted screening appears to be cost-effective compared to universal screening.