Prevalence of Diabetes in Low‐ and Middle‐Income Countries' Incarcerated Populations: A Systematic Review and Meta‐Analysis

Background and Aims Diabetes mellitus (DM) is a major global health concern, particularly in resource‐constrained settings. Incarcerated populations in low‐ and middle‐income countries (LMICs) face increased health risks and limited care access, yet remain understudied. No prior systematic review has synthesized prevalence data for this group. This review aims to estimate diabetes prevalence among incarcerated individuals in LMICs. Methods We conducted a systematic review and meta‐analysis per PRISMA and GATHER guidelines. PubMed, Scopus, Embase, Web of Science and regional databases were searched till May 2025. Random‐effects models with Freeman‐Tukey double arcsine transformation were applied in R (version 4.5.1). Publication bias was assessed using DOI plots. Study quality was evaluated with the Joanna Briggs Institute tool, and certainty of evidence was appraised using an adapted GRADE approach. Results Twenty‐three studies (107,473 participants) were included. Mean participant ages varied from 29 ± 9.6 to 40.5 ± 9.8 years. The tentative pooled estimate of diabetes prevalence was 3.7% (95% CI: [2.5%–5.1%], Prediction Interval [0%–12.2%]), with high heterogeneity ( I 2 = 95.4%). The pooled prevalence estimate was highest in Asia (5.2%), particularly Pakistan (9.8%), and lowest in South America (2.3%). The DOI plot revealed major asymmetry, indicating a high possibility of publication bias. Overall certainty of evidence was rated as very low for diabetes prevalence due to risk of bias and publication bias. Interpretation of the pooled estimate is limited by substantial heterogeneity and limitations in the primary studies, including differences in diabetes ascertainment methods. Conclusion Diabetes is an important health concern among incarcerated populations in LMICs. Differences in diabetes ascertainment methods, participant characteristics and correctional healthcare systems may have potentially contributed to the variability of the pooled estimates. A hypothesis is that there may also be potentially limited detection or underdiagnosis which possibly could have contributed to the observed prevalence estimates.