Prediabetes represents a critical transitional stage between normoglycaemia and type 2 diabetes mellitus (T2DM) and is particularly important among obese adults, who represent a high-risk group for accelerated metabolic dysregulation and progression to T2DM. Despite this, prediabetes remains underdiagnosed in sub-Saharan Africa, including Nigeria, limiting opportunities for early detection and prevention. This study assessed the prevalence of prediabetes and its associated risk factors among obese individuals in Benin City, Edo State, Nigeria. A cross-sectional analytical study was conducted among 131 obese adults residing in Benin City. Data on demographic, behavioural, anthropometric, and clinical characteristics were analysed using descriptive statistics and multivariable logistic regression with average marginal effects. Model diagnostics included variance inflation factors and the Hosmer–Lemeshow goodness-of-fit test, while model discrimination was evaluated using the area under the receiver operating characteristic curve (AUC). A Random Forest classifier was applied as an exploratory machine learning approach to assess the relative importance of predictors. The prevalence of prediabetes was 27 (20.61%). Multivariable analysis identified age (OR = 1.043; 95% CI 0.992–1.097), family history of diabetes (OR = 3.016; 95% CI 0.957–9.505; dy/dx = 0.1536, p < 0.05), and smoking (OR = 22.001; 95% CI 0.911–531.142; dy/dx = 0.4301, p < 0.05) as relevant factors. The large effect estimate for smoking should be interpreted cautiously, given the modest sample size and potential for sparse data bias. Diagnostic tests indicated no multicollinearity and good model fit (Hosmer–Lemeshow p = 0.354), with acceptable discrimination (AUC = 0.67). The Random Forest model identified age, body mass index, and waist circumference as the most influential predictors, although overall predictive performance was modest. Prediabetes is relatively common among obese adults in Benin City and is associated with life-course, behavioural, and familial risk factors. These findings may inform targeted screening and early detection strategies among high-risk obese populations. However, results should be interpreted in light of the cross-sectional design, modest sample size, and limited predictive performance of the machine-learning model.
Adejumo et al. (Tue,) studied this question.