Patterns of complication burden in people with diabetes: a latent class analysis approach

AIMS: To identify patterns of complication burden among individuals with diabetes mellitus based on sociodemographic, behavioral, and clinical characteristics, and to examine their co-occurrence with diabetes-related comorbidities. METHODS: This cross-sectional study was conducted during a diabetes health campaign in a municipality in southern Bahia, Brazil, involving 1,542 patients. Data were obtained through a standardized questionnaire and ophthalmological examinations. Latent class analysis was applied to identify subgroups with similar clinical characteristics. Models with two to four classes were estimated, with the two-class model presenting the most parsimonious and interpretable solution according to BIC. Associations between classes and comorbidities were estimated using Poisson regression with robust variance, adjusted for age and sex. RESULTS: Two classes were identified. Class 1 (86.6%) showed lower complication burden, with preserved vascular and sensory function. Class 2 (13.4%) was characterized by a higher frequency of ulceration, amputation, absent peripheral pulses, and impaired protective sensation. Individuals in Class 2 presented higher prevalence of cardiovascular disease (PR = 1.47), myocardial infarction (PR = 1.64), neurological disease (PR = 1.67), and retinopathy (PR = 1.63). CONCLUSION: The identified classes primarily reflect differences in peripheral complication burden, with higher co-occurrence of vascular and microvascular conditions in the more affected group. These findings describe patterns of complication clustering within a screening population and may support population-level strategies for identifying individuals with greater healthcare needs.