Abstract
IMPORTANCE: No clear process exists for categorizing social risks in a way that informs effective social risk screening and intervention development. OBJECTIVE: To investigate social risk profiles and associations of those profiles with clinical outcomes in adults with diabetes using latent profile analysis. DESIGN, SETTING, AND PARTICIPANTS: For this cross-sectional study, a latent profile analysis was conducted using data for adults with type 2 diabetes collected at 2 primary care clinics in the Southeastern US from 2013 to 2014. Data were analyzed from November to December 2023. MAIN OUTCOMES AND MEASURES: Participants completed validated questionnaires for 26 social risk factors within 5 domains of social determinants of health: socioeconomic, neighborhood, education, food, and social and community context. In addition, participants completed questions that assessed psychological risk and behavioral risk. A 3-step latent profile analysis was used to identify different subgroups within the sample. Profiles were then regressed on outcomes of hemoglobin A1c (HbA1c), blood pressure, and quality of life. RESULTS: A total of 615 participants (mean [SD] age, 61.3 [10.9] years; 379 men [61.6%]) were included. Five latent class profiles were identified. The lowest risk group had significantly higher mental health-related quality of life compared with a group with higher neighborhood risk (β, 1.11; 95% CI, 0.67 to 1.55). The second group had low economic risk but high neighborhood risk and served as the reference group. The third group had high economic and neighborhood risk and had significantly higher blood pressure (β, 8.08; 95% CI, 2.16 to 14.01) compared with the reference. The fourth group had high psychological and behavioral risks but low socioeconomic and neighborhood risks. This group had significantly higher HbA1c (β, 0.47; 95% CI, 0.01 to 0.92) and lower mental health-related quality of life (β, -1.83; 95% CI, -2.41 to -1.24) compared with the reference. The highest risk group indicated high risk in all domains, had significantly higher HbA1c (β, 1.07; 95% CI, 0.50 to 1.63), and had lower mental health-related quality of life (β, -2.15; 95% CI, -2.87 to -1.42) compared with the reference. CONCLUSIONS AND RELEVANCE: These findings suggest that social risk profiles can be identified according to social, psychological, and behavioral risk domains and the health outcome of concern among adults with diabetes. Future work should consider the use of social risk profiles in intervention development and testing.