Resilience and diabetes-related distress as factors for future glycemic trajectories following insulin initiation in type 2 diabetes mellitus

Psychosocial factors, particularly diabetes-related distress and resilience, play a crucial role in glycemic outcomes among individuals with type 2 diabetes (T2D). However, their impact during the initiation of insulin therapy remains underexplored. This study investigated whether baseline diabetes distress and psychological resilience predict longitudinal glycemic outcomes in insulin-naïve adults with T2D. A 24-month prospective observational study was conducted among 60 adults with T2D initiating insulin for the first time. Diabetes distress and resilience were assessed using the Problem Areas in Diabetes (PAID) and the Connor–Davidson Resilience Scale-10 (CD-RISC-10), respectively. HbA1c was measured at baseline and every 3 months. Linear mixed-effects models were used to examine associations between psychosocial variables and HbA1c trajectories, adjusting for baseline resilience and using clustered standard errors. Participants had a mean age of 58±10 years and baseline HbA1c level of 10.6±1.9%. Mean PAID and resilience scores were 33.3±15.4 and 28.3±6.7, respectively. HbA1c improved significantly after insulin initiation (mean decline=−2.2%, p<0.001). Distress, whether modeled dichotomously (PAID≥40 vs <40) or in tertiles, was not significantly associated with HbA1c slope (p>0.5). However, moderate distress (25.1–40.0) predicted persistently higher HbA1c compared with low distress (β=1.22, p<0.001). Resilience consistently predicted lower HbA1c across all models (β≈0.05 per point, p<0.001). While baseline diabetes distress independently did not predict the rate of HbA1c improvement, higher psychological resilience emerged as a consistent determinant of more favorable glycemic outcomes. Additionally, individuals with moderate levels of diabetes distress demonstrated the poorest long-term glycemic control, indicating that a tertile-based categorization of PAID scores may more effectively capture subgroups at risk for suboptimal outcomes.