Abstract
AIMS: Elevated depressive symptoms are common among adults with type 2 diabetes (T2D). In a secondary analysis from an RCT of a diabetes self-management support intervention that did not target depressive symptoms, we sought to determine if depressive symptoms were reduced by the intervention (i.e., depressive symptoms an outcome) or, alternatively, if intervention effects on hemoglobin A1c were lesser among persons with clinically elevated depressive symptoms (i.e., depressive symptoms an effect modifier). METHODS: We evaluated a text messaging intervention, REACH, in a diverse (half non-white, half underinsured) sample of N = 506 adults with T2D. Participants completed the Patient Health Questionnaire-8 (PHQ) and A1c tests at baseline and 6 months. We conducted a factor analysis to identify somatic- and cognitive-affective symptoms on the PHQ. We tested our hypotheses with regression models, using interaction terms and subgroup analyses. RESULTS: REACH improved depressive symptoms among participants with lower baseline A1c (<8.5%; β = -0.133, p = .007; cognitive β = -0.107, p = .038; somatic β = -0.131, p = .014) but not among participants with higher baseline A1c (≥8.5%; β = 0.040, p = .468). Baseline depressive symptoms did not modify the effect on A1c. CONCLUSIONS: We found support for the hypothesis that depressive symptoms - both somatic- and cognitive-affective - may be an outcome, rather than an effect modifier, of effective diabetes self-management support interventions.