An Analysis of Change Mechanisms in the Integrating Depression and Diabetes Treatment Randomized Clinical Trial: A Longitudinal Structural Equation Model

Background: Individuals with multiple chronic diseases are at greater risk for healthcare challenges because they require complex care from multiple specialists. Further, the healthcare system is siloed which creates challenges with care coordination for these populations. Stigma, healthcare costs, and lack of access to specialty healthcare providers are additional barriers that exacerbate healthcare challenges for people with chronic diseases, especially for those in low and middle-income countries. This study is a secondary analysis from a randomized clinical trial in India that adapted a collaborative care model (CoCM) to provide high-quality, culturally tailored, integrated care for patients with diabetes, depression, and/or cardiometabolic risk factors.

Methods: The present study performed a secondary analysis of change mechanisms within the randomized clinical trial. The study randomized 404 individuals to either treatment as usual (208) or treatment as usual plus the adapted CoCM intervention (196) with 371 completing the trial. The present study examines whether depressive symptoms (using the Symptom Checklist (SCL)-20), health behaviors, and metabolic profiles mediate the effect of the CoCM intervention on diabetes outcomes (HbA1c). A longitudinal structural equation model was utilized to assess the effect of the CoCM intervention on diabetes outcomes and to analyze the aforementioned mediators. Before performing the mediation model, a confirmatory factor analysis (CFA) was performed on each of the latent variables individually and collectively to ensure an acceptable model fit.

Results: The CFA for the final model produced an acceptable model fit (X²(2985, N= 371) =5111.33, RMSEA= .042_(.040-.046); SRMR=.049). The SCL-20 was utilized to represent depressive symptoms with factor loadings ranging from .62-.87. There were also strong factor loadings for health behaviors (.40-.81) and metabolic profile (.63-.91). The mediation model showed a significant indirect effect where the CoCM was associated with decreased HbA1c from baseline to 6 months (f2= -0.11, p<.01). The standardized coefficient was of similar magnitude at 12 months but was no longer statistically significant (f3=-.10, p=.11). Depressive symptoms significantly mediated the relationship between the CoCM model and HbA1c outcomes. The path between depression and HbA1c was significant at 6 months (b2=.11, p=.02) but the association was no longer significant at 12 months (b3=.04, p=.51) indicating that depression was associated with elevated HbA1c but that the relationship weakened from 6-12 months. The path between the CoCM intervention and depression was significant at 6 months (a2=-.24, p<.01) and strengthened at 12 months (a3=-.34 p<.01) demonstrating a significant negative association between the CoCM and depressive symptoms. There was also a significant path between depression and health-promoting behaviors where increased depressive symptoms were associated with decreased engagement with positive health behaviors at 6 months (d2=-.12, p<.01) and at 12 months (d3=-.11, p<.01).

Conclusions and Implications: The CoCM intervention improved HbA1c for patients within a 6-month time frame. Both depression and health behaviors were significant change mechanisms influencing the effect of the CoCM with HbA1c. Social Workers act as care managers within CoCMs and can work to create and sustain positive treatment effects for patients with complex, comorbid conditions.