Mechanisms of patient health behavior change in a randomized controlled trial of a spouse-assisted intervention.

Spouse-assisted interventions can improve health behaviors, but mechanisms of action are unknown. This study evaluated mediators of dietary and physical activity outcomes during a spouse-assisted intervention to improve low-density lipoprotein cholesterol. This is a secondary analysis of data from a randomized controlled trial comparing usual care (n = 128) to a spouse-assisted lifestyle change intervention (n = 127) comprising nine monthly goal setting telephone calls to participants and support planning calls to spouses over 11 months. Structural equation modeling was used to examine if the intervention influenced the putative mediators of participant self-efficacy and perceived spousal support at 6 months (i.e. action test); if changes in putative mediators at 6 months were associated with changes in diet and physical activity outcomes at 11 months (i.e. conceptual test); and if treatment condition effects on outcomes at 11 months were mediated by its effects on the 6-month putative mediators (i.e. indirect effects test). Participants were 94.9% male, 64.9% white and were 61.3 years old on average. The action test showed that the intervention increased dietary self-efficacy (p < .001) and perceived spousal support for diet (p < .001) and physical activity (p < .01) at 6 months. The conceptual test showed that increases in participant physical activity self-efficacy at 6 months were associated with increases in physical activity frequency (p = .01) and duration (p = .04) at 11 months; other putative mediators were not associated with changes in outcomes at 11 months. The indirect effects tests did not support a mediating role for self-efficacy or perceived spousal support. Intervention-induced changes in spousal support and dietary self-efficacy did not translate into behavior change. Other mechanisms may be driving behavior change.

Duke Scholars

Author Heather Alyse King Population Health Sciences

Author Cynthia Jan Coffman Biostatistics & Bioinformatics, Division of Biostatistics