In this study, Social Learning Theory was used to generate four psychosocial variables that are thought to influence adherence to the diabetes self-care regimen. These were: perception of self-efficacy, confidence in outcomes, social support and knowledge of the diabetes regimen. The purpose of this study was to examine the extent to which these psychosocial variables and selected demographic variables (age, duration of diabetes, presence of documented complications) effected glycemic control by altering participation in self-care.

A convenience sample of 118 inner-city black women with Type II-NIDDM receiving outpatient care at a large urban hospital were asked to complete measures of each of the psychosocial variables on two occasions, separated by an interval of 4-5 months. Body measurements (BMI; WHR) were obtained on all study participants.

Review of the body measurement data and diet logs identified the study sample as a high risk group based on their pattern of android obesity and high fat consumption. Bivariate analyses at Time 1, demonstrated that the measures of self-efficacy and confidence in outcomes were more strongly associated with the self-care regimen than knowledge and social support, with diet and exercise being more predictable than medications and home-testing. At Time 1 self-efficacy alone was able to explain 24% of the variance in diet, 53% of the variance in exercise, and 18% of the variance in home-testing. At Time 2, although self-efficacy remained fairly constant in its ability to explain home-testing (18%) the association between self-efficacy and diet completely dropped out (0.0%) and the association between self-efficacy and exercise diminished by slightly less than half (29%) suggesting variability within individuals in sense of self-efficacy over time and inconsistencies in the effect of this variable over the different regimen areas. At both Times 1 and 2, no effect on adherence behaviors is seen with social support. No clear relationship between the psychosocial predictors or adherence behaviors and glycemic control could be demonstrated through either bivariate or multivariate analyses.