OCR Text |
Show on the relevant self-management behavior as well as the model chi-square to assess the overall goodness-of-fit of the model for each self-management behavior. This analysis was chosen to see if membership in a certain subgroup significantly increased (or decreased) the likelihood of participation in a self-management behavior. For example, to what degree were males more or less likely to conduct daily blood glucose monitoring than females and how this impact might vary in strength and direction vis-avis daily feet checking, following a meal plan, and exercising regularly? Analysis of the simultaneous effect of the respective covariates across a range of four self-management behaviors enables observation of the degree of selectivity or consistency in the strength and direction of effects. Also analyzed was which covariates exert statistically significant effects on a specific self-management behavior. FINDINGS Table 1 shows the results obtained from the multivariate logistic regression analysis of the four diabetes self-management behaviors. The findings point to a high degree of selectivity in the effects of covariates. No covariate exerts a statistically significant effect on all of the self-management behaviors considered in this study. Better self-reported health increases the likelihood of engaging in each self-management behavior; however, in no case is the impact statistically significant. The most prevailing effect is observed for gender, exerting statistically significant effects on three of the four self-management behaviors. However, the direction of effect varies across the behaviors. Being male significantly lowers the odds of daily blood glucose monitoring and following a meal plan by nearly one-half (odds ratio of 0.52; p .01 and 0.56; p .01, respectively), but nearly doubles the likelihood of exercising at least three times per week (odds ratio = 1.94; p .01). Four covariates (age, diagnosis duration, education, and diabetes education in the past 12 months) each exert a statistically significant effect on two of the behaviors. Each of these covariates has a statistically significant influence on daily blood glucose monitoring. Three of the four covariates (age, education, and diabetes education) also exert a statistically significant effect on daily feet checking, whereas the fourth (diagnosis duration) exerts its second effect on following a meal plan. However, there are some notable variations in direction across self-management behaviors. A fuller comparative discussion of the nature of these effects across the self-management behaviors follows. Although age and education significantly impact both daily blood glucose monitoring and feet checking, there are some interesting variations in the patterning. Being 50 or more years of age is linked to a statistically significant lowering of the odds of daily blood glucose monitoring by more than one-half (odds ratio=0.45; p .01), but a statistically significant near doubling of the likelihood of daily feet checking (odds ratio=1.91; p .05) relative to their counterparts who are less than 50 years of age. Most interesting is the counterintuitive finding that those adults with diabetes who have a college degree or higher are significantly less likely to conduct daily blood glucose monitoring and feet checking than their counterparts with less than a college degree. Having a college degree or higher lowers the odds of daily blood glucose monitoring by about two-fifths (odds ra-tio=0.59; p .05) and daily feet checking by one-half (odds ra-tio=0.50; p .01) vis-a-vis those who are not college graduates. Duration of diagnosis also impacts two of the self-management behaviors, namely, daily blood glucose monitoring and following a meal plan, but the direction varies. Being five or more years post-diagnosis significantly elevates the likelihood of daily blood glucose monitoring by nearly two-thirds (odds ra-tio=1.65; p .05), but significantly lowers the likelihood of following a meal plan by nearly one-half (odds ratio=0.54; p .01). Diabetes education significantly and positively increases the likelihood of both daily blood glucose monitoring and feet checking. In particular, diabetes education nearly triples the odds of daily blood glucose monitoring (odds ratio=2.71; p .001) and nearly doubles the odds of daily feet checking (odds ra-tio=1.87; p .05). Although being associated with a slight increase in the likelihood of following a meal plan and exercising, the effect of diabetes education on these two behaviors is not statistically significant. Not being currently treated for depression has a statistically significant positive impact on just one of the four self-management behaviors; namely, exercising. Not being depressed significantly doubles the odds of exercising at least three times per week (odds ratio=1.96; p .05). The model chi-square is statistically significant for all four self-monitoring behaviors suggesting a good fit in each case. The level of significance is below .001 level of probability for daily blood glucose monitoring and below the .05 level of probability in the case of the other three behaviors. There are five statistically significant effects on daily blood glucose monitoring. Being a younger adult, female, five plus years being post-diagnosis, not having a college degree, and having had diabetes education in the past year all significantly increase the likelihood of engagement in daily blood glucose monitoring. There are two statistically significant effects for each of the remaining self-management behaviors. Three statistically significant effects are evident in thee case of daily feet checking. Being an older adult, having fewer years of education, but having had diabetes education within the past year significantly increase the odds of daily feet checking. Females and those less than five years post-diagnosis are significantly more likely to follow a meal plan than males and those five or more years post-diagnosis. Finally, being male and not depressed significantly increases the odds of exercising three or more times a week. CONCLUSIONS The main objectives of this study were to explore the influence of a set of seven covariates across four important diabetes self-management behaviors. The findings point to a complex patterning of effects across the four self-management behaviors. Statistically significant effects varied across the behaviors. Moreover, consistency in strength of effect did not necessarily imply consistency in direction. The greatest degree of differentiation was observed for daily blood glucose monitoring, one of the more complex regimens with attendant high costs. There were five statistically significant covariates, and the model provided the best fit of the four self-management behaviors. Nonetheless, even feet checking, with its simplicity and low effort, elicited three statistically significant effects. The results of this analysis also suggest that strategies aimed at increasing daily blood glucose monitoring need to pay particular attention to older adults, males, those who are less than five years post-diagnosis, and, surprisingly, those with a high level of formal education. At the same time, diabetes education does play a very positive role in increasing the likelihood of daily blood glucose monitoring. Therefore, increased efforts should be placed in expanding diabetes education to insure greater adherence to blood glucose monitoring. Diabetes educa- 38 Utah's Health: An Annual Review Volume IX |