Impact of Depression on Co-morbid Diabetes in Primary Care

Abstract

This thesis was based on three interrelated studies and offers the following contributions to the epidemiology of depression and diabetes in Canada:

1. The first study centres around a validation study to determine the sensitivity and specificity of the CPCSSN depression case finding algorithm in patients with diabetes and to estimate the true prevalence of depression in patients with diabetes. The sensitivity of the depression algorithm was found to be 50.7% (39.1-62.1%) and specificity 92.5% (86.6-96.0%). The prevalence of depression in patients with diabetes using the CPCSSN case finding algorithm was 19.8% and the validation sample produced a higher estimate of this prevalence rate of 34.5% (28.4-41.2%). Different sensitivities were observed for males and females, and for seniors and those <65 years of age.

2. The objective of the second study was to determine, among Canadians with diabetes, what is the relative risk of having poor blood sugar control for those patient with depression compared to those without depression. Factors that impact this relationship and potential treatment disparities for those with depression were also explored. Patients with depression were more likely to have poor blood sugar control with a crude relative risk of 1.28 (1.16, 1.42) and a fully adjusted relative risk of 1.22 (1.10, 1.34). Covariates positively associated with poor control included being male, the number of years with diabetes, and having comorbid dementia. Covariates associated with better control were age over 65 years, having comorbid hypertension, osteoarthritis, or Parkinson’s. Effect modification was observed for both being obese and being a senior. No evidence of diabetes treatment disparities were found for patients with depression compared to those without depression.

3. The final study adjusts the relative risks from the second study for the misclassification observed in the validation study. The sensitivity was used to estimate the potential strength and direction of the relative risk had the misclassification not occurred. The crude corrected estimate of relative risk of poor blood sugar control for patients with diabetes with and without recent depression was 5.99 (5.47, 6.55), which was much higher than the crude relative risk estimated in the second study.