Using linked health data to explore the epidemiology and impact of mental health and health behaviours in multiple sclerosis

Few population-based, methodologically rigorous studies have evaluated the association between mental health and health behaviours in MS. The goal of this dissertation is to contribute to the broader understanding of these relationships and their potential impact on MS. This dissertation was based on two main cohorts: 1) a multi-site clinic-based longitudinal cohort from across Canada; 2) a population-based health administrative and clinical cohort in British Columbia. A large (n=949) sample of MS patients were recruited from four Canadian MS clinics. Participants completed a series of questionnaires at three visits over two years. The prevalence of psychiatric comorbidities (depression [35%] and anxiety [54%]), and adverse health behaviours (smoking [24%] and non-adherence to disease-modifying therapies (DMTs) [22%]) was high. Alcohol dependence was associated with increased odds of anxiety (Odds Ratio (OR):1.84;95% confidence interval (CI):1.32–2.58) and depression (OR:1.53;95%CI:1.05–2.23), as was smoking (anxiety OR:1.29;95% CI:1.02–1.63; depression OR:1.37;95%CI:1.04–1.78). Non-adherence (<80% of expected doses reported as taken in the previous 30 days) was associated with alcohol dependence (OR:2.14;95% CI:1.23–3.75). When compared to adherence rates estimated from prescription dispensation information in pharmacy records in the prior year, self-reported non-adherence was found to be highly specific (0.96), but only moderately sensitive (0.38). Those who self-reported non-adherence were at high risk of non-adherence over the following year. We identified cases of mood or anxiety disorders using a validated algorithm applied to health administrative data. The presence of a mood or anxiety disorder was associated with significantly increased neurologic disability, measured by the Expanded Disability Status Scale (β-coefficient:0.45; p<0.0001) among 1250 incident cases of MS followed for an average of 9 years. Last, we identified incident cases of MS who did not access an MS clinic, using a validated algorithm applied to administrative data. Sex and socioeconomic status distributions were similar, but non-MS clinic users were older (46 vs 41 years, p