Journal Club: Effect of comorbidity on mortality in multiple sclerosisMultiple sclerosis (MS) is a chronic demyelinating disease of the CNS estimated to reduce life expectancy by 7-14 years compared to demographically similar groups in the general population.1 While several disease complications pose a potential risk for mortality, results from large observational studies suggest that comorbidities substantially influence survival in MS. 2,3 In a recent Neurology ® article, Marrie et al. 4 applied an effective methodologic approach to assess the contribution of comorbidity to excess mortality in the MS population. Besides its considerable epidemiologic merit, the study has important implications for the management of patients with MS, particularly if closer surveillance and targeted interventions may reduce the burden of comorbidity among them and improve survival. performed a retrospective matched cohort study using population-based administrative data. The authors stratified cohorts by birth year in order to study temporal changes in survival, hypothesizing that the latter improves in the MS population over time. They also hypothesized that comorbidity shortens the survival of persons with MS. Cox regression analysis employed for statistical analysis provides us with a thorough insight into the effect of multiple covariates on the time to death, whereas some other studies traditionally use logistic regression, which estimates the probability of binary survival response (died vs alive) over the period of observation. An alternative method with time-to-event outcome is the log-rank test, which does not allow adjustment for confounding factors.
5METHODS The authors combined data extracted from 2 data sources. The first was administrative (health) data from the province of Manitoba, Canada, covering 98% of the population. The second data source was the Manitoba Vital Statistics Death Database, encompassing information on all deaths in Manitoba, including date and cause of death classified according to ICD-9 or ICD-10, depending on the time period. Using a validated administrative case definition, the authors identified all MS cases and up to 5 controls for each case, matched on sex, exact year of birth, and region of residence, from April 1, 1984, to March 31, 2012 To analyze the survival in both populations, the authors used univariate Cox regression with age as the time scale. Unlike the standard approach of using time on study as the time scale in a Cox regression (which means estimating time from entry into the study until the event of interest, typically death), and adjusting the model for age, this alternative approach to use age as the time scale allows a straightforward adjustment for multiple effects of aging process, and is recommended for analyzing epidemiologic cohort data. 6 On the other hand, the authors addressed a common source of bias in applications of survival analysis known as left truncation, which occurs when the individuals who have already passed the event of interest prior to study...