Attributable Risk Estimation in Cohort Studies

Population attributable risk (PAR) or etiologic fraction is the proportion of a disease that could be prevented by elimination of a causal risk factor from the population. PAR is likely to be a function of time because both the prevalence of a risk factor and its effect on exposed individuals may change over time, as may the underlying risk of disease. In cohort studies with a wide range of follow-up, it may be important to account for this time dependency. Estimating the full PAR curve on some time scale of interest (e.g. age or time on study) can be more informative than considering a single value. Time-specific PAR can be estimated based on cumulative incidence adjusted for the competing risk of death. Cox models with time-dependent covariates can be used to obtain PAR estimates adjusted for confounders. The unique value of this approach is illustrated with examples that arose from our studies of heart failure in rheumatoid arthritis.