Hypothetical interventions to prevent stroke: an application of the parametric g-formula to a healthy middle-aged population

The effects of interventions on multiple lifestyle and metabolic risk factors, initiated in midlife or later in a healthy population, on the long-term risk of first-ever stroke is not known. A particular methodological challenge in observational studies is to estimate the unbiased effect of a time-varying exposure in presence of time-varying confounders, if those confounders are affected by prior exposure. In such cases, the parametric g-formula can be applied to estimate an unbiased effect. We applied the parametric g-formula to estimate the 18-years (1994-2012) cumulative stroke risk under different scenarios of hypothetical interventions on levels of blood pressure, cholesterol, weight, physical activity, smoking and alcohol intake; and compared these to the observed scenario, to calculate the population risk ratios and risk differences. Among 14,796 eligible participants in the prospective, population-based Tromsø study (baseline mean age 46.1 years, 51% women), the observed 18-years stroke risk was 5.9%. A feasible joint hypothetical intervention on six lifestyle and metabolic risk factors would reduce the 18-year stroke risk by 32% (95% confidence interval 16, 44). A combination of more intensive interventions reduced the estimated 18-years stroke risk by 64% (95% confidence interval 40, 80). Blood pressure reduction and quitting smoking significantly reduced the risk when applied separately.