Effects of Hypothetical Interventions on Ischemic Stroke Using Parametric G-Formula

Background and Purpose- Standard analytic approaches (eg, logistic regression) fail to adequately control for time-dependent confounding and, therefore, may yield biased estimates of the total effect of the exposure on the outcome. In the present study, we estimate the effect of body mass index, intentional physical activity, HDL (high-density lipoprotein) cholesterol, LDL (low-density lipoprotein) cholesterol, hypertension, and cigarette smoking on the 11-year risk of ischemic stroke by sex using the parametric g-formula to control time-dependent confounders. Methods- Using data from the MESA (Multi-Ethnic Study of Atherosclerosis), we followed 6809 men and women aged 45 to 84 years. We estimated the risk of stroke under 6 hypothetical interventions: maintaining body mass index <25 kg/m², maintaining normotension (systolic blood pressure <140 and diastolic <90 mm Hg), quitting smoking, maintaining HDL >1.55 mmol/L, maintaining LDL <3.11 mmol/L, and exercising at least 210 minutes per week. The effects of joint hypothetical interventions were also simulated. Results- In men, the 11-year risk of ischemic stroke would be reduced by 85% (95% CI, 66-96) for all 6 hypothetical interventions. In women, this same effect was estimated as 55% (95% CI, 6-82). Conclusions- The hypothetical interventions explored in our study resulted in risk reduction in both men and women.