We propose to continue our efforts to apply and further develop analytic methods for estimating the effect of public health interventions using complex longitudinal data from observational studies. Our focus will be the estimation of the causal effects and attributable risks of various interventions on the risk of coronary heart disease and related outcomes in prospective studies. Examples of hypothetical interventions are """"""""inactive individuals increase their activity to at least 30 minutes brisk walking per day,"""""""" """"""""everybody drinks 2 cups of coffee daily,"""""""" and """"""""statin therapy is initiated after a diagnosis of diabetes is made."""""""" We will estimate the effect of these hypothetical interventions using methods that overcome some key limitations of standard statistical methods. That is, the methods we propose allow us to estimate attributable risks of exposure, appropriately adjust for time-dependent confounding due to intermediate variables or variables affected by exposure, estimate the effect of joint interventions on multiple risk factors, and estimate the effect of dynamic interventions. Specifically, we will implement recently developed methods based on inverse probability of dynamic marginal structural models, and on g-estimation of nested structural models, to estimate the effect of hypothetical interventions under the assumption of no unmeasured confounding, and to assess the sensitivity of the estimates to violations of the assumption that can arise because of subclinical disease (""""""""reverse causation"""""""" bias) and exposure measurement error. The main reasons for the lack of widespread use of these methods is their perceived technical complexity, the scarcity of research on how to translate their theoretical development into practical applications, and the lack of standard software. We will adapt and further develop these methods to address realistic questions in cardiovascular epidemiology, and produce user-friendly and computationally-efficient software to apply them. Our research will produce state-of-the-art estimates of the effects of public health interventions on risk factors for coronary heart disease that use (i) some of the best observational cohorts available, and (ii) the most powerful methods available for causal inference from complex longitudinal data.
Public health policy requires estimating the effects of hypothetical interventions on risk factors for cardiovascular diseases and other health outcomes. Unfortunately, standard regression methods may result in biased effect estimates when applied to complex longitudinal data. We propose to develop and implement methods that, under assumptions less restrictive than those of standard methods, can be used to estimate the effect of hypothetical interventions. We will also develop software for these methods, and will make it publicly available.
Showing the most recent 10 out of 52 publications
