Handling time varying confounding in observational research

Abstract

Many exposures of epidemiological interest are time varying, and the values of potential confounders may change over time leading to time varying confounding. The aim of many longitudinal studies is to estimate the causal effect of a time varying exposure on an outcome that requires adjusting for time varying confounding. Time varying confounding affected by previous exposure often occurs in practice, but it is usually adjusted for by using conventional analytical methods such as time dependent Cox regression, random effects models, or generalised estimating equations, which are known to provide biased effect estimates in this setting. This article explains time varying confounding affected by previous exposure and outlines three causal methods proposed to appropriately adjust for this potential bias: inverseprobability-of-treatment weighting, the parametric G formula, and G estimation. Many longitudinal studies aim to estimate the overall causal effect of a time varying exposure on the outcome, which requires adjustment for time varying confounding. However, in certain clinical scenarios one or more time varying confounders are affected by past exposure. In our clinical example, serum testosterone level is the time varying confounder because doctors are likely to use the value of testosterone to titrate treatment. The level of testosterone after baseline is, however, affected by baseline testosterone treatment. This pattern, for lack of a shorter term, is referred to as "time varying confounding affected by past exposure."The problem: controlling for time varying confounding affected by past exposure Several methods can control confounding in observational studies at the design and analysis stages, including restriction, stratification, regression modelling, matching, and propensity scoring. Summary poinTS• Many exposures of epidemiological interest are time varying, and time varying confounding affected by past exposure often occurs in practice• Time varying confounding affected by past exposure is often adjusted for by using conventional analytical methods such as time dependent Cox regression, random effects models, or generalised estimating equations in clinical research, which are known to provide biased effect estimates in this setting• Three causal methods have been proposed to appropriately adjust for time varying confounders that are affected by past exposure: inverse-probability-oftreatment weighting, parametric G formula, and G estimation

Population attributable fraction

“…Cohort studies are observational and thus liable to confounding,56 so crude (unadjusted) RR should not be used. An adjusted RR can be used in the Miettinen formula to estimate a valid PAF.…”

Some methodological issues in the design and analysis of cluster randomised trials

“…Randomised trials are subject to random (chance) confounding 8 as randomisation does not prevent confounding by baseline risk factors, but it only makes confounding random 1 2. The risk of random confounding is generally greater in cluster randomised trials than individual randomised trials as the number of clusters is often small 3.…”

Family structure effects on U.S. children’s well-being? Re-examining the family instability hypothesis

“…But it is the double nature of these "intermediate" variables, as effects of changes in family life, as well as causes of further changes in family life, that is both as confounders and as mediators, what generates the bias when unadjusted for or inapropriately adjusted for in statistical analysis. The only known set of methods that can overcome this insidious form of bias are referred to as G-methods (see Mansournia et al (2017) and Hernán and Robins (2006)). Lee and McLanahan (2015) discussed this issue and claimed that their estimates of negative effects on the socio-emotional domain among White children in a cohort of urban American families were robust to exposure-confounder feedback.…”