Relevant causal question - Epi Inquiry Accolade 01/22/2006: Kurth et al.

[Epidemiologic Inquiry 2006, 1: 8]

In epidemiology, we often launch an investigation to examine the simple causal association between exposure and disease outcome. However, too often, we can lose sight of the more important purpose- to determine "the relevant causal question"...

This week in AJE, Kurth et al. investigates the comparative results between multivariable adjustment of confounders via various methods of propensity score adjustment and propensity score weighting (inverse probability weighting standardization, aka marginal structural model), in the observational setting studying the use of a pharmacologic agent and risk of death.They found the overall results varied dramatically between the method of adjustment, ranging from RR=1 (conventional propensity adjustment) to RR=10 (IPW) for the same disease association. However, they discovered that IPW did report consistent RR=1 among those with propensity greater than 5% probability of treatment...

This article demonstrates the importance of the "relevant causal question" in epidemiology and clinical medicine, as those contraindicated with probability of drug treament less than 5% would not likely receive treatment in a realistic population. Thus, forcing every individual to take the drug in an entire population, regardless of clinical contraindications, would not be the relevant causal question in epidemiology, nor relevant to clinical practice. (That said, there are also important advantages of IPW standardization for time-dependent confounding, not discussed here.)

Thus, for clearly highlighting such key methodologic issues in epidemiologic research, the editors select as the Epidemiologic Inquiry Accolade: Investigation of the Week...

Results of Multivariable Logistic Regression, Propensity Matching, Propensity Adjustment, and Propensity-based Weighting under Conditions of Nonuniform Effect
Tobias Kurth, Alexander M. Walker, Robert J. Glynn, K. Arnold Chan, J. Michael Gaziano, Klaus Berger, and James M. Robins
American Journal of Epidemiology 2006 163(3):262-270