Invited commentary: Identifying the improbable, the value of incremental insights

Abstract

There has been a long-standing debate about whether birth weight directly affects adult blood pressure, or whether the association is entirely mediated through current weight. In this issue of the Journal, Chiolero et al. (Am J Epidemiol. 2014;179(1):4-11) quantitatively evaluate whether bias from an unmeasured confounder of the relationship between current weight and current blood pressure could artificially create the appearance of a direct effect of birth weight on current blood pressure. Their results suggest that the conditions required to induce such a bias are improbable, given their assumptions. This insight moves the debate forward, and the next step is to evaluate their assumptions with similar quantitative rigor. It is also useful at this stage to reflect on the origins of the research question and the substantive implications of the debate. In particular, is birth weight actually a causal actor or a proxy for some other factor, such as fetal environment? In other words, would an intervention that changed birth weight affect current blood pressure? Speculation can produce many plausible scenarios, but our intuitions are often misleading. It is quantitative evaluation of these scenarios that provides incremental insights. Chiolero et al. illustrate an elegant application of theoretical methods to a substantive question. © 2013 The Author.