MRC Centre for Causal Analyses in Translational Epidemiology University of Bristol Bristol, United Kingdom Julia.Mackay@bristol.ac.uk (Smith)
To the Editor:
In their excellent review on negative controls,1 Lipsitch and colleagues1 state they “are not aware of an example of the use of a negative control exposure to detect confounding” in an epidemiologic setting. One such explicit application of this approach relates to the use of paternal exposures as a negative control exposure for maternal exposures considered to have an intrauterine influence on offspring outcomes.2 Paternal smoking, as a negative control, may show associations with offspring outcomes similar to those of maternal smoking if the associations are generated by shared familial confounding factors or by parental genotypes transmitted to the offspring. If, however, there is an intrauterine influence, then only the maternal exposure would be expected to show an independent association with the outcome.
Figure 1 demonstrates that the effect of maternal smoking during pregnancy on offspring birthweight is considerably greater than that of paternal smoking during pregnancy, and mutual adjustment (to take account of associative mating by smoking) attenuates the paternal effect to zero. This is in line with the considerable body of evidence that maternal smoking has a causal effect on offspring birthweight. There has been enthusiasm for the proposition that fetal exposure to maternal smoking leads to increased obesity in offspring.3 Although maternal smoking during pregnancy does indeed demonstrate the expected association, the strength of association with paternal smoking during pregnancy is similar before and after mutual adjustment (Fig. 2).4 This casts doubt on the causal nature of the association between intrauterine exposure to maternal smoking and offspring adiposity.
As Lipsitch and colleagues1 argue, the use of negative controls could be usefully expanded in epidemiology. If associations are found with such controls, this does not invalidate the observation under interrogation but does encourage further intense scrutiny of potential biases and confounding that may underlie what is seen.
George Davey Smith
MRC Centre for Causal Analyses in Translational Epidemiology
University of Bristol
Bristol, United Kingdom
1. Lipsitch M, Tchetgen Tchetgen E, Cohen T. Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology. 2010;21:383–388.
2. Davey Smith G. Assessing intrauterine influences on offspring health outcomes: can epidemiological findings yield robust results? Basic Clin Pharmacol Toxicol. 2008;102:245–256.
3. Power C, Jefferis BJ. Fetal environment and subsequent obesity: a study of maternal smoking. Int J Epidemiol. 2002;31:420–421.
© 2012 Lippincott Williams & Wilkins, Inc.
4. Leary S, Davey Smith G, Ness A. ALSPAC Study team. Smoking during pregnancy and components of stature in the offspring. Am J Hum Biol. 2006;18:502–512.