Aims:

A positive association between maternal smoking in pregnancy and childhood obesity has been confirmed in a number of meta-analyses of observational studies (Oken, Levitan et al. 2008, Ino 2010, Weng, Redsell et al. 2012). This empirical evidence for a causal association between intrauterine exposure to nicotine and childhood overweight later in life, however, has been questioned because of potential residual confounding (Raum, Kupper-Nybelen et al. 2011, Yang, Decker et al. 2013). Major concerns regarding possible residual confounding were based on the observation that children who were exposed to paternal or secondhand smoking while in utero as well as after pregnancy also had an increased risk of being overweight, and that this risk was similar to that observed in children with maternal intrauterine exposure to much higher levels of nicotine or other smoked products (Leary, Smith et al. 2006, Kleiser, Schaffrath Rosario et al. 2009, Plachta-Danielzik, Kehden et al. 2012, Harris, Willett et al. 2013). These results were surprising given that second-hand environmental tobacco smoke (ETS) exposure is not necessarily considered an intrauterine exposure, and the associated exposure dose of nicotine is much lower compared to that seen with direct maternal smoking (Florescu, Ferrence et al. 2007). Although mutual adjustment did not eliminate the effect of maternal smoking (Leary, Smith et al. 2006, Apfelbacher, Loerbroks et al. 2008, von Kries, Bolte et al. 2008, Kleiser, Schaffrath Rosario et al. 2009), the persistence of a similar effect size from paternal smoking either during or after pregnancy on childhood overweight raises concern about the potential of both ETS and maternal smoking in pregnancy to affect childhood overweight, and might be a reflection of a common unidentified family characteristic accounting for residual confounding, such as genetic or environmental factors, which could explain the effects of both paternal and maternal smoking during pregnancy.

An alternative explanation might be that there exists a very low threshold of smoke exposure from either father or mother leading to a priming effect that increases a child's risk to become overweight later on in life: in this case, the low exposure from ETS or paternal smoking would be sufficient for the priming effect and a linear dose effect of maternal smoking in pregnancy would not be observed. Indeed, it is still unclear whether or not a linear dose effect of maternal smoking on childhood overweight exists. Some studies suggest that a linear dose effect does exist (Power and Jefferis 2002, Hill, Shen et al. 2005), however, most confidence limits of the dose-related effect estimates overlap widely. The associations observed in other studies suggest instead a threshold effect, with a steep increase in effect at low exposure levels that flattens out at higher exposure levels (Reilly, Armstrong et al. 2005, Chen, Pennell et al. 2006).

Assuming a threshold effect, exposure to paternal smoking and ETS levels above the threshold could yield similar effects on childhood obesity as maternal smoking exposure in pregnancy, which would be compatible with the concept that low dose smoking exposure leads to an intrauterine priming effect which increases the offspring's risk to become overweight later on in life. This individual patient data (IPD) meta-analysis aims to identify whether a linear dose-effect relationship exists for maternal smoking during pregnancy on childhood obesity.