This commentary is on the original article by Holst et al. on pages 725-732 of this issue.Over the last quarter of a century, numerous reports on adverse outcomes associated with maternal febrile exposures have been published. Although results have been inconsistent, the general trend is suggestive of a harmful effect -a hypothesis supported by earlier animal and human studies. Furthermore, the central nervous system (CNS) seems especially susceptible to the detrimental effects of such fever.
1,2However, most of the studies which have assessed the effects of maternal febrile exposure on offspring are limited to outcomes that are evident at birth or shortly after. Few have evaluated longer-term developmental effects, including those reporting an association between maternal fever and developmental delay 3 or poor academic performance.
4In their study, Holst et al. 5 add to this body of knowledge by evaluating motor development in children in relation to in utero febrile exposure.The biological plausibility of an effect of thermal exposure during pregnancy on subsequent development is supported by studies on the offspring of guinea pigs that were exposed to hyperthermia.1 These offspring demonstrated significant reductions in brain stem and cerebellar size, and also displayed learning deficits proportional to the reductions in their brain weight. As with many biological systems, one would assume that if maternal fever caused physical changes in the CNS, then a dose-response relationship may exist. However, Holst et al. 5 failed to identify such an association, thus replicating a trend reported in a recent systematic review.2 Holst et al. stated that this suggests that it is not the fever but the underlying causes of the fever that might be the harmful exposure.Although possible, this assumption fails to consider limitations in design and the heterogeneity of the exposures and effects that are often reported. For instance, data on febrile episodes are usually ascertained in a retrospective fashion (regardless of any prospective recruitment and/or follow-up that may be part of the study design). Therefore, data on the dose or 'severity' of the exposure, often reported in terms of the 'highest temperature recorded', may suffer from recall bias. However, thermal dose is a function not just of the temperature but also of the duration and frequency of exposures.6 Retrospective ascertainment of the duration and frequency, along with highest temperature attained, is difficult and often results in missing data points or unreliable estimates.There are also other issues to consider. When, during embryogenesis, did the febrile exposure occur? What about effect modification caused by antipyretic use? If medications were used during pregnancy, were there any unintended long-term harmful effects of these on the offspring? What role do other factors, such as vitamin use, cigarette smoke, or infective agents, play? Are there compensatory mechanisms that hinder the observation of any effects that are temporally too removed from the...