Stuebe A, Lee K.
Pediatrics. 2006 May;117(5):1848-9
Letters to the Editor
In their recent Pediatrics article, Hauck et al1 reported on the association between pacifier use at last sleep and sudden infant death syndrome (SIDS) and recommended universal pacifier use to reduce the risk of SIDS.
It is far from clear, however, that pacifiers play a causal role in SIDS. Each of the 9 studies in the meta-analysis conducted by Hauck et al used a case-control design to assess whether a pacifier was used during the last sleep before SIDS death, or the sleep among control infants. In such a study design, there is potential for differential recall bias between parents of infants with SIDS and those of control infants on the basis of both the timing of the questioning and the parents’ perceptions of whether pacifier use is “right.” If an infant usually used a pacifier, it seems unlikely that control parents could recall whether the infant did or did not accept a pacifier in an otherwise unremarkable sleep period. For a “usual-pacifier” infant, we suspect that they would be biased toward reporting that a pacifier was used.
In addition, the putative exposure, “no pacifier in last sleep,” is not a homogenous entity. For example, consider an infant with a viral illness and a decreased arousal threshold who then falls asleep in a caregiver’s arms without his or her pacifier. Later that night, with arousal dampened by the illness, the infant dies as a result of SIDS. In this case, “no pacifier in last sleep” is a marker for decreased arousal. What if an infant rejects the pacifier at last sleep because of an abnormal neurologic phenomenon that is, in fact, a harbinger of SIDS? Or, what if an infant dies and, in an agonal tongue thrust, ejects the pacifier? It is unclear from the methods reported in the reference studies whether any or all of these scenarios would be considered SIDS cases with no pacifier in last sleep. In none of them would the lack of pacifier use have caused SIDS.
In their discussion, Hauck et al reject the hypothesis that infant illness results in decreased pacifier use, noting that 1 of the 9 studies included in the meta-analysis controlled for symptoms of illness.2 But what if the “symptom” was merely falling asleep more easily than usual? Would parents report that as “illness?” We question whether it is possible to control for such a subtle confounder in a retrospective case-control study.
Hauck et al go on to argue that there is a dose-response relationship between pacifier use and SIDS risk, because “use for all sleep periods is necessary.” This is not a dose-response effect. A dose-response effect implies that more frequent pacifier use confers greater protection, a finding which would require assessment of the duration and frequency of pacifier use among cases and noncases. Such data do not appear anywhere in the study.
In fact, it may be just as reasonable to conclude, on the basis of the finding that pacifier nonuse at last sleep increases the risk of SIDS, that no infants should be given pacifiers, because “pacifier-dependent” infants may be more likely to succumb to SIDS when the pacifier is inadvertently forgotten.
The authors also invoke biological plausibility to support a causal relationship between pacifier use and SIDS. One of their primary references, an article on arousal threshold in pacifier users versus non–pacifier users, assesses a convenience sample of 56 infants at a median age of 10 weeks.3 Infants who used pacifiers had lower arousal thresholds than infants who did not use pacifiers. Among infants who were breastfed, this difference was not significant, although the sample size was small. It is important to note that, because these infants were not randomly assigned to pacifier use, it is possible that infants who slept more deeply never “needed” a pacifier. Thus, pacifier use at 10 weeks may be a marker for a fussier, more wakeful infant, not for a protection against deep sleep.
The consistent association between pacifier use at last sleep and reduced SIDS risk is interesting. However, we disagree with the authors’ assertion that there is adequate evidence to recommend universal pacifier use in the absence of prospective cohort studies or randomized clinical trials showing an inverse association between pacifier use and SIDS.
Given the tenuous relationship between pacifiers and SIDS protection, we are concerned that the American Academy of Pediatrics chose to include a recommendation supporting pacifier use in their revised SIDS guidelines.4 The medical literature is littered with well-intentioned interventions born of case-control data and plausible explanations. National organizations need to proceed cautiously before making recommendations that can have a major public health impact.