Sudden Infant Death Syndrome, Bedsharing, Parental Weight, and Age at Death

DISCUSSION

The suggestion that some SIDS-like deaths are attributable to adult-infant bedsharing and consequent suffocation is not new, and 2 recent reports using data from the Consumer Product Safety Commission have stimulated heightened interest.^30,31 The first of these described 2178 mechanical suffocation deaths of infants during the 16 years, 1982 to 1995, most often related to bed structure or bedding, but in 180 attributed to parental overlying on the death certificates.³⁰ The second report concerned only the deaths of 515 children <2 years old while sleeping in adult beds, 1990 through 1997.³¹ Of these, 394 were ascribed to mechanical entrapment and 121 to overlying by a bedsharing adult. Of interest in relation to our report is that in both of these studies, the ages at death attributed to overlying were significantly lower than those resulting from entrapment and other causes. Although these reports are uncontrolled, the statements regarding overlying at death provided by the parents leave little doubt that the majority, if not all, of these deaths were caused by suffocation. Indeed, as the authors note, the numbers of cases are likely to be grossly underestimated. In this population we found that 30 (41%) of the 74 infants for whom bedsharing status was known were sleeping with 1 or more persons at the time of death. Because there have been no published reports on the prevalence of this practice in this or a comparable US population <6 months old, the significance of this observation is unclear.

A recent population-based study from 5 regions in England compared 325 SIDS cases with 1300 healthy controls using data obtained by parental interviews.³⁵ Univariate analysis showed a significantly increased risk associated with cosleeping on a bed or sofa with a parent. However, the authors found that the significance of the risk of bedsharing but not that of sofa sharing disappeared after multivariate adjustment for certain parental and environmental factors. However, this lack of significance is subject to challenge because the multivariate analysis adjusted for factors associated with bedsharing but not directly related to SIDS, such as parental fatigue, alcohol use, and household overcrowding. These factors enhance the likelihood of bedsharing or of failing to recognize overlying of the infant, but cannot be considered causes of SIDS by themselves. An additional problem is that 20% of cases reported as sudden unexplained death in infancy by local authorities were eliminated from the study for reasons that were not specified.

This study conclusively demonstrates significantly different survival curves and significantly younger mean age at death of the bedsharing groups within this cohort. Hoffman and Hillman have stated that “since the age at death distribution of SIDS is one of its most characteristic features, it is important to examine other risk factors by age at death. Any risk factor found to shift the age at death to either an earlier or later mean age could have important etiologic implications.”⁷ In view of this, these results suggest that bedsharing by an infant and an adult may be a risk factor for SIDS-like deaths. Our evidence provides an epidemiologically plausible construct based on 3 interrelated observations. First, SIDS-like deaths occurred significantly earlier in the presence of parent-infant bedsharing than in the absence of bedsharing. Second, the data suggest that SIDS in association with bedsharing occurred earlier if the bedsharing parent was heavy. Third, large mothers of SIDS infants were significantly more likely to be bedsharing at death than were small mothers of SIDS babies. It is important to recognize that none of these observations supports our conclusion by itself; it is the combination of the 3 that is meaningful. This construct is additionally supported by the lack of a demonstrated association between young age at death in the presence of large maternal size without bedsharing and only a marginal association between bedsharing in the absence of large maternal size. Assuming that these results reflect causation in some instances, the most likely mode of death is respiratory restriction caused by overlying, and/or suffocation. Thus, the results meet criteria for causation including consistency of the results of different analytic approaches to these data, the strength of the association which makes chance unlikely, and coherence (biological and clinical plausibility).⁴³ Additional support for the bedsharing hypothesis is the lack of alternative explanations for our results, particularly explanations that would fit the strong relationships found among bedsharing, age at death, and maternal weight.

We believe that these relationships make other explanations unlikely; as an example, there might be a propensity for mothers to take younger infants to bed with them, and accordingly younger victims of idiopathic SIDS would be more apt to be bedsharing at death. However, the fact that bedsharing was associated significantly more often with early death in infants of heavy mothers but not with infants of small mothers makes this explanation highly unlikely. We discarded the possibility that, for some unknown reason, infants of obese mothers are at enhanced risk of SIDS including its occurrence at an earlier age because of the disparity in age at death between infants of bedsharing and nonbedsharing heavy mothers. Another explanation considered and rejected is the use of drugs and/or alcohol during pregnancy, in part because it was reported significantly more often among smaller mothers than among heavier bedsharing mothers of SIDS infants. Substance abuse is associated with lower prepregnancy weight and poor weight gain in pregnancy.^44,45 Although substance use among bedsharing women of lower weights might also predispose to overlying or suffocation because of impaired maternal responses to stimuli, the data do not show a significant difference in age at death associated with drug use. Because of the absence of any drug or alcohol users in the group of larger bedsharing women, this study provides no information on the potential combined effect of size and substance related impairment.

The relationship of race and bedsharing is difficult to assess in this study because race-related prevalence data on bedsharing for this age group do not exist. However, these findings are in general agreement with those reported by Lozoff et al⁴⁶ who found that black mothers in Cleveland were more likely to start bedsharing when their infants were newborns. Although the average age of SIDS is reported to be 5 to 7 days earlier for black infants than for white infants,⁴⁷ no significant race-related difference was observed in this study. The facts that blacks were more likely to sleep with their infants but that race does not confound the relationship between bedsharing and age at death raises the question of whether these results reflect a cultural practice among a population at high risk of SIDS or whether the practice itself contributes to that high risk. The former is unlikely to explain the observed phenomenon given the demonstrated weight relationship. The relationship to weight has additional implications, in that there is strong evidence, at least in the United States, of secular increases in obesity in adults of childbearing age.⁴⁸

A limitation of these results is that the population studied is urban and disproportionately minority and low income. This is clearly an important high-risk group for SIDS, but extrapolation to other populations requires caution. Additionally, for some data, such as substance use, maternal self-reporting had to be relied on in the majority of cases. The effect, if any, of classifying 10 infants as nonbedsharing in the absence of specific information on this factor would be to produce a conservative statistical estimate by creating a bias toward the null.⁴⁰

An additional limitation of this study is the lack of a control group and hence the inability to estimate the magnitude of the increased risk introduced by bedsharing. However, a major problem in determining whether parent-infant bedsharing sometimes results in suffocation of the infant is the difficulty in selecting a meaningful control group. Undoubtedly, the vast majority of parents bedshare with their infants with frequencies ranging from rarely to always and for periods of time varying from a few minutes to the entire night, thus blurring the distinction between bedsharing and nonbedsharing situations. Therefore, the practice is not a dichotomous variable, and would be nearly impossible to quantitate in controls.³¹ However, in the absence of classical controls, as pointed out by Freedman,⁴⁸ descriptive epidemiologic studies often provide data to answer specific questions.

The size of the cohort studied was limited, but it was sufficient to produce statistically significant results, and did comprise all cases of SIDS deaths during the study period. The accrual of additional cases was compromised by the initiation of aggressive local educational efforts, which altered ascertainment and classification of these deaths. The 88 infant deaths in which bedsharing was identified as a possible contributor since the cutoff date for this study cannot be included for these reasons.

Another consideration is whether the lack of maternal interviews by study personnel weakens the conclusions. However, it is established practice for the initial attendants at such unfortunate and unexpected events as SIDS to collect pertinent information. It is unlikely that this information, obtained immediately, is inaccurate. The results obtained from later interviews, in contrast, might be understandably colored by guilt, denial, or other sources of bias.

Although this study indicates that bedsharing is related to a proportion of SIDS-like deaths, it does not explain the mechanism of death. At least 4 possibilities exist singly or in combination, including direct airway obstruction as a result of full or partial overlying, restrictive compression of the infant's chest by the weight of the adult, hypoxia and/or hypercapnea from rebreathing, and overheating; the first 2 being the most probable.

Our interpretation of these results does not conflict with previous findings, such as the role of prone sleeping, and is supported by other studies. Furthermore, our data are consonant with studies regarding smoking and SIDS. We believe that our study provides additional evidence that the spectrum of deaths classified under the rubric of “SIDS” is not a single entity. Instead, there are cases of SIDS-like deaths that have alternative explanations that require supplementing clinical and pathologic studies with thorough pursuit of the circumstances surrounding the death. We do not suggest that idiopathic SIDS is nonexistent, but rather that a proportion of SIDS deaths is related to external factors including bedsharing as well as sleep position, and may be preventable. Furthermore, although it may be desirable to designate these deaths as SIDS and thus provide comforting reassurance to the parents, misclassification compromises pursuit of the causes of true SIDS.⁴⁹