The Contribution of Prone Sleeping Position to the Racial Disparity in Sudden Infant Death Syndrome: The Chicago Infant Mortality Study

DESIGN AND METHODS

RESULTS

Four hundred sixty-nine infants met the study inclusion criteria and were enrolled in the study. The leading cause of death was SIDS, found for 260 (55%) of the infants. The other causes of death were as follows: undetermined (50, 11%), respiratory tract infections (34, 7%), asphyxia (25, 5%), congenital anomalies (19, 4%), other infections (13, 3%), other injuries (12, 3%), malnutrition attributable to neglect (12, 3%), dehydration (10, 2%), prematurity-related causes (9, 2%), and fire-related causes (7, 1%). (Totals do not add up to 100% because of rounding.) The 260 SIDS infants are the subject of this study.

By definition, all infants who died from SIDS had a full autopsy, scene investigation, and medical record review. Infants who appeared to have died from SIDS but for whom scene investigations could not be completed were designated as undetermined. Ninety-eight percent of autopsies were completed within 1 day of pronouncement of death, and 100% of death scene investigations were completed within 3 days, with the median time to completion 1 day. The 2-week follow-up family interview was conducted for 198 (76%) of the case infants. The median follow-up time for the family interview was 15 days from the date of death. The main reason for not conducting the family interview was inability to locate the primary caregiver (n = 41, 16%), followed by parental refusal (n = 21, 8%). Responders and nonresponders were similar in age, marital status, parity, education, Kessner Index, and infant’s age at death.

The majority of the SIDS infants were African American (75%, n = 195), 13% were white, Hispanic (n = 34), and 12% were white, non-Hispanic (n = 31). There was a perfect match on race-ethnicity for control infants. The age difference between case and control infants was within 2 weeks for 88% of the infant pairs. Only 1 control infant was over the 28-day target (29 days younger than the case infant to whom he was matched). There were no differences in the closeness of age matching by the age at death of the case infants. The mean age of the case infants was 89 days (standard deviation [SD]: 57) and of the control infants 86 (SD: 57, difference not significant). Likewise, matching by birth weight category was very close. Among the case infants, 72 (28%) were <2500 g, 179 (69%) were 2500 to 4000 g, and 9 (3%) were >4000 g. The comparable figures for the control infants were 27%, 70%, and 4%, respectively (difference not significant). The mean birth weights of the cases and controls were 2813 g (SD: 705) and 2919 g (SD: 597), respectively (P = .06).

The majority (88%) of SIDS deaths occurred within 4 months of age, with a peak between 1 to 3 months. Seven percent occurred in the first month, and 2 infants (0.8%) died in the first week (ages 5 and 6 days). The distribution for African American infants is skewed slightly to the younger ages compared with the other infants (Fig 1), but this difference is not statistically significant (P = .67). The mean age of the African American SIDS victims was 12.1 weeks (SD: 7.8) and for the others it was 14.4 weeks (SD: 9.1; P = .048). Differences in gender or gestational age did not explain this difference. However, infants of all race-ethnicities who bed shared were found to be younger than those who did not bed share during the last/reference sleep, and a greater proportion of African American infants bed shared (58.0%) compared with the others (29.2%; P = .001). The typical seasonal pattern of SIDS deaths was found, with a fall-winter predominance (64%).

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Age distribution of SIDS case infants by race-ethnicity.

The relationship between several potentially confounding sociodemographic factors and SIDS is shown in Table 1. The control mothers were slightly older than case mothers (24.8 ± 6.5 years vs 23.2 ± 5.4, respectively; P < .002). The lowest risk category for maternal age was 30 years or older, and the highest risk was found for mothers between 20 and 24 (OR: 2.8; 95% CI: 1.6–4.8). Single marital status was associated with greater than a threefold increased risk (OR: 3.5; 95% CI: 2.1–5.7), and less than a high school education and inadequate prenatal care were each associated with a sixfold increased risk (OR: 5.7; 95% CI: 3.4–9.4 and 6.5, 3.5–12.0, respectively). Other sociodemographic factors found to be associated with increased SIDS risk were higher parity, intermediate or low job prestige score, chronic maternal unemployment, and receiving public assistance (Table 1). Although there was a slight predominance of males among the SIDS victims, the OR (1.3) did not achieve statistical significance (95% CI: 0.9–1.8).

The ORs for prone sleeping did not vary significantly by study year, so all years were combined for subsequent analysis. There was over a twofold increased risk associated with being placed for last sleep in the prone position and SIDS compared with the nonprone positions (OR: 2.4; 95% CI: 1.6–3.7; Table 2). Adjusting for the 4 confounders (maternal education, age, marital status, and adequacy of prenatal care) changed the OR and CI minimally (OR: 2.5; 95% CI: 1.5–4.3). No significantly increased risk associated with the side position over supine position was observed (unadjusted OR: 1.0; 95% CI: 0.6–1.6 and adjusted OR: 1.2; 95% CI: 0.7–2.3). In a multivariate model that included other factors associated with increased or decreased risk of SIDS, the OR for prone sleeping over nonprone sleeping was even higher (OR: 4.0; 95% CI: 1.8–8.8). These factors were soft bedding, maternal smoking during pregnancy, pillow use, bed sharing, upper respiratory infection in the previous 2 days, and pacifier use when put down for the last sleep. There was no interaction observed between maternal smoking and prone sleeping position in the risk of SIDS; namely, prone sleeping was a risk factor for infants regardless of their mother’s smoking status.

The proportion of SIDS infants who had been placed prone for last sleep increased with increasing age; this was not found among the control infants (Table 3). Prone sleeping was not associated with a significantly increased risk among the infants under 1 month of age, but was for the infants 1 to 3 months old and ≥4 months (OR: 2.1; 95% CI: 1.2–3.5 and 4.3, 1.4–13.3, respectively).

The OR for prone sleeping differed significantly according the race-ethnicity of the infant. For African Americans, the unadjusted OR was 1.8 (95% CI: 1.2–2.6), whereas for all others it was 10.3 (3.2–33.8) (Table 4). The respective adjusted ORs were 1.7 (95% CI: 1.1–2.8) and 12.2 (95% CI: 2.0–73.2) (Table 4). Although the rates of prone sleeping for both groups was almost identical among the SIDS infants, 58% among African Americans and 55% among the others, the rates for the control infants were very different, 43% compared with 12%, thus accounting for the different ORs.

The population attributable risk for prone sleeping was 31% based on the multivariate model. That is, 31% of the SIDS cases can be attributed to the risk factor of prone sleeping. The attributable risk was 19% for African Americans and 12% for the others.

Infants usually placed for sleep in the prone position (defined as the usual position placed for sleep in the previous 2 weeks) were at almost twice the risk for SIDS compared with infants sleeping in other positions (unadjusted OR 1.9; 95% CI: 1.2–2.9 and adjusted OR: 1.8; CI: 1.03–3.1; Table 5). No additional increase in SIDS risk was found for infants who were usually placed nonprone and had been placed prone for last sleep compared with those who had been placed prone usually and for last sleep (Table 5). The sample size for these analyses was smaller (n = 198 matched pairs) because there were fewer SIDS mothers participating in the follow-up family interview during which usual sleep position questions were asked.

Mothers were asked during the follow-up interview how they decided what sleep position to use for their infant. This information is available for 193 case mothers and their matched control mothers. A higher proportion of the control mothers (64%) compared with case mothers (46%) reported being advised about sleep position by a doctor or nurse after delivery (P < .001; Table 6). For both groups, the side position was the most commonly advised, followed by the prone position. For cases and controls combined, a greater proportion of African American mothers than non-African American mothers reported being told to use the prone position (25% vs 7%, respectively; P = .01). Over 94% of case and control mothers reported that they had followed the medical advice. For those mothers who were not told what position to use after delivery or did not follow their doctor’s or nurse’s advice, “past experience” was the most important factor influencing their choice. The advice of friends or relatives and the comfort of their infants were also common reasons for both case and control mothers.

DISCUSSION

In this comprehensive case-control study of SIDS in a major urban center, the majority of SIDS deaths were among African American infants. Demographic and descriptive characteristics of the infants were similar to those found in other populations and settings both within and outside the United States. These include the majority of deaths occurring by 4 months of age, cold season predominance, and several maternal factors including younger age, single marital status, lower education, less adequate prenatal care, higher parity, and lower job prestige.^{11,27–29,36–40} Although this sample demonstrated an earlier peak than the typical 2- to 4-month peak, there are indications from other reports that the SIDS age distribution may be changing to a higher proportion of younger cases.⁴¹ Our finding that the age of death is younger for African Americans compared with Caucasians is consistent with other studies.^42,43 Bed sharing was associated with lower age at death for all ethnic groups in this sample, similar to findings from Norway.⁴¹ Because this practice was more common among the African Americans, as reported also by others,⁴⁴ it may account in part for their younger age.

The unadjusted OR of 2.4 for prone sleep, primarily determined by the African American majority, is similar to that of international studies published in 1992 or earlier, when prone sleeping in the general population was highest.⁴⁵ In this study, higher rates of prone sleeping among the African American control infants persisted as late as 1996, 2 years after the launching of the national “Back to Sleep” campaign. Our finding of higher rates of prone sleeping among African Americans compared with Caucasians is consistent with other studies.^46–50 This study additionally provides evidence directly linking prone sleeping to SIDS among African Americans. The population attributable risk for prone sleeping was higher for African American infants than for others, taking other risk factors for SIDS into account. Thus, prone sleeping could be responsible, in part, for the higher prevalence of SIDS observed among African Americans.

In the current study, prone sleeping became a stronger risk factor with increasing age of the infant. It was not associated with a significantly increased risk in the neonatal period, which could be attributable to the small number of deaths in this age group. Consistent with our data, others have reported highest risk for SIDS related to prone sleep during specific age periods, including 10 weeks or more in England⁵¹ and 13 to 24 weeks in the Nordic countries.³⁷ This implies that infants may be most vulnerable to SIDS from prone sleeping during a critical period of maturation of physiologic functioning,^37,52–54 and emphasizes the need to maintain the recommendation for supine sleeping until at least 6 months of age, and perhaps throughout infancy.⁵⁵

In contrast to other studies, we did not find that infants who usually sleep nonprone are at higher risk for SIDS when placed prone for last sleep compared with those who usually sleep prone.^37,56 This analysis was based on a smaller sample size because the questions about usual sleep habits were asked in the follow-up interview in which participation was reduced, and lack of statistical power could have contributed to our finding. It is unlikely that the results were biased by nonresponse, because responders and nonresponders were similar on a number of demographic and maternal factors (marital status, parity, education, prenatal care index, maternal age, maternal smoking, and age of infant at death) as well as infant care factors (bedding softness, breastfeeding, prone sleeping, and bed sharing). Additional research is needed to better understand the risk of SIDS associated with placing infants prone who are more accustomed to nonprone sleeping.

Side sleeping was not found to be associated with a significantly increased risk of SIDS, similar to findings from a Tasmanian study.⁵⁷ The OR of 2.3 found in multivariate analysis, however, was similar to other studies showing ORs associated with side sleeping of ∼2.^58–61 Because the current study was designed during a time when side sleeping had not yet been identified as a risk factor, the sample size calculations did not take this into account, and thus our sample may have been too small to detect a statistically significant effect.

As in any retrospective study, recall bias may occur if mothers of SIDS infants recall exposures more thoroughly than mothers of unaffected, healthy infants, thus resulting in an apparent association when there is none. However, prospectively collected data on sleep position have confirmed results from previous retrospective studies, indicating that recall bias has not been a major problem in case-control studies of SIDS.^60,62 Furthermore, it has been shown that the length of time lapsed between the exposure and the recall has a greater influence on recall accuracy.⁶³ In this study, parents of both SIDS victims and control infants were interviewed about their infant’s sleep position within a short time of the sleep period. Thus, we do not believe that recall bias was a problem in this study.

There were some early delays in interviewing control families in relation to the timing of the matched case infant’s death. This preceded the start of the “Back to Sleep” campaign, and was well before interventions in Chicago were initiated, particularly those targeting minority communities. Therefore, we believe that there was little likelihood that this would have biased the results, further supported by an analysis that revealed no differences in prone rates by the time between the infant death and control interview.

The control selection process presented a number of challenges, particularly in light of the well-known difficulty in recruiting persons of color into research studies,^64–70 as well as the study’s requirement to match on race-ethnicity, birth weight, and age. A number of methods to enroll control infants were tested before choosing the method selected as the most feasible and effective. Because the number of potential controls contacted was large compared with those who were actually enrolled, the possibility of nonresponse bias exists. However, the analyses comparing the nonresponders to responders are reassuring, with differences noted in only 3 variables, ie, the responders were more educated, had better prenatal care, and were of lower parity. Appropriate adjustments to outcomes of interest were made to control for these differences.

What can explain the greater use of the prone position among the case infant families? Although the choice of sleep position is likely multifactorial, the medical community seems to have played a significant role. Fewer case mothers than control mothers reported being advised about sleep position in the hospital after the birth of their infants. Side position was most often recommended for both groups, reflecting the recommendations at the time (the recommendation to use the supine position only was made in 1996),⁷¹ as well as the probable continued fear among nursing personnel to recommend the supine position because of perceived aspiration risk. This was demonstrated in a survey of Iowa maternity hospitals, in which 90% recommended either the side or supine position for infant sleeping, with fear of aspiration given as the major reason for this policy.⁷² In the current study, almost all the mothers reported that they followed the advice given, indicating the strong influence of medical personnel. This was similar to findings by Ray et al⁷³ comparing white higher income private patients with African American lower income clinic patients that a majority in both groups followed medical advice given in the hospital, although fewer of the latter mothers had received instructions to use a nonprone position. Likewise, Johnson and colleagues⁴⁸ found in a survey of inner-city parents of color that medical personnel were influential in promoting the supine sleeping position, although a high proportion of mothers were still resistant. In the current study, African American mothers of both case and control infants were more likely to have used the prone position than other racial-ethnic groups.

CONCLUSION

Prone sleeping position does seem to explain, in part, the higher SIDS rates among blacks in Chicago, and it is likely that this conclusion can be generalized to the disparities in SIDS rates found elsewhere in the United States. Dispersion of the “back to sleep” message, both directly to parents and caregivers and to medical personnel and others who influence parental infant care choices, is necessary to increase supine sleeping among African American infants. Because contact with hospital personnel after normal deliveries is brief, sleep position and other related recommendations must be delivered consistently and reinforced at each medical encounter. It is hoped that these measures will help reduce the gap found in the rates of SIDS among African American infants compared with other racial-ethnic groups in the United States.