Objectives. The Back to Sleep campaign has been credited with recent declines in the incidence of sudden infant death syndrome. Using survey data for the 1996–1998 birth cohorts, this epidemiologic study examines infant sleep position in a large, population-based sample.
Data and Methods. Data concerning infant sleep position are drawn from the 1996–1998 Pregnancy Risk Assessment Monitoring System for 15 states. Weighted multiple logistic regression analysis is used to examine correlates of infant sleep position.
Results. The prevalence of prone infant sleeping significantly declined between 1996 and 1998 (adjusted odds ratio [AOR] = 0.70; 95% confidence interval [CI] = [0.63, 0.78]). African Americans were more likely than non-Hispanic whites to sleep prone, (AOR = 1.45; 95% CI = 1.33,1.59), and were less likely to sleep supine (AOR = 0.52; 95% CI = 0.48, 0.57). Hispanic/Latinos were less likely overall than non-Hispanic whites to sleep prone (AOR = 0.81; 95% CI = 0.69, 0.95), but were also less likely to sleep supine (AOR = 0.78; 95% CI = 0.69, 0.87). Adherence to sleep position recommended by the American Academy of Pediatrics increased sharply among Hispanic/Latino infants. Very low birth weight infants and infants in larger families were less likely to sleep in the recommended supine position. Infants born between 1001 and 1500 g (AOR = 0.67; 95% CI = 0.57, 0.79), and extremely low birth weight infants between 500 and 1000 g (AOR = 0.57; 95% CI = 0.45, 0.72) were especially unlikely to sleep supine. Infants in households with more than 3 other children (AOR = 1.72; 95% CI = 1.08, 2.74) were more likely to sleep prone.
Conclusions. The prevalence of supine infant sleep increased between 1996 and 1998. Low adherence to sleep position recommendations of the American Academy of Pediatrics among African Americans, very low birth weight infants, and infants in large families remain public health concerns.
Sudden infant death syndrome (SIDS) is the most common cause of postneonatal infant death in the United States. In 1992, the American Academy of Pediatrics (AAP) issued a recommendation that “healthy term infants” be placed on their backs (supine) to sleep and to avoid the prone sleeping position.1 This Back to Sleep campaign has been credited with reduced SIDS incidence during the 1990s.2–5 Although the overall population prevalence of prone infant sleep has declined, adherence to Back to Sleep recommendations has varied by race/ethnicity, parental education, and other factors.3
Recommendations for sleep positions among low birth weight (LBW) infants born weighing <2501 g have been less clear. Prone sleep provides several potential benefits for preterm infants that must be weighed against heightened SIDS risks. As noted by Martin and colleagues,6 these benefits include improved pulmonary function and improved oxygenation. Supine sleeping has been associated with less quiet sleep and greater variability in heart rate.7
Out of concern that premature infants might incur special risks from supine sleeping, the initial AAP recommendation did not apply to “premature infants experiencing respiratory distress.”1 However, large subsequent declines in SIDS incidence encouraged a different view. The revised AAP policy statements specifically removed the exception for premature infants, stating that the benefits of supine sleeping outweigh the potential risks, except in cases of relatively rare specific diagnoses.8 The SIDS Global Strategy Task Force (Developmental Physiology Working Group) has also emphasized the importance of supine sleeping for preterm infants, including neonatal intensive care unit patients, except for those with specific contraindications to supine sleeping.9
LBW infants experience significantly higher SIDS incidence, a risk tightly correlated with prematurity.10–14 High SIDS incidence among LBW infants, very low birth weight (VLBW) infants born weighing <1501 g, and extremely low birth weight (ELBW) infants born weighing <1001 g, persists despite increased overall survival within these infant populations.13 We hypothesize that infants in these LBW categories are less likely than normal birth weight infants to sleep supine.
Race/ethnicity has also been identified as an important SIDS risk. African Americans experience more than twice the SIDS incidence of non-Hispanic whites.15 Much of the disparity is associated with other risk factors such as economic status.15 Yet these results are consistent with prior studies indicating that African Americans are less likely than non-Hispanic whites or Hispanic/Latinos to sleep supine.2,16,17 Young mothers, and mothers with fewer years of completed education, appear especially likely to place their infants prone.5,13,18 Pollack and Frohna15 investigate how race/ethnic disparities changed between the 1989 and 1996 birth cohorts. These authors indicate that relative risks of SIDS failed to decline for African Americans; in fact, the relative risk increased slightly over this time period.
In contrast to African Americans, the rate of SIDS deaths among Hispanic/Latino infants is significantly below that found among non-Hispanic whites.11,12,19,20 We hypothesize that African Americans are less likely to sleep supine, while Hispanic/Latinos are more likely to sleep in the AAP-recommended position, compared with non-Hispanic whites. We explore these questions regarding changes in infant sleep position using a large, population-based study of the 1996–1998 infant birth cohorts.
Data regarding infant sleep position are drawn from the 1996, 1997, and 1998 Pregnancy Risk Assessment Monitoring System (PRAMS) study. PRAMS is a large, population-based study of infant health status and behavioral risks run by the Centers for Disease Control and Prevention. Each month, PRAMS surveys a random sample of mothers who have given birth during the previous 2 to 6 months, using birth certificates. Fifteen states provided pertinent infant sleep position data for at least 1 year during the 1996–1998 sample period. Most states oversample mothers of LBW infants. Some states oversample women of selected racial/ethnic groups. The survey is administered in written form, with 2 subsequent mailings to nonresponders. Phone call follow-up is done for those not responding to the mailings. Regarding sleep position, mothers are asked, “How do you put your new infant down to sleep most of the time?” Potential responses include: back, stomach, or side.2,21
Data for 55 263 live infants were included in the descriptive analysis. We examined the changes in prone and supine sleep position for a number of subgroups within the PRAMS dataset. We augment the descriptive analysis with a multiple logistic regression analysis to control for potential confounding effects. A total of 50 904 observations included complete data for all covariates used in the multivariate analysis.
Given the complex study design of the PRAMS study, the STATA (Stata Press, College Station, TX) software package is used to compute weighted multiple logistic regression analysis, with corrected standard errors to account for sampling and design effects. Because parental characteristics and infant health factors might have different implications for different choices of infant sleep placement, we run separate models of prone and supine sleeping. (A somewhat more complex, multinomial logistic regression model yields similar results.) We computed means and adjusted odds ratios (AORs) using the SVY (survey estimation) procedures in the STATA software package, version 6.22
We estimated a main-effect specification to indicate the impact of basic sociodemographic and infant health factors. We also estimated a time-interacted specification to identify changes in infant risk over the study period. Within the latter specification, we interacted year with race/ethnicity, birth weight, maternal education status, and maternal smoking during pregnancy. These variables are of clinical and policy interest, and also appeared most likely to reveal important patterns based on previously published research.
Within the overall PRAMS population, the reported prevalence of prone sleep declined from 26.9% to 19.1% over the 3-year period. Figure 1 shows the prevalence of prone sleep for the overall population and for 11 important subgroups in each year of data. African American infants had a higher initial baseline of prone sleep and remained higher over the 3 years. Although rates of prone sleeping for non-Hispanic whites were similar to those for Hispanic/Latinos in 1996, by 1998 the Hispanic/Latino prevalence was much lower than other groups.
Consistent with AAP recommendations, the prevalence of prone sleep declined in each subgroup shown in Fig 1, though the reported decline for infants whose mothers smoked during pregnancy was notably smaller than that observed in other groups. In addition to the decline seen for Hispanic/Latinos, proportional declines were especially sharp among infants born to college graduates.
The prevalence of supine sleep, the position recommended by the AAP, increased from 30.9% to 48.9% from 1996 to 1998. Figure 2 shows the descriptive data for supine sleeping, with the prevalence of supine sleep increasing in each population subgroup. However, baseline prevalence and subsequent prevalence changes notably differed across the categories. For example, the prevalence of supine sleep increased much more sharply among Hispanic/Latinos than among African Americans. ELBW infants had the lowest reported supine sleep prevalence.
Factors Predicting Prone Sleeping Position
Results of the multiple logistic regression analysis are shown in Table 1. Controlling for many potential confounders, prone sleeping prevalence sharply declined between 1996 and 1998 (AOR = 0.70; P < .001). African Americans were more likely than nonHispanic whites to sleep prone (AOR = 1.45; P < .001). African American infants were more likely to live in large families, which is also a risk factor for prone sleep. Hispanic/Latino infants were less likely than non-Hispanic whites to sleep prone (AOR = 0.81; P < .01).
Girls were less likely to sleep prone (AOR = 0.88; P < .001). Infants with multiple siblings were significantly more likely than only children to sleep prone. Point estimates increased steadily with family size, so that children with 3 or more siblings had an AOR of 1.72 (P < .05). State effects also mattered. Infants from Arkansas, Florida, and Oklahoma were more likely to sleep prone.
Maternal education, maternal age and marital status, and infant birth weight had only a small impact on the likelihood of prone sleeping. Maternal smoking during pregnancy—another important SIDS risk factor—appeared unrelated to the likelihood of prone infant sleep. Contrary to our initial hypothesis, LBW, VLBW, and ELBW were not associated with prone sleep. We found a small opposite effect among infants weighing between 1501 and 2500 g.
Factors Predicting Supine Sleeping Position
The second column of Table 1 shows the parallel analysis for the supine position. Most notably, the prevalence of supine sleeping increased sharply between 1996 and 1998 (AOR = 2.07; P < .001). However, African Americans were significantly less likely to sleep in the recommended position (AOR = 0.52; P < .001).
Contrary to previous expectation, Hispanic/Latinos were less likely than non-Hispanic whites to sleep in the AAP-recommended position (AOR = 0.78; P < .001). Aside from the overall changes in the prevalence of prone and supine sleeping, the most striking feature of the PRAMS data are the remarkable change in sleep position among Hispanic/Latino infants. Hispanic/Latino infants in the 1996 birth cohort were more likely than the general population to sleep prone. By the 1998 birth cohort, prone sleeping had declined in prevalence by more than half, while the prevalence of supine sleep had doubled.
Infants in large families were less likely than others to sleep supine, although our estimates were less precise than we obtained for the prone sleep model. Maternal educational level was a small, though statistically significant predictor of supine sleeping. Infants born to younger mothers were less likely to sleep supine.
Although LBW was not a significant predictor of prone sleep, VLBW and ELBW infants were significantly less likely to sleep supine. Infants born weighing between 1001 and 1500 g (AOR = 0.67; P < .001) and infants born weighing between 500 and 1000 g (AOR = 0.57; P < .001) were especially unlikely to sleep supine.
Changing Risk Factors Over Time
Table 2 shows the results of the interaction specification, in which race/ethnicity, maternal smoking during pregnancy, maternal educational status, and birth weight are interacted with year of the PRAMS sample. As in Table 1, the left column shows results for prone infant sleeping and the right column shows results for supine sleeping.
The most striking feature of Table 2 is the changed point estimates for Hispanic/Latino infants. Controlling for overall trends and other risk factors during the study period, the AOR corresponding to Hispanic/Latino status declined from 1.02 to 0.66 in our prone sleep specification. The difference between these point estimates was significant (P = .021). Within the analysis of supine sleep, the AOR for Hispanic/Latino status increased from 0.58 to 0.89. This difference was again highly significant (P = .007). At the beginning of the survey period, Hispanic/Latino infants were more likely than comparable non-Hispanic whites to sleep prone, and were significantly less likely to sleep supine. By 1998, Hispanic/Latinos were still significantly less likely to sleep prone, while differences in supine sleeping were no longer statistically significant.
African American infants displayed quite different results. In 1996, African Americans were significantly more likely than non-Hispanic whites to sleep prone (AOR = 1.48). The AOR was virtually identical within 1998 data (AOR = 1.45; P = .83). Relative disparities between African Americans and non-Hispanic whites in supine sleep slightly widened over the study period. (AOR = 0.56 in 1996; AOR = 0.51 in 1998).
Aside from race/ethnicity, several other variables yielded large or statistically significant interactions. In our analysis of prone sleep, infants born to women college graduates showed rapid reduction in prone sleep. The change in AOR was significant (P < .036). Both high school and college graduates showed rapid increases in the prevalence of supine sleep (P < .007 for both variables).
The prevalence of supine sleep among infants born weighing between 1501 and 2500 g also increased (P < .007) over the sample period. However, no similar change in AOR was observed in our prone sleep specification. Controlling for overall trends, we observed no statistically or substantively significant change in sleep placement for either VLBW or ELBW infants.
Mothers who used tobacco during pregnancy were initially less likely than other mothers to place their infants prone (AOR = 0.799; P < .05). By 1998, however, these mothers were no less likely than other mothers to place their infants prone (AOR = 1.03). The difference between these coefficients only approached statistical significance (P < .07). However, the point estimates suggest that smokers were less responsive than others to the Back to Sleep message.
Our analysis yields a robust set of findings to evaluate recent SIDS prevention efforts. Within the 1996–1998 birth cohorts, we find high and increasing prevalence of supine infant sleep, in accordance with AAP recommendations. The prevalence of prone sleeping also declined notably in the overall population and within most demographic groups. Both of these changes are consistent with the large observed decline in SIDS incidence following the Back to Sleep campaign. The success of this clear, simple, and consistent message provides a useful model for other public health efforts.
Our results also indicate significant shortcomings of existing efforts. Although the prevalence of supine sleeping increased (and the prevalence of prone sleeping decreased) among African American infants, racial disparities in sleep position remained pronounced. The rate of change in sleep position among African Americans was slower than that observed in other racial/ethnic groups and was not fast enough to meet public health goals or to narrow disparities with other groups. These patterns match results from the National Infant Sleep Position Study that documents large racial disparities in infant sleep patterns.5 In both univariate and multivariate models, large black-white differences persisted, with little evidence of proportional convergence over the 3 PRAMS survey years.
In contrast to the small change seen among African Americans, changes in sleep patterns were quite pronounced among Hispanic/Latinos. Reported SIDS incidence was quite low within this group, even within the first 1996 birth cohort. The small baseline differences in sleep position between Hispanic/Latinos and non-Hispanic whites are puzzling, given the low reported 1996 SIDS incidence among Hispanic infants. However, rapid declines in prone sleeping, and rapid increases in supine sleeping among Hispanic/Latinos are consistent with the 23% decline in observed SIDS incidence within this group over the study period.
Perhaps surprisingly, maternal risk behaviors during pregnancy had little predictive value in predicting either prone or supine sleeping. Controlling for other factors, women who smoked during pregnancy were no more likely than others to place their infants prone, and were no less likely to place their infants in the supine position. The initial timing of prenatal care was also unrelated to reported positions of infant sleep. These noncorrelations are noteworthy because they suggest that patterns of infant sleep position are unlikely to confound known studies of SIDS incidence. Many studies of SIDS incidence establish prenatal smoking and delayed prenatal care as risk factors.11,15,20,23 Because these risk factors appear uncorrelated or weakly correlated with prone infant sleep, existing studies of standard risk factors are unlikely to be confounded by correlations with infant sleep patterns.
The trend in infant sleep placement was less favorable for infants born to women who smoked during pregnancy. Because prenatal smoking itself is so strongly discouraged, it is possible that patients who persist in smoking may be less receptive than other pregnant patients to advice regarding other avoidable health risks.
Parents of VLBW and ELBW infants are less likely than others to place their infants in the supine position. Infants weighing <1501 g at birth appear no more likely than normal birth weight infants to sleep prone, but are more likely to sleep on their sides rather than in the recommended supine position. Reasons for this pattern are unknown.
The low prevalence of supine sleep among infants with older siblings is also striking. The correlation between sleep position and family size partly accounts for the higher prevalence of prone sleep among African Americans. Reasons for this correlation are not known. Habits of sleep placement formed with older children may play an important role.
This study has important limitations that must be considered in evaluating the results. As in most studies of infant sleep position, our data are based on maternal report rather than on direct observation. If mothers view supine sleep as the most socially accepted survey response, they may underreport other placement choices.
Our data are also based on 15 states, and thus may not be generalizable to other populations. In part, because of the specific states included in our PRAMS data, we did not specifically examine infant sleep placement within important ethnic minorities such as Native Americans or Asian Americans. In part, because our results are consistent with previous studies,4,5,16 we do not believe that the PRAMS sampling strategy has biased our results.
Our study also lacks information regarding exposure to specific SIDS prevention efforts. Thus, we cannot relate changes in infant sleep placement to specific interventions such as public service announcements or office-based interventions. Data of the sort provided by Moon and colleagues24 may allow more specific program evaluation in this area.
Our results provide additional evidence for the success of SIDS prevention efforts. However, they also identify important risk groups that require focused interventions. The persistent black-white disparity requires development of additional culturally competent public health programs to further advance public health messages among different racial/ethnic groups. New initiatives, such as those launched by the partnership between the National Black Child Development Institute, the National Institute of Child Health and Human Development, and the Health Resources and Services Administration to reduce SIDS among African Americans, are especially important.
Dr Pollack’s research was funded, in part, by the Centers for Disease Control and Prevention through the Prevention Research Center of Michigan, School of Public Health, University of Michigan (grant U48/CCU51577).
Sarah Mayberry, MPH, and Myron Wegman, MD, offered useful comments. Data and comments were provided by Mary Lyn Gaffield, Christopher Johnson, and the PRAMS Working Group, including Melissa Baker, Debora Barnes-Josiah, Lois Bloebaum, Jo Bouchard, Paul Buescher, Margaret Brozicevic, Diana Cheng, Darci Cherry, Loretta Fuddy, Kristen Helms, Martha Henson, Richard Hopkins, Tonya Johnson, Suzanne Kim, Fabienne Laraque, Linda Lohdefinck, Richard Lorenz, Michael Medvesky, Susan Nalder, Kathy Perham-Hester, Gina Redford, Bruce Steiner, and Rhonda Stephens.
- ↵American Academy of Pediatrics, Task Force on Infant Positioning and SIDS. Positioning and SIDS. Pediatrics.1992;89 :1120– 1126
- ↵Willinger M, Hoffman HJ, Harford RB. Infant sleep position and risk for sudden infant death syndrome: report of meeting held January 13 and 14, 1994, National Institutes of Health, Bethesda, MD. Pediatrics.1994;93 :814– 819
- ↵Martin RJ, Herrell N, Rubin D, Fanaroff A. Effect of supine and prone positions on arterial oxygen tension in the preterm infant. Pediatrics.1979;63 :528– 531
- ↵Goto K, Mirmiran M, Adams MM, et al. More awakenings and heart rate variability during supine sleep in preterm infants. Pediatrics.1999;103 :603– 609
- ↵American Academy of Pediatrics, Task Force on Infant Positioning and SIDS. Positioning and sudden infant death syndrome (SIDS): update. Pediatrics.1996;98 :1216– 1218
- ↵Hunt CE. Expanded “Back-to-Sleep” recommendations: hospital-based safe sleeping practices. J SIDS Infant Mortal.1998;2 :223– 224
- ↵Oyen N, Markestad T, Skaerven R, et al. Combined effects of sleeping position and prenatal risk factors in sudden infant death syndrome: The Nordic Epidemiological SIDS Study. Pediatrics.1997;100 :613– 621
- ↵Bigger HR, Silvestri JM, Weese-Mayer DE. Sudden infant death syndrome (SIDS) death rates in normal birth weight ([NBW] >2500 g), low birth weight ([LBW] 1499–2500 g), and very-low birth weight ([VLBW] 500–1499 g) infants: comparison of rates before and after implementation of the Back to Sleep initiative. Pediatr Res.2000;47 :143A . Abstract 839
- ↵Bigger HR, Silvestri JM, Shott SB, Weese-Mayer DE. Changes in sudden infant death syndrome (SIDS) rates: gender comparisons by birth weight before and after implementation of the Back to Sleep initiative. Pediatr Res.2000;47 :143A . Abstract 840
- ↵Bigger HR, Silvestri JM, Shott S, Weese-Mayer DE. Influence of increased survival in very low birth weight, low birth weight, and normal birth weight infants on the incidence of sudden infant death syndrome in the United States: 1985–1991. J Pediatr.1998;133 :73– 78
- ↵Mathews TJ, Curtin SC, MacDorman MF. Infant mortality statistics from the 1998 period linked birth/infant death data set. National Vital Statistics Report.2000;48 :1– 25
- ↵STATA. Stata Reference Manual Release 6. College Station, TX: Stata Press;1999
- ↵American Academy of Pediatrics, Task Force on Infant Sleep Position and Sudden Infant Death Syndrome. Changing concepts of sudden infant death syndrome: implications for infant sleeping environment and sleep position. Pediatrics.2000;105 :650– 656
- ↵Moon RY, Biliter WM, Croskell SE. Examination of state regulations regarding infants and sleep in licensed child care centers and family child care settings. Pediatrics.2001;107 :1029– 1036
- Copyright © 2002 by the American Academy of Pediatrics