Objective. To estimate the changes in birth weight- and gestational age-specific sudden infant death syndrome (SIDS) mortality rates since the publication of the sleep-positioning recommendations by the American Academy of Pediatrics Task Force on Infant Positioning and SIDS.
Methods. This is a historical cohort study using US vital statistic linked birth and infant death certificate files for the years 1991 and 1995. SIDS deaths were identified as any death attributed toInternational Classification of Diseases, Ninth Revisioncode 7980, occurring between the 28th and 365th days of life.
Results. There were 4871 deaths attributed to SIDS in 1991 for a postneonatal mortality rate of 1.2/1000 postneonatal survivors compared with 3114 deaths in 1995 for a rate of .8/1000. This represents a 33% drop in the postneonatal SIDS mortality from 1991 to 1995. Between 1991 and 1995, SIDS rates declined 38%, 38%, 35%, and 32% for birth weight groupings of 500 to 999 g, 1000 to 1499 g, 1500 to 2499 g, and ≥2500 g, respectively. There were no SIDS deaths attributed to infants weighing <500 g. The SIDS rates declined 27%, 21%, 40%, and 23% for gestational age groups of <29 weeks, 29 to 32 weeks, 33 to 36 weeks, and ≥37 weeks. The rate of decline did not differ significantly across birth weight- or gestational age-specific categories. There was a significant increase in the black:non-black postneonatal SIDS mortality ratio from 2.00 to 2.28, reflecting a smaller decline in birth weight- and gestational age-specific mortality for blacks than observed for the non-black population.
Conclusion. Postneonatal SIDS mortality decreased significantly across all broad birth weight and gestational age categories. If the decline in the prevalence of prone positioning that has been reported since 1992 has occurred across all birth weight and gestational age, these data support the hypothesis that supine or side sleep positioning is effective in preterm/low birth weight infants as well as term infants.
In 1992, the American Academy of Pediatrics Task Force on Infant Positioning and Sudden Infant Death Syndrome (SIDS) issued its first recommendation that infants be put to sleep in a supine or side position.1 This recommendation was based, for the most part, on case–control studies performed outside of the United States that suggested a significant relationship between prone sleep positioning and the risk for SIDS.2–8 The Back to Sleep campaign was initiated in 1994 to inform the public about the risks associated with prone sleep positioning of infants, with the goal of further reducing the prevalence of prone positioning in the United States.9 A national telephone survey conducted in the United States between 1992 and 1996 documented a decline in the prevalence of prone sleep positioning from 70% in 1992 to 24% in 1996.10 Along with this drop in prone sleep positioning prevalence during that period, there was a 38% drop in the overall SIDS mortality rate for the United States. It is believed that part, if not all, of the decline in SIDS rates in the United States during this period may be related to the dramatic drop in the prevalence of prone sleep positioning. Whether this drop in the SIDS rate has been shared across all gestational age and birth weight categories, however, has not been reported. The objective of this study was to determine whether the decline in the overall SIDS mortality rate occurred across all birth weight and gestational age categories.
Data were obtained from United States linked birth and infant death certificate tapes for the years 1991 and 1995. The 1991 files are birth cohort files composed of infants born in 1991 who died in 1991 or 1992 before their first birthday. The 1995 files are period files composed of all infant deaths occurring in 1995, some of whom were born in 1994 linked with their respective birth certificate files.11 The change in linking procedures adopted by the National Center for Health Statistics in 1995 does not affect the validity of comparisons between years because there is no overlap of these cohorts or periods. SIDS cases were identified byInternational Classification of Diseases, Ninth Revisioncode 7980. Only deaths occurring in the postneonatal period (after 27 days) were analyzed. Maternal demographic variables were obtained from infant birth certificate data. Race was dichotomized into black and non-black categories based on maternal race classification. Birth weight in grams and gestational age in weeks were obtained from birth certificate information and categorized. Although gestational ages reported for both survivors and SIDS cases were as low as 17 weeks, none of these infants weighed <500 g. It must be assumed that there were misclassification or transcription errors on the part of those reporting this information.
All analyses were performed using SAS (SAS Institute, Cary, NC).12 The significance of changes in rates between years were determined using standard χ2 tests of homogeneity. Changes in SIDS rates across birth weight- and gestational age-specific categories were estimated and tested for statistical significance using a Poisson model for vital rates.13Logistic regression analysis was used to evaluate the change in SIDS rates between years, independent of maternal race, age, education, total pregnancies, tobacco use, birth number, infant sex, birth weight, and gestation. Logistic regression was also used to determine the significance of interaction between year of birth and race.
There were 4 088 221 infants who survived the neonatal period, including the 4871 SIDS deaths in 1991 for a postneonatal SIDS rate of 1.19 SIDS/1000 postneonatal survivors. In 1995, there were 3 885 119 postneonatal survivors and 3114 postneonatal SIDS deaths for a postneonatal SIDS rate of .80. This represents a 33% decline in the SIDS rate over this time span that was statistically significant (P < .001). Between 1991 and 1995, there were small shifts in the distribution of race, maternal age, maternal education, total pregnancies, tobacco use, birth number, and infant sex that attained statistical significance (P values for differences were all <.001). The associations between these characteristics and SIDS, however, were significant in both years (Table 1).
Gestational age-specific SIDS rates showed significant reductions in SIDS rates between the 2 years (Table 2). Although the percent decline did not vary significantly between gestational age-specific categories, the percent decline seemed to be greatest among infants in the 33- to 36-week gestational age category. Birth weight-specific SIDS rates also showed significant declines in SIDS rate between the 2 years (Table 2). There were no significant differences in the percent decline across the birth weight categories and no point estimate for a category stood out particularly.
Postneonatal SIDS rates for blacks were at least 2 times higher than for non-blacks, and the black:non-black SIDS postneonatal mortality ratio increased significantly between 1991 and 1995 from 2.00 to 2.28 (P = .004; Table 3). The higher black:non-black mortality ratio is reflected in the lower percent decline in SIDS rates for blacks compared with non-blacks across all gestational age categories (Table 3).
There were small, but statistically significant, shifts in population distributions of some risk factors for SIDS between 1991 and 1995. We attempted to adjust for the possible confounding effect that these population distribution shifts might make on the relationship between year of birth and postneonatal SIDS mortality through the use of logistic regression analysis (Table 4). The adjusted odds ratio for year of birth (.92) suggests an adjusted average annual decline in postneonatal SIDS rates of 8% per year from 1991 to 1995. Adjustments for the selected population characteristics that were available, all of which are significantly associated with SIDS, did not negate the significant decline in postneonatal SIDS rates between 1991 and 1995.
This analysis documents a significant decline in postneonatal SIDS mortality between 1991 and 1995. This decline seems to have been shared across all gestational age and birth weight categories and is occurring independent of any shifts in population characteristics known to be risk factors for SIDS. Although we cannot guarantee that there has not been a shift in what is classified as SIDS, the continued decline in the overall postneonatal mortality rate between 1991 and 1995 from 3.4 per 1000 live births to 2.7 argues against reclassification as an explanation for the decline in SIDS rates.14 That is, because the overall number of postneonatal deaths has dropped, it is unlikely that SIDS deaths have been reclassified. If the decline in SIDS rates were attributable to reclassification from the SIDS category to another category, the overall postneonatal mortality rate would not be expected to drop. What factors then might be responsible for this decline?
In 1992, the American Academy of Pediatrics Task Force on Infant Positioning and SIDS issued a recommendation to place infants to sleep in a supine or side position.1 This recommendation was based on studies observing a relationship between prone sleep positioning and SIDS in other countries.2–8 Further information on the positive results of national campaigns in other countries to modify infant sleep position and reduce the occurrence of SIDS was reported in 1994 at a conference convened by the National Institute of Child Health and Human Development.15–18This resulted in the production of the Back to Sleep campaign by the US Public Health Service in conjunction with the American Academy of Pediatrics, the SIDS Alliance, and the Association of SIDS and Infant Mortality Programs. The goal of the campaign was to substantially reduce the prevalence of prone sleep positioning among infants and reduce the incidence of SIDS.
A national telephone survey conducted between 1992 and 1996 documented a substantial decline in the prevalence of prone positioning from 70% in 1992 to 24% in 1996.10 This was accompanied by a 28.5% decline in the overall SIDS rate from 1992 to 1995.10 Whether this overall decline in the SIDS rate in the United States was experienced across all gestational age and birth weight categories has not been previously reported.
We report a 33% decline in postneonatal SIDS rates between 1991 and 1995 that was apparent across all gestational age and birth weight categories. We have no direct evidence that the decline was associated with a change in the prevalence of prone positioning among the lower gestational age and birth weight categories. However, based on the national telephone interview, there is no evidence that infants weighing <2500 g are at greater risk of being placed prone when taken home.10 Changes between 1991 and 1995 in demographic and environmental risk factors known to be associated with SIDS, such as race, maternal age and education, parity, and tobacco use,19 do not account for the drop in SIDS rate. Thus, we are left to speculate that the change may be related to the decline in the prevalence of prone positioning across all gestational age and birth weight categories.
The less impressive decline in postneonatal SIDS rates across gestational age- and birth weight-specific categories for blacks indirectly supports the argument for infant sleep positioning as the causal factor behind the drop in SIDS rates. The prevalence of prone positioning among blacks in 1996 was reported to be 43%, compared with 22% among whites.10 It follows that the less substantial decline in the prevalence of prone positioning among blacks may be related to the slower decline in the black postneonatal SIDS rate and the increasing black:non-black disparity.
That preterm and low birth weight infants would benefit from positioning for sleep in a nonprone position has been somewhat uncertain. The initial American Academy of Pediatrics Task Force recommendations hedged on the issue, stating, “For premature infants with respiratory distress… . prone may well be the position of choice.”1 The uncertainty of positioning preterm infants supine may stem in part from reports of improved ventilation among preterms in the supine position.20,,21 These observations were made among infants that had postconceptional ages <36 weeks and may have been experiencing some degree of respiratory distress. Most preterm infants, however, are discharged at postconceptional ages of 36 or more weeks and are no longer experiencing any respiratory distress. Studies in preterm infants of >36 weeks of postconceptional age who are no longer experiencing respiratory distress, as well as in term infants, have not demonstrated improved oxygenation in the prone position.22–24 Thus, there should be no particular advantage to prone sleeping for preterm infants who have resolved their respiratory distress at the time of discharge. In addition, the observations that prone sleeping is associated with deeper levels of sleep may put the somewhat weaker and less mobile preterm at greater risk of being trapped in a position from which it cannot extract itself.21,,25,26 Although reported that it is less likely that the preterm infant will move to a direct face down position, the converse may be that if accidentally obtaining a face down position, the preterm may be less capable of moving from such a position.27
There are several weaknesses of this analysis that we must acknowledge. First, as we stated earlier, we have no direct evidence that the decline in SIDS rates that we report is associated with a change in infant sleep position. This is because infant sleep position is not reported on infant birth or death certificates. We have, however, attempted to control for demographic changes that might have contributed to the change in rates during the 2 periods. Second, we cannot say that a decline in the occurrence of risk factors that might make a preterm infant at greater risk for SIDS has not accounted for the decline in SIDS rates among preterm. However, there is little evidence to support that there are specific medical risk factors associated with SIDS among preterms for which we could control and no vital statistic registry available that records these potential medical risk factors. Finally, we must acknowledge the fallibility of vital statistic data. Errors in gestational age classification, birth weight recording, and cause of death classification make the use of such data somewhat tenuous. Nevertheless, the large sample sizes available and the national representation they afford provide the best estimates of SIDS rates in the United States.
In summary, this analysis has demonstrated that the decline in SIDS rates since the publication of nonprone sleeping recommendations has been shared across all gestational age- and birth weight-specific categories. Although we cannot prove that this decline is a direct result of placing preterm infants in supine or side positions to sleep, the indirect evidence is suggestive and should reassure us that the Back to Sleep recommendations can be applied to preterm infants as well as healthy term infants.
- Received June 11, 1999.
- Accepted September 13, 1999.
Reprint requests to (M.H.M.) Department of Pediatrics, University of Texas Medical Branch, Galveston, TX 77555-0526. E-mail:
This analysis was presented in part at the annual meeting of the Society for Pediatric Research; May 3, 1999; San Francisco, CA.
- SIDS =
- sudden infant death syndrome
- American Academy of Pediatrics, Task Force on Infant Positioning and SIDS
- ↵Forggatt P. Epidemiologic aspects of the Northern Ireland study. In: Bergman AS, Beckwith JB, Ray CG, eds. Sudden Infant Death Syndrome. Seattle, WA: University of Washington Press; 1970:32–46
- ↵Nicholl JP, O'Cathain A. The epidemiology of different groups of babies dying from the sudden infant death syndrome: report to the Foundation for the Study of Sudden Infant Deaths; April 15, 1988; United Kingdom
- Lee NN,
- Chan YF,
- Davies DP,
- Lau E,
- Yip DC
- American Academy of Pediatrics, Task Force on Infant Positioning and SIDS
- ↵National Center for Health Statistics. Linked Birth/Infant Death Data Set: 1995 Period Data. Hyattsville, MD: Public Health Service; 1997
- ↵SAS Institute Inc. SAS User's Guide: Statistics, Version 5. Cary, NC: SAS Institute Inc; 1985
- ↵Freeman D. Applied Categorical Data Analysis. New York, NY: Marcel Dekker, Pub; 1987;42–47
- ↵National Center for Health Statistics. Health United States, 1996–97 and Injury Chartbook. Hyattsville, MD: National Center for Health Statistics; 1997:101
- Willinger M,
- Hoffman HJ,
- Hartford RB
- Mitchell EA,
- Grunt JM,
- Evard C
- Levene S,
- McKenzie SA
- Kahn A,
- Groswasser J,
- Sottiaux M,
- Rebuffat E,
- Franco P,
- Dramaix M
- Goto K,
- Mirmiran M,
- Adams M,
- et al.
- Copyright © 2000 American Academy of Pediatrics