Trends in Postneonatal Aspiration Deaths and Reclassification of Sudden Infant Death Syndrome: Impact of the “Back to Sleep” Program

METHODS

Data were obtained from linked birth and infant death statistic tapes for the United States for the years 1991, 1995, and 1996. In addition, US vital statistic natality and mortality tapes were used for the years 1992, 1993, and 1994 because of the absence of linked files for those years. For the years in which linked files were not available, denominator data (number of births) were obtained from the natality tapes and numerator data (underlying cause of death) were obtained from mortality tapes. The International Classification of Disease, Ninth Revision (ICD-9) codes for underlying causes of death that were investigated included the following: SIDS (7980), aspiration (E911), asphyxia (7990), respiratory failure (7991), and accidental suffocation in bed/cradle (E913.0). Additional broad categories of underlying causes of death were formed to determine the major causes of death replacing the drop in the SIDS proportionate mortality ratio. These broad categories included the following: infection (1011–1399), neoplasms (1400–2399), endocrine disorders (400–2799), hematologic disorders (2800–2899), mental disorders (2900–3199), neurologic disorders (3200–3899), circulation disorders (3900–4599), respiratory disorders (4600–5199), gastrointestinal disorders (5200–5799), genitourinary disorders (5800–6299), birth complications (6300–6769), dermatologic disorders (6800–7099), musculoskeletal disorders (7100–7399), congenital anomalies (7400–7599), perinatal disorders (7600–7799), ill-defined conditions (7800–7999), and injury-related disorders (E800–E999).

Demographic and infant birth information were used from the 1995 and 1996 linked file data to examine the differences in characteristics of infants who died of SIDS compared with infants who died of suffocation. Information used from the linked birth certificate file included maternal race, age, education, parity, birth number, smoking status, and infant birth weight.

Only infants who weighed 500 g or more were used in the analyses. Postneonatal mortality was defined as infant deaths that occurred at 28 days or more of life, and postneonatal mortality rates were calculated by dividing postneonatal deaths by the number of live births. Proportionate mortality ratios were calculated by dividing the number of postneonatal deaths for a specific underlying cause of death by the total number of postneonatal deaths and then multiplying by 100 to obtain a percentage of total mortality. Significant trends in postneonatal mortality rates and postneonatal mortality ratios over time were determined using χ² tests for trends. Logistic regression was used to determine significant independent population characteristics of infants who died of SIDS compared with infants who died of suffocation. All analyses were run in SAS (SAS, Inc, Cary, NC). P ≤ .05 was arbitrarily selected as the level indicative of statistical significance.

RESULTS

Postneonatal mortality decreased 21.9% during the 6-year period (1991–1996) from 319 postneonatal deaths per 100 000 livebirths to 249 (P < .001; Table 1). The SIDS postneonatal mortality rate declined 38.9% from a rate of 118 deaths per 100 000 livebirths in 1991 to 72 in 1996 (P < .001). In addition, the neonatal SIDS mortality rate, which composes only 6% of the total SIDS rate, declined approximately 28% from 0.07 deaths per 1000 livebirths in 1991 to 0.05 in 1996. This suggests that there has been no shift to an earlier age of death for SIDS. Postneonatal aspiration-related deaths declined significantly during the 6-year period from 1.39 postneonatal deaths per 100 000 live births in 1991 to 0.93 in 1996 (P = .01). The trend in postneonatal deaths attributed to suffocation in a bed or a cradle increased from 2.82 postneonatal deaths per 100 000 live births in 1991 to 3.26 in 1996 but failed to reach statistical significance (P = .056). There were no significant trends in postneonatal mortality for deaths attributed to asphyxia or respiratory failure.

The proportion of the postneonatal mortality accounted for by SIDS declined from 37.1% in 1991 to 28.8% in 1996 (P < .001; Table 2). This represents an 8.3% decline in the proportion of the postneonatal mortality contributed by SIDS during this time period. Of the causes of death for which SIDS might be reclassified, only suffocation in bed or cradle showed a significant increase in its proportion of postneonatal mortality (P < .001). Altogether, the causes of death to which SIDS might be reclassified accounted for <2% of the postneonatal mortality annually during the 6-year period. Approximately 93% of the 8.3% decline in the SIDS postneonatal proportionate mortality ratio was accounted for by increases in the postneonatal proportionate mortality ratios for congenital anomalies, injuries, ill-defined conditions, disorders of circulation, disorders of the digestive system, and suffocation.

The only alternative underlying cause of death to which SIDS might be reclassified that demonstrated an increase in the postneonatal proportionate mortality ratio was death by suffocation. The differences in population characteristics between infants who died of suffocation and SIDS thus were examined. Logistic regression was used to identify population characteristics independently associated with infants who died of SIDS compared with infants who died of suffocation. Characteristics examined included maternal race, age, education, parity, birth number, smoking status, and infant birth weight. Only maternal race and smoking status were significantly different between the 2 groups. Thirty percent of infants who died of SIDS were black compared with 21% of infants who died of suffocation. The adjusted odds ratio for being black and dying of SIDS was 1.61 (95% confidence interval: 1.17–2.21). Thirty percent of mothers of infants who died of SIDS reported smoking cigarettes during pregnancy compared with 39% of mothers of infants who died of suffocation. The adjusted odds ratio for being an infant who died of suffocation and whose mother smoked compared with being an infant of a smoker and dying of SIDS was 1.36 (95% confidence interval: 1.02–1.81).

DISCUSSION

Fear of vomiting or choking remains a common concern among women who choose the lateral or prone position to place their infants to sleep.⁴ Countries in which a campaign has been mounted to promote supine positioning for infant sleep, however, have not documented any increase in problems associated with supine positioning.^5,6 Ponsonby et al⁵ found no increase in reported episodes of infant cyanosis, pallor, breath-holding, or breathing difficulties in Tasmania. These investigators, in fact, reported a greater incidence of reports of cyanosis among infants who were placed in the prone position. Hunt et al⁶ reported no increases in choking, wheezing, or breathlessness among infants who were followed longitudinally after the initiation of a “Back to Sleep” campaign in Avon, England. In addition, Bayard and Beal⁷ in Australia, reporting on 196 cases of infant and early childhood deaths, found no cases of aspiration death among infants who were placed on their side or back but 3 cases among infants who were placed in the prone position.

The results of this study support the observations of the studies cited above; that is, during a period when the prevalence of supine positioning increased in the United States, the mortality rate for aspiration-related deaths dropped. Willinger et al² reported that the prevalence of prone positioning of infants for sleep in the United States dropped from 70% in 1992 to 24% in 1996, whereas supine positioning increased from 13% to 35%. Postneonatal aspiration-related deaths dropped significantly from 1.39 per 100 000 live births in 1991 to 0.93 in 1996.

The concern about aspiration in the supine position among clinicians seems to stem from literature related to gastroesophageal reflux. Orenstein et al,⁸ in a commentary on the American Academy of Pediatrics Task Force on Infant Positioning recommendations, sited a number of studies that argued for an increased risk of reflux-related problems for infants placed in the supine position. These studies included ones that suggested that gastroesophageal reflux was increased among infants who were placed in the supine position.^9–13 In addition, studies were cited that suggested that supine positioning might cause life-threatening events in infants with no history of unusual regurgitation.^14–18

Other investigators have cited data that suggest the supine position poses no greater risk for aspiration than the prone position and, in fact, may confer even less risk. Kahn et al¹⁹ evaluated acid reflux in 40 infants who were alternately placed in supine and prone positions. Jeffery et al²⁰ recorded swallowing rate and level of arousal among 10 term infants who were given 0.4 mL of water into the oropharynx and alternated into the supine and prone positions. Swallowing rate was significantly reduced in the prone position, and there was no compensatory increase in arousal. The authors suggested that the reduction in airway protective reflexes in the prone position during active sleep might increase the risk of SIDS. Other investigators have further explored arousability and an association with position. Most recently, Horne et al²¹ reported significantly higher arousal thresholds for infants at 2 to 3 weeks and 2 to 3 months of age in the prone position compared with the supine position. Whether the level of arousability would make an infant more or less vulnerable to the possibility of aspiration remains highly speculative.

The reduction of the proportionate postneonatal mortality ratio for SIDS and the replacement with causes of death not likely to be confused with SIDS is reassuring. The majority of the decrease in the SIDS proportionate mortality is accounted for by the increase in the proportionate mortality associated with congenital anomalies and injuries during the study period. This shift to a greater proportion of the total deaths being attributable to congenital anomalies and injuries does not indicate an increase in number of deaths for these causes but simply that they represent a greater proportion of the total deaths. This observation helps to validate the contention that the “Back to Sleep” campaign may be associated with an absolute reduction in SIDS deaths and not simply a reclassification of deaths.

The problem of identifying a SIDS death remains a challenge. The necessity of death-scene investigations in the final diagnosis of unexplained infant deaths remains an ideal that has not been implemented across the United States. In a 1996 report from the Centers for Disease Control and Prevention only 4 states (California, Minnesota, Missouri, and New Mexico) had detailed written protocols for death-scene investigation.²² Bass et al²³ demonstrated the utility of death-scene investigation in a review of 26 deaths ascribed a diagnosis of SIDS. Death-scene investigations were conducted within 1 week of the presumed SIDS death. In 6 cases, the investigators observed strong circumstantial evidence of accidental death, and in 18 other cases they discovered various possible causes other than SIDS that included accidental asphyxiation by an object in the crib or bassinet, smothering by overlying, hyperthermia, and shaken infant syndrome.

In the analysis reported in this article, although a significant increase was observed in the mortality rate associated with suffocation in bed or cradle, the rate remains extremely low compared with the SIDS rate, 0.0093 per 1000 live births versus 0.72 per 1000 live births in 1996. Movement away from classifying deaths as SIDS to classifying unexplained deaths as suffocation does not seem to be happening at a rapid rate and/or death-scene investigations may not be used as extensively as recommended.

The increase in deaths attributed to suffocation provided the stimulus to examine how the characteristics of the mothers of infants who died of suffocation compared with the characteristics of infants who died of SIDS. These 2 causes of death may well be confused with one another, and finding characteristics unique to either cause might help in identifying misclassified deaths. In addition, as the rebreathing hypothesis for SIDS and the threat of suffocation from bedding were more widely publicized, pathologists and other physicians who complete death certificates may have become more aware of these mechanisms as potential causes of death and altered their coding.

A greater proportion of infants who died of SIDS were black than among infants who died of suffocation, and a greater proportion of mothers of infants who died of suffocation smoked than among mothers of infants who died of SIDS. As a matter of speculation, the racial disparity may be a proxy for socioeconomic conditions that predispose for classification of unexplained infant death as SIDS over that of suffocation. That racial or socioeconomic characteristics may bias classification systems is not without precedent.^24–26 The greater prevalence of smoking among mothers of infants who died of suffocation may offer new insight into the relationship between smoking and SIDS; that is, smoking may be a proxy for behavior associated with overlying or situations in which an infant is likely to suffocate. Thus, those SIDS deaths associated with mothers who smoke may be misclassified. Blair et al²⁷ reported that 84% of infants who died of SIDS while sharing a bed had mothers who smoked. Overlying and infant suffocation have been suggested as a potential risk associated with cobedding.²⁸ Maternal smoking may then simply be an additional flag for behavior that puts an infant at additional risk of suffocation.

There are several limitations to the current study. Vital statistic data do not provide information on whether a death-scene investigation was completed as part of the process for classifying the cause of death. This may allow for an overestimation of the number of SIDS deaths that are recorded and an underestimation of other related causes of death, such as asphyxia and accidental suffocation. There is also no way to validate maternal and infant characteristics and no information on infant positioning before or at the time of death. Despite these limitations, the use of vital statistic data provides the best opportunity to obtain a national overview of SIDS and related causes of death.

CONCLUSION

Analysis of vital statistic data before and after the implementation of the “Back to Sleep” campaign shows no evidence to substantiate the fear that aspiration-related deaths might increase as more infants are placed in the supine position to sleep. In addition, although there is some evidence that suffocation in bed or cradle is being identified more frequently as a cause of death, it composes an extremely small proportion of the postneonatal mortality compared with SIDS. Whether a larger proportion of SIDS deaths will be reclassified as deaths related to suffocation or asphyxia may well be dependent on the success of implementing death-scene investigations as a requirement for reporting an unexplained infant death.