Changes in the Classification of Sudden Unexpected Infant Deaths: United States, 1992–2001
http://www.100md.com
《小儿科》
Department of Pediatrics, University of Texas Medical Branch, Galveston, Texas
Division of Vital Statistics, National Center for Health Statistics, Hyattsville, Maryland
ABSTRACT
Background. Sudden infant death syndrome (SIDS) makes up the largest component of sudden unexpected infant death in the United States. Since the first recommendations for supine placement of infants to prevent SIDS in 1992, SIDS postneonatal mortality rates declined 55% between 1992 and 2001.
Objective. The objective of this analysis was to examine changes in postneonatal mortality rates from 1992 to 2001 to determine if the decline in SIDS was due in part to a shift in certification of deaths from SIDS to other causes of sudden unexpected infant death. In addition, the analysis reviews the change in mortality rates attributed to the broad category of sudden unexpected infant death in the United States since 1950.
Methods. US mortality data were used. The International Classification of Diseases (ICD) chapters "Symptoms, Signs, and Ill-Defined Conditions" and "External Causes of Injury" were considered to contain all causes of sudden unexpected infant death. The following specific ICD (ninth and tenth revisions) underlying-cause-of-death categories were examined: "SIDS," "other unknown and unspecified causes," "suffocation in bed," "suffocation-other," "aspiration," "homicide," and "injury by undetermined intent." The average annual percentage change in rates was determined by Poisson regression. An analysis was performed that adjusted mortality rates for changes in classification between ICD revisions.
Results. The all-cause postneonatal mortality rate declined 27% and the postneonatal SIDS rate declined 55% between 1992 and 2001. However, for the period from 1999 to 2001 there was no significant change in the overall postneonatal mortality rate, whereas the postneonatal SIDS rate declined by 17.4%. Concurrent increases in postneonatal mortality rates for unknown and unspecified causes and suffocation account for 90% of the decrease in the SIDS rate between 1999 and 2001.
Conclusions. The failure of the overall postneonatal mortality rate to decline in the face of a declining SIDS rate in 1999–2001 raises the question of whether the falling SIDS rate is a result of changes in certifier practices such that deaths that in previous years might have been certified as SIDS are now certified to other non-SIDS causes. The observation that the increase in the rates of non-SIDS causes of sudden unexpected infant death could account for >90% of the drop in the SIDS rates suggests that a change in classification may be occurring.
Key Words: sudden infant death sudden unexpected infant death SUID sudden infant death syndrome SIDS postneonatal mortality
Abbreviations: ICD, International Classification of Diseases SIDS, sudden infant death syndrome ICD-9, International Classification of Diseases, Ninth Revision ICD-10, International Statistical Classification of Diseases, Tenth Revision CI, confidence interval
Tracking changes in the occurrence of sudden infant death in the United States is accomplished through the use of the national vital statistics system.1 Cause-of-death data are classified according to the International Classification of Diseases (ICD), which is revised periodically to take into account advances in medical knowledge. The way deaths are classified in vital statistics depends on which revision of the ICD is in effect at the time of death.2,3 Cause-specific mortality trends can be affected by a variety of factors including changes between ICD revisions,4 changes in definitions of causes of death (such as sudden infant death syndrome [SIDS]),5,6 and changes in certifier interpretation of and adherence to cause-of-death definitions.7 These factors affect all causes but are particularly important for causes such as SIDS with a nonspecific definition. The tracking and interpretation of trends in causes of sudden infant death in the United States have proven difficult because of these very issues.
Between 1992 and 1999 the total SIDS rate in the United States dropped from 1.2 per 1000 live births to 0.67 (44% decrease).8–10 Much of this decline has been attributed to recommendations issued in 1992 for placing infants supine during sleep.8,9 The Back to Sleep campaign has been a success in terms of changing behavior for the positioning of infants for sleep. The prevalence of prone positioning of infants in the United States dropped from 70% in 1992 to 20% in 1998.11
Increasingly, however, the reliability of SIDS certification is being questioned. Mitchell et al7 question the reliability of SIDS certification because of observations in Australia of increasing numbers of deaths being classified into other causes such as asphyxia while the SIDS rate falls. Others suggest that pathologists may have changed their diagnostic preferences.12 A review of 81 cases of sudden infant death judged to be cases of infanticide reported that more than half of the cases had originally been certified as SIDS.13 Beckwith14 suggests that "SIDS remains a diagnosis of exclusion," subject "to the whims of the diagnosing pathologist." An expert panel of forensic pathologists has most recently proposed a categorization of SIDS deaths in an attempt to remove some of the diagnostic ambiguity.15
A question then exists concerning the "true" magnitude of the decline in SIDS since 1992, the time at which the first supine-sleeping recommendations were issued in the United States.8 Because SIDS and other causes of sudden unexpected infant death occur predominantly in the postneonatal period, we attempted to address this question by examining changes in the overall postneonatal mortality rate, changes in SIDS rates, and changes in the rates of other sudden unexpected causes of infant death occurring in the postneonatal period in the United States for the period of 1992–2001. In addition, we were interested in observing the trend in sudden unexpected deaths reported before the definition of SIDS in 1969.
METHODS
Data were obtained from US mortality data sets from 1950 to 2001, which are available from mainframe data tapes, CD-ROMs, CDC Wonder, and various publications.16–18 Age of death for this analysis was limited to the postneonatal period: 28 to 364 days. In this analysis, the 2 ICD chapters "Symptoms, Signs, and Ill-Defined Conditions" and "External Causes of Injury" are considered to contain all of the sudden unexpected infant deaths. We examined the number of deaths at 5-year intervals from 1950 to 2001 attributed to these 2 ICD chapters, 1 of which eventually contained the SIDS diagnosis ("Symptoms, Signs, and Ill-Defined Conditions"). This was done because before 1973 there was no specific ICD code for SIDS. Comparability of cause-of-death classification over time for these large ICD chapters was evaluated and found to be reasonable (analysis available on request). For the 1992–2001 analysis, cause-of-death data were classified according to ICD-9 (International Classification of Diseases, Ninth Revision) from 1992 to 19982 and ICD-10 (International Statistical Classification of Diseases, Tenth Revision) from 1999 to 2001.3 The codes for underlying causes of death that were evaluated are specified in Table 1.
Beginning in 1999, a variable was added to public-use mortality files to identify deaths that were pending investigation. "Pending investigation" is marked on a death certificate when the physician, medical examiner, or coroner does not have all the information needed to certify the cause of death accurately at the time of death-certificate completion. The "pending" designation is used most commonly when no other cause of death is listed; however, a certifier may list a tentative cause of death and mark the case "pending investigation." In such cases, an amendment to finalize the cause of death should be submitted when additional information becomes available; however, such changes are not always received by the time that the National Center for Health Statistics closes the data file (usually 10–12 months after the end of the data year). Trends since 1999 in death records marked "pending investigation" were examined to evaluate the possible effect of these cases on SIDS trends.
Trends in rates of underlying causes of death were evaluated by Poisson regression. An average annual percentage change in rates was generated by using the point estimate generated by the Poisson regression equation, exponentiating the natural log to that of , subtracting that exponentiated from 1.0, and then multiplying that remainder by 100 to achieve the average annual percentage change in rate. Ninety-five percent confidence intervals (CIs) were generated similarly by using the Wald lower and upper 95% CIs of the value. All analyses were conducted by using Proc Genmod in SAS (SAS Institute, Cary, NC).
In the move from ICD-9 to ICD-10, there is the possibility that cause-specific mortality rates calculated under ICD-9 might not be comparable to rates calculated under ICD-10.4 For example, under ICD-9, when SIDS was listed on the death certificate together with a "well-defined" cause of death, the well-defined cause of death was preferred. However, under ICD-10, this rule was changed to give SIDS and other well-defined causes an equal probability of being selected, with the final determination being made based on the ICD selection and modification rules. To evaluate this possibility, comparability ratios were used to adjust the rates for the period of 1992–1998 to make them comparable to rates that would be observed under ICD-10 rules in effect from 1999 to 2001.4 The adjusted rates for 1992–1998 then were combined with the 1999–2001 rates to determine if the rate of change of the mortality rate for the specified underlying cause fell within the 95% CIs of the rate of change for the unadjusted rates.
RESULTS
Before 1970 the ICD chapter listing symptoms and ill-defined conditions, which eventually was to house the diagnosis of SIDS, represented 5% of all postneonatal mortality, whereas the ICD chapter containing diagnoses associated with external causes of injury contained 10% of all postneonatal mortality (Fig 1). That began to change in 1970 as the ICD chapter on symptoms and ill-defined conditions grew to represent 15% of all postneonatal mortality. From that point on, the symptoms chapter continued to grow in size, peaking in 1990 and representing 41% of all postneonatal mortality, whereas the external-causes chapter dropped to 6% of all postneonatal mortality. When examined as a proportion of total sudden unexpected infant deaths (symptoms plus external-causes chapters), the symptoms chapter accounted for 31% to 36% of all sudden unexpected infant deaths in 1950–1965. This proportion then increased to a high of 87% in 1985 and 1990 and subsequently declined to 71% in 2001.
1950 to 2001, however, the overall postneonatal mortality rate dropped from 852.7 to 231.1 deaths per 100000 live births. It is interesting to note that despite increases from 1955 to 1975 and declines from 1980 to 2001, the overall postneonatal mortality rate for the symptoms and external-causes ICD chapters combined was quite similar in 2001 to that in 1950 and 1955 (Fig 1). In contrast, if the rates in 1950 and 2001 are compared, the "residual" category of causes of death not associated with sudden unexpected infant deaths declined by 83%.
Since the recommendations were made for supine placement of infants for sleep in 1992, the total postneonatal mortality rate dropped from 314.4 to 231.1 per 100000 live births in 2001 (Table 1), an average annual rate of decline of 3.8% (95% CI: –4.1, –3.6). The postneonatal SIDS rate dropped at an average annual rate of 8.6% (95% CI: –8.9, –8.2). Other causes of sudden unexpected infant death have increased, although they represent a much smaller proportion of postneonatal mortality (see Table 1). The postneonatal mortality rate from unknown causes increased 3.3% annually; suffocation in bed, suffocation-other, and injury due to unknown intent in the "External Causes of Injury" chapter have likewise shown increases over the period from 1992 to 2001. Suffocation in bed increased at an average annual rate of 11.2% (more in the latter part of the time period), and suffocation-other increased at an average annual rate of 3.2%. Other causes of sudden unexpected infant death in the "External Causes of Injury" chapter such as homicide have not changed significantly. In contrast, aspiration-related deaths have decreased over this 10-year period.
In the more recent period from 1999 to 2001, there was no significant change in the total postneonatal mortality rate, whereas the postneonatal SIDS rate continued to decline (Table 2). During this period, the number of deaths attributed to the unknown- or unspecified-cause category within the symptoms chapter increased significantly, and a larger and larger percentage of these deaths were designated as "pending investigation." Since 1999 the mortality rate from the pending diagnoses of the unknown and unspecified category has increased significantly from 6.8 to 12.6 per 100000. In 2001 the pending diagnoses in the unknown- and unspecified-cause category represented 91% of all the pending diagnoses for all postneonatal deaths. The percentage of pending diagnoses classified as SIDS decreased between 1999 and 2001.
If the possibility were entertained that the pending unknown or unspecified diagnoses would have been attributed to SIDS had updated information been available, the SIDS rate in 1999–2001 would have declined at an average annual rate of 3.7% rather than the rate of 9.1% (Table 3). If death-certifier preference has shifted such that previously classified SIDS deaths are now classified as "suffocation," the inclusion of these suffocation deaths and pending unknown or unspecified deaths with SIDS deaths then accounts for 90% of the decline in the SIDS rate observed between 1999 and 2001 and results in a nonsignificant decline in SIDS.
Finally, to validate the trends, we examined the possible effect of different ICD revisions on trends in cause-specific mortality rates. We used comparabil-ity ratios to adjust the rates for the major underlying causes of death that were of interest in this analysis for the period of 1992–1998 under ICD-9 to be comparable to the ICD-10 rates4 (Table 4). For the causes of death examined, the estimate of the rate of change using the adjusted rates fell just at the upper or lower 95% CI of the rate of change obtained by using the unadjusted rates.
DISCUSSION
Sudden unexpected infant death has been a hazard of infancy since the dawn of man.19 The ability to track and explain these deaths has been a persistent problem. This difficulty seems to be associated with ICD-coding changes and the practices of medical certifiers in attributing deaths to these causes. Before 1969 there was no definition of SIDS, and before 1973 there was no ICD classification for SIDS.19,20 Before 1970, the majority of sudden unexpected infant deaths fell under the "External Causes of Injury" chapter of the ICD. After the definition of SIDS was put forward and an ICD code was established, the number of deaths attributed to diagnoses in the ICD chapter "Symptoms, Signs, and Ill-Defined Conditions," in which the SIDS diagnosis resided, increased dramatically from 6% of postneonatal mortality in 1965 to 15% in 1970 and 32% in 1975. Because of the changes in ICD revisions and certifier practices, examining the change over time in the large ICD chapters and the combination of those chapters that contain these deaths may provide a better perspective. In general, although increasing from 1955 to 1975 and then declining from 1980 to 2001, the mortality rate for diagnoses falling under the ICD chapters "Symptoms, Signs, and Ill-Defined Conditions" and "External Causes of Injury" combined was similar in 2001 to that in 1950 and 1955. In contrast, the "residual" category of causes of death not associated with sudden unexpected infant deaths declined by 83% between 1950 and 2001. The similarity of the combined rates between 1950 and 2001, however, may in part be coincidental, because some sudden unexpected infant deaths before the 1970s may have been misclassified into the "residual" category of deaths. It is impossible at this point to assess the magnitude of this effect. As awareness of sudden unexpected infant deaths as a public health problem increased, the precision of the diagnoses also improved. Thus, the increase and declines in sudden unexpected infant deaths from 1975 to 2001 were driven primarily by changes in SIDS rates, as discussed in detail below.
Since the establishment of a unique ICD code in 1973 to identify SIDS deaths, SIDS rates rose rapidly until 1980 and then fell dramatically from 1992 to 2001. The surge in the diagnosis of SIDS and its peak incidence in the 1980s may be attributed to the initial definition of the syndrome and the attention given to attempts at developing a unifying hypothesis for its cause.19,21
The 1969 definition of SIDS guided the certification of sudden infant deaths over the next 20 years. In 1989 the National Institute of Child Health and Human Development convened a group of experts to set new goals for SIDS research and refine the definition.6 The 1989 definition resulting from this meeting defines SIDS as "the sudden death of an infant under 1 year of age which remains unexplained after a thorough case investigation, including performance of a complete autopsy, examination of the death scene, and review of the clinical history." This definition restricted the age, and the implicit requirement of a death-scene investigation was made explicit. More recently an expert panel of forensic pathologists proposed a definition of SIDS that includes several categories.15 This approach defines SIDS generally as "the sudden unexpected death of an infant less than 1 year of age, with the onset of the fatal episode apparently occurring during sleep that remains unexplained after a thorough investigation including performance of a complete autopsy and review of the circumstances of death and the clinical history."
Despite the development of more specific definitions of SIDS, a uniform procedure for collecting and evaluating information on sudden unexplained infant deaths in the United States needs additional development. In 1992 the US Congress recommended that the Department of Health and Human Services develop such a procedure, and a workshop was convened by the Center for Disease Control and National Institute of Child Health and Human Development to do so. The result was the development of the Sudden Unexplained Infant Death Investigation Report Form (SUIDIRF).22 The purpose of this form was to provide a generic model protocol for investigating sudden unexplained infant deaths; assist state and local death-investigation jurisdictions in developing a uniform approach; ensure that all information is considered in the investigation; document the extent of investigation of a scene of sudden unexplained infant death; and provide information useful to the pathologist during autopsy.22 The impact of the Sudden Unexplained Infant Death Investigation Report Form guidelines remains uncertain. At the time of the publication of the guidelines in 1996, only 4 states had written protocols for sudden unexplained infant death-scene investigations.22 As of December 31, 2001, the American Academy of Pediatrics, Division of State Government Affairs, recorded 27 states as having passed legislation mandating autopsies and death-scene investigations for child deaths.23 Child-death–review programs are active in 48 states and offer a potential liaison for the review and classification of unexpected infant deaths.24 The majority of review teams, however, perform retrospective reviews, losing the opportunity for input into the classification of an unexpected infant death.24
Death-scene investigations have proven to be important in differentiating SIDS from other sudden infant deaths. Bass et al25 conducted death-scene investigations in 26 consecutive cases brought to the emergency department in which the presumptive diagnosis was SIDS. They found that 6 were accidental and 18 had causes of death other than SIDS. Their primary conclusion was that many sudden infant deaths have a definable cause that can be revealed by careful investigation of the death scene. Meadows came to a similar conclusion based on the review of 81 cases in England.13 The extent of the use of death-scene investigations in the diagnosis and differentiation of SIDS remains uncertain.
Based on the analysis of vital statistics data of the United States from 1992 to 2001, there has been a decrease in the number of deaths attributed to SIDS. What proportion of this decline can be attributed to the Back to Sleep campaign remains uncertain.9,26–28 The 26% decrease in the total postneonatal mortality rate from 1992 to 1999, however, lends support to the notion that the Back to Sleep campaign had a real and substantial impact during this time period. Nevertheless, the stagnation of the postneonatal mortality rate from 1999 to 2001 in the face of a persistent decrease in the SIDS rate suggests that the decrease in the SIDS rate during this 3-year period resulted from changes in the certification of some sudden unexpected infant deaths from SIDS into other categories. The failure to observe an additional substantial decline in the prevalence of prone positioning between 1999 and 2001, however, may have also contributed to the stagnation of the postneonatal mortality rate. The prevalence of prone positioning as determined by an ongoing national telephone survey was 13.5%, 13.2%, and 12.0% for the years 1999, 2000, and 2001, respectively.29
Overpeck et al30 reported on the impact of assigning deaths to the ICD-9 category of "unknown cause" (ICD-9 code 799) on the underascertainment of sudden unexpected infant deaths. Although only 10.9% of the total deaths analyzed during the study period from 1983 to 1991 and 1995 to 1996 were attributed to unknown cause, it was estimated that 50% of those might be attributed to SIDS based on the timing of death.
Postneonatal mortality rates attributable to injury for a period from 1988 to 1998 in the United States have been reported to have decreased slightly.31 Within that group, however, mechanical suffocation rates, although numerically much smaller than SIDS, were reported to have increased 47.9% over that 11-year period. Our analysis demonstrates a 100% increase in rate from 1992 to 1998 and a 95% increase in rate over the 4-year period from 1998 to 2001. Have the risks for mechanical suffocation increased so dramatically during this period, or does this increase reflect better case investigation or certifier preference Reallocation of the increase in these deaths to the category of SIDS along with the pending unknown-cause deaths for the period from 1999 to 2001 would change the rate of decline of SIDS to a nonsignificant rate of 1.3 (95% CI: –3.7, 1.2) and would account for 90% of the decline in the SIDS rate observed during this period.
The limitations of this analysis using US vital statistics data point specifically to the limitations of certifying sudden unexpected infant deaths. National vital statistics data have not contained information on whether an autopsy was done since 1994.32 There has never been information on death certificates as to whether a death-scene investigation was done. Whether a classification of death is pending has only been available on public-use vital statistics files since 1999.33 Sudden unexpected infant deaths fall within a category of death requiring a medicolegal investigation, and the quality of that certification process varies considerably. Hanzlick and Combs34 reported that 48% of the US population is served by medical examiner systems, whereas the remainder of the population is served by coroner systems. Coroners are usually elected laypersons who rely on available medical personnel to assist in investigations and perform autopsies. The development of a standard death-scene investigation protocol for sudden unexpected infant deaths has thus been a critical challenge and goal taken on by the Centers for Disease Control and Prevention.22 A recent survey of coroners and medical examiners conducted by the US Health Resources and Services Administration confirmed the variation in perspective, policies, and procedures regarding unexplained infant death and the use of the SIDS diagnosis.35 Without standardization in the process of certifying sudden unexpected infant deaths, the reporting of trends for these categories of deaths will remain suspect.
CONCLUSIONS
Sudden unexpected infant death as defined by infant deaths falling into the broad ICD chapters "Symptoms, Signs, and Ill-Defined Conditions" and "External Causes of Injury" remains a major cause of infant death in the United States, with 4271 deaths attributed to these 2 ICD chapters in 2001. Since the supine-sleep recommendations were issued in the United States in 1992, there have been substantial reductions in SIDS deaths, most likely related to the marked reduction in prone-sleep-position prevalence. However, the lack of decline in the postneonatal mortality rate from 1999 to 2001 suggests that reclassification of SIDS cases is now occurring. In addition, the failure to further diminish the prevalence of prone positioning in the United States for these years may also be contributing to stagnation of the postneonatal mortality rate. Attempts to further refine the definition of SIDS to identify a group of infants who may suffer a common vulnerability to sudden death will continue to be a challenge. The efficacy of the implementation of guidelines for the certification of deaths under the current definition of SIDS remains elusive. Until agreement on a more standardized approach to certifying sudden infant death can be attained, monitoring progress on reducing sudden unexpected infant death should include observing trends in the broad ICD-10 chapters "Symptoms, Signs and Abnormal Clinical and Laboratory Findings, Not Elsewhere Classified" and "External Causes of Morbidity and Mortality."
FOOTNOTES
Accepted Dec 20, 2004.
No conflict of interest declared.
REFERENCES
Mathews TJ, Fay M, MacDorman MF. Infant Mortality Statistics From the 2001 Period Linked Birth/Infant Death Data Set. National Vital Statistics Reports. Vol 52, No. 2. Hyattsville, MD: National Center for Health Statistics; 2003
World Health Organization. International Classification of Diseases, Ninth Revision. Geneva, Switzerland: World Health Organization; 1979
World Health Organization. International Statistical Classification of Diseases and Related Health Problems, Tenth Revision. Geneva, Switzerland: World Health Organization; 1992
National Center for Health Statistics. US Public Use ICD-9/ICD-10 Comparability File [on CD-ROM]. Hyattsville, MD: National Center for Health Statistics; 2003
Beckwith JB. Discussion of terminology and definition of the sudden infant death syndrome. In: Bergman AB, Beckwith JB, Ray CG, eds. Sudden Infant Death Syndrome: Proceedings of the Second International Conference on the Causes of Sudden Death in Infants. Seattle, WA: University of Washington Press; 1970:14 –22
Willinger M, James LS, Catz C. Defining the sudden infant death syndrome (SIDS): deliberations of an expert panel convened by the National Institute of Child Health and Human Development. Pediatr Pathol. 1991;11 :677 –684
Mitchell E, Krous HF, Donald T, Byard RW. Changing trends in the diagnosis of sudden infant death. Am J Forensic Med Pathol. 2000;21 :311 –314
Kattwinkel J, Brooks J, Myerberg D. American Academy of Pediatrics AAP Task Force on Infant Positioning and SIDS: Positioning and SIDS [published correction appears in Pediatrics. 1992;90:264]. Pediatrics. 1992;89 :1120 –1126
Malloy MH, Freeman DH. Age at death, season, and day of death as indicators of the effect of the back to sleep program on sudden infant death syndrome in the United States, 1992–1999. Arch Pediatr Adolesc Med. 2004;158 :359 –365
Willinger M, Ko CW, Hoffman HJ, Kessler RC, Corwin MJ. Factors associated with caregivers' choice of infant sleep position, 1994–1998: the National Infant Sleep Position Study. JAMA. 2000;283 :2135 –2142
Byard RW, Beal SM. Has changing diagnostic preference been responsible for the recent fall in incidence of sudden infant death syndrome in South Australia J Paediatr Child Health. 1995;31 :197 –199
Meadow R. Unnatural sudden infant death. Arch Dis Child. 1999;80 :7 –14
Beckwith JB. Defining the sudden infant death syndrome. Arch Pediatr Adolesc Med. 2003;157 :286 –290
Krous HF, Beckwith JB, Byard RW, et al. Sudden infant death wyndrome and unclassified sudden infant deaths: a definitional and diagnostic approach. Pediatrics. 2004;114 :234 –238
National Center for Health Statistics. Multiple cause of death public use data file, selected years. Hyattsville, MD: US Public Health Service; issued annually
National Center for Health Statistics. Vital Statistics of the United States. Vol II, Mortality, Part A. Washington, DC: Public Health Service; published annually
Bergman AB. The "Discovery" of Sudden Infant Death Syndrome. New York, NY: Praeger Publishers; 1986
MacDorman MF, Rosenberg HM. Trends in Infant Mortality by Cause of Death and Other Characteristics, 1960–88. Vital Health Statistics. Vol 20. Hyattsville, MD: National Center for Health Statistics; 1993
Byard RW, Krous HF. Sudden infant death syndrome—a change in philosophy. In: Byard RW, Krous HF, eds. Sudden Infant Death Syndrome: Problems, Progress and Possibilities. London, England: Arnold Press; 2001:1 –3
Centers for Disease Control and Prevention. Guidelines for death scene investigation of sudden, unexplained infant deaths: recommendations of the interagency panel of sudden infant death syndrome. MMWR Morb Mortal Wkly Rep. 1996;45 (RR-10):1–22
American Academy of Pediatrics, Division of State Government and Chapter Affairs. Review of Child Death Investigations: State Statutes, 2003. Elk Grove Village, IL: American Academy of Pediatrics; 2003
Webster RA, Schnitzer PG, Jenny C, Ewigman BG, Alario AJ. Child death review. The state of the nation. Am J Prev Med. 2003;25 :58 –64
Bass M, Kravath RE, and Glass L. Death-scene investigation in sudden infant death. N Engl J Med. 1986;315 :100 –105
Willinger M, Hoffman HJ, Hartford 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
Kattwinkel J, Brooks J, Keenan ME, Malloy M. Infant sleep position and sudden infant death syndrome (SIDS) in the United States: joint commentary from the American Academy of Pediatrics and selected agencies of the Federal Government. Pediatrics. 1994;93 :820
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
Overpeck MD, Brenner RA, Cosgrove C, Trumble AC, Kochanek K, MacDorman M. National underascertainment of sudden unexpected infant deaths associated with deaths of unknown cause. Pediatrics. 2002;109 :274 –283
Tomashek KM, Hsia J, Iyasu S. Trends in postneonatal mortality attributable to injury, United States, 1988–1998. Pediatrics. 2003;111 (5 pt 2):1219–1225
Anderson RN, Kochanek KD, Murphy SL. Report of final mortality statistics, 1995. Mon Vital Stat Rep. 1997;45 (suppl 2):11
National Center for Health Statistics. Multiple Cause of Death Public Use Data File for 1999. Hyattsville, MD: US Public Health Service; 2001
Hanzlick R, Combs D. Medical examiner and coroner systems: history and trends. JAMA. 1998;279 :870 –874
Rusinko PS. Variation in SIDS Diagnosis and Procedures: Final Report. Rockville, MD: Health Resources and Services Administration, Maternal and Child Health Bureau; 2003(Michael H. Malloy, MD, MS)
Division of Vital Statistics, National Center for Health Statistics, Hyattsville, Maryland
ABSTRACT
Background. Sudden infant death syndrome (SIDS) makes up the largest component of sudden unexpected infant death in the United States. Since the first recommendations for supine placement of infants to prevent SIDS in 1992, SIDS postneonatal mortality rates declined 55% between 1992 and 2001.
Objective. The objective of this analysis was to examine changes in postneonatal mortality rates from 1992 to 2001 to determine if the decline in SIDS was due in part to a shift in certification of deaths from SIDS to other causes of sudden unexpected infant death. In addition, the analysis reviews the change in mortality rates attributed to the broad category of sudden unexpected infant death in the United States since 1950.
Methods. US mortality data were used. The International Classification of Diseases (ICD) chapters "Symptoms, Signs, and Ill-Defined Conditions" and "External Causes of Injury" were considered to contain all causes of sudden unexpected infant death. The following specific ICD (ninth and tenth revisions) underlying-cause-of-death categories were examined: "SIDS," "other unknown and unspecified causes," "suffocation in bed," "suffocation-other," "aspiration," "homicide," and "injury by undetermined intent." The average annual percentage change in rates was determined by Poisson regression. An analysis was performed that adjusted mortality rates for changes in classification between ICD revisions.
Results. The all-cause postneonatal mortality rate declined 27% and the postneonatal SIDS rate declined 55% between 1992 and 2001. However, for the period from 1999 to 2001 there was no significant change in the overall postneonatal mortality rate, whereas the postneonatal SIDS rate declined by 17.4%. Concurrent increases in postneonatal mortality rates for unknown and unspecified causes and suffocation account for 90% of the decrease in the SIDS rate between 1999 and 2001.
Conclusions. The failure of the overall postneonatal mortality rate to decline in the face of a declining SIDS rate in 1999–2001 raises the question of whether the falling SIDS rate is a result of changes in certifier practices such that deaths that in previous years might have been certified as SIDS are now certified to other non-SIDS causes. The observation that the increase in the rates of non-SIDS causes of sudden unexpected infant death could account for >90% of the drop in the SIDS rates suggests that a change in classification may be occurring.
Key Words: sudden infant death sudden unexpected infant death SUID sudden infant death syndrome SIDS postneonatal mortality
Abbreviations: ICD, International Classification of Diseases SIDS, sudden infant death syndrome ICD-9, International Classification of Diseases, Ninth Revision ICD-10, International Statistical Classification of Diseases, Tenth Revision CI, confidence interval
Tracking changes in the occurrence of sudden infant death in the United States is accomplished through the use of the national vital statistics system.1 Cause-of-death data are classified according to the International Classification of Diseases (ICD), which is revised periodically to take into account advances in medical knowledge. The way deaths are classified in vital statistics depends on which revision of the ICD is in effect at the time of death.2,3 Cause-specific mortality trends can be affected by a variety of factors including changes between ICD revisions,4 changes in definitions of causes of death (such as sudden infant death syndrome [SIDS]),5,6 and changes in certifier interpretation of and adherence to cause-of-death definitions.7 These factors affect all causes but are particularly important for causes such as SIDS with a nonspecific definition. The tracking and interpretation of trends in causes of sudden infant death in the United States have proven difficult because of these very issues.
Between 1992 and 1999 the total SIDS rate in the United States dropped from 1.2 per 1000 live births to 0.67 (44% decrease).8–10 Much of this decline has been attributed to recommendations issued in 1992 for placing infants supine during sleep.8,9 The Back to Sleep campaign has been a success in terms of changing behavior for the positioning of infants for sleep. The prevalence of prone positioning of infants in the United States dropped from 70% in 1992 to 20% in 1998.11
Increasingly, however, the reliability of SIDS certification is being questioned. Mitchell et al7 question the reliability of SIDS certification because of observations in Australia of increasing numbers of deaths being classified into other causes such as asphyxia while the SIDS rate falls. Others suggest that pathologists may have changed their diagnostic preferences.12 A review of 81 cases of sudden infant death judged to be cases of infanticide reported that more than half of the cases had originally been certified as SIDS.13 Beckwith14 suggests that "SIDS remains a diagnosis of exclusion," subject "to the whims of the diagnosing pathologist." An expert panel of forensic pathologists has most recently proposed a categorization of SIDS deaths in an attempt to remove some of the diagnostic ambiguity.15
A question then exists concerning the "true" magnitude of the decline in SIDS since 1992, the time at which the first supine-sleeping recommendations were issued in the United States.8 Because SIDS and other causes of sudden unexpected infant death occur predominantly in the postneonatal period, we attempted to address this question by examining changes in the overall postneonatal mortality rate, changes in SIDS rates, and changes in the rates of other sudden unexpected causes of infant death occurring in the postneonatal period in the United States for the period of 1992–2001. In addition, we were interested in observing the trend in sudden unexpected deaths reported before the definition of SIDS in 1969.
METHODS
Data were obtained from US mortality data sets from 1950 to 2001, which are available from mainframe data tapes, CD-ROMs, CDC Wonder, and various publications.16–18 Age of death for this analysis was limited to the postneonatal period: 28 to 364 days. In this analysis, the 2 ICD chapters "Symptoms, Signs, and Ill-Defined Conditions" and "External Causes of Injury" are considered to contain all of the sudden unexpected infant deaths. We examined the number of deaths at 5-year intervals from 1950 to 2001 attributed to these 2 ICD chapters, 1 of which eventually contained the SIDS diagnosis ("Symptoms, Signs, and Ill-Defined Conditions"). This was done because before 1973 there was no specific ICD code for SIDS. Comparability of cause-of-death classification over time for these large ICD chapters was evaluated and found to be reasonable (analysis available on request). For the 1992–2001 analysis, cause-of-death data were classified according to ICD-9 (International Classification of Diseases, Ninth Revision) from 1992 to 19982 and ICD-10 (International Statistical Classification of Diseases, Tenth Revision) from 1999 to 2001.3 The codes for underlying causes of death that were evaluated are specified in Table 1.
Beginning in 1999, a variable was added to public-use mortality files to identify deaths that were pending investigation. "Pending investigation" is marked on a death certificate when the physician, medical examiner, or coroner does not have all the information needed to certify the cause of death accurately at the time of death-certificate completion. The "pending" designation is used most commonly when no other cause of death is listed; however, a certifier may list a tentative cause of death and mark the case "pending investigation." In such cases, an amendment to finalize the cause of death should be submitted when additional information becomes available; however, such changes are not always received by the time that the National Center for Health Statistics closes the data file (usually 10–12 months after the end of the data year). Trends since 1999 in death records marked "pending investigation" were examined to evaluate the possible effect of these cases on SIDS trends.
Trends in rates of underlying causes of death were evaluated by Poisson regression. An average annual percentage change in rates was generated by using the point estimate generated by the Poisson regression equation, exponentiating the natural log to that of , subtracting that exponentiated from 1.0, and then multiplying that remainder by 100 to achieve the average annual percentage change in rate. Ninety-five percent confidence intervals (CIs) were generated similarly by using the Wald lower and upper 95% CIs of the value. All analyses were conducted by using Proc Genmod in SAS (SAS Institute, Cary, NC).
In the move from ICD-9 to ICD-10, there is the possibility that cause-specific mortality rates calculated under ICD-9 might not be comparable to rates calculated under ICD-10.4 For example, under ICD-9, when SIDS was listed on the death certificate together with a "well-defined" cause of death, the well-defined cause of death was preferred. However, under ICD-10, this rule was changed to give SIDS and other well-defined causes an equal probability of being selected, with the final determination being made based on the ICD selection and modification rules. To evaluate this possibility, comparability ratios were used to adjust the rates for the period of 1992–1998 to make them comparable to rates that would be observed under ICD-10 rules in effect from 1999 to 2001.4 The adjusted rates for 1992–1998 then were combined with the 1999–2001 rates to determine if the rate of change of the mortality rate for the specified underlying cause fell within the 95% CIs of the rate of change for the unadjusted rates.
RESULTS
Before 1970 the ICD chapter listing symptoms and ill-defined conditions, which eventually was to house the diagnosis of SIDS, represented 5% of all postneonatal mortality, whereas the ICD chapter containing diagnoses associated with external causes of injury contained 10% of all postneonatal mortality (Fig 1). That began to change in 1970 as the ICD chapter on symptoms and ill-defined conditions grew to represent 15% of all postneonatal mortality. From that point on, the symptoms chapter continued to grow in size, peaking in 1990 and representing 41% of all postneonatal mortality, whereas the external-causes chapter dropped to 6% of all postneonatal mortality. When examined as a proportion of total sudden unexpected infant deaths (symptoms plus external-causes chapters), the symptoms chapter accounted for 31% to 36% of all sudden unexpected infant deaths in 1950–1965. This proportion then increased to a high of 87% in 1985 and 1990 and subsequently declined to 71% in 2001.
1950 to 2001, however, the overall postneonatal mortality rate dropped from 852.7 to 231.1 deaths per 100000 live births. It is interesting to note that despite increases from 1955 to 1975 and declines from 1980 to 2001, the overall postneonatal mortality rate for the symptoms and external-causes ICD chapters combined was quite similar in 2001 to that in 1950 and 1955 (Fig 1). In contrast, if the rates in 1950 and 2001 are compared, the "residual" category of causes of death not associated with sudden unexpected infant deaths declined by 83%.
Since the recommendations were made for supine placement of infants for sleep in 1992, the total postneonatal mortality rate dropped from 314.4 to 231.1 per 100000 live births in 2001 (Table 1), an average annual rate of decline of 3.8% (95% CI: –4.1, –3.6). The postneonatal SIDS rate dropped at an average annual rate of 8.6% (95% CI: –8.9, –8.2). Other causes of sudden unexpected infant death have increased, although they represent a much smaller proportion of postneonatal mortality (see Table 1). The postneonatal mortality rate from unknown causes increased 3.3% annually; suffocation in bed, suffocation-other, and injury due to unknown intent in the "External Causes of Injury" chapter have likewise shown increases over the period from 1992 to 2001. Suffocation in bed increased at an average annual rate of 11.2% (more in the latter part of the time period), and suffocation-other increased at an average annual rate of 3.2%. Other causes of sudden unexpected infant death in the "External Causes of Injury" chapter such as homicide have not changed significantly. In contrast, aspiration-related deaths have decreased over this 10-year period.
In the more recent period from 1999 to 2001, there was no significant change in the total postneonatal mortality rate, whereas the postneonatal SIDS rate continued to decline (Table 2). During this period, the number of deaths attributed to the unknown- or unspecified-cause category within the symptoms chapter increased significantly, and a larger and larger percentage of these deaths were designated as "pending investigation." Since 1999 the mortality rate from the pending diagnoses of the unknown and unspecified category has increased significantly from 6.8 to 12.6 per 100000. In 2001 the pending diagnoses in the unknown- and unspecified-cause category represented 91% of all the pending diagnoses for all postneonatal deaths. The percentage of pending diagnoses classified as SIDS decreased between 1999 and 2001.
If the possibility were entertained that the pending unknown or unspecified diagnoses would have been attributed to SIDS had updated information been available, the SIDS rate in 1999–2001 would have declined at an average annual rate of 3.7% rather than the rate of 9.1% (Table 3). If death-certifier preference has shifted such that previously classified SIDS deaths are now classified as "suffocation," the inclusion of these suffocation deaths and pending unknown or unspecified deaths with SIDS deaths then accounts for 90% of the decline in the SIDS rate observed between 1999 and 2001 and results in a nonsignificant decline in SIDS.
Finally, to validate the trends, we examined the possible effect of different ICD revisions on trends in cause-specific mortality rates. We used comparabil-ity ratios to adjust the rates for the major underlying causes of death that were of interest in this analysis for the period of 1992–1998 under ICD-9 to be comparable to the ICD-10 rates4 (Table 4). For the causes of death examined, the estimate of the rate of change using the adjusted rates fell just at the upper or lower 95% CI of the rate of change obtained by using the unadjusted rates.
DISCUSSION
Sudden unexpected infant death has been a hazard of infancy since the dawn of man.19 The ability to track and explain these deaths has been a persistent problem. This difficulty seems to be associated with ICD-coding changes and the practices of medical certifiers in attributing deaths to these causes. Before 1969 there was no definition of SIDS, and before 1973 there was no ICD classification for SIDS.19,20 Before 1970, the majority of sudden unexpected infant deaths fell under the "External Causes of Injury" chapter of the ICD. After the definition of SIDS was put forward and an ICD code was established, the number of deaths attributed to diagnoses in the ICD chapter "Symptoms, Signs, and Ill-Defined Conditions," in which the SIDS diagnosis resided, increased dramatically from 6% of postneonatal mortality in 1965 to 15% in 1970 and 32% in 1975. Because of the changes in ICD revisions and certifier practices, examining the change over time in the large ICD chapters and the combination of those chapters that contain these deaths may provide a better perspective. In general, although increasing from 1955 to 1975 and then declining from 1980 to 2001, the mortality rate for diagnoses falling under the ICD chapters "Symptoms, Signs, and Ill-Defined Conditions" and "External Causes of Injury" combined was similar in 2001 to that in 1950 and 1955. In contrast, the "residual" category of causes of death not associated with sudden unexpected infant deaths declined by 83% between 1950 and 2001. The similarity of the combined rates between 1950 and 2001, however, may in part be coincidental, because some sudden unexpected infant deaths before the 1970s may have been misclassified into the "residual" category of deaths. It is impossible at this point to assess the magnitude of this effect. As awareness of sudden unexpected infant deaths as a public health problem increased, the precision of the diagnoses also improved. Thus, the increase and declines in sudden unexpected infant deaths from 1975 to 2001 were driven primarily by changes in SIDS rates, as discussed in detail below.
Since the establishment of a unique ICD code in 1973 to identify SIDS deaths, SIDS rates rose rapidly until 1980 and then fell dramatically from 1992 to 2001. The surge in the diagnosis of SIDS and its peak incidence in the 1980s may be attributed to the initial definition of the syndrome and the attention given to attempts at developing a unifying hypothesis for its cause.19,21
The 1969 definition of SIDS guided the certification of sudden infant deaths over the next 20 years. In 1989 the National Institute of Child Health and Human Development convened a group of experts to set new goals for SIDS research and refine the definition.6 The 1989 definition resulting from this meeting defines SIDS as "the sudden death of an infant under 1 year of age which remains unexplained after a thorough case investigation, including performance of a complete autopsy, examination of the death scene, and review of the clinical history." This definition restricted the age, and the implicit requirement of a death-scene investigation was made explicit. More recently an expert panel of forensic pathologists proposed a definition of SIDS that includes several categories.15 This approach defines SIDS generally as "the sudden unexpected death of an infant less than 1 year of age, with the onset of the fatal episode apparently occurring during sleep that remains unexplained after a thorough investigation including performance of a complete autopsy and review of the circumstances of death and the clinical history."
Despite the development of more specific definitions of SIDS, a uniform procedure for collecting and evaluating information on sudden unexplained infant deaths in the United States needs additional development. In 1992 the US Congress recommended that the Department of Health and Human Services develop such a procedure, and a workshop was convened by the Center for Disease Control and National Institute of Child Health and Human Development to do so. The result was the development of the Sudden Unexplained Infant Death Investigation Report Form (SUIDIRF).22 The purpose of this form was to provide a generic model protocol for investigating sudden unexplained infant deaths; assist state and local death-investigation jurisdictions in developing a uniform approach; ensure that all information is considered in the investigation; document the extent of investigation of a scene of sudden unexplained infant death; and provide information useful to the pathologist during autopsy.22 The impact of the Sudden Unexplained Infant Death Investigation Report Form guidelines remains uncertain. At the time of the publication of the guidelines in 1996, only 4 states had written protocols for sudden unexplained infant death-scene investigations.22 As of December 31, 2001, the American Academy of Pediatrics, Division of State Government Affairs, recorded 27 states as having passed legislation mandating autopsies and death-scene investigations for child deaths.23 Child-death–review programs are active in 48 states and offer a potential liaison for the review and classification of unexpected infant deaths.24 The majority of review teams, however, perform retrospective reviews, losing the opportunity for input into the classification of an unexpected infant death.24
Death-scene investigations have proven to be important in differentiating SIDS from other sudden infant deaths. Bass et al25 conducted death-scene investigations in 26 consecutive cases brought to the emergency department in which the presumptive diagnosis was SIDS. They found that 6 were accidental and 18 had causes of death other than SIDS. Their primary conclusion was that many sudden infant deaths have a definable cause that can be revealed by careful investigation of the death scene. Meadows came to a similar conclusion based on the review of 81 cases in England.13 The extent of the use of death-scene investigations in the diagnosis and differentiation of SIDS remains uncertain.
Based on the analysis of vital statistics data of the United States from 1992 to 2001, there has been a decrease in the number of deaths attributed to SIDS. What proportion of this decline can be attributed to the Back to Sleep campaign remains uncertain.9,26–28 The 26% decrease in the total postneonatal mortality rate from 1992 to 1999, however, lends support to the notion that the Back to Sleep campaign had a real and substantial impact during this time period. Nevertheless, the stagnation of the postneonatal mortality rate from 1999 to 2001 in the face of a persistent decrease in the SIDS rate suggests that the decrease in the SIDS rate during this 3-year period resulted from changes in the certification of some sudden unexpected infant deaths from SIDS into other categories. The failure to observe an additional substantial decline in the prevalence of prone positioning between 1999 and 2001, however, may have also contributed to the stagnation of the postneonatal mortality rate. The prevalence of prone positioning as determined by an ongoing national telephone survey was 13.5%, 13.2%, and 12.0% for the years 1999, 2000, and 2001, respectively.29
Overpeck et al30 reported on the impact of assigning deaths to the ICD-9 category of "unknown cause" (ICD-9 code 799) on the underascertainment of sudden unexpected infant deaths. Although only 10.9% of the total deaths analyzed during the study period from 1983 to 1991 and 1995 to 1996 were attributed to unknown cause, it was estimated that 50% of those might be attributed to SIDS based on the timing of death.
Postneonatal mortality rates attributable to injury for a period from 1988 to 1998 in the United States have been reported to have decreased slightly.31 Within that group, however, mechanical suffocation rates, although numerically much smaller than SIDS, were reported to have increased 47.9% over that 11-year period. Our analysis demonstrates a 100% increase in rate from 1992 to 1998 and a 95% increase in rate over the 4-year period from 1998 to 2001. Have the risks for mechanical suffocation increased so dramatically during this period, or does this increase reflect better case investigation or certifier preference Reallocation of the increase in these deaths to the category of SIDS along with the pending unknown-cause deaths for the period from 1999 to 2001 would change the rate of decline of SIDS to a nonsignificant rate of 1.3 (95% CI: –3.7, 1.2) and would account for 90% of the decline in the SIDS rate observed during this period.
The limitations of this analysis using US vital statistics data point specifically to the limitations of certifying sudden unexpected infant deaths. National vital statistics data have not contained information on whether an autopsy was done since 1994.32 There has never been information on death certificates as to whether a death-scene investigation was done. Whether a classification of death is pending has only been available on public-use vital statistics files since 1999.33 Sudden unexpected infant deaths fall within a category of death requiring a medicolegal investigation, and the quality of that certification process varies considerably. Hanzlick and Combs34 reported that 48% of the US population is served by medical examiner systems, whereas the remainder of the population is served by coroner systems. Coroners are usually elected laypersons who rely on available medical personnel to assist in investigations and perform autopsies. The development of a standard death-scene investigation protocol for sudden unexpected infant deaths has thus been a critical challenge and goal taken on by the Centers for Disease Control and Prevention.22 A recent survey of coroners and medical examiners conducted by the US Health Resources and Services Administration confirmed the variation in perspective, policies, and procedures regarding unexplained infant death and the use of the SIDS diagnosis.35 Without standardization in the process of certifying sudden unexpected infant deaths, the reporting of trends for these categories of deaths will remain suspect.
CONCLUSIONS
Sudden unexpected infant death as defined by infant deaths falling into the broad ICD chapters "Symptoms, Signs, and Ill-Defined Conditions" and "External Causes of Injury" remains a major cause of infant death in the United States, with 4271 deaths attributed to these 2 ICD chapters in 2001. Since the supine-sleep recommendations were issued in the United States in 1992, there have been substantial reductions in SIDS deaths, most likely related to the marked reduction in prone-sleep-position prevalence. However, the lack of decline in the postneonatal mortality rate from 1999 to 2001 suggests that reclassification of SIDS cases is now occurring. In addition, the failure to further diminish the prevalence of prone positioning in the United States for these years may also be contributing to stagnation of the postneonatal mortality rate. Attempts to further refine the definition of SIDS to identify a group of infants who may suffer a common vulnerability to sudden death will continue to be a challenge. The efficacy of the implementation of guidelines for the certification of deaths under the current definition of SIDS remains elusive. Until agreement on a more standardized approach to certifying sudden infant death can be attained, monitoring progress on reducing sudden unexpected infant death should include observing trends in the broad ICD-10 chapters "Symptoms, Signs and Abnormal Clinical and Laboratory Findings, Not Elsewhere Classified" and "External Causes of Morbidity and Mortality."
FOOTNOTES
Accepted Dec 20, 2004.
No conflict of interest declared.
REFERENCES
Mathews TJ, Fay M, MacDorman MF. Infant Mortality Statistics From the 2001 Period Linked Birth/Infant Death Data Set. National Vital Statistics Reports. Vol 52, No. 2. Hyattsville, MD: National Center for Health Statistics; 2003
World Health Organization. International Classification of Diseases, Ninth Revision. Geneva, Switzerland: World Health Organization; 1979
World Health Organization. International Statistical Classification of Diseases and Related Health Problems, Tenth Revision. Geneva, Switzerland: World Health Organization; 1992
National Center for Health Statistics. US Public Use ICD-9/ICD-10 Comparability File [on CD-ROM]. Hyattsville, MD: National Center for Health Statistics; 2003
Beckwith JB. Discussion of terminology and definition of the sudden infant death syndrome. In: Bergman AB, Beckwith JB, Ray CG, eds. Sudden Infant Death Syndrome: Proceedings of the Second International Conference on the Causes of Sudden Death in Infants. Seattle, WA: University of Washington Press; 1970:14 –22
Willinger M, James LS, Catz C. Defining the sudden infant death syndrome (SIDS): deliberations of an expert panel convened by the National Institute of Child Health and Human Development. Pediatr Pathol. 1991;11 :677 –684
Mitchell E, Krous HF, Donald T, Byard RW. Changing trends in the diagnosis of sudden infant death. Am J Forensic Med Pathol. 2000;21 :311 –314
Kattwinkel J, Brooks J, Myerberg D. American Academy of Pediatrics AAP Task Force on Infant Positioning and SIDS: Positioning and SIDS [published correction appears in Pediatrics. 1992;90:264]. Pediatrics. 1992;89 :1120 –1126
Malloy MH, Freeman DH. Age at death, season, and day of death as indicators of the effect of the back to sleep program on sudden infant death syndrome in the United States, 1992–1999. Arch Pediatr Adolesc Med. 2004;158 :359 –365
Willinger M, Ko CW, Hoffman HJ, Kessler RC, Corwin MJ. Factors associated with caregivers' choice of infant sleep position, 1994–1998: the National Infant Sleep Position Study. JAMA. 2000;283 :2135 –2142
Byard RW, Beal SM. Has changing diagnostic preference been responsible for the recent fall in incidence of sudden infant death syndrome in South Australia J Paediatr Child Health. 1995;31 :197 –199
Meadow R. Unnatural sudden infant death. Arch Dis Child. 1999;80 :7 –14
Beckwith JB. Defining the sudden infant death syndrome. Arch Pediatr Adolesc Med. 2003;157 :286 –290
Krous HF, Beckwith JB, Byard RW, et al. Sudden infant death wyndrome and unclassified sudden infant deaths: a definitional and diagnostic approach. Pediatrics. 2004;114 :234 –238
National Center for Health Statistics. Multiple cause of death public use data file, selected years. Hyattsville, MD: US Public Health Service; issued annually
National Center for Health Statistics. Vital Statistics of the United States. Vol II, Mortality, Part A. Washington, DC: Public Health Service; published annually
Bergman AB. The "Discovery" of Sudden Infant Death Syndrome. New York, NY: Praeger Publishers; 1986
MacDorman MF, Rosenberg HM. Trends in Infant Mortality by Cause of Death and Other Characteristics, 1960–88. Vital Health Statistics. Vol 20. Hyattsville, MD: National Center for Health Statistics; 1993
Byard RW, Krous HF. Sudden infant death syndrome—a change in philosophy. In: Byard RW, Krous HF, eds. Sudden Infant Death Syndrome: Problems, Progress and Possibilities. London, England: Arnold Press; 2001:1 –3
Centers for Disease Control and Prevention. Guidelines for death scene investigation of sudden, unexplained infant deaths: recommendations of the interagency panel of sudden infant death syndrome. MMWR Morb Mortal Wkly Rep. 1996;45 (RR-10):1–22
American Academy of Pediatrics, Division of State Government and Chapter Affairs. Review of Child Death Investigations: State Statutes, 2003. Elk Grove Village, IL: American Academy of Pediatrics; 2003
Webster RA, Schnitzer PG, Jenny C, Ewigman BG, Alario AJ. Child death review. The state of the nation. Am J Prev Med. 2003;25 :58 –64
Bass M, Kravath RE, and Glass L. Death-scene investigation in sudden infant death. N Engl J Med. 1986;315 :100 –105
Willinger M, Hoffman HJ, Hartford 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
Kattwinkel J, Brooks J, Keenan ME, Malloy M. Infant sleep position and sudden infant death syndrome (SIDS) in the United States: joint commentary from the American Academy of Pediatrics and selected agencies of the Federal Government. Pediatrics. 1994;93 :820
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
Overpeck MD, Brenner RA, Cosgrove C, Trumble AC, Kochanek K, MacDorman M. National underascertainment of sudden unexpected infant deaths associated with deaths of unknown cause. Pediatrics. 2002;109 :274 –283
Tomashek KM, Hsia J, Iyasu S. Trends in postneonatal mortality attributable to injury, United States, 1988–1998. Pediatrics. 2003;111 (5 pt 2):1219–1225
Anderson RN, Kochanek KD, Murphy SL. Report of final mortality statistics, 1995. Mon Vital Stat Rep. 1997;45 (suppl 2):11
National Center for Health Statistics. Multiple Cause of Death Public Use Data File for 1999. Hyattsville, MD: US Public Health Service; 2001
Hanzlick R, Combs D. Medical examiner and coroner systems: history and trends. JAMA. 1998;279 :870 –874
Rusinko PS. Variation in SIDS Diagnosis and Procedures: Final Report. Rockville, MD: Health Resources and Services Administration, Maternal and Child Health Bureau; 2003(Michael H. Malloy, MD, MS)