Unexplained Infant Deaths Without Unsafe Sleep Factors: 2011 to 2020

We selected cases from the SUID and Sudden Death in the Young (SDY) Case Registry (hereafter Registry), a multi-jurisdictional, population-based Centers for Disease Control and Prevention surveillance system built on the National Center for Fatality Review and Prevention’s child death review (CDR) program. The Registry is not national, but states and jurisdictions participating in the Registry use the National Fatality Review Case Reporting System (NFR-CRS), a database also used by all participants of CDR.¹³ We examined SUIDs (<1 year of age) that occurred from 2011 to 2020 among residents of states and jurisdictions participating in the Registry (Alaska, Arizona, Colorado, Cook County, Illinois, Delaware, Georgia, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Nevada, New Hampshire, New Jersey, New Mexico, Pennsylvania, San Francisco, California, Tennessee, the Tidewater Region, Virginia, Utah, Pierce County, Washington, and Wisconsin).

The Registry SUID definition includes any of the following in the death certificate cause-of-death section: unknown, undetermined, SIDS, SUID, unintentional sleep-related asphyxia, suffocation, or strangulation, unspecified suffocation, cardiac or respiratory arrest without other well-defined causes, or ill-defined causes with potentially contributing unsafe sleep factors. Intentional homicides were excluded. Trained Registry staff classified SUIDs into categories on the basis of detailed, documented case evidence from the death investigation, medical and social history, and autopsy findings per the Registry Classification System.¹⁴ Analyses were limited to SUIDs classified as unexplained, no unsafe sleep factors (U-NUSF).

U-NUSF included infants who were (1) awake, under supervision, and witnessed to become unresponsive (hereafter referred to as witnessed) or (2) found unresponsive in a safe sleep environment after an apparent period of sleep (hereafter referred to as unwitnessed). A safe sleep environment includes a supine position on a firm, non-inclined surface, including a crib, bassinet, or portable crib (hereafter referred to as crib), and a sleep surface free of soft or loose bedding, objects, adults, children, or pets.⁴ All U-NUSF remained unexplained after a complete investigation (ie, documentation of the position of the infant and location found and autopsy with toxicology, imaging, and other ancillary tests).¹⁵ U-NUSF may or may not have other potentially fatal findings, concerning conditions, or competing causes of death identified during the investigation; however, how these factors contributed to death was uncertain, and the cause was reported as unknown or SIDS on the death certificate.¹⁴ 

We examined deaths by infant demographic, birth and environmental characteristics, death circumstances, infant medical history, and death investigation findings. We analyzed discrete variables, developed composite variables (combining 2 or more discrete variables), and created new variables based on a qualitative review of all text and narrative fields. Infant demographics and birth characteristics included age in months, sex, gestational age in weeks, Medicaid insured, twin or not, and the composite variable of race and ethnicity (combining a variable for race and a variable for ethnicity).¹⁶ We acknowledge that race and ethnicity are social constructs. For this analysis, per the NFR-CRS data dictionary,¹⁶ CDR teams are instructed to abstract race and ethnicity from the death certificates. Environmental characteristics included exposure to maternal smoking during pregnancy and ever breastfed. Death circumstances included location (eg, home, day care) and place (eg, crib, car seat) in which the death occurred, as well as a new qualitative variable for activity at the time of death. It is possible for an infant to be found unresponsive in one location and pronounced dead in another. For this analysis, for the variables location in which death occurred and place in which death occurred, we referred to the location and place the infant went unresponsive regardless of where death was pronounced. Infant medical history included medications or remedies in the 72 hours before death, symptoms within 72 hours of death, and pre-existing medical conditions, which included the text field “other, specify.” Text responses from the “other, specify” field were grouped, and select responses were reported as examples but did not include an exhaustive list of all text responses. Infant medical history also included medical visits, a qualitative variable defined as having had sick or well visits or hospital birth within 1 week of death. The death investigation findings section included a composite variable, documented signs or history of abuse or neglect. This variable was created by combining the discrete variables of an open child protective services (CPS) case, history of child maltreatment, and documented previous abuse. Text or narrative fields without information were treated as “no response.” Abnormal autopsy findings and brief return of spontaneous rhythm following resuscitative efforts were qualitative variables.

To develop qualitative variables, all text and narrative fields were analyzed by using an iterative approach.¹⁷ First, a 2-person team reviewed all responses to identify common themes not reflected in existing NFR-CRS variables. Emerged themes, such as activity at death and medical information, became labels for new variables. Text responses were then independently grouped by each reviewer using those themes. New themes, such as medical visit within 1 week of death, were added during the grouping process. Finally, all authors discussed themes and collapsed them into final variables. In accordance with NFR-CRS data suppression rules regarding small cell sizes (1–5), some findings were not stratified by witnessed versus unwitnessed.

We calculated the frequencies and percentages of infant demographics, birth and environmental characteristics, death circumstances, infant medical history, and death investigation findings stratified by witnessed or unwitnessed. Because of suppression rules, missing (when a question was skipped during data entry or not discussed or mentioned during review) and unknown (when information was discussed but not available) variables were often combined into 1 variable.¹⁸ Analyses were conducted using the SAS system for Windows version 9.4 (SAS Institute, Cary, NC) and Excel. Research involving deceased individuals is not human subjects research according to 45 CFR 46.102(e1) and does not require institutional review board approval.

Of the 7474 SUIDs reported by jurisdictions participating in the Registry from 2011 to 2020, 117 (1.5%) were classified as U-NUSF. Of these, 65 (56%) were witnessed, and 52 (44%) were unwitnessed. Overall, 74% of U-NUSF occurred before 4 months of age (71% witnessed; 78% unwitnessed), including 22% who were neonates (<28 days; 28% witnessed; 15% unwitnessed). Witnessed deaths most commonly occurred at <1 month of age (28%), whereas unwitnessed deaths most commonly occurred at 2 to 3 months of age (44%; Table 1). Most infants were male (58% overall; 51% witnessed; 67% unwitnessed) and born at term (75% overall; 71% witnessed; 81% unwitnessed). Fifteen percent were Hispanic (25% witnessed; <6% unwitnessed), 33% were non-Hispanic Black (29% witnessed; 38% unwitnessed), and 41% were non-Hispanic white (35% witnessed; 48% unwitnessed). More than half (60%) were insured by Medicaid (62% witnessed; 58% unwitnessed). Of 117 U-NUSF, there were <6 twin births among witnessed and unwitnessed deaths, with no higher-order multiples.

One-quarter of U-NUSF (26%) were exposed to maternal tobacco smoke during pregnancy (12% witnessed; 42% unwitnessed; Table 1). Additionally, 61% were ever breastfed (62% witnessed; 60% unwitnessed).

Among U-NUSF, 69% occurred in the infant’s home (62% witnessed; 77% unwitnessed; Table 2). All unwitnessed deaths occurred in a crib. Among witnessed deaths, 54% occurred in caregiver arms, 18% traveling in a car seat, and 15% on a bed, couch, adult chair, or floor. All infants experiencing unwitnessed deaths were reportedly sleeping at the time of death. Among witnessed deaths, infants were in distress, with labored breathing or vomiting (38%), being soothed (28%), sitting (18%), or being fed or nursed (15%). Fewer than 6 infants experiencing witnessed deaths traveling in a car seat were en route to a medical provider because of signs of distress (data not shown).

Twenty percent of infants were given medication or home remedies within 72 hours of death (20% witnessed; 19% unwitnessed; Table 2). Examples of write-in responses to the open-ended “other” text field on medication or home remedy use included vitamin D, saline, vitamins, probiotics, gripe water, or allergy medication. Forty-four percent of infants exhibited 1 or more of the listed symptoms within 72 hours of death (51% witnessed; 35% unwitnessed), including fussiness (11%), difficulty breathing or apnea (10%), vomiting (9%), and “other” (20%). Examples of “other” symptoms included rhinitis, cough, conjunctivitis, or gagging.

Twenty-seven percent of all infants had 1 or more reported pre-existing medical conditions (26% witnessed; 29% unwitnessed; Table 2). The most common conditions were “other” (17%), followed by cardiac abnormalities for witnessed (9%) and infection or allergies for unwitnessed (percentage suppressed to maintain confidentiality). The most common “other” conditions included seizures, reflux, and jaundice. There were a large number of “not specified” responses for conditions and symptoms (32% overall; 34% witnessed; 31% unwitnessed) and pre-existing medical conditions (39% overall: 46% witnessed; 31% unwitnessed). Finally, 41% of all infants had a medical visit (sick, well, or hospital birth) within 1 week of death (52% witnessed; 27% unwitnessed).

Signs or history of abuse or neglect were documented in 15 of 117 deaths (15% witnessed; 10% unwitnessed; Table 2). Abnormal autopsy findings were reported in 46% of all deaths (49% witnessed; 42% unwitnessed; Table 2). Among infants with abnormal autopsy findings, central nervous system abnormalities (30%), visceral congestion or edema (mostly in the lungs; 30%), and cardiovascular abnormalities (30%) were most frequently reported (Table 2). Among infants with central nervous system abnormalities, 9 had cerebral hypoxia-ischemia, 4 had findings that indicated possible trauma, and 3 infants had hippocampal anomalies (data not shown). The most common cardiovascular findings were congenital anomalies, including patent foramen ovale or patent ductus arteriosus, septal defects, and bridging coronary artery (11 of 16), followed by cardiomegaly and ventricular hypertrophy (5 of 16; data not shown). Twelve of the 13 infants with respiratory abnormalities had inflammatory conditions (eg, tracheitis, pneumonitis, pneumonia; data not shown). Autopsy findings were able to be stratified. In 15% of infants, there was a brief return of spontaneous rhythm after resuscitative efforts (26% overall; witnessed; 0% unwitnessed; Table 2).

Deaths included in the Registry provide a unique opportunity to investigate rare U-NUSF deaths (∼1% of all SUIDs in the Registry). Identifying SUIDs occurring in a safe sleep environment or while the infant is awake may inform new etiologic hypotheses.

The proportion of neonatal deaths (<28 days) among U-NUSF was twice that of all SUIDs in the Registry (22% vs 11%).² Although the clinical importance of these findings is unclear, neonatal deaths in this subpopulation have been previously attributed to undiagnosed congenital disorders, genetic causes (eg, metabolic disease or cardiac channelopathy),^19,20 and sudden unexplained postnatal collapse.²¹ Pathologists who specialize in pediatrics can perform comprehensive anatomic and metabolic clinicopathological studies to potentially improve understanding of early neonatal deaths.^22,23 

We also described the environmental characteristics of prenatal exposure to cigarette smoke and breastfeeding among U-NUSF. Numerous studies link prenatal cigarette exposure to increased risk of SIDS and impaired arousal patterns, autonomic function, and cardiovascular reflexes.^24–27 In our study, 1 in 4 infants had exposure to maternal tobacco smoke during pregnancy, which is higher than the national average of 4.6%.²⁸ For unwitnessed infants, 42% had prenatal tobacco exposure. Breastfeeding is a protective factor for SIDS.²⁹ In our study, more than half of infants were reported to have been ever breastfed; this is lower than the national average of ∼80%.³⁰ Data on breastfeeding duration were unavailable. Improved understanding of the roles of maternal tobacco smoke exposure during pregnancy²⁷ and breastfeeding duration²⁹ could inform the etiology of some unexplained infant deaths, including U-NUSF.

Among witnessed U-NUSF, death circumstances (location, place, and activity at the time of death) offer insight into how these deaths occurred. The location of most witnessed deaths was home. For place, witnessed deaths occurred most often in a caregiver’s arms, followed by in a car seat traveling. For activity at the time of death, the largest proportion were in distress, with labored breathing, or vomiting. Finally, 1 in 4 witnessed deaths had a heart rhythm briefly restored following resuscitative efforts.

Information indicating symptoms within 24 hours of death and pre-existing medical conditions were commonly reported among U-NUSF; however, many were grouped into the “other” category, which included a heterogeneous group of nonspecific illnesses and health conditions. The authors of multiple studies^31–33 have documented that some SIDS infants exhibit signs of illness before death and are more likely to have been recently seen by a medical provider; however, none have determined whether such findings factor into death. The witnessed group had more symptoms, and more reported medical visits than the unwitnessed group. It may be that among witnessed deaths, caregivers were more vigilant in supervising their infants’ activities because of concerns about a health issue; this hypothesis warrants further study.

The incidence of infanticide among sudden infant deaths ranges from 1% to 10%; abuse and neglect can lead to sudden deaths in infancy that may be undetected homicides.^34–36 We found that 15 of 117 U-NUSF had documented signs or history of abuse or neglect; all of these deaths included external injuries (mostly scalp or face contusions). Some infant histories included multiple blunt force injuries and signs of neglect. SUIDs are complex to investigate, often with no abnormal autopsy findings; therefore, the cause and manner of death can remain unexplained amid suspicion of abuse or neglect. A recent study revealed a strong association between previous contact with CPS and subsequent death.³⁷ Therefore, to understand the role of abuse and neglect, standardized, thorough death investigations combined with multidisciplinary review of case information, including CPS reports, are important. CDR and the Registry are one way to bring professionals together to thoroughly discuss each infant’s medical and social history. This can lead to informed changes to practices in agencies that support families, like CPS and home visiting, and help inform improvements to death investigation practices.

In our study, nearly half of all infants had abnormal autopsy findings. Abnormal autopsy findings do not always translate into diagnostic findings.³⁸ For example, increased brain weights from edema with no diagnostic relevance have been reported.³⁹ Individually, abnormal autopsy results may be insignificant, but studies like ours that aggregate findings from these rare deaths may lead to new etiologic hypotheses. To determine the cause of death, the National Association of Medical Examiners’ Panel on Sudden Unexpected Death in Pediatrics⁴⁰ recommends that SUID autopsies include genetic testing and testing for metabolic and infectious disease, combined with a review of medical and social history, and understanding of family history of sudden cardiac death.⁴¹ 

Our analysis had some limitations. First, data are retrospective and dependent on the availability and accuracy of information documented during death investigations and autopsies and in medical records, which may not be systematic or consistent over time.⁴⁰ Second, “other, specify” fields frequently resulted in heterogeneous groups, and it was not feasible in this small study to break them down into more informative groupings. Third, the description of circumstances at death relied on caregiver reports and witness interviews, typically documented by law enforcement, coroners, or medicolegal investigators during the investigation.⁴² Scenes are often chaotic, and accounts of circumstances can be affected by recall bias or fear of repercussion from law enforcement or CPS.⁴³ Finally, some data had to be suppressed, limiting the full understanding of some data elements. The extent to which these limitations contribute to inaccurate or incomplete data cannot be known. The strengths of our analysis include an aggregate of >100 cases of a rare type of death derived from 11 years of population-based surveillance. Additionally, the Registry represents one-third of SUID in the United States, with wide geographic diversity. Data allowed for a qualitative review to supplement discrete variables and the creation of new and composite variables. Finally, we examined 117 U-NUSF with descriptions of infant demographics, birth and environmental characteristics, death circumstances, infant medical history, and death investigation findings that are not comprehensively detailed in other available data sources like death certificates.

The primary approach to reducing SUIDs since the mid-1990s has focused on safe sleep promotion.⁴ Infant deaths in our study were not preventable through safe sleep. A better understanding of potential physiologic, genetic, and environmental risk factors and their mechanisms for deaths occurring in a safe sleep environment or while the infant is awake is needed. Causes for these deaths are likely multi-factorial and, in many cases, accompanied by additional factors, such as symptoms of illness and prenatal smoke exposure. Characterizing these deaths is key to advancing knowledge of SUID, developing new etiologic hypotheses, and informing risk reduction strategies.^11,44 Continued research will be most informative when backed by a detailed medical history and thorough autopsy and death investigation data.⁴⁰ 

We acknowledge the SUID and SDY Case Registry awardees in Arizona, Colorado, Georgia, Cook County, Illinois, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Nevada, New Hampshire, New Jersey, New Mexico, Pennsylvania, Tennessee, Utah, and Wisconsin and the Health Resources Services Administration-funded National Center for Fatality Review and Prevention.

CDR

child death review

CPS

child protective services

NFR-CRS

National Fatality Review Case Reporting System

SDY

sudden death in the young

SIDS

sudden infant death syndrome

SUID

sudden unexpected infant death

U-NUSF

unexplained, no unsafe sleep factors