We used data from the Centers for Disease Control and Prevention’s SUID Case Registry (the Registry),¹⁶ a multi-jurisdictional, population-based surveillance system. The Registry builds on existing child death review programs and protocols and has been previously described.^16–18 Briefly, multidisciplinary child death review teams review and compile information on child deaths from multiple sources (eg, death certificates, autopsy reports, medical records) and make prevention recommendations based on their findings. Review information and recommendations are entered into the National Fatality Review-Case Reporting System (NFR-CRS).¹⁷

We studied 8192 SUID that occurred during 2011 to 2020 among residents of Registry jurisdictions including: Alaska; Arizona; San Francisco County, California; Colorado; Delaware; Georgia; Cook County, Illinois; Indiana; Kentucky; Louisiana; Maryland; Michigan; Minnesota; New Hampshire; New Jersey; New Mexico; Nevada; Pennsylvania; Tennessee; Utah; Tidewater Region of Virginia; Pierce County, Washington; and Wisconsin. SUID included deaths with any of the following causes reported on the death certificate: 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. We defined unsafe sleep as prone or side position, shared sleep surface, sleep surface other than a crib or bassinet, any bedding other than a fitted sheet, or soft objects in the sleep area.⁵ We excluded cases that had not undergone Registry data quality control procedures¹⁹ (n = 153) and those with missing or unknown information about surface sharing (n = 444). After exclusions, 7595 SUID remained.

Infant demographic and birth characteristics included age at death, sex, race and ethnicity, gestational age at birth, insurance type, and plurality. Sleep environment included infant’s found position and location, presence of soft bedding (excluding the sleep surface), and number of unsafe sleep factors other than surface sharing (including soft bedding, not in a crib, and non-supine position), and, among surface sharing infants, with whom the infant was surface sharing. Other characteristics included exposure to prenatal maternal cigarette smoking, ever breastfed, primary caregiver a parent, caregiver age, supervisor a parent, supervisor impaired by drugs or alcohol at time of death, open child protective services case for the infant, and having a crib or bassinet in the infant’s home. Variables are described in the NFR-CRS data dictionary.²⁰

Infants were designated as “sharing” if they were sleeping with another person (ie, infant, child, or adult) or animal on any surface (eg, adult bed, crib, couch) at time of death. The NFR-CRS data dictionary guides child death review team members to ascertain infant race and ethnicity from the death certificate.^20,21 We acknowledge that race and ethnicity are social constructs and not genetic or biological categories.^22,23 We choose to report race and ethnicity because race and racism are embedded in our culture, societal structures, and systems supporting and affecting families and their understanding and implementation of safe sleep practices. We refer to cribs, bassinets, and portable cribs as “crib.” We quantified “unsafe sleep factors other than surface sharing” by combining multiple fields (ie, objects in the child’s sleep environment, found location, and found position). Because evidence about the safety of in-bed sleepers is limited, we grouped infants found in portable bassinets placed on an adult bed (n < 6) as “adult bed” for found location. Although “fed human milk” is more inclusive, we used “breastfed” for consistency with the NFR-CRS. “Primary caregiver” is the person who had responsibility for the infant’s care a majority of the time.²⁰ “Supervisor” is the person who had responsibility for the infant’s care at time of death.²⁰ “Parent” includes biological, adoptive, or step-parent.

The Registry Classification System category and suffocation mechanism were assigned by trained Registry staff using the SUID Case Registry Classification System and Algorithm.^12,24 We collapsed the categories unexplained-incomplete case information and unexplained-no autopsy or death scene investigation into unexplained-incomplete information.

Among SUID, we calculated frequencies and percentages by surface sharing status for infant demographic and birth characteristics, sleep environment, other characteristics, Registry Classification System category, and, for SUID categorized as explained or possible suffocation, suffocation mechanism. We calculated χ-square tests of independence to determine if each variable was associated with surface sharing status. To limit the chance of erroneous associations, we excluded missing results from the χ-square analyses. Magnitude or direction of associations were not estimated. Most variables were significantly associated with surface sharing status (P < .05). In large study populations such as ours, statistical significance can emerge with small quantitative differences, complicating interpretation.²⁵ Thus, we highlight clinically meaningful differences of at least 5 percentage points between sharing and nonsharing infants. We conducted analyses using SAS 9.3 (SAS Institute, Cary, NC). Each jurisdiction signed a data-use agreement allowing inclusion of de-identified, aggregated data.

Of the 7595 SUID, 59.5% were sleep surface sharing and 40.5% were not at time of death (Table 1).

Overall, 59.5% of SUID were surface sharing when they died; 40.5% were not. These percentages are similar to other studies of SIDS and SUID (49.6% to 64.1%).^6,13,14 Surface sharing among live infants ranges from 10.1% to 61.4% depending on study population.^2,6,26–30

Compared with nonsharing infants, sharing infants were more often 0 to 3 months old, non-Hispanic Black, publicly insured, found supine, in an adult bed or chair/couch, with a higher number of unsafe sleep factors (in addition to surface sharing) present, were exposed to prenatal maternal cigarette smoking, were supervised by a parent at time of death, or had a supervisor who was impaired by drugs or alcohol at time of death. Compared with sharing, nonsharing infants were more often >3 months old, non-Hispanic white, privately insured, found prone, in a crib, or had soft bedding in the sleep environment.

Many factors associated with surface sharing among living infants are similar to characteristics we described for surface-sharing SUID. Surface sharing among living infants has been shown to vary by measures of poverty and the following: non-Hispanic Black or racial or ethnic minorities, lower parental education, teenage motherhood, lower income, breastfeeding, maternal smoking, and residential mobility (ie, moved at least once since birth).^26–33

The prevalence of SUID exposed to prenatal maternal cigarette smoking (36.5% among SUID in the Registry; 41.4% among sharing and 30.5% among nonsharing infants) was higher than the 2020 US rate of 5.5% among all births.³⁴ Maternal smoking is a known risk factor for SIDS and SUID, and the risk of SIDS associated with surface sharing increases when 1 or both parents smoke or when the infant’s mother smoked during pregnancy.^{4–6,35–39} Surface sharing-related risk for SIDS increases 10-fold when surface sharing occurs with a current smoker or if the pregnant parent smoked during pregnancy.^{1,4,5,40–44}

Breastfeeding is a protective factor against SIDS and mother-infant surface sharing has been encouraged by some to facilitate breastfeeding,^4,45–47 despite American Academy Pediatrics (AAP)’s recommendation of nonshared infant sleep surfaces.^1–5,48 Among SUID in our study, there was <5% difference between sharing and nonsharing in the proportion of infants ever breastfed. Interpretation of this finding is limited because “ever breastfed” is typically abstracted from the birth certificate, which only documents breastfeeding initiation, and not exclusivity and duration.⁴⁹

Multiple births are more likely to be preterm or have low birth weight, increasing the risk of SIDS.⁴ AAP recommends multiples sleep on separate surfaces.⁵⁰ However, we found among SUID, surface sharing was more common among multiples than singletons, most often in an adult bed followed by the same crib. Other studies have similarly found multiples more commonly surface share than singletons.^51,52 Parents with multiples cite space and financial constraints as reasons for placing their infants to sleep on a shared surface.⁵³ This finding has important implications because multiples are over-represented among SUID, both in our study and US death data. During 2011 to 2020, 5.9% of US SUID were multiples,¹⁵ whereas 3.4% of US births were multiples.³⁴

Although some characteristics were more common among surface sharing or nonsharing infants, and future research may be necessary to identify the etiology of those differences, most SUID had at least 1 unsafe factor in their sleep environment regardless of surface sharing status. Surface sharing in the absence of other unsafe sleep factors was rare. Furthermore, nonsharing infants were commonly in both an unsafe sleep position and with soft bedding in their sleep environment. Thus, surface-sharing in and of itself may not be what caregiver education should focus on. These results support efforts to provide comprehensive safe sleep messaging and not focus solely on not surface sharing, for all families at every encounter.

Clinicians can use evidence from this and previous studies to shape conversations on safe sleep guidance, including understanding motivations for surface sharing^2,54,55 and the impact of modeling behavior and giving advice to encourage safe sleep practices.^26,56,57 Previously reported reasons for surface sharing included breastfeeding, facilitating better sleep for the infant or mother, calming a fussy infant, convenience, keeping a close watch over the infant, and protection from environmental dangers.^2,54,55 African American mothers reported privacy, concern about becoming accustomed to always sleeping in the parents’ bed, and fear about suffocation as reasons for not surface sharing.⁵⁵

Most infants in our study were being cared for by a parent when they died. This finding is relevant because parental practices may be influenced by practices observed in the hospital⁵⁶ or advice from healthcare providers (eg, safe sleep recommendations or smoking cessation)^26,57 can impact behavior. Thus, it is critical for healthcare providers to appropriately model and discuss planned and actual infant sleep practices during prenatal visits, birth hospitalization, and postnatal and well-child visits. Engaging parents in discussions about their sleep practices and helping them make decisions to address their concerns and also reduce SUID risk is valuable.

As surface sharing infants more commonly did not have a crib in the home and more often relied on public insurance, when appropriate, pediatricians and other healthcare providers can consider connecting caregivers with free crib distribution programs. These programs can improve safe sleep knowledge and practice.^58–60 Additional research is needed to understand how socioeconomic and other social determinants of health influence infant sleep environments and how best to support families in practicing safe infant sleep.^61–63

Our analysis has several limitations. First, sleep environment data depends on availability and accuracy of information documented during death investigation, which relies on witness reports of an often chaotic scene.⁶⁴ Surface sharing and other unsafe sleep practices may be underreported because of caregiver awareness of safe sleep recommendations and social desirability bias.⁵ Caregiver reasons for surface sharing were not available. Second, varying data collection methods and bias may influence information documented in the Registry.⁶⁵ For example, there was no standard assessment (eg, blood or breathalyzer) or documentation of drug and alcohol impairment of infant supervisors. Therefore, bias is possible in drug screening if, for example, low-income or nonwhite caregivers were differentially screened for substance use.^66,67 Third, our study population was limited to 23 US states and jurisdictions, which may limit generalizability. However, the Registry represents a third of US SUID and has wide geographic diversity. Fourth, cautious interpretation of crib availability is warranted because of a high number of unknown responses. Finally, we were unable to determine risk because the Registry includes only infant deaths and thus, we lacked an appropriate comparison group (eg, living infants).

Characteristics of surface sharing and nonsharing infants among SUID varied by age at death, race and ethnicity, infant insurance type, and presence of unsafe sleep factors. However, most SUID had multiple unsafe sleep factors present regardless of sharing status. The safest place for an infant to sleep is supine, on a nonshared sleep surface, in a crib or bassinet, and without soft bedding.⁵ Supporting families in following the AAP recommendations for reducing sleep-related infant deaths⁵ is complex. Our findings support comprehensive safe sleep counseling for every family at every encounter beyond just asking where an infant is sleeping.