Trends in Infant Bed Sharing in the United States, 1993-2000: Title and subTitle BreakThe National Infant Sleep Position Study FREE

BED SHARING between infants and family members, particularly parents, is common in many cultures.^{1 - 2} Given the influence of the sleep environment on the risk for sudden infant death syndrome (SIDS), the benefits and hazards of bed sharing require scrutiny and further investigation. An argument in favor of bed sharing is that the frequency and duration of breastfeeding is increased under bed-sharing conditions.^{3 - 4} In addition, researchers have suggested that increased infant arousals in quiet sleep, and the entrainment of arousals with the mother during bed sharing may protect infants from SIDS.^{5 - 6} However, no case-control studies have demonstrated a protective effect for SIDS.^{7 - 9} In fact, some case-control studies have shown a risk for SIDS associated with bed sharing under the following conditions: maternal cigarette smoking, recent maternal alcohol consumption, infant covered by a duvet, and parental tiredness.^{10 - 13} The Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI) reported that the risk for SIDS when bed sharing on a sofa is very high (multivariate odds ratio [OR] = 48.99).¹²

In addition, several reports from cohorts of sudden infant deaths describe the hazards of bed sharing and/or sleeping on an adult bed, sofa, or other nonjuvenile furniture, that include suffocation from overlaying, entrapment or head covering, and strangulation from entrapment.^{14 - 18} The American Academy of Pediatrics (AAP) Task Force on Infant Positioning and SIDS has recommended that an infant not be placed to sleep on a sofa, and if a mother chooses to have an infant sleep in her bed, care should be taken to avoid entrapment, soft bedding, and loose covers.⁸

While it is clear that these hazards exist, it is difficult to estimate the probability that adverse outcomes will occur, based on the available evidence. The current report attempts to advance the literature in these respects by presenting population-based data on the frequency of infant bed sharing and predictors of infant bed sharing based on data collected in the National Infant Sleep Position (NISP) study.

Datastat Inc (Ann Arbor, Mich) conducted telephone interviews annually each May starting in 1992, by randomly sampling households with infants younger than 8 months from a list sample of households obtained from Metromail (Lincoln, Neb).¹⁹ The list is based on public information from birth records, infant photography companies, and infant formula companies, and is compiled to have appropriate geographic representation for the 48 coterminous states based on the number of births in each state. Metromail guarantees a 50% eligibility rate with the targeted subject population. Interviews were completed in households that responded affirmatively to the following question: "Is there an infant in this house who was born in the last 7 months; that is, on or after [date]?" Interviews were requested and completed with the nighttime caretaker (80%-85% of those interviewed were mothers) of the infant. The target number of households was 1000 per year.

Beginning in 1993, households in which the mother of the infant did not complete high school were oversampled after the national sample was complete, to achieve a total of 100 households in this education category. Oversampling for households in which the mother had less than a high school education added approximately 35 infants each year to those in the national sample.

The number of calls (national sample plus the oversample of those with less than a high school education) completed annually from 1993 to 2000 was: 1048 (1070 infants, because some were twins), 1041 (1053 infants), 1043 (1052 infants), 1038 (1048 infants), 1033 (1061 infants), 1050 (1065 infants), 1134 (1147 infants), and 1066 (1081 infants), respectively. The estimated response rates calculated as previously described¹⁰ were 81.7%, 81.6%, 86.4%, 84.8%, 87.5%, 83.4%, 73.7%, and 62.9%, respectively.

The interview was developed specifically for the NISP study, and averages 10 minutes.¹⁹ Information was obtained in the following order: (1) characteristics of the infant and the sleep environment; (2) infant sleep position; (3) reasons for sleep position choice; (4) sleep position recommendations from specific sources; (5) sociodemographic information about the mother and the household. Once eligibility was confirmed, 7% did not complete the interview.

Regarding where the infant slept, the respondents were instructed in the following manner: "I am going to read a list of places where infants often sleep. After I finish reading the list, please tell me where the infant usually slept at night during the past 2 weeks." The list was as follows: a crib, a bassinet, a cradle, a carry cot or traveling bed, an adult bed or mattress, a sofa, a playpen, a car or infant seat, or someplace else (specify). If the caregiver did not respond "adult bed or mattress," they were then asked, "During the past 2 weeks, how often has [he/she] slept on an adult bed or mattress . . . half the time, less than half the time, or never?"

To obtain information on bed sharing, all the respondents were then asked, "Does the baby usually sleep alone or share [the usual sleep place] with another person or child?" If the caregiver replied that they share, they were then probed to specify with whom (parent or guardian, another adult, another child), and all responses were recorded.

All analyses were conducted on the combination of the national sample and the oversample for mothers with less than a high school education. The baseline year for the analyses is 1993, when the "less than high school" oversample was initiated. χ² Tests were used to test for differences in proportions (Table 1). The simple analysis of time trends in bed sharing and use of a quilt were evaluated with the χ² test for trend. Survey year, maternal and infant characteristics, characteristics of the sleep environment, and position placed were used in univariate logistic regression modeling to determine their individual influence on routine sharing of an adult bed or mattress at night during the preceding 2 weeks. The same factors were used in a multivariate logistic regression model for bed sharing to determine their independent contributions. All analyses were conducted with SAS 6.12 (Statistical Analysis Software 6.12; SAS Institute, Cary, NC).²⁰

This study was reviewed by the institutional review board of Boston University School of Medicine (Boston, Mass) and was considered an exempt study. There were no personal identifiers in the sample. Respondents had the opportunity to decline the survey at any point during the interview, or to refuse to answer specific questions.

The median infant age in the survey sample (1993-2000) was 134 days, the 10th percentile was 65 days, and the 90th percentile was 195 days. These did not vary significantly from year to year.

For the years 1993 through 2000 combined, 69.7% of infants were usually placed to sleep in a crib, 11% in a bassinet, 9.2% in an adult bed, 5.9% in a cradle, 1.0% in a car seat, 0.8% in a playpen, 0.2% on a sofa, and 1.6% in other places.

While the frequency of routine sleep on an adult bed was low (9.2%), NISP respondents reported that 44.7% of all infants spent some time on an adult bed at night during the preceding 2 weeks (Table 1). Furthermore, it was more common for infants of nonwhite mothers to sleep on an adult bed for half or more of the time (41.8%, 25.5%, and 33% of infants of black, Hispanic, or Asian/other mothers, respectively, for the years 1993-2000 combined) than infants of white mothers (12%). Sleeping on an adult bed half or more of the time was substantially higher for infants of mothers younger than 18 years (36.4%) than infants of mothers aged 18 to 29 years (16.5%) or 30 years or older (12.9%). In addition, the proportion of infants routinely sleeping on an adult bed doubled between 1993-1994 and 1999-2000.

All NISP respondents were asked whether the infant usually slept alone or shared the usual sleep place with another person or child. The majority of infants who were usually placed on an adult bed (91.6%) or sofa (76.5%) at night slept with their parent(s). Only a small percentage of infants usually placed on an adult bed at night slept alone (3.7%), with other adults (4.2%), or with other children (0.3%). Conversely, almost all infants who usually shared with a parent (97.7%) were reported to usually sleep at night in the last 2 weeks on an adult bed.

Among the 121 sets of multiple births, there was a high correlation for where the infant slept within multiple birth sets, with about 90% sleeping in a crib. Seven of the multiple-birth infants usually slept on a sofa (41% of all sofa sleepers in the NISP study), 5 shared with a parent, and 2 shared with each other. None of the 15 multiple-birth infants who usually slept on an adult bed shared the bed with parents.

The proportion of infants routinely sharing an adult bed or mattress increased between 1993 and 2000, from 5.5% to 12.8%. Over time, there was a significant increase in bed sharing among those with maternal age of 18 years or older; those whose maternal race or ethnicity was reported as white or Asian/other; those living in the South, Mid-Atlantic, and Midwest; infants who were 8 weeks or older; and infants with normal birthweight or infants carried to term (Table 2).

The doubling in the proportion of infants bed sharing during the interval 1993 to 2000 was independent of all maternal, infant, and sleep environment characteristics examined (Table 3). In general, those characteristics significantly associated with bed sharing remained significant independent determinants on multivariate analysis, with the exception of maternal education and parity. Infants of mothers aged less than 18 were twice as likely as other infants to routinely sleep in a bed-sharing situation, while black infants were 4 times more likely to bed share as white infants and Asian/other almost 3 times more likely than white infants. Low household income also increased the likelihood of bed sharing.

Infant bed sharing was found to decrease significantly with the age of the infant. In addition, infants with low birthweight and infants who were preterm were less likely to bed share.

Infants who bed shared were 2.9 times more likely than those who did not bed share to usually sleep beneath more than 2 bed covers, and they were almost twice as likely to be covered with a quilt, independent of the room temperature or geographic region where the infant resided (Table 3). Analysis of trends over time shows that the use of a quilt or comforter declined between 1993-1994 and 1999-2000, from 39.1% to 20.5% for infants who slept alone, and from 53.4% to 40.5% for infants who usually shared an adult bed. These time trends are similar, and each is statistically significant.

The prevalence of the usual prone sleep position declined between 1993-1994 and 1999-2000, from 33.9% to 8.9% for bed-sharing infants, and from 52.1% to 14.1% for infants who usually slept alone.

In the last decade, it has been recognized that the sleep environment of infants contributes significantly to infant health. Prone sleep position, soft bedding, and high-thermal microenvironments increase the risk of SIDS.⁸ Public health campaigns targeted at reducing the prevalence of these risk factors have been successful. In the United States, SIDS rates have declined by almost half, from 1.2 of 1000 live births in 1992, to 0.67 of 1000 in 1999, as the nonprone sleep positions have become the norm.^{8 ,19 ,21 - 23} However, the role of bed sharing in infant outcome is unclear. There is a perception that bed sharing is common, but definitive information is lacking. To our knowledge to date, this article provides the first national data on the use of an adult bed as a sleep location for infants.

We found that sleeping on an adult bed is, in fact, more common than we expected at the onset of the investigation, with nearly 50% of infants in the NISP study spending at least some time during the preceding 2 weeks sleeping on an adult bed at night, and about 20% doing so half the time or more, while an adult bed is the "usual" nighttime sleep environment for 13% of infants in recent years. The NISP data also document an increase in routine bed sharing during the past decade from 5.5% in 1993-1994 to 12.8% in 1999-2000.

The increased trend in routine bed sharing was widespread, as defined by many maternal and infant characteristics, including among infants placed in all sleep positions. Thus, the increase in use of nonprone positioning during this period^{19 ,21} cannot directly account for the increase in bed sharing. However, it is possible that the greater attention given to infants and their sleep environment by the AAP recommendation, and national and local "Back to Sleep" public education campaigns to promote the back-sleep position may have had an indirect influence on this behavior. In addition, public health initiatives to promote breastfeeding²⁴ may influence trends in bed sharing. The National Center for Health Statistics reported an increase in breastfeeding between 1993 and 1999, from 56% to 67% in the postpartum period, and from 19% to 30% at 6 months of age.²⁵

There is some evidence for increasing trends elsewhere. An increase in bed sharing has been observed in Norway, with 4% of caregivers among the control group reporting routine bed sharing from 1984 to 1989, 7% from 1990 to 1992, and 15% from 1993 to 1998.¹³ During this period, there was also a concomitant increase in infants who died from SIDS who routinely shared a bed, as well as the proportion found dead while bed sharing.

There are limitations in the information provided by the NISP study. The primary objective of this study was to monitor choice of infant sleep position. In the interests of keeping respondent burden to a minimum, we did not obtained information on the common behavior of bed sharing for parts of the night, reasons for bed sharing, or on some potentially hazardous behaviors associated with bed sharing, such as cigarette smoking or alcohol use.

In addition, the survey population was a random sample of nighttime caregivers of infants younger than 8 months from a purchased list, which was based on public information limited to households with telephones. As with other telephone surveys, the economically disadvantaged are underrepresented. Relative to US natality statistics, this sample overrepresents white infants and underrepresents mothers with low education and young age.²¹

Given the inability to contact households without telephones, or homeless families, the prevalence of usual bed sharing is likely to be underestimated. A case-control study of SIDS conducted in California between 1989 and 1992, reported that the proportion of routine bed sharing during the night was 14.7% among white subjects, 55% among black subjects, and 41% among Latinos.² Similarly, in a study conducted in the District of Columbia in 1995 to 1996, 47% of mothers of 3- to 7-month-old infants, who were predominantly black, reported that the infant usually shared a bed with a parent.²⁶ A more recent study of black mothers recruited from a Medicaid managed-care clinic in St Louis found that 61% of their infants spent at least one night of bed sharing during the preceding 2 weeks, and 49% had shared a bed the night before.²⁷ These prevalences are higher than those of black infants in the NISP study for usual bed sharing (28%), but they are similar to those of the 67% who spent some time on an adult bed at night during the preceding 2 weeks, and the 42% who reported bed sharing half the time or more than half the time.

Black infants were 5 times more likely to routinely share a bed as compared with white infants, and after adjustment for sociodemographic and infant characteristics, the probability remained high (adjusted OR = 4.04). Similar results were observed with Asian infants (univariate OR = 3.4; adjusted OR = 2.72). While infants of young mothers and low income households were also more likely to routinely bed share, the magnitude of the independent contribution of race or ethnicity was greater, suggesting a strong cultural influence on this practice. Similar results were obtained in a prospective study of 10 355 mother-infant pairs in Massachusetts and Ohio. At 3 months of age, the strongest predictors for bed sharing were of black or Asian race or ethnicity, followed by being breastfed, having a mother younger than 18 years or unmarried, and low household income.²⁸

The use of quilts to cover the infant, which has been declining since 1994, was more common among bed-sharing infants in all years. Head covering by a quilt has been reported as a risk for SIDS for infants placed in either the prone or supine sleep position.^{29 - 31} Blair et al¹² demonstrated that the SIDS risk associated with bed sharing was in part explained by use of a duvet. They suggested that the risk associated with bed sharing may be due to other known modifiable risk factors that are present in the bed-sharing environment.

In summary, the NISP study shows that the adult bed is a common location for infants to sleep at night, that bed sharing as a routine practice is growing in the United States, and that cultural factors play an important role in bed sharing. Given the desire by some parents to engage in this practice, more research is needed to understand the range of practices and their potential benefits or hazards. Such research could inform us about strategies for bed sharing that would diminish the hazards and preserve the potential benefits and cultural traditions.

Corresponding author: Marian Willinger, PhD, Pregnancy and Perinatology Branch, Center for Research on Mothers and Children, National Institute of Child Health and Human Development, 6100 Executive Blvd, Room 4B03, MSC 7510, Bethesda, MD 20892-7510 (e-mail: mw75q@nih.gov).

Accepted for publication August 30, 2002.

This study was supported in part by grant HD-2-9067 from the United States Public Health Service (Rockville, Md). We thank Francisco Lazo for survey management.

Bed sharing between infants and family members is believed to be common; however there is much controversy regarding the risks and benefits associated with this behavior. While prior studies have investigated associations with SIDS risk, or potential correlations with breastfeeding or accidental suffocation, this is the first article to examine national trends of the use of an adult bed as a place of nighttime sleep for infants, and of bed sharing with parents. The adult bed is a common place for the infant to sleep at night. Although a smaller proportion of respondents routinely slept with their infants, this proportion has doubled throughout the last decade. Bed sharing was common among certain racial or ethnic groups when taking measures of socioeconomic factors into account, suggesting that there are strong cultural influences on this practice.