Abstract
Mother–infant bed-sharing has been associated with an increased risk of sleep-related infant deaths, and thus, health messaging has aimed to discourage this behavior. Despite this messaging, bed-sharing remains a common practice in the United States, especially among minority families. Moreover, rates of accidental suffocation and strangulation in bed (often related to bed-sharing) are on the rise, with Black infants at two to three times greater risk than Whites. Multiple studies have identified risk factors for bed-sharing, but a gap remains between findings and translation into interventions. The socio-ecological model (SEM) has been suggested as a way to study and design interventions addressing complex public health issues. This article reconceptualizes the literature on mother–infant bed-sharing using the SEM. PubMed, POPLINE, ERIC, and Psych Info were searched for articles that (a) included bed-sharing as the outcome variable, (b) were published between 2000 and 2013, (c) were conducted in the United States, and (d) included quantitative comparison of more than one factor. The following data were extracted: sample characteristics, bed-sharing definition, methods, factors examined, key findings, and conclusions. Data were summarized into five SEM levels—infant, maternal, family and household, and community and society, nested within the historical context of race. Sixteen studies met inclusion criteria. Significant factors associated with bed-sharing were present within each SEM level of influence. Educational interventions may increase efficacy by attending to multiple levels of the SEM, especially when implementing such interventions within minority subpopulations. Using a harm reduction approach to reducing the risk around bed-sharing may be one way to account for the multiple influences on bed-sharing. The science and practice of minimizing mother–infant bed-sharing may be advanced through use of the SEM.
Mother–infant bed-sharing has been associated with sleep-related infant deaths (Carpenter et al., 2013; Vennemann et al., 2012), and bed-sharing is highly discouraged (American Academy of Pediatrics [AAP], 2005, 2011; National Institute of Child Health and Human Development, 2007). However, a recent study suggests that U.S. bed-sharing rates are increasing (Colson et al., 2013), with Blacks reporting higher bed-sharing rates than Whites (Brenner et al., 2003; Broussard, Sappenfield, & Goodman, 2012; Fu, Colson, Corwin, & Moon, 2008; Lahr, Rosenberg, & Lapidus, 2007; McCoy et al., 2004; Shields, Hunsaker, Muldoon, Corey, & Spivack, 2005; Willinger, Ko, Hoffman, Kessler, & Corwin, 2003). Likewise, national rates of sleep-related infant death are increasing, with Black infants at twice the risk of death than Whites (Mathews & MacDorman, 2013; Shapiro-Mendoza, Kimball, Tomashek, Anderson, & Blanding, 2009).
Research has identified factors associated with bed-sharing; however, translating findings into effective interventions is challenging. It has been suggested that infant sleep interventions address the unique needs and influences of the target population (Baddock, Galland, Bolton, Williams, & Taylor, 2006; Gettler & McKenna, 2011; Horsley et al., 2007; Volpe, Ball, & McKenna, 2013), yet systematic approaches to developing theory-driven interventions remain underutilized. The socio-ecological model (SEM) has the potential to remedy this gap. The purpose of this article is to explore application of the SEM to maternal–infant bed-sharing. Development of a more nuanced understanding of bed-sharing may help researchers and clinicians narrow racial disparities in bed-sharing, which indirectly could reduce disparities in sleep-related infant deaths.
The SEM posits that individuals dynamically interact with their environment across time by actively shaping, and being shaped by, their environments (Bronfenbrenner, 1977, 1986; Glass & McAtee, 2006; Lewin, 1936; Lounsbury & Mitchell, 2009). Thus, behavioral interventions focused only on changing behavior may be less effective than those that also attend to the individual’s social context (Gettler & McKenna, 2010; Glass & McAtee, 2006). The SEM helps address complex public health problems that require comprehensive prevention approaches, especially when addressing health disparities (Department of Health & Human Services, 2013; Smedley & Syme, 2000; Stokols, 1996; Thomas, Quinn, Butler, Fryer, & Garza, 2011). The SEM has been applied to address complex research questions such as father involvement with children (Gavin et al., 2002), the impact of long-term infant hospitalization (Miles, Holditch-Davis, Schwartz, & Scher, 2007), child growth, adolescent maternal–fetal attachment, child wellness (Reifsnider, Gallagher, & Forgione, 2005), physical activity in children with autism spectrum disorders (Obrusnikova & Miccinello, 2012), and tobacco use in adolescent girls (DiNapoli, 2009). Additionally, the SEM can inform design and implementation of health promotion activities (Stokols, 1996). For example, Stokols (1996) suggests the SEM can help (a) identify “high-impact leverage points and intermediaries,” (b) combine health promotion components at the person-centered and environmental level to maximize the potential of success in health promotion activities, and (c) measure the impact of health promotion activities.
Current theories that have been applied to infant health include the life-course perspective, which suggests that birth outcomes are not just a result of the 9 months of pregnancy, but of the mother’s entire life course (including her own experiences in utero; Lu & Halfon, 2003; Lu et al., 2010). The life-course perspective has been especially useful in helping explain the higher incidence of poor birth outcomes among U.S. Blacks (Lu & Halfon, 2003; Lu et al., 2010). The SEM complements the life-course perspective by situating those maternal experiences within multiple levels of influence. The SEM has been applied to help explain racial disparities in perinatal mortality (Alio, Richman, et al., 2010). Alio, Richman, et al. (2010) suggest that birth outcomes are influenced at every level of the SEM and that the historical context of racism helps to explain the racial disparities existent in these outcomes.
While reviews of bed-sharing literature exist (Ball, 2012; Ball & Volpe, 2013; Buswell & Spatz, 2007; Horsley et al., 2007; Thoman, 2006), theoretical models have yet to be applied to bed-sharing. Ball and Volpe (2013) suggest that differentiating between infant-care practices, parental behaviors, and cultural beliefs would strengthen risk-reduction interventions. Current infant sleep literature focuses primarily on infant- and family-level factors (Dahl & El-Sheikh, 2007; Germo, Chang, Keller, & Goldberg, 2007; Goldberg & Keller, 2007; Hiscock, 2010; McKenna et al., 1993; Sadeh, Mindell, Luedtke, & Wiegand, 2009; Sadeh, Titkotzky, & Scher, 2010). The SEM is a meaningful and systematic approach to clarify multiple levels of influence to increase the efficacy of interventions beyond their current levels.
Method
A narrative synthesis method was applied to a systematic review of the literature on bed-sharing (Popay et al., 2006). PubMed, POPLINE, ERIC, and Psych Info were searched for studies using the following search terms individually and in combination: “bed share,” “bed sharing,” “co sleep,” “co sleeping,” and “infant.” Studies were included if they met the following criteria: (a) bed-sharing as the outcome variable, (b) published between 2000 and 2013, (c) conducted in the United States, and (d) including a quantitative comparison of more than one influential factor. Abstracts and methods were reviewed to determine eligibility, and for those that were eligible, sample characteristics, bed-sharing definition, methods, factors, key findings, and conclusions were extracted.
Results
The initial search yielded 413 results (Figure 1); 16 studies met inclusion criteria (see Table 1). Study samples included large data sets such as the Pregnancy Risk Assessment and Monitoring System (Blabey & Gessner, 2009; Broussard et al., 2012; Lahr et al., 2007), or multistate samples (Hauck, Signore, Fein, & Raju, 2008; McCoy et al., 2004; Willinger et al., 2003), while other studies included clinic-based (Flick, White, Vemulpalli, Stulac, & Kemp, 2001; Morgan & Johnson, 2001; Norton & Grellner, 2011), or smaller, locally based samples (Brenner et al., 2003; Fu, Moon, & Hauck, 2010; Glenn & Quillin, 2007; Shields et al., 2005; Weimer et al., 2002). Sample sizes ranged from N = 33 (Glenn & Quillin, 2007) to N = 18,986 (Colson et al., 2013), with the majority of samples including multiple racial and ethnic groups (Blabey & Gessner, 2009; Brenner et al., 2003; Broussard et al., 2012; Colson et al., 2013; Hauck et al., 2008; Lahr et al., 2007; Shields et al., 2005; Weimer et al., 2002; Willinger et al., 2003). Results from the review were synthesized using a SEM consisting of five levels of influence: infant, maternal, family and household, and community and society, with each level nested within the historical context of race (Figure 2). Each level of the SEM is populated with concepts relating to the issue of bed-sharing based on the literature.
Results of the article selection process.
Studies Examining Factors Associated With Mother–Infant Bed-Sharing in the United States, 2000-2013.
Note. PRAMS = Pregnancy Risk Assessment and Monitoring System; PNC = prenatal care; SIDS = sudden infant death syndrome; WIC = Women, Infant, and Child services; SES = socioeconomic status; IPV = intimate partner violence; ER = emergency room; NICU = neonatal intensive care unit; LBW = low birth weight.
Socio-ecological model of mother–infant bed-sharing.
As applied to bed-sharing, the smallest unit within the SEM is the infant. Our model suggests that the infant level includes factors, such as infant characteristics, that influence bed-sharing. Among the 12 studies that included infant-level factors, findings were inconclusive for age, birth weight, preterm birth, and health status. A higher likelihood of bed-sharing was found for infants younger than 1 month in age (Fu et al., 2010) and 4 months (Colson et al., 2013; Morgan & Johnson, 2001; Willinger et al., 2003), while Fu et al. (2008) found higher bed-sharing rates for infants older than 1 month, and Weimer et al. (2002) found no significance. Normal (greater than 2,500 g) birth weight infants (Norton & Grellner, 2011; Willinger et al., 2003) were more likely to bed-share, but three studies found no significance (Brenner et al., 2003; Lahr et al., 2007; McCoy et al., 2004). Willinger et al. (2003) found a higher likelihood of bed-sharing among normal birth weight and full-term infants, while Colson et al. (2013) found a higher likelihood among preterm (less than 37 weeks gestation) infants. Bed-sharing was less likely for infants admitted to the neonatal intensive care unit (Norton & Grellner, 2011), but not significant for overall health (Fu et al., 2008; Norton & Grellner, 2011), an emergency room visit, or a hospital stay (Norton & Grellner, 2011). Gender (Brenner et al., 2003; Lahr et al., 2007; Norton & Grellner, 2011; Shields et al., 2005) and sleep problems (Weimer et al., 2002) were not significant.
The maternal level includes factors such as maternal characteristics, behaviors, and infant-rearing practices, and was noted in 13 studies. A higher likelihood of bed-sharing was found for infant bedding practices, but findings were inconclusive for maternal age, breastfeeding, depression, sleep position, and substance use. Bed-sharing was associated with using a quilt or comforter (Colson et al., 2013; Willinger et al., 2003), soft bedding (such as a pillow or soft mattress; Flick, White, et al., 2001; Fu et al., 2010), multiple layers under the infant (Flick, White, et al., 2001), and more than two covers, regardless of how hot or cold the temperature of the room (Willinger et al., 2003). Mothers younger than 18 (Willinger et al., 2003), 20 (Brenner et al., 2003; Colson et al., 2013), and 25 years (Broussard et al., 2012) were more likely to bed-share. Two studies found mothers older than 20 (Fu et al., 2008; Fu et al., 2010) and 34 years (Broussard et al., 2012) were more likely to bed-share, but age was not significant in four studies (Blabey & Gessner, 2009; Hauck et al., 2008; Lahr et al., 2007; Shields et al., 2005). Breastfeeding was significant in five studies (Broussard et al., 2012; Hauck et al., 2008; Lahr et al., 2007; McCoy et al., 2004; Norton & Grellner, 2011), but not in four others (Brenner et al., 2003; Fu et al., 2008; Fu et al., 2010; Shields et al., 2005). Depression was associated with bed-sharing for younger infants (3 to 7 months), but not for older infants (Brenner et al., 2003), and Broussard et al. (2012) found no significance.
Nonsupine sleep position increased likelihood of bed-sharing in five studies (Blabey & Gessner, 2009; Flick, White, et al., 2001; Fu et al., 2008; Morgan & Johnson, 2001; Shields et al., 2005), decreased the likelihood in one study (Willinger et al., 2003), and was not significant in six studies (Brenner et al., 2003; Broussard et al., 2012; Colson et al., 2013; Fu et al., 2010; Hauck et al., 2008; Lahr et al., 2007). Alcohol and marijuana use were associated with bed-sharing in one study (Blabey & Gessner, 2009), but not in two others (Brenner et al., 2003; Broussard et al., 2012). While maternal tobacco use was significant in three studies (Blabey & Gessner, 2009; Fu et al., 2010; Hauck et al., 2008), it was not in five others (Brenner et al., 2003; Broussard et al., 2012; Fu et al., 2008; Lahr et al., 2007; Shields et al., 2005). Stress (Brenner et al., 2003; Broussard et al., 2012), pregnancy intention (Broussard et al., 2012), pacifier use (Fu et al., 2010), infant clothing layers (Willinger et al., 2003), and survey mode (Lahr et al., 2007) were not significant.
The family and household level includes paternal, familial, and household factors that affect the infant, such as household structure and family characteristics, and was included in 13 studies. Moving more than once since the infant’s birth was associated with a higher likelihood of bed-sharing (Brenner et al., 2003). Single motherhood was significant in five studies (Blabey & Gessner, 2009; Brenner et al., 2003; Broussard et al., 2012; Hauck et al., 2008; Lahr et al., 2007; Mollborg, Wennergren, Norvenius, & Alm, 2011), but not in one (Fu et al., 2010). Having less than three rooms for sleeping was associated with bed-sharing (Weimer et al., 2002), but household crowding was not significant in two other studies (Brenner et al., 2003; McCoy et al., 2004), nor was the infant having his or her own room (Weimer et al., 2002). Broussard et al. (2012) found that experiencing partner-related stress increased the likelihood of bed-sharing. A “poor” social environment, defined by Norton and Grellner (2011) as including reports of substance use, intimate partner violence, or family services involvement, decreased the likelihood of bed-sharing. Parity and/or birth order (Brenner et al., 2003; Broussard et al., 2012; Colson et al., 2013; Hauck et al., 2008; Lahr et al., 2007; McCoy et al., 2004; Shields et al., 2005; Willinger et al., 2003), household smoking (Brenner et al., 2003; Lahr et al., 2007; McCoy et al., 2004), abuse (Broussard et al., 2012), no father on birth certificate (Broussard et al., 2012), and substance use in the home (Brenner et al., 2003) were not significant.
The SEM defines the community and society level as factors beyond the infant and family, including neighborhood conditions, and access to opportunities and resources such as education, employment, and health care (Rimer & Glanz, 2005). Maternal education, income level, and prenatal care are included in this level because they serve as a proxy for other variables, such as issues of access. Community and society level was included in 15 studies. Completion of less than the recommended number of well-child visits was associated with bed-sharing (Norton & Grellner, 2011). Higher rates of bed-sharing were associated with living in the Northern or Southwestern regions (Blabey & Gessner, 2009), the Southern region (Colson et al., 2013; Willinger et al., 2003), or the Western region (Colson et al., 2013) of the United States. Bed-sharing was associated with an income less than 185% of the federal poverty level (Hauck et al., 2008), earning less than $20,000 (Willinger et al., 2003) and less than $50,000 annually (Colson et al., 2013; Lahr et al., 2007), but not significant in three studies (Brenner et al., 2003; McCoy et al., 2004; Weimer et al., 2002). Lower socioeconomic status was associated with bed-sharing when defined as enrollment in Medicaid (Morgan & Johnson, 2001; Norton & Grellner, 2011), Women, Infant, and Child (WIC) services or public insurance (Norton & Grellner, 2011), or measured by the Hollingshead Index of Social Position (Glenn & Quillin, 2007); however, two studies found no significance for type of insurance at delivery (Broussard et al., 2012; Lahr et al., 2007) or WIC (Broussard et al., 2012; Lahr et al., 2007). Lower education (12 years or less) was significant in five studies (Blabey & Gessner, 2009; Brenner et al., 2003; Colson et al., 2013; Fu et al., 2008; Weimer et al., 2002), but not in six others (Broussard et al., 2012; Fu et al., 2010; Hauck et al., 2008; Lahr et al., 2007; McCoy et al., 2004; Willinger et al., 2003). Neutral or no physician advice on bed-sharing was significant in one study (Colson et al., 2013), but not in another (Weimer et al., 2002). Prenatal care defined as trimester of care (Brenner et al., 2003; Broussard et al., 2012; Lahr et al., 2007), adequacy of care (Lahr et al., 2007), type of prenatal care (Lahr et al., 2007), and prenatal education (Broussard et al., 2012) were not significant, nor were type of well-child care (Fu et al., 2008; Lahr et al., 2007), maternal employment (Shields et al., 2005), residence type (urban vs. rural; Lahr et al., 2007; Norton & Grellner, 2011), father’s socioeconomic status (Glenn & Quillin, 2007), and birth hospital (Lahr et al., 2007).
The historical context of race incorporates the impact racism has had on U.S. Blacks (Alio, Richman, et al., 2010). In a developmental context, Blacks gained voting and land ownership rights much later in U.S. history than Whites, and as a result, the trajectory of historical accumulation of wealth and privilege has been shorter for Blacks (Dominguez, Dunkel-Schetter, Glynn, Hobel, & Sandman, 2008). The structural and personally mediated aspects of racism continue to impact Blacks today (Alio, Richman, et al., 2010). In particular, experiences of racial discrimination have been linked to poor birth outcomes such as low birth weight, preterm birth, and, consequently, infant mortality (Collins, David, Handler, Wall, & Andes, 2004; Collins et al., 2000; David & Collins, 2007; Dominguez, 2011; Dominguez et al., 2008; Mustillo et al., 2004). The historical context of race takes into account these social processes. Ten studies found that being in a minority racial or ethnic group was significantly associated with bed-sharing (Blabey & Gessner, 2009; Brenner et al., 2003; Broussard et al., 2012; Colson et al., 2013; Fu et al., 2008; Hauck et al., 2008; Lahr et al., 2007; McCoy et al., 2004; Shields et al., 2005; Willinger et al., 2003).
Five studies examined bed-sharing within a racial context either by differentiating results by race/ethnicity (Broussard et al., 2012; McCoy et al., 2004), or by including a sample made up of mostly Blacks (Flick, White, et al., 2001; Fu et al., 2008; Fu et al., 2010). Among infant level factors, only infant age was significant, and findings were inconclusive with one study finding a higher likelihood of bed-sharing for infants younger than 1 month (Fu et al., 2010) and one finding a higher likelihood for infants older than 1 month (Fu et al., 2010).
All four studies included maternal level factors for Blacks. Younger maternal age—less than 25 years (Broussard et al., 2012)—and older maternal age—25 years or older (Fu et al., 2008) or 35 years or older (Broussard et al., 2012)—were associated with bed-sharing. More layers of bedding and use of soft bedding were associated with bed-sharing (Flick, White, et al., 2001; Fu et al., 2010). Findings were mixed regarding the influence of breastfeeding—while Broussard et al. (2012) found breastfeeding predicted bed-sharing at a higher level than for Whites, McCoy et al. (2004) found breastfeeding was less highly associated with bed-sharing than for Whites, and two other studies found no significance (Fu et al., 2008; Fu et al., 2010). Among Blacks who reported depression, bed-sharing was more than seven times higher than among those who did not report depression, while no significance was found for Whites (Broussard et al., 2012). Findings were mixed regarding sleep position, with two studies finding a higher likelihood of bed-sharing for those placing infants nonsupine to sleep (Flick, White, et al., 2001; Fu et al., 2008), and two finding no significance (Broussard et al., 2012; Fu et al., 2010). Findings were mixed for maternal tobacco use where two studies found it to be associated with a higher likelihood of bed-sharing (Fu et al., 2010; McCoy et al., 2004) and two found no significance (Broussard et al., 2012; Fu et al., 2008). Pacifier use (Fu et al., 2010), emotional stress, pregnancy intention, substance use, and traumatic stress (Broussard et al., 2012) were not significant.
Three studies examined race-related family and household-level factors. In one study, single Black mothers were almost twice as likely to bed-share than their married counterparts, while White single mothers were only slightly more likely to bed-share than married White mothers (McCoy et al., 2004); in two studies, marital status was not significant for Blacks (Broussard et al., 2012; Fu et al., 2010). Abuse, father’s name on birth certificate, parity, and partner-associated stress (Broussard et al., 2012) were not significant.
Four studies examined the community and society level within the context of race. Findings regarding maternal education were mixed; Fu et al. (2008) found that high school or less education was associated with a higher likelihood of bed-sharing for Blacks, but three studies found education was not significant (Fu et al., 2010; McCoy et al., 2004). Findings regarding prenatal care were also mixed—Broussard et al. (2012) found that Blacks who received late or no prenatal care were almost four times more likely to bed-share than Blacks who received prenatal care in the first trimester, but Fu et al. (2010) found prenatal care use was not significant. Financial stress, insurance at delivery, prenatal care education (Broussard et al., 2012), WIC (Broussard et al., 2012; McCoy et al., 2004), type of well-child care (Fu et al., 2008), and income level (McCoy et al., 2004) were not significant.
Discussion
The SEM explains complex public health problems requiring comprehensive prevention approaches (Department of Health & Human Services, 2013; Smedley & Syme, 2000; Stokols, 1996; Thomas et al., 2011). The SEM provides an insightful reconceptualization into the interaction of these levels in mother–infant bed-sharing. In this review, significant factors were present at each SEM level. Infant age and health status were associated with an increased likelihood of bed-sharing in some studies, with the remaining factors being nonsignificant. Research literature on reasons for bed-sharing (not included in this review) suggest parents bed-share to monitor ill infants, and in response to infant crying (Chianese, Ploof, Trovato, & Chang, 2009; Hackett & Simons, 2013; Hauck et al., 2008; Lee & Gay, 2011; McKenna & Volpe, 2007; Ramos, 2003; Rowe, 2003; Weimer et al., 2002).
Among maternal-level factors, bed-sharing was associated with more and softer bedding use, which is especially concerning due to the increasing rates of accidental suffocation and strangulation in bed (Shapiro-Mendoza et al., 2009). While findings were inconclusive for breastfeeding, it has been identified as a reason for bed-sharing (Ateah & Hamelin, 2008; Baddock, Galland, Taylor, & Bolton, 2007; Chianese et al., 2009; Hackett & Simons, 2013; Hauck et al., 2008; Joyner, Oden, Ajao, & Moon, 2010; Kendall-Tackett, Cong, & Hale, 2010).
Among family/household factors, being unmarried, experiencing partner-related stress, and social environment were associated with likelihood of bed-sharing. These findings are in line with the literature on environmental reasons for bed-sharing, including no space for a crib (Bettegowda, Manzano, & Boyd, 2004; Joyner et al., 2010; Weimer et al., 2002), protection from vermin or violence (Chianese et al., 2009; Culver, 2008; Joyner et al., 2010), or to keep the infant warm (Baddock et al., 2007; Bettegowda et al., 2004). Additionally, family members were identified as sources for advice on infant sleep (Germo et al., 2007; Kendall-Tackett et al., 2010; Lathen, 2009; Ramos, 2003; Tomori, 2011).
Among community/society-level factors, lower income level, lower socioeconomic status, and less than the recommended number of well-child visits was associated with a higher likelihood of bed-sharing, and findings were mixed for maternal education, physician advice, and region of the United States. These findings suggest the presence of community- and society-level factors that may influence bed-sharing. In addition to family members, sources of bed-sharing information include health care providers (Hauck et al., 2008; Kendall-Tackett et al., 2010; Tomori, 2011), friends (Kendall-Tackett et al., 2010; Ramos, 2003), and media (Chung, Oden, Joyner, Sims, & Moon, 2012; Culver, 2008; Ramos & Youngclarke, 2006).
When examining the historical context of race, especially the structural and personally mediated impacts that racism has had on U.S. Blacks, only four studies differentiated findings for Blacks, and these findings were limited. Infant-level factors were not examined in the reviewed literature. Among maternal-level factors, soft and/or more layers of bedding were associated with bed-sharing. Findings regarding age, breastfeeding, sleep position, and tobacco use were inconclusive, with one study finding depression to be associated with a higher likelihood of bed-sharing. The reviewed literature did not specifically attend to the impact that experiences of racism may have on maternal-level factors. Experiences of racism may be important to explore in future research, especially since racism (a) has been identified as a stressor (Clark, Anderson, Clark, & Williams, 1999; Sawyer, Major, Casad, Townsend, & Mendes, 2012), (b) impacts trust of health care providers (Acegbembo, Tomar, & Logan, 2006), and (c) impacts both mother and infant health (Collins et al., 2000; Collins et al., 2004; Geronimus, 1992; Giscombe & Lobel, 2005; Mays, Cochran, & Barnes, 2007; Mustillo et al., 2004; Nuru-Jeter et al., 2009).
Among family- and household-level factors for Blacks, findings were inconclusive, with one study suggesting single motherhood increased the likelihood of bed-sharing, and no significance in two studies. Abuse, father’s name on birth certificate, parity, and partner-associated stress were not significant. Among community- and society-level factors, findings were also inconclusive, with one study finding lower maternal education increased bed-sharing for Blacks, but three others finding no significance. Lower prenatal care usage was linked in one study with bed-sharing, but not significant in another study.
These results suggest that the most conclusive evidence for influence among Blacks may be maternal-level factors such as bedding use when bed-sharing. Soft and more bedding has been associated with a higher risk of sleep-related infant deaths, and thus use of bedding may contribute to the racial disparities in these outcomes (AAP, 2011). Though not included in this current review, Ajao, Oden, Joyner, and Moon (2011) and Chianese et al. (2009) suggest a limited understanding among some Blacks of the concept of “firm” bedding. The limited number of studies (four) examining race-based differences suggests a need for additional research focused on the historical context of race on bed-sharing. Furthermore, the differential findings by race and ethnicity also suggest that different cultural beliefs and practices may influence bed-sharing behaviors (Broussard et al., 2012; Fu et al., 2010; Lahr et al., 2007; McCoy et al., 2004; Shields et al., 2005; Weimer et al., 2002; Willinger et al., 2003). For example, many cultures believe that bed-sharing with the infant produces a more emotionally secure child (McKenna & Volpe, 2007).
In the area of health promotion practice, this review suggests strategies should attend to each level of the SEM. The presence of significant differences between racial groups suggests that health promotion strategies should be specifically tailored to the unique needs, experiences, and beliefs of minority groups, which includes understanding the effects of the historical context of race and racism (Campbell & Quintiliani, 2006; Resnicow, Baranowski, Ahluwahalia, & Braithwite, 1999). Instead of one standard message to “not” bed-share, messaging should be tailored to specific populations to account for the differential factors influencing their decision making around bed-sharing (Ahlers-Schmidt, Kuhlmann, Kuchlmann, Schunn, & Rosell, 2014; Ball & Volpe, 2013; Blabey & Gessner, 2009; Chianese et al., 2009; Gettler & McKenna, 2010; Hackett & Simons, 2013; Hiscock, 2010). The literature has further suggested a “harm reduction” or “harm minimization” approach to safe sleep education, in which the goal is to provide additional information around reducing the risk of infant death beyond just discouraging bed-sharing (Ahlers-Schmidt et al., 2014; Ball & Volpe, 2013; Blabey & Gessner, 2009; Chianese et al., 2009; Gettler & McKenna, 2010; Hackett & Simons, 2013; Hiscock, 2010). Current safe sleep messages risk stigmatizing families who do choose to bed-share in their given circumstances (Ball & Volpe, 2013; Fetherston & Leach, 2012; Mesich, 2005; Morgan, Groer, & Smith, 2006; Owens, 2008), resulting in underreporting of bed-sharing (Ball & Volpe, 2013; Broussard et al., 2012; Gurbutt & Gurbutt, 2007; Weimer et al., 2002), which can then result in families attempting to reduce the risk of death in their own (not necessarily safer) ways, such as placing pillows around the infant (Ajao et al., 2011; Flick, White, et al., 2001; Fu et al., 2010). Indeed, the AAP’s 2011 policy statement includes language regarding particularly dangerous factors in combination with bed-sharing, including bed-sharing with smokers, under the influence of alcohol or other medications, with infants less than 3 months of age, while excessively tired, with a nonparent or multiple persons, on a soft surface, or with soft bedding (AAP, 2011). Harm reduction messaging could destigmatize bed-sharing, encourage open discussion among caregivers and their providers, and allow discussion of ways to decrease risks when bed-sharing.
A multilevel intervention could address the maternal-level through home visiting programs, which have demonstrated strong potential in improving maternal and infant outcomes (Avellar & Supplee, 2013). The home visitor could tailor messaging to address the mother’s personal beliefs and preferences, for example, by educating her about the AAP’s additional recommendations to reduce risk if she does choose to bed-share (AAP, 2011). To help address the family-level, home visitors could work to intentionally engage the father and other family members in the intervention (Alio, Salihu, Kornosky, Richman, & Marty, 2010; Flick, Vemulapalli, Stulac, & Kemp, 2001), as well as help address any environmental reasons (such as pest infestation or lack of a crib) that may lead them to bed-share. At the community level, the intervention might engage the faith-based community to help educate the community (Maternal and Child Health Bureau, 2004) and to help tailor messaging to the unique needs and experiences of specific minority groups (Campbell & Quintiliani, 2006; Resnicow et al., 1999). Also at the community level, the intervention may engage local child care organizations on safe sleep and to serve as peer educators to parents (Moon, Calabrese, & Aird, 2008; Rimer & Glanz, 2005). At the society level, the intervention may petition local media to depict accurate safe sleep environments (Joyner, Gill-Bailey, & Moon, 2009) and use social marketing and mass media campaigns to share tailored messages about safe sleep.
It is important to note that significant factors may not be directly causal (Glass & McAtee, 2006). Research is needed to help elucidate potential causal factors and to determine the directions of associations between different factors, for example, to determine if breastfeeding leads to bed-sharing, or if another phenomenon (such as belief in attachment parenting) leads to both breastfeeding and bed-sharing. Future research can also add to the literature on risk models among subpopulations, for example, how the historical context of race interacts with other levels of the SEM, as well as how additional factors (i.e., the added stress of racism), might influence bed-sharing.
Nine of the 16 studies included in this review used logistic regression to identify a model of risk factors for increased likelihood of bed-sharing. The significance of a factor is reliant on the other factors included in the model. In our review, factors were presented individually (not part of the whole model), and thus, findings should be interpreted with caution. Regardless, this review provides potential factors to consider incorporating into health promotion strategies. Future research might consider structuring hierarchical regression models based on the SEM to help identify which levels have the most potential impact on bed-sharing. Another limitation of this review is that it focused on quantitative literature; however, qualitative research on reasons for bed-sharing should also be considered in future descriptions of bed-sharing within the SEM (Chianese et al., 2009; Germo et al., 2007; Joyner et al., 2010; Powell Kennedy, Gardiner, Gay, & Lee, 2007). Additionally, this current SEM model is limited to the findings in the current review. This conceptualization of bed-sharing within the SEM is a starting point for building more systematic approaches to addressing the issue of sleep-related infant deaths. Future work may refine and expand the concepts that populate each level of the SEM.
The SEM helps reconceptualize the phenomenon of bed-sharing and provides additional insight into more efficacious intervention design. Communities, counties, and states can use a SEM approach to develop systematic multilevel interventions that both attend to each level individually as well as to the interaction of factors across levels that ultimately influence infant sleep safety. Furthermore, the SEM approach holds great potential for complementing the life-course perspective to reducing disparities in birth outcomes. Layering the SEM approach to infant bed-sharing onto the life-course perspective can help illuminate nuances and connections that may otherwise remain hidden. Innovative new approaches to interventions may also be developed through a more sophisticated understanding of bed-sharing arising from the union of these approaches.
However, the literature remains inconclusive regarding the influence of several factors within each level, such as infant age, birth weight, maternal age, household crowding, and socioeconomic status. Despite conflicting findings, the current literature does suggest that bed-sharing can be influenced by factors at every level of the SEM, and thus interventions should address the multiple levels of influence on bed-sharing. To better inform development of multilevel interventions, more conclusive research is needed (Brenner et al., 2003; Fu et al., 2008; Hauck et al., 2008; Lahr et al., 2007; Morgan & Johnson, 2001; Shields et al., 2005; Weimer et al., 2002; Willinger et al., 2003). In-depth, qualitative studies examining the context of infant sleep among different racial groups may help add clarity to this issue while also informing tailoring of interventions. In turn, tailored interventions addressing the needs of specific racial and ethnic groups may increase the safety of infant sleep environments, decrease the risk for sleep-related infant death, and, ultimately, reduce racial disparities in infant mortality.
Footnotes
Acknowledgements
We gratefully acknowledge direction and support from Mary Kay Madsen, PhD, Professor Emeritus, University of Wisconsin-Milwaukee College of Health Sciences, in earlier versions of this work.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.