Is “Bed Sharing” Beneficial and Safe during Infancy? A Systematic Review

Background. There is conflicting evidence regarding the safety and efficacy of bed sharing during infancy—while it has been shown to facilitate breastfeeding and provide protection against hypothermia, it has been identified as a risk factor for SIDS. Methods. A systematic search of major databases was conducted. Eligible studies were observational studies that enrolled infants in the first 4 weeks of life and followed them up for a variable period of time thereafter. Results. A total of 21 studies were included. Though the quality of evidence was low, bed sharing was found to be associated with higher breastfeeding rates at 4 weeks of age (75.5% versus 50%, OR 3.09 (95% CI 2.67 to 3.58), P = 0.043) and an increased risk of SIDS (23.3% versus 11.2%, OR 2.36 (95% CI 1.97 to 2.83), P = 0.025). Majority of the studies were from developed countries, and the effect was almost consistent across the studies. Conclusion. There is low quality evidence that bed sharing is associated with higher breast feeding rates at 4 weeks of age and an increased risk of SIDS. We need more studies that look at bed sharing, breast feeding, and hazardous circumstance that put babies at risk.


Background
In olden days, mothers and babies were separated after birth to prevent infectious diseases and to keep newborns in a safe and controlled environment. The practice of "rooming-in" (keeping the mother and the baby in same room) started when it was realized that the separation had serious implications for the emotional and psychological development of both mothers and newborns [1]. Rooming-in is now recommended by the World Health Organization (WHO) and UNICEF as part of the Baby Friendly Hospital Initiative (BFHI) programme to promote breastfeeding [2].
During rooming-in, the infant is placed close to the mother either by bed sharing, by an attached side-car crib, or by her bedside in a standalone cot. Of these, the "ideal" method would be "bed sharing" in which the newborn is kept on the same bed as that of the mother. The "bed sharing" behavior, however, is controversial in public health parlance. Some consider it a significant risk factor for sudden infant death syndrome (SIDS) and argue for its wholesale elimination [3][4][5]. Others disagree, finding little or no scientific evidence for an association with SIDS, except among smoking mothers [6][7][8].
When research into SIDS began in the 1980's, researchers had little idea of why these babies died. In the 1990's, bed sharing related to infant sleeping environment was identified as one of the risk factors. By the end of this decade, a stronger interaction between bed sharing and smoking was observed. One recent study exposed an interaction between bed sharing and alcohol/drugs [9]. This study also partially explained the difference in SIDS rates between cultures where cosleeping is the usual practice. Bed sharing is common in certain cultures where the prevalence of SIDS is high, including the African black populations in the United States, Maori, and Aboriginal populations. Bed sharing is common in certain cultures where the prevalence of SIDS is low, including Asian 2 International Journal of Pediatrics communities (Japan, Hong Kong, Bangladesh, and those in UK) and Pacific Islander communities in New Zealand. Actually, it is not the bed sharing that distinguishes these cultures, but other factors (e.g., smoking and use of alcohol/ drugs) which in conjunction with cosleeping may put infants at risk [9].
With this background, we conducted present systematic review to evaluate the efficacy and safety of bed sharing compared to no bed sharing during infancy, to provide updated evidence to the World Health Organization (WHO) which is in the process of reviewing its recommendations on postnatal care of new born infants and their mothers.

Types of Studies.
We intended to include randomized and quasirandomized trials that compared bed sharing of mothers and their infants with no bed sharing. If randomized studies were not available, we planned to include observational studies after applying the following eligibility criteria: (1) cohort studies that enrolled infants in the first 4 weeks of life and followed them up for a variable period of time thereafter were included; (2) case-control studies in which (i) infants being exclusively breastfed at 4-6 wk/3-4 months/6 months of age or those who died of SIDS were the "cases"; (ii) the exposure (i.e., bed sharing) occurred in the neonatal period for at least a few "cases"; if this is not explicitly mentioned in the study, we assumed this if the study had reported the exposure status, that is, bed sharing, as a "routine" or "usual" practice; (3) studies that did not include neonates and measured the exposure in the "last sleep" were excluded; (4) cross-sectional studies that measured the association between breastfeeding status at different time points and bed sharing starting from the neonatal period until 1 year of age were included.
We excluded the studies on "rooming-in" if they had not reported bed sharing separately. When there were multiple publications from one study, only the publication with the most relevant information to the systematic review was used.
2.2. Exposure or Intervention. Bed sharing of mother and her neonate was the exposure studied. Terminology regarding bed sharing, room sharing, and cosleeping used both in the scientific literature and the popular literature is inconsistent and potentially problematic. For the purposes of this paper, "bed sharing" term was used if the newborn was kept on the same bed as that of the mother.

Outcome Measures and Their
Definitions. The primary outcomes were the (a) proportion of infants being breastfed (any or exclusive) at 4-6 wk, 3-4 months, and 6 months of age and (b) proportion of infants dying of SIDS in the first year of life. The secondary outcomes were the incidence of hypothermia, all-cause mortality during neonatal period, and incidence of neonatal sepsis.
Exclusive breastfeeding was defined as an infant receiving only breast milk with no additional foods or liquids, not even water except multivitamin supplements or medications, and any breastfeeding means that an infant receives any amount of breast milk regardless of supplements. SIDS was defined as the unexplained death without warning of an apparently healthy infant usually during sleep. Hypothermia was defined as skin temperature <36.5 ∘ C. Neonatal mortality referred to deaths due to all causes occurring in the first 28 days of postnatal life while neonatal sepsis was defined as the clinical syndrome of bacteremia with systemic signs and symptoms of infection in the first 28 days of postnatal life with or without culture positivity. Searches were limited to human studies. There were no language restrictions. For further identification of ongoing trials, the website http://www.clinicaltrials.gov/ was searched and relevant trials were screened for eligibility of inclusion in the review. We also checked the cross-references of relevant articles. We treated side-car crib exposure (used in some studies) as no bed sharing.

Data
Extraction. Data extraction was done using a standardized data extraction form that was designed and pilot tested a priori. Three authors (Rashmi Ranjan Das, M. Jeeva Sankar, and Ramesh Agarwal) independently extracted data from included studies, including year, setting (country, type of population, socioeconomic status, baseline neonatal mortality, baseline practice of rooming-in or bedding-in, gestation, and birth weight of infants), exposure/intervention (bed sharing, routine or last night), and results (outcome measures, effect, significance). Disagreements in extracted data were resolved through discussion.

Assessment of Risk of Bias in Included
Studies. Two review authors (M. Jeeva Sankar, Rashmi Ranjan Das) independently assessed the methodological quality of the selected studies. Quality assessment was undertaken using the Newcastle Ottawa Scale (NOS) for observational studies. This scale assesses the quality under three major headings, namely, selection of the studies (representativeness, exposure assessment/control selection), comparability (adjustment for main/ additional confounders), and outcome/exposure (adequacy of outcome measured, exposure measured versus self-report) [10]. Any disagreement was resolved through discussion with the third author (Ramesh Agarwal).

Data Analysis.
For each outcome measure, the total number of participants and the number of participants experiencing the event and the adjusted odds ratio (aOR) provided by the study authors were extracted. We intended to use only aOR for pooling the results. If adjusted ratios were not provided in a given study, we used the unadjusted OR. Meta-analysis was done by the generic inverse variance using the user written command "metan" in Stata 11.2 (Stata-Corp, College Station, TX). We used the natural logarithm converted values of aOR and their confidence intervals (CI) for computing the pooled estimates. The heterogeneity between the studies was quantified by using a measure of the degree of inconsistency in their results ( 2 statistic). Given that all the included studies were observational and had an inherent risk of heterogeneity between them, we planned to use fixed-effect model for pooling their results irrespective of the degree of heterogeneity. For the outcome of breastfeeding status, we planned to have only one summary resultbreastfeeding status at 4 to 6 weeks of age.

Breastfeeding.
Only one study reported exclusive breastfeeding rates while the others reported any breastfeeding rates [15]. Mailed questionnaires were used to collect data in all studies except one which used face-to-face interview [15]. All except the one study were from developed countries [15]. Loss to follow-up rate varied from 2.2% to 50% in the included studies. The risk of bias in these studies was moderate to high (Table 1). Almost all the studies showed a consistent beneficial effect on any/exclusive breastfeeding rates at the three different time points, that is, at 4 to 6 weeks, 3 to 4 months, and at 6 months of age. The effect sizes of the individual studies varied from 1.5 to 6.7 (Table 1). We could pool the results of two studies that reported breastfeeding rates at 4-6 weeks of age [13,14]. Pooled analysis showed 3-fold odds of being exposed to bed share during the neonatal period in those who were breastfed compared to those who were not (pooled OR        3.09; 95% CI (2.67, 3.58); 2 = 75.6%; Figure 2). Since all were observational studies, the quality of evidence as assessed by GRADE criteria was found to be low (Table 4).

SIDS.
All except one study that evaluated the association between bed sharing and SIDS were from developed countries [26]. The attrition rate varied from 2.2% to 50% in all but two studies where it was unclear [21,27]. The attrition rate for both the groups was calculated separately by percentage of those who responded to the total questionnaires distributed for each group. Five studies reported routine or usual bed sharing [9,12,20,26,28], whereas; rest reported bed sharing on a particular night (last sleep for cases and reference sleep for controls). When we separately studied routine/usual versus last/reference sleep bed sharing, we found similar increase in risk (Figure 3). This means that pattern of bed sharing does not affect the SIDS risk. More than half of the studies showed significant association between bed sharing and SIDS in the enrolled infants (Table 2). Pooled analysis of all 13 studies demonstrated 2.4-fold odds of being exposed to bed share in those who die of SIDS compared to the controls (aOR 2.36; 95% CI (1.97, 2.83); 2 = 48.5%; Figure 3). Given that all were observational studies, the quality of evidence as assessed by GRADE criteria was found to be low (Table 4).

Secondary Outcomes.
None of the included studies had reported the incidence of hypothermia or sepsis, NMR, among the enrolled infants.

Discussion
"Bed sharing" has been commonly used interchangeably with terms like "bedding-in" or "cosleeping. " "Bed sharing" is a topic of common interest in regard to its controversial association with sudden infant death syndrome (SIDS) on one hand and breastfeeding on the other hand. There are many reasons that make parents prefer bed sharing with their infants, such as ease in breastfeeding, enjoying the time spent with the infant, comforting the infant in case he gets fussy, and to put him to sleep, attending the infant quickly in case of any mishap or during any illness, promoting love, affection, or bonding, and so forth [12].
Breastfeeding has been proposed to be one of the most prominent reasons for bed sharing. But studies have found difficulties in establishing whether successful breastfeeding leads to bed sharing or bed sharing is because of breastfeeding. Based on the demographic and health surveys conducted between years 2002 and 2008, WHO presented data on indicators assessing infant feeding practices for 46 countries [31]. The data showed that breastfeeding rates are the highest in early infancy, which is also the time with the highest prevalence of bed sharing. In a systematic review based on crosssectional studies in infancy beyond the neonatal period the authors found a positive correlation between bed sharing and breastfeeding [32]. However, in a longitudinal study a twoway, complex, interdependent, and temporal relationship was found which was itself insufficient to distinguish the independent role of bed sharing in breastfeeding [16]. A common perception is that mothers who breastfeed commonly also bed share frequently in order to avoid any interruption in breastfeeding at night.
The present systematic review demonstrates significantly association between bed sharing during the neonatal period and breastfeeding at 4-6 weeks of age and at 6 month of age. Admittedly, the quality of evidence was low. Although the data generated from observational studies are of low quality, observational studies are still the most common source of information available to SIDS researchers. The studies were also heterogeneous, but we could not carry out either subgroup or metaregression analyses to investigate the cause of heterogeneity because of the small number of studies.
The AAP task force on SIDS recommends that bed sharing should be avoided when the infant is younger than 3 months irrespective of parents smoking status. They also describe that breastfeeding is associated with a reduced risk of SIDS, and if possible, mothers should exclusively breastfeed for 6 months as per the recommendations of the AAP [33]. These statements are perplexing and might be difficult for the parents to follow.

Comparison with the Existing Literature.
A recently done meta-analysis on relationship between bed sharing and SIDS included case-control studies with predefined criteria (an adequate definition for SIDS; autopsies performed in >95% of cases; an appropriate description of SIDS ascertainment in the study population; a clear description of the process of control selection; and sufficient data to calculate ORs and 95% CIs or the actual ORs and 95% CIs were provided) [34].
The combined OR for SIDS in all bed share versus nonbed share infants was 2.89 (95% CI 1.99, 4.18). The risk was the highest when maternal smoking was present (OR, 6.27 (95% CI 3.94, 9.99)), and the infant was <12 weeks old (OR, 10.37 (95% CI 4.44, 24.21)). In this meta-analysis bed sharing as an exposure was studied at any time period during infancy and included both routine as well as last night bed share. In contrast, we studied bed sharing as an exposure starting during the neonatal period and continued thereafter for a variable time period. Besides this, we also studied breastfeeding simultaneously which was not done in the abovementioned meta-analysis.
The included studies in present meta-analysis consistently aimed to identify the prevalence of known or potential risk factors for SIDS. Three studies more specifically aimed to investigated bed sharing and SIDS [19,27,28]. One study [30] reported bed sharing including those sleeping on the mattress as well as sofa, while another studied bed sharing during the daytime also (this was not used in the present review) [19]. Definitions of sleeping location, whether bed sharing or nonbed sharing, were heterogeneous. For the outcome of SIDS, the most frequently investigated interaction with bed sharing was smoking (most commonly by the mother either during pregnancy or postpartum). We reported the outcomes after adjusting for smoking. An important point that needs mentioned here is that none of the studies reviewed distinguish between planned and accidental bed sharing, a feature that is probably one of the biggest factors involved in the studies identifying an increased risk of this practice.
Only one study from developing country was included in the present meta-analysis [26]. This study reported rate of bed sharing to be 14% in the control subjects, while the SIDS rate in the country was low (0.3 per 1000 live births). Asian countries like Japan and China, though developed, have a lower rate of SIDS despite a higher rate of bed sharing [35,36]. A bed sharing rate of 37% was reported in Japan in year 2006 when the rate of SIDS was only 0.16 per 1000 live births, and the corresponding figures in China were 24% and 0.16, respectively, during year 2002. Considering these figures, a general recommendation of restricted bed sharing might not be appropriate for developing (including Asian) countries in contrast to developed countries where other known risk factors might play important role.
Published studies from developing countries and particularly from Asian community show a low rate of SIDS in spite of a higher rate of bed sharing. This lower prevalence of SIDS is whether due to the protective effect of breastfeeding or is just a reflection of the demographic that chooses breastfeeding is a matter of debate. One study supported these findings and showed that exclusive breastfeeding at 1 month of age halved the risk, and even partial breastfeeding at 1 month of age reduced the risk [37]. They also found a high incidence of SIDS below 6 months age, and based on these findings they recommended that breastfeeding should be continued until at least 6 months of age when the risk of SIDS diminishes. A recent meta-analysis including 18 studies showed that breastfeeding is protective against SIDS, and this effect is stronger when breastfeeding is exclusive [38]. While both the studies identified a direct protective effect of breastfeeding on SIDS, neither of them examined the impact of bed sharing on the individual outcomes. But studies examining the risk factors for SIDS have found the risk of bed sharing to be so profound that the protective effect of breastfeeding did not significantly influence the magnitude of the risk associated with bed sharing [39][40][41]. This has been hypothesized to be due to the lifestyle and socioeconomic class that decides both breastfeeding and SIDS rates supported by findings from developed versus developing or Asian countries [40,42]. Finally, it is actually difficult to isolate breastfeeding and bed sharing from other risk factors related to SIDS.
We did not find any relevant studies, with a contemporaneous comparison, examining the effect of bed sharing in relation to other outcomes of interest (the incidence/ prevalence of hypothermia, mortality, and sepsis rate in bed sharing neonates or infants).

Strengths and Weaknesses of Review.
The strength of present systematic review is that it included studies having exposure to bed share during neonatal period (and continued thereafter) and addressed both the benefits (breastfeeding) and risks (of SIDS) of bed sharing. But there are some potential limitations that merit attention: (a) methodological issues: all the studies were observational and were conducted mostly in developed country settings; the age group included from the neonatal period and up to 2 years of age, thus making it difficult to generalize the result (as SIDS is meant for during infancy only); the control groups were also variably defined in all the studies (as shown in Table 2); (b) dramatic change in SIDS epidemiology during the last decade [43]; (c) since we included only studies having exposure to bed share during neonatal period and excluded those having exposure during last sleep only, our results cannot be generalized to SIDS as a whole in infancy, as last night bed sharing that happened after the neonatal period has not been captured in our review. Also, we were not able to evaluate the "net" beneficial effect of the intervention as none of the studies had reported both breastfeeding and SIDS rates together.

Further Area of Research.
A total of 21 observational studies (breastfeeding and bed sharing = 8; SIDS and bed sharing = 13) were included in the present review. None of the studies reported both the primary outcomes (SIDS and breastfeeding) in bed sharing population, so we were not able to evaluate the "net" effect of bed sharing. Most of the included studies were undertaken more than a decade ago, and the prevalence of bed sharing might have changed as a result of guideline recommendations or societal factors. Bed sharing is a well-known cultural practice in developing countries, but only one study was found to be eligible for inclusion in the review. Control groups were also variably defined in all the included studies. Because prospective studies of SIDS are not possible given the now rarity of these deaths, more detailed retrospective studies that look at bed sharing, breastfeeding, and the hazardous circumstances that put babies at risk are needed.

Conclusion
There is low quality evidence that bed sharing is associated with higher breastfeeding rates at 4 weeks of age and increased risk of SIDS irrespective of maternal smoking. Due to paucity of studies, it is difficult to predict whether neonates are at a more risk than older infants (>1 month age). We need more detailed studies that look at bed sharing, breastfeeding and hazardous circumstance that put babies at risk.