Description of the condition
In humans, routine mother-infant separation shortly after birth is unique to the 20th century. This practice diverges from evolutionary history, where neonatal survival depended on close and virtually continuous maternal contact. Although from an evolutionary perspective skin-to-skin contact (SSC) is the norm, separating the newborn from its mother soon after birth has now become common practice in many industrialized societies. Therefore, for the purpose of this review, SSC has to be the experimental intervention. Ironically, and importantly, the experimental intervention in studies with all other mammals is to separate newborns from their mothers.
Description of the intervention
Early SSC is the placing of the naked baby prone on the mother's bare chest at birth or soon afterwards. In the evolutionary context, this would have been "immediate and continuous". In the current care context, initiation and duration are not defined. The concept of "care" does not change; only the place where such care is provided changes. Further, although a dose-response effect has not been documented in randomized controlled trials (RCTs), the general belief is that SSC should continue until the end of the first successful breastfeeding to show an effect and to enhance early infant self-regulation (Widstrom 2011).
How the intervention might work
The rationale for SSC comes from animal studies in which some of the innate behaviors of neonates that are necessary for survival are shown to be habitat dependent (Alberts 1994). In mammalian biology, maintenance of the maternal milieu following birth is required to elicit innate behaviors from the neonate and the mother that lead to successful breastfeeding, and thus survival. Separation from this milieu results in immediate distress cries (Alberts 1994) and "protest-despair" behavior. Human infants placed in a cot cry 10 times more than SSC infants.Their cry is similar to the vocalizations of separated rat pups (Michelsson 1996). In rodent studies, the pups who had the least attentive contact from their mothers were the ones whose health and intelligence were compromised across the lifespan (Francis 1999; Liu 1997; Liu 2000; Meaney 2005; Plotsky 2005). Also in the report by Liu 2000 a cross-fostering study provided evidence for a direct relationship between maternal behavior and hippocampal development in the offspring.
Healthy, full term infants employ a species-specific set of innate behaviors immediately following delivery when placed in SSC with the mother (Righard 1990; Varendi 1994; Varendi 1998; Widstrom 1987; Widstrom 1990). They localize the nipple by smell and have a heightened response to odor cues in the first few hours after birth (Porter 1999; Varendi 1994; Varendi 1997). More recently Widstrom 2011 described the sequence of nine innate behaviors as the birth cry, relaxation, awakening and opening the eyes, activity (looking at the mother and breast, rooting, hand to mouth movements, soliciting sounds), a second resting phase, crawling towards the nipple, touching and licking the nipple, suckling at the breast and finally falling asleep. This 'sensitive period' predisposes or primes mothers and infants to develop a synchronous reciprocal interaction pattern, provided they are together and in intimate contact. Infants who are allowed uninterrupted SSC immediately after birth and who self-attach to the mother's nipple may continue to nurse more effectively. Effective nursing increases milk production and infant weight gain (De Carvalho 1983; Dewey 2003). Anderson 2004a used SSC as an intervention for 48 healthy mother/full term infant dyads with breastfeeding problems identified between 12 to 24 hours postbirth. SSC was provided during the next three consecutive breastfeedings. Breastfeeding was successful, even in this racially disparate sample (Chiu 2008) and was exclusive in 81% of these dyads at hospital discharge, 73% at one week, and 52% at one month postbirth. Temperatures were taken before (baseline), during, and after each SSC breastfeeding. Baseline temperatures reached, and remained in thermoneutral range (Chiu 2005) suggesting that mothers have the ability to modulate infant temperature if given the opportunity to breastfeed in SSC. Because these mothers and their infants were having breastfeeding difficulties, hospital staff and parents can logically be reassured that healthy newborn infants, with or without breastfeeding difficulties, may safely breastfeed in SSC so far as temperature is concerned. In a study of infrared thermography of the whole body during the first hour postbirth, Christidis 2003 found that SSC was as effective as radiant warmers in preventing heat loss in healthy full term infants.
SSC through sensory stimuli such as touch, warmth, and odor is a powerful vagal stimulant, which among other effects releases maternal oxytocin (Uvnas-Moberg 1998; Winberg 2005). Oxytocin causes the skin temperature of the mother's breast to rise, providing warmth to the infant (Uvnas-Moberg 1996). When operating in a safe environment, oxytocin, and direct SSC stimulation of vagal efferents, are probably part of a broader neuro-endocrine milieu (Porges 2007). A global physiological regulation of the autonomic nervous system is achieved, supporting growth and development, (homeorhesis). Under conditions perceived by the newborn to be dangerous, stress mechanisms come into operation, with the focus on survival rather than development (allostasis). The theory of allostasis is the relationship between psycho-neurohormonal responses to stress and physical and psychological manifestations of health and illness (McEwen 1998; Shannon 2007). Allostasis is necessary, and it can be viewed as beneficial, because its goal is to bring aberrant physiology closer to normal; however, an allostatic response comes with a physiological cost referred to as allostatic load. The higher the allostatic load the greater the damage from stress, because allostatic load is cumulative. SSC also lowers maternal stress levels. Handlin 2009 found a dose-response relationship between the amount of SSC and maternal plasma cortisol two days postbirth. A longer duration of SSC was correlated with a lower median level of cortisol (r = - 0.264, P = 0.044).
Oxytocin antagonizes the flight-fight effect, decreasing maternal anxiety and increasing calmness and social responsiveness (Uvnas-Moberg 2005). During the early hours after birth, oxytocin may also enhance parenting behaviors (Uvnas-Moberg 1998; Winberg 2005). SSC outcomes for mothers suggest improved bonding/attachment (Affonso 1989); other outcomes are increased sense of mastery and self-enhancement, resulting in increased confidence. Sense of mastery and confidence are relevant outcomes because they predict breastfeeding duration (Dennis 1999). Women with low breastfeeding confidence have three times the risk of early weaning (O'Campo 1992) and low confidence is also associated with perceived insufficient milk supply (Hill 1996).
Marin 2010 found that time to expulsion of the placenta was shorter (M = 409 + 245 sec.) in mothers of SSC infants than in control mothers (M = 475 + 277 sec., P = 0.05). When SSC on the mother's abdomen, the infant's knees and legs press into her abdomen in a massaging manner which would logically induce uterine contractions and thereby reduce risk of postpartum hemorrhage. Mothers who experience SSC have reduced bleeding (Dordevic 2008) and more rapid delivery of the placenta (Marin 2010).
In previous meta-analyses with full term infants, early contact was associated with continued breastfeeding (Bernard-Bonnin 1989; Inch 1989; Perez-Escamilla 1994). Just altering hospital routines can increase breastfeeding levels in the developed world (Rogers 1997). Conde-Agudelo 2011 conducted a Cochrane review of 16 randomized clinical trials of kangaroo mother care (KMC), a strategy of continuous or intermittent SSC with exclusive or nearly exclusive breastfeeding and early hospital discharge of infants less than 2500 g at birth in settings with limited resources. KMC was associated with reductions in several clinically important adverse infant outcomes, including mortality at hospital discharge and at latest follow-up, nosocomial infection/sepsis at hospital discharge and severe infection/sepsis at latest follow-up, hypothermia and hospital length of stay. SSC mothers were more satisfied with the method of care, and more likely to be exclusively breastfeeding at hospital discharge. In another meta-analysis of 24 studies (13 case-series, five RCT's, one cross-over and four cohort), Mori 2010 evaluated outcomes in both low and normal birthweight infants up to 28 days old. Infant body temperature increased 0.22 ºC during and 0.14 ºC after SSC (P < 0.001, 21 studies); heart rate increased 2.04 beats per minute (bpm) during (P = 0.05) and decreased 0.07 bpm after SSC (P = 0.95, 12 studies); oxygen saturation decreased 0.60% (three/fifths of 1%) during (P = 0.01) and 0.48% (essentially one-half of 1%) after SSC (P = 0.06, 10 studies). These decreases in oxygen saturation are too small to be of clinical significance.
Why it is important to do this review
Separation of mothers from their newborn infants at birth has become standard practice, despite mounting evidence that this may have harmful effects. However, delivery room and postpartum hospital routines may significantly disrupt early maternal-infant interactions including breastfeeding (Anderson 2004a; Odent 2001; Winberg 1995). The possibility exists that postnatal separation of human infants from their mothers is stressful (Anderson 1995) and might result in harmful effects that persist across the lifespan, if the studies with laboratory animals cited earlier hold true for humans. This possibility needs careful evaluation using the allostatic theoretical framework (McEwen 1998) as well as new epigenetic findings (Meaney 2005).
A concurrent widespread decline in breastfeeding is of major public health concern. Although more women are initiating breastfeeding, fewer are breastfeeding exclusively. Using data from the Infant Feeding Practices Study II conducted in the United States by the Food and Drug Administration in 2005 to 2007, Grummer-Strawn 2008 found that 83% of mothers initiated breastfeeding, but only 48% exclusively breastfed during their hospital stay. These innate behaviors can be disrupted by early postpartum hospital routines as shown experimentally by Widstrom 1990 and in descriptive studies by Gomez 1998; Jansson 1995 and Righard 1990. Gomez 1998 found that infants were eight times more likely to breastfeed spontaneously if they spent more than 50 minutes in SSC with their mothers immediately after birth, and concluded that the dose of SSC might be an essential component regarding breastfeeding success. Bramson 2010 showed a clear dose-response relationship between SSC in the first three hours postbirth and exclusive breastfeeding at discharge in a large (N = 21,842 mothers) hospital-based cohort study, (odds ratio (OR) for exclusive breastfeeding = 1.665 if in SSC for 16 to 30 minutes, and OR = 3.145 for more than 60 minutes of SSC).
The purpose of this review is to examine the available evidence of the effects of early SSC on breastfeeding exclusivity and duration and other outcomes in mothers and their healthy full term and late preterm newborn infants. Although our intent is to examine all clinically important outcomes, breastfeeding is the predominant outcome investigated so far in healthy newborns. Hence, our emphasis is on breastfeeding, although we also will examine maternal-infant physiology and behavior. Because the focus of this review is on mothers and their healthy infants, potential effects of early SSC on father-infant attachment and also the resistance of staff to this intervention are beyond the scope of this review. Maternal feelings about early SSC and satisfaction with the birth experience are important and relevant but require more qualitative methods. The focus of this review is on randomized controlled trials used to test the effects of early SSC. This is an update of a Cochrane review first published in 2003 and previously updated in 2007.
To assess the effects of early skin-to-skin contact for healthy newborn infants compared to standard contact (infants held swaddled or dressed in their mothers arms, placed in open cribs or under radiant warmers).
The three main outcome categories include:
a) establishment and maintenance of breastfeeding/lactation;
b) infant physiology - thermoregulation, respiratory, cardiac, metabolic function, neurobehavior;
c) maternal-infant bonding/attachment.
Criteria for considering studies for this review
Types of studies
All randomized controlled trials in which the active encouragement of early skin-to-skin contact (SSC) between mothers and their healthy newborn infants was compared to usual hospital care. We have not included quasi-randomized trials (e.g. where assignment to groups was alternate or by day of the week, or by other non-random methods).
Types of participants
Mothers and their healthy full term or late preterm newborn infants (34 to less than 37 completed weeks' gestation) having early SSC starting less than 24 hours after birth, and controls undergoing standard patterns of care.
Types of interventions
Early SSC for term or late preterm infants can be divided into several subcategories.
(a) In 'birth SSC', the infant is placed prone skin-to-skin on the mother's abdomen or chest during the first minute postbirth. The infant is suctioned while on the mother's abdomen or chest, if medically indicated, thoroughly dried and covered across the back with a prewarmed blanket. To prevent heat loss, the infant's head may be covered with a dry cap that is replaced when it becomes damp. Ideally, all other interventions are delayed until at least the end of the first hour postbirth or the first successful breastfeeding.
(b) In 'very early SSC', beginning approximately 30 to 40 minutes postbirth, the naked infant, with or without a cap, is placed prone on the mother's bare chest. A blanket is placed across the infant's back.
(c) 'Early SSC' can begin anytime between one and 24 hours postbirth. The baby is naked (with or without a diaper and cap) and is placed prone on the mother's bare chest between the breasts. The mother may wear a blouse or shirt that opens in front, or a hospital gown worn backwards, and the baby is placed inside the gown so that only the head is exposed. What the mother wears and how the baby is kept warm and what is placed across the baby's back may vary. What is most important is that the mother and baby are in direct ventral-to-ventral SSC and the infant is kept dry and warm.
In the future these groups may be analyzed separately. However, at present, not enough studies are available for subgroup analysis.
Standard contact includes a number of diverse conditions, infants held swaddled or dressed in their mothers arms, or infants placed in open cribs or under radiant warmers in the mother's room or elsewhere with no holding allowed.
Types of outcome measures
- Number of mothers breastfeeding (any breastfeeding) one month to four months postbirth
- Duration of breastfeeding
- Infant stabilization during the transition to extra-uterine life Measured by the SCRIP score (e.g. stability of the cardio-respiratory system – a composite score of heart rate, respiratory status and arterial hemoglobin oxygen saturation (SaO2), range of scores = 0-6 (Fischer 1998)
- Blood glucose levels during/after SSC compared to standard care
- Infant thermoregulation = temperature changes during/after SSC compared to standard care (measured by axillary temperature)
Breastfeeding outcomes (secondary)
- Maternal breast temperature during and after SSC
- Breast engorgement
Infant outcomes (secondary)
- Infant heart rate during/after SSC compared to standard care
- Respiratory status - respiratory rate during/after SSC compared to standard care
- Neonatal intensive care unit admissions
- Infant weight changes/rate of growth in gm/kg/day (daily weight change, change in weight over days of study) (Hill 2007)
- Length of hospital stay
- Duration of infant crying
- Maternal perceptions of bonding/connection to her infant
- Maternal pain post cesarean
- Maternal sensitivity to her infant’s cues
- Maternal anxiety
- Maternal parenting confidence
Search methods for identification of studies
We searched the Cochrane Pregnancy and Childbirth Group’s Trials Register by contacting the Trials Search Co-ordinator (30 November 2011).
The Cochrane Pregnancy and Childbirth Group’s Trials Register is maintained by the Trials Search Co-ordinator and contains trials identified from:
- quarterly searches of the Cochrane Central Register of Controlled Trials (CENTRAL);
- weekly searches of MEDLINE;
- weekly searches of EMBASE;
- handsearches of 30 journals and the proceedings of major conferences;
- weekly current awareness alerts for a further 44 journals plus monthly BioMed Central email alerts.
Details of the search strategies for CENTRAL, MEDLINE and EMBASE, the list of handsearched journals and conference proceedings, and the list of journals reviewed via the current awareness service can be found in the ‘Specialized Register’ section within the editorial information about the Cochrane Pregnancy and Childbirth Group.
Trials identified through the searching activities described above are each assigned to a review topic (or topics). The Trials Search Co-ordinator searches the register for each review using the topic list rather than keywords.
Searching other resources
The first three review authors have been active trialists in this area and have personal contact with many groups in this field including the International Network for Kangaroo Mother Care, based in Trieste (see Appendix 1).
We did not apply any language restrictions.
Data collection and analysis
For the methods used when assessing the trials identified in the previous version of this review, see Appendix 2.
For this update we used the following methods when assessing the reports identified by the updated search.
Selection of studies
Two review authors independently assessed for inclusion all the potential studies we identified as a result of the search strategy. First, we screened titles and abstracts of all the retrieved studies. Two review authors independently assessed full text articles for inclusion in the review. We resolved any disagreement through discussion or, if required, we consulted a third individual. We have listed studies that did not meet the inclusion criteria for the review in the Characteristics of excluded studies tables along with the reasons for their exclusion.
Data extraction and management
We designed a form to extract data. For eligible studies, at least two review authors extracted the data using the agreed form. We resolved discrepancies through discussion or, if required, we consulted a third person. We entered data into Review Manager software (RevMan 2011) and checked for accuracy.
When information regarding any of the above was unclear, we attempted to contact authors of the original reports to provide further details.
Assessment of risk of bias in included studies
Two review authors independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We resolved any disagreement by discussion or by involving a third assessor. We used the electronic 'Risk of bias' form in RevMan 2011 to describe study methodological quality.The following criteria were assessed.
(1) Sequence generation (checking for possible selection bias)
We described for each included study the method used to generate the allocation sequence in sufficient detail to allow an assessment of whether it produced comparable groups.
We assessed the method as:
- low risk of bias (any truly random process, e.g. random number table; computer random number generator);
- high risk of bias (any non-random process, e.g. odd or even date of birth; hospital or clinic record number);
- risk of bias unclear.
(2) Allocation concealment (checking for possible selection bias)
We described for each included study the method used to conceal the allocation sequence and determined whether intervention allocation could have been foreseen in advance of, or during recruitment, or changed after assignment.
We assessed the methods as:
- low risk of bias (e.g. telephone or central randomization; consecutively numbered sealed opaque envelopes);
- high risk of bias (open random allocation; unsealed or non-opaque envelopes, alternation; date of birth);
- risk of bias unclear.
(3) Blinding (checking for possible performance and detection bias)
We described for each included study the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant received. We considered that studies were at low risk of bias if they were blinded, or if we judged that the lack of blinding could not have affected the results. We assessed blinding separately for different outcomes or classes of outcomes.
We assessed the methods as:
- low, high or unclear risk of bias for participants;
- low, high or unclear risk of bias for personnel;
- low, high or unclear risk of bias for outcome assessors.
(4) Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations)
We described for each included study, and for each outcome or class of outcomes, the completeness of data including attrition and exclusions from the analysis. We stated whether attrition and exclusions were reported, the numbers included in the analysis at each stage (compared with the total randomized participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups or were related to outcomes. Where sufficient information was reported, or could be supplied by the trial authors, we re-included missing data in the analyses. We assessed methods as:
- low risk of bias where loss was low and was balanced across groups;
- high risk of bias where attrition was high or unbalanced across groups;
- unclear risk of bias.
(5) Selective reporting bias
We described for each included study how we investigated the possibility of selective outcome reporting bias and what we found.
We assessed the methods as:
- low risk of bias (where it was clear that all of the study’s prespecified outcomes and all expected outcomes of interest to the review had been reported);
- high risk of bias (where not all the study’s prespecified outcomes had been reported; one or more reported primary outcomes were not prespecified; outcomes of interest were reported incompletely and so could not be used; study failed to include results of a key outcome that would have been expected to have been reported);
- unclear risk of bias.
(6) Other sources of bias
We described for each included study any important concerns we had about other possible sources of bias.
We assessed whether each study was free of other problems that could put it at risk of bias:
- low, high or unclear risk of other bias.
(7) Overall risk of bias
We made explicit judgements about whether studies were at high risk of bias, according to the criteria given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). With reference to (1) to (6) above, we assessed the likely magnitude and direction of the bias and whether we considered it is likely to impact on the findings. We explored the impact of the level of bias through undertaking sensitivity analysis.
Measures of treatment effect
For dichotomous data, we have presented results as summary risk ratio with 95% confidence intervals.
For continuous data, we used the mean difference if outcomes were measured in the same way between trials. We used the standardized mean difference to combine trials that measured the same outcome, but used different methods.
Unit of analysis issues
The unit of analysis was the healthy newborn infant receiving SSC or standard care.
We did not identify any cluster-randomized trials in this version of the review. If we identify such trials in future updates we will include them in the review along with individually randomized trials.
To take account of design effect, provided sufficient information is available, we will adjust the sample sizes and event rates from cluster-randomized studies using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We will use an estimate of the intracluster correlation co-efficient (ICC) derived from the trial (if possible), or from another source. If ICCs from other sources are used, we will report this and will conduct sensitivity analyses to investigate the effect of variation in the ICC.
We will consider that it is reasonable to combine the results from both individually and cluster-randomized trials if there is little heterogeneity between the study designs and the interaction between the effect of intervention and the choice of randomization unit is considered to be unlikely.
Dealing with missing data
For included studies, we noted levels of attrition and have described in the Characteristics of included studies tables the levels of loss to follow up at each data collection point.
For all outcomes, we carried out analyses, as far as possible, on an intention-to-treat basis, i.e. we attempted to include all participants randomized to each group in the analyses, and all participants were analyzed in the group to which they were allocated, regardless of whether or not they received the allocated intervention. The denominator for each outcome in each trial was the number randomized minus any participants whose outcomes were known to be missing.
Assessment of heterogeneity
We assessed statistical heterogeneity in each meta-analysis using the I² and Chi² statistics. We regarded heterogeneity as substantial if either I² was greater than 40% or there was a low P value (less than 0.10) in the Chi² test for heterogeneity.
Assessment of reporting biases
If there were 10 or more studies in the meta-analysis we planned to investigate reporting biases (such as publication bias) using funnel plots. In this version of the review insufficient studies contributed data to allow us to carry out this analysis for all but one of the outcomes. In future updates, if more studies become available we will assess funnel plot asymmetry visually, and will use formal tests for funnel plot asymmetry. For continuous outcomes, we will use the test proposed by Egger 1997, and for dichotomous outcomes, we will use the test proposed by Harbord 2006. If asymmetry is detected in any of these tests or is suggested by a visual assessment, we will perform exploratory analyses to investigate it.
We carried out statistical analysis using the Review Manager software (RevMan 2011). We used fixed-effect meta-analysis for combining data when it was reasonable to assume that studies were estimating the same underlying treatment effect: i.e. where trials were examining the same intervention, and the trials’ populations and methods were judged sufficiently similar. If there was clinical heterogeneity sufficient to expect that the underlying treatment effects differed between trials, or if substantial statistical heterogeneity was detected, we used random-effects meta-analysis to produce an overall summary if an average treatment effect across trials was considered clinically meaningful. The random-effects summary was treated as the average range of possible treatment effects and we have discussed the clinical implications of treatment effects differing between trials. If the average treatment effect was not clinically meaningful, we did not combine trials.
If we used random-effects analyses, the results have been presented as the average treatment effect with its 95% confidence interval, along with estimates of I².
Subgroup analysis and investigation of heterogeneity
For primary outcomes we planned subgroup analysis by:
- gestational age at birth; infants born at term (greater than 37 weeks) versus later preterm (greater than 34 to 37 weeks);
- type of SSC (at birth, very early SSC, and early SSC).
We planned to assess differences between subgroups by using the interaction tests available in RevMan 2011. In this version of the review insufficient studies contributed data to allow us to carry out the planned analysis. In future updates, as more data become available, we hope to be able to look for possible differences between subgroups.
We planned to carry out sensitivity analysis to look at whether the methodological quality of studies had an impact on results; however, none of the included studies met all criteria for low risk of bias and we therefore did not carry out this analysis in this version of the review. In view of the mixed methodological quality of trials we advise caution in the interpretation of results. For our two primary outcomes there were high levels of heterogeneity; much of the variation was due to a single study; we therefore carried out sensitivity analysis excluding this study from the analysis to examine the impact on results (Sosa 1976a). For infant physiological outcomes, we also carried out sensitivity analysis to explore high levels of heterogeneity.
Description of studies
Results of the search
Thirty-four studies with 2177 mother-infant dyads met the inclusion criteria. None of the 34 studies met all of the methodological quality criteria (see Figure 1 and Figure 2). The total sample sizes in the studies ranged from eight to 204 mother-infant pairs. The studies represented very diverse populations in Canada, Chile, Germany, Guatemala, Iran, Israel, Italy, Japan, Nepal, Poland, Russia, South Africa, Spain, Sweden, Taiwan, Thailand, the United Kingdom, and the United States. All but four of the 34 studies included only healthy full term infants. Four studies (Anderson 2003; Bergman 2004; Chwo 1999; Syfrett 1996) were done with healthy late preterm infants who were assigned to the normal newborn nursery. Three studies (Gouchon 2010; McClellan 1980; Nolan 2009) were conducted with mothers scheduled for repeat cesarean birth using regional anesthesia. One study (Huang 2006) was conducted with hypothermic, but otherwise healthy, newborns postcesarean birth with spinal anesthesia. One paper reported results for studies carried out in three different sites and we have treated these as three different studies in the data and analysis (Sosa 1976a; Sosa 1976b; Sosa 1976c). A large number of outcomes (31) have been reported in the analysis, but only 15 included multiple trials and for many of the other outcomes only a relatively small number of studies (two or three) contributed data. Four of the 34 included studies did not report data on our prespecified primary and secondary outcomes and data from these studies have not been included in the analysis (Curry 1982; Fardig 1980; Ferber 2004; Hales 1977). Details of all included studies are set out in the Characteristics of included studies tables.
|Figure 1. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.|
|Figure 2. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
The characteristics of the intervention varied greatly between studies. Duration of skin-to-skin (SSC) ranged from approximately 15 minutes (De Chateau 1977; Svejda 1980; Thomson 1979; Vaidya 2005) to a mean of 37 of 48 hours (84%) of continuous SSC (Syfrett 1996); in this study all dyads received 24 minutes of SSC before randomization. Hake-Brooks 2003 (under Anderson 2003) reported that SSC mothers gave SSC 22% of the time and held their wrapped infants for 11.6% of the observation period. Although SSC began by 0 to 15 minutes postbirth in 18 of the 34 studies, the SSC dyads in the study by Shiau 1997 could not begin until four hours postbirth because of hospital policy. SSC did not begin until a mean of 21.3 hours postbirth in the study by Chwo 1999 of late preterm infants 34 to 36 weeks' gestational age. In 22 of the 34 studies the infants were given the opportunity to suckle during SSC but only five studies (Carfoot 2004; Carfoot 2005; Gouchon 2010; Khadivzadeh 2008; Moore 2005) documented the success of the first breastfeeding using a validated instrument, the Infant Breastfeeding Assessment Tool. The amount of assistance the mothers received with breastfeeding during SSC was unclear in many of the research reports.
Substantial differences were found between studies in the amount of separation that occurred in the control group. In eight studies (Chwo 1999; Hales 1977; Huang 2006; Mizuno 2004; Shiau 1997; Sosa 1976a; Sosa 1976b; Sosa 1976c), infants were removed from their mothers immediately postbirth and reunited 12 to 24 hours later. In five studies (Carlsson 1978; Craig 1982; Gouchon 2010; Svejda 1980; Thomson 1979), the mothers held their swaddled infants for about five minutes soon after birth and then were separated from their infants. Control mothers held their swaddled infants six times for 60 minutes in Chwo 1999, 20 minutes in Kastner 2005, 60 minutes in Moore 2005 and for two hours in the recovery room in Punthmatharith 2001. The swaddled control infants in Khadivzadeh 2008 were reunited with their mothers after the episiotomy repair. Control infants in Nolan 2009 were separated from their mothers for a mean of 21 minutes and in Gouchon 2010 for a mean of 51 minutes post-cesarean birth. There were four groups in the study by Bystrova 2003; an SSC group, a mother's arms group where the infants were held swaddled or dressed, a nursery group and a reunion group where the infants were taken to the nursery immediately postbirth for 120 minutes but reunited with their mothers for rooming-in on the postpartum unit. In Hake-Brooks 2003 (under Anderson 2003) control mothers held their wrapped infants 13.9% of the time (M = 6.67 hours).
Fifty studies were assessed and excluded from the review. The primary reason for exclusion was that the investigators did not state that the infants in the intervention group received early SSC with their mothers. When the information in the research report was unclear, where possible we contacted the investigators, to determine whether the early contact was indeed skin-to-skin (see the table of Characteristics of excluded studies).
New studies found at this update
Five randomized controlled trials have been added to the review.Two of the new studies (Gouchon 2010; Nolan 2009) were conducted with mothers scheduled for repeat cesarean birth using regional anesthesia. One study (Huang 2006) was conducted with hypothermic, but otherwise healthy, newborns post-cesarean birth with spinal anesthesia. The results from four additional reports involving the data set from Bystrova 2003, two additional reports from Anderson 2003 and one additional report from Bergman 2004 have been added to this update.
Risk of bias in included studies
The overall methodological quality of the included studies was mixed and many studies did not provide clear information on the way the that the randomization sequence was generated, or on the methods used to conceal group allocation at the point of randomization. In 18 of the 34 studies the way the randomization sequence was produced was not clearly described and in 22 of the 34 studies, not enough information was provided to determine if the method of random assignment was robust before allocation of the participants to groups occurred. In two studies (De Chateau 1977; McClellan 1980), allocation concealment was at high risk of bias because the researchers used an open table of random numbers. In seven studies (Anderson 2003; Bergman 2004; Chwo 1999; Moore 2005; Punthmatharith 2001; Shiau 1997; Syfrett 1996), allocation concealment was controlled by using a computer program to assign women to groups (the minimization method). Although the Syfrett 1996 study was small (n = 8), the recruiter was naive to the minimization method of random assignment. In 12 studies (Carlsson 1978; Christensson 1992; Christensson 1995; Hales 1977; Kastner 2005; Khadivzadeh 2008; Mazurek 1999; Mizuno 2004; Svejda 1980; Thomson 1979; Vaidya 2005; Villalon 1993), the researchers indicated that women were randomly assigned to groups but no further information was provided about the randomization method. In eight studies (Bystrova 2003; Carfoot 2004; Craig 1982; Curry 1982; Nolan 2009; Sosa 1976a; Sosa 1976b; Sosa 1976c), sealed envelopes were used but the investigators do not state whether the envelopes were sequentially numbered, although Bystrova 2003 noted that the envelopes were opened consecutively.
Blinding women, clinical staff and outcome assessors to treatment group is extremely difficult for this type of intervention, and we found it difficult to judge the impact of lack of blinding on particular outcomes. We have provided a single assessment for performance blinding and detection bias for these studies taking into account both the information provided by investigators, and our assessments of the potential impact of lack of blinding on the particular outcomes measured (i.e. we considered for example, that lack of blinding may have had more impact on outcomes where unblinded outcome assessors observed maternal behavior and less impact on laboratory investigations). In 28 studies we rated the impact of lack of blinding as unclear; in studies where an unblinded investigator provided clinical care for the treatment group and also collected outcome data, we assessed that results may have been more susceptible to observer bias. In the Characteristics of included studies tables we have described for each study any attempts investigators made to blind women, staff and outcome observers.
None of the research reports stated that the delivery and postpartum staff were unaware of the group assignment of the mothers. Ferber 2004, however, stated that the nursery staff were blind to patient group assignment. Therefore, in the majority of studies control for provider performance bias was difficult to determine. In the 10 studies that evaluated infant physiological outcomes (Bergman 2004; Bystrova 2003; Christensson 1992; Fardig 1980; Gouchon 2010; Huang 2006; Mazurek 1999; Nolan 2009; Syfrett 1996; Villalon 1993), however, patient or provider performance bias may not be as significant an issue as it might be with maternal attachment and breastfeeding outcomes. Surprisingly, there was more effort made to control for patient performance bias than for provider performance bias. In seven studies (Carlsson 1978; Craig 1982; Curry 1982; Ferber 2004; Kastner 2005; Svejda 1980; Thomson 1979), it was reported that the women were not aware that they were receiving an experimental treatment and/or they were not informed about the true purpose of the study. Adequate control for patient performance is problematic in the more recent studies because of Institutional Review Board requirements that investigators disclose the true purpose of the study or the experimental conditions, or both. In 14 of the 34 studies, outcome assessors (whenever possible) were reported to be unaware of the woman's group assignment. In several studies, when infant physiological or crying data were obtained by observation during SSC (Bergman 2004; Bystrova 2003; Christensson 1992; Christensson 1995; Fardig 1980; Gouchon 2010; Huang 2006; Mazurek 1999; Syfrett 1996; Villalon 1993), the outcome assessors could not be masked.
Incomplete outcome data
In all but one study (Carlsson 1978), outcome data were either obtained on all the enrolled women or reasons were provided for women who withdrew or had to be withdrawn. Six investigators (Anderson 2003; Bergman 2004; Bystrova 2003; Carfoot 2005; Gouchon 2010; Moore 2005), utilized the Consort Guidelines (Moher 2001; Moher 2010) to document the flow of participants through their clinical trial.
Effects of interventions
All the studies reviewed were randomized controlled trials. Most of our primary and secondary outcomes were measured in only one or two studies and where more studies contributed data there was some heterogeneity between trials. Where we identified moderate or high statistical heterogeneity (I² greater than 40%), we have drawn attention to this in the text and would urge caution in the interpretation of these results which show the average treatment effect.
Primary outcomes - breastfeeding rates/duration
Early SSC resulted in better overall performance on several measures of breastfeeding status, although there was heterogeneity between studies.
More SSC dyads were still breastfeeding one to four months postbirth (average risk ratio (RR) 1.27, 95% confidence interval (CI) 1.06 to 1.53). This meta-analysis included 13 studies and involved 702 mother-infant pairs. In 11 of the 13 studies, SSC dyads were more likely to be breastfeeding at one to four months postbirth, although the difference reached statistical significance in only two studies (Sosa 1976c; Thomson 1979). Overall, there were differences in the size of the treatment effect between studies leading to moderate heterogeneity for this outcome (T² = 0.04, P = 0.03, I² = 47%) ( Analysis 1.1). Much of the heterogeneity was due to a single study (Sosa 1976a) (the study author speculated that variation in treatment effect was due to the particular population attending the study hospital). We carried out a sensitivity analysis removing this study from the analysis; when this study was excluded there was no evidence of statistical heterogeneity (T
|Figure 3. Funnel plot of comparison: 1 Skin-to-skin versus standard contact healthy infants, outcome: 1.1 Breastfeeding 1 month to 4 months postbirth.|
Seven studies with 324 mother/infant pairs reported data on the duration of breastfeeding in days. Six of the seven studies (De Chateau 1977; Mizuno 2004; Shiau 1997; Sosa 1976b; Sosa 1976c; Svejda 1980) found a longer duration of breastfeeding in the SSC dyads (mean difference (MD) 42.55 days, 95% CI -1.69 to 86.79). However, overall, the difference between groups did not reach statistical significance. Results from this meta-analysis should be interpreted with caution; there was considerable heterogeneity for this outcome and some of the findings are from studies with very small sample sizes, and although mean duration of breastfeeding was reported this may not have been an appropriate measure in studies where the distribution of scores was unlikely to have been normal (heterogeneity: T² = 2216, P = 0.007, I² =66%) ( Analysis 1.2). Again, it was clear from visual examination of the forest plot that much of the heterogeneity was due to the Sosa 1976a study. Sensitivity analysis excluding this study removed heterogeneity (I
Infant primary outcomes
Infant physiological stability in the hours following birth
Bergman 2004 utilized SCRIP scores (a measure of infant cardio-respiratory stability in preterm infants that evaluates infant heart rate, respiratory rate and oxygen saturation) (Fischer 1998) to compare SSC in healthy late preterm 31 SSC infants (mean gestational age (GA) = 34.2 weeks) with 13 late preterm control infants (mean GA = 35.3 weeks) placed in a servo-controlled incubator next to their mothers. SSC infants had higher SCRIP scores during the first six hours postbirth, indicating better stabilization (MD 2.88, 95% CI 0.53 to 5.23) ( Analysis 1.3). A subset of infants below 1800 grams birthweight also demonstrated better stabilization (MD 4.92, 95% CI -1.67 to 11.51) but this result did not reach statistical significance ( Analysis 1.4).
Blood glucose 75 to 90 minutes following the birth was measured in two studies with 94 infants; blood glucose was higher in SSC infants (MD 10.56 mg/dL, 95% CI 8.40 to 12.72) and this result was statistically significant ( Analysis 1.5).
Infant axillary temperature at 90 minutes to two hours after the birth was reported in three studies including a total of 168 dyads. In the studies by Christensson 1992 and Christensson 1995, infants had SSC or were placed in a 'cot' (bassinet) next to the mother during the first 90 minutes postbirth. Neither group of infants was fed. In Villalon 1993 control infants were taken to the nursery. Due to heterogeneity between studies we did not combine results in meta-analysis but results for the individual studies are set out in Analysis 1.6. In the Christensson 1992 and Christensson 1995 studies, results favored the SCC group (RR 0.40, 95% CI 0.19 to 0.61, and RR 0.50, 95% CI 0.17 to 0.83 respectively) whereas in the study by Villalon 1993 temperatures were on average slightly higher for the control group at this time point (RR -0.10, 95% -0.24 to 0.04) (although at other time points reported results favored the intervention group and therefore, in view of these inconsistencies findings for this study, are difficult to interpret).
Two studies with 57 women reported the number exclusively breastfeeding at hospital discharge; there was no evidence of a difference between groups receiving SSC compared with routine care (RR 0.99, 95% CI 0.66 to 1.47) ( Analysis 1.7).
Three studies with 245 women examined breastfeeding status (using the Index of breastfeeding status (IBS) at one month postpartum. The IBS is a single item indicator and consists of three major levels of breastfeeding exclusivity -- full, partial, and token breastfeeding. Full breastfeeding is divided into two sub-categories exclusive and almost exclusive. In almost exclusive breastfeeding, the infant is given water, juice, vitamins and minerals infrequently in addition to breast-milk. Partial breastfeeding is divided into three sub-categories – high, medium and low. Token breastfeeding is occasional and irregular, less than 15 minutes a day.There was no clear evidence of differences between groups for this outcome, and results varied considerably between studies therefore the overall average treatment effect should be interpreted with caution (mean difference (MD) 0.86, 95% CI -0.73 to 2.44) (heterogeneity: T² = 1.70, P < 0.0001, I² = 90%) ( Analysis 1.8).
More infants were exclusively breastfeeding up to three to six months postbirth in three studies (n = 149) (RR 1.97, 95% CI 1.37 to 2.83) ( Analysis 1.9).
Two studies reported breastfeeding at one year postbirth.There were no statistically significant differences between groups (RR 6.91, 95% CI 0.82 to 46.78) ( Analysis 1.10).
Two studies with 54 women examined breastfeeding effectiveness scores and those in the SCC group had higher mean scores (MD in IBFAT scores 1.79, 95% CI 0.24 to 3.35) ( Analysis 1.11). The Infant Breastfeeding Assessment Tool (IBFAT) evaluates four parameters of infant suckling competence infant state of arousal or readiness to feed; rooting reflex; latch-on; and suckling pattern.The infant can receive a score of 0-3 on each item for a maximum total score of 12 indicating adequate suckling competence (Matthews 1988; Matthews 1991) .
Carfoot 2004, Carfoot 2005 and Khadivzadeh 2008 found that infants held SSC were slightly more likely to breastfeed successfully during their first feeding postbirth than those who were held swaddled in blankets by their mothers, although he evidence of a difference between groups was not statistically significant and there was considerable variability between findings in these three studies (n = 315) (average RR 1.36, 95% CI 0.95 to 1.95; heterogeneity T² = 0.07, P = 0.03, I² = 72%) ( Analysis 1.12). These findings were obtained using a modification of the IBFAT.
In a single study with data for 88 women, Bystrova 2003 reported the number of infants that suckled within two hours of the birth; there was no clear evidence of differences between groups (RR 1.06, 95% CI 0.83 to 1.35) ( Analysis 1.13).
Maternal breast temperature
Bystrova 2003 found significant between group differences in the mean variation in breast temperature 30 to 120 minutes postbirth as measured by the interquartile range between mothers who held their infants SSC and those who were separated from their infants (MD 0.60, 95% CI 0.34 to 0.86) ( Analysis 1.14). Duration of SSC was 95 minutes. The researchers suggested that these variations in maternal breast temperature in the SSC group may regulate infant temperature more effectively than stable breast temperatures and help prevent neonatal hypothermia.
Breast engorgement pain (measured by the self-reported Six Point Breast Engorgement Scale (Hill 1994) or by the mother's perception of tension/hardness in her breasts) was less for SSC than non-SSC mothers on day three postbirth (standardized mean difference (SMD) -0.41, 95% CI -0.76 to -0.06) ( Analysis 1.15) (Bystrova 2003; Shiau 1997).
Infant physiological outcomes
Infant heart rate and respiratory rate
Four studies (Christensson 1992; Mazurek 1999; Nolan 2009; Villalon 1993) obtained data on infant respiratory rate 75 minutes to two hours postbirth. and three studies obtained data on infant heart rate. SSC infants had a lower mean heart rate than control infants who were separated from their mothers although the evidence of a difference between groups did not reach statistical significance and there was high heterogeneity for this outcome (MD -3.05 beats per minute (BPM), 95% CI -7.84 to 1.75; 183 infants); (heterogeneity: T² = 15.26, P = 0.0005, I² 87%) ( Analysis 1.16). Results also favored SCC infants for respiratory rate but again these results did not reach statistical significance and there was considerable variability in findings between studies (MD -3.12 RPM, 95% CI -6.61 to 0.37; 215 infants) (heterogeneity T² = 9.24, P = 0.004, I² = 77%) ( Analysis 1.17). Heterogeneity was mainly due to findings from the Villalon 1993 study; as stated above, findings at different time points varied considerably in this study. We carried out sensitivity analysis where results for this study were excluded; for both heart rate and respiratory rate, removal of findings for Villalon 1993 resulted in statistically significant findings favoring the SCC groups and there was no longer any evidence of between study heterogeneity (heart rate MD -5.77, 95% CI -7.46 to -4.11; respiratory rate MD -4.76, 95% CI -6.12 to 3.41) ( Analysis 1.33; Analysis 1.34).
Bergman 2004 compared the number of infants in the two groups who did not exceed physiological parameters for stability requiring medical attention. The five parameters were infant skin temperature less than 35.5 ºC on two consecutive occasions, heart rate less than 100 or more than 180 BPM on two consecutive occasions, apnea more than 20 seconds, oxygen saturation less than 87% on two consecutive occasions, blood glucose less than 2.6 mmol/L and FIO2 up to 0.6 with continuous positive airways pressure (CPAP) up to 5 cm of water pressure. Fifteen of the 18 SSC and one of the 13 control infants did not exceed parameters (RR 10.83, 95% CI 1.63 to 72.02). The most common reasons for exceeding parameters in control infants were hypothermia, hypoglycemia, and respiratory problems ( Analysis 1.18).
Neonatal intensive care unit (NICU) admissions
There were no significant differences between groups in infant admissions to the NICU in Bergman 2004. Two SSC infants and one control infant required CPAP and they were transferred to the NICU (RR 1.44, 95% CI 0.15 to 14.29) ( Analysis 1.19).Two studies with 42 infants (Chwo 1999; Syfrett 1996) examined hospital length of stay in late preterm infants 34 to 36 weeks' GA and found no significant between group differences and there was high heterogeneity for this outcome (MD -95.30 hours, 95% CI -368.50 to 177.89) ( Analysis 1.22).
Infant body weight change
No statistically significant differences were found in infant body weight change day 14 postbirth; this outcome was reported in two studies with 43 infants (MD -8.00 grams, 95% CI -175.60 to 159.61) ( Analysis 1.20) (Chwo 1999; Moore 2005). Infant weigh change per kilogram per day was not reported in any of the included studies.
Christensson 1995 found that 12 of the 14 SSC infants cried no more than one minute during the 90-minute observation compared with only one of the 15 control infants (RR 12.86, 95% CI 1.91 to 86.44) ( Analysis 1.23). Mazurek 1999 found that SSC infants cried for a shorter length of time during a 75-minute observation period than control infants (MD -8.01 minutes, 95% CI -8.98 to -7.04) ( Analysis 1.24).
Bystrova 2003 used The Parent-Child Early Relational Assessment (PCERA) in a study with data for 61 women.The PCERA (Clark 1985; Clark 1999) has eight sub-scales evaluating maternal and infant behavior and interaction. Bystrova 2003 found no evidence of significant between group differences for maternal positive affective involvement at 12 months postbirth (MD 1.90, 95% CI -1.14 to 4.94) ( Analysis 1.25) however, SSC dyads appeared more mutual and reciprocal (MD 1.30, 95% CI 0.24 to 2.36) than those who were separated immediately postbirth and later reunited for rooming-in ( Analysis 1.26). Bigelow 2010 under Bergman 2004 found significant effects of SSC during the first 24 hours postbirth on maternal sensitivity to her infant's behavioral cues, which demonstrated a dose-response effect. However it was a small subsample (12 of 31 randomized), and the effect was related to the 24-hour dose, with participants randomized to the control condition later receiving SSC, hence excluded from this review.
Mothers who held their infants SSC indicated a strong preference for the same type of postdelivery care in the future (86%) whereas only 30% of mothers who held their infants swaddled indicated that they would most certainly prefer this type of care in the future (RR 2.82, 95% CI 2.08 to 3.82) ( Analysis 1.28) (Carfoot 2005). Mothers who held their infants SSC displayed less state anxiety day three postbirth (MD -5.00, 95% CI -9.00 to -1.00) ( Analysis 1.29). Parenting confidence scores were measured in a single study with data for 20 women; there was no evidence of significant differences between groups (MD 5.60, 95% CI -6.24 to 17.44) (Moore 2005) at one month postbirth between mothers who held their infants SSC or swaddled ( Analysis 1.30).
A large number of additional outcomes were measured in the included studies. Most of these outcomes were measured in single studies The clinical importance of results for many such outcomes is difficult to determine. Outcomes which appeared similar were measured in a range of different ways, in addition, many outcomes were reported at different or multiple time points and results may not have been consistent within or between studies. Non-prespecified outcomes reported include observed mother and infant behavior during the first few hours after birth, outcomes relating to breastfeeding (e.g. duration of first feed and number of breastfeeding problems) and a range of outcomes relating to mother-child interaction.
Summary of main results
The results of this review demonstrated a statistically significant positive effect of skin-to-skin contact (SSC) on the following primary outcomes: breastfeeding one month to four months postbirth, SCRIP score first six hours postbirth, and blood glucose mg/dL at 75 to 90 minutes postbirth, We did not identify significant between group differences in duration of breastfeeding, and results relating to infant axillary temperature at 90 minutes to two hours postbirth were difficult to interpret due to high heterogeneity.
We found a statistically significant and positive effect of SSC on the following secondary outcomes: success of the first breastfeeding (IBFAT score), mean variation in maternal breast temperature 30 to 120 minutes postbirth, infant did not exceed physiological parameters for stability, number of babies not crying for more than one minute during a 90-minute observation, amount of crying in minutes during a 75-minute observation period and PCERA dyadic mutuality and reciprocity 12 months postbirth. We did not identify significant between group differences in successful first breastfeeding (IBFAT score 10 to 12 or BAT score 8 to 12), infant heart rate 75 minutes to two hours postbirth, infant respiratory rate 75 minutes to two hours postbirth, infant body weight change (grams) day 14 postbirth, transfers to the neonatal intensive care unit, infant hospital length of stay in hours, or PCERA maternal positive affective involvement and responsiveness 12 months postbirth.
No negative outcomes associated with SSC were reported in any of the studies except Sosa 1976a, who reported a longer duration of breastfeeding in the control group.
In summation, the totality of significant outcomes relating to breastfeeding, infant physiology and maternal neurobehavior supports the use of SSC in the early period after birth. However, this overall finding should be treated with some caution: for many outcomes only one or two studies contributed data, and for those outcomes where several studies were combined in meta-analysis there was considerable heterogeneity between individual studies. At the same time, some of those results that did not reach statistical significance were derived from small studies which did not have the statistical power to demonstrate differences between groups.
Only two breastfeeding meta-analyses contained more that three studies. Thirteen studies (702 infants) reported breastfeeding rates between one and four months postbirth ( Analysis 1.1) demonstrating that mothers in the SSC group were more likely to be breastfeeding than those in the control group. The only other outcome with more than three studies (seven studies, 324 infants) was breastfeeding duration ( Analysis 1.2). Infants in the SSC group breastfed an average of 42.55 days longer than control infants and when a study with inconsistent results was removed from the analysis the difference between groups was statistically significant. Evidence for breastfeeding exclusivity was conflicting, being no different at hospital discharge ( Analysis 1.7, 2 studies) but significantly greater at three to six months postbirth ( Analysis 1.9, three studies).The findings of improved breastfeeding for the two largest meta-analyses in this review were obtained in diverse countries and among women of low and high socio-economic class.
Results for IBFAT scores for the first breastfeeding postbirth were conflicting with one meta-analysis which treated this outcome as a dichotomous variable ( Analysis 1.12) demonstrating no significant between group differences and another meta-analysis which used interval level data ( Analysis 1.11) finding a significant effect of early SSC. Moore 2005 also found that SSC and the mother's nipple protractility contributed equally to the variance in infant IBFAT scores.The mother's nipple protractility was important in relation to the infant's ability to establish competent suckling. Dewey 2003 reported that suboptimal breastfeeding behavior during the first 24 hours postbirth was associated with the mother's flat or inverted nipples (RR 1.56). These infants were also 2.6 times more likely to have excessive weight loss.
Timing of when this outcome is measured may be critical because most healthy full term infants spontaneously grasp the nipple and begin to suckle by approximately 55 minutes postbirth. During the first 30 minutes, they may only lick the nipple. Widstrom 2011 found that some infants may take up to 45 minutes to latch after crawling towards and reaching the nipple and recommended that this process should not be disturbed or forced. Also, the intervention will be more successful if a clinician reassures the mother that healthy full term babies are able to crawl to the breast and begin to nurse on their own without assistance when they are ready. After the first two hours postbirth, infants often become sleepy and difficult to arouse.
Babies breastfed more successfully during SSC immediately postbirth than if they were held swaddled in blankets, probably because of the extra tactile, odor, and thermal cues provided by SSC, but this result did not translate into significantly more mothers breastfeeding at one to four months postbirth in two studies by the same investigator (Carfoot 2004; Carfoot 2005). Carfoot 2005 stated that barriers to long-term breastfeeding, such as returning to work, and breastfeeding problems contributed to the minimal effect that early SSC had on this outcome. Early SSC appears to have less of an effect on breastfeeding exclusivity or duration in studies where control infants are held swaddled by their mothers or placed swaddled or clothed on their mother's naked chest and given the opportunity to breastfeed soon after birth than in studies where control infants are separated from their mothers for 12 to 24 hours immediately postbirth. Given the strong evidence of the negative impact of early mother-infant separation, it is noteworthy that in some hospitals usual care still includes this practice for healthy full term newborns (Mizuno 2004).
Moore 2005 suggested that barriers to long-term breastfeeding that exist in the United States, especially the customary absence of, or very brief, paid maternity leave, attenuated the effectiveness of early SSC on breastfeeding status day 28 to one month postbirth ( Analysis 1.8). The mothers in Punthmatharith 2001 delivered in a Baby Friendly Hospital in Thailand with 24-hour rooming-in. Control infants were cup fed if they needed supplementation. In addition, most of the SSC took place in extremely warm, un-air conditioned eight-bed postpartum rooms with frequent visitors so that contextual issues, such as body warmth and modesty, may have changed SSC desirability and also effectiveness.
Such factors as room temperature, lack of privacy, modesty, overcrowding, supplemental bottle or pacifier use, and 24-hour rooming-in may play a role in the effectiveness of SSC. Early SSC may not have as strong an effect on long-term breastfeeding in countries with a widespread bottle feeding culture compared to countries with cultures that are supportive of breastfeeding. In the studies by Carfoot 2004, Carfoot 2005 ,and Moore 2005, mothers in the control group received extra assistance with breastfeeding, which is not always available with usual hospital care. In Moore 2005, the investigator was an experienced lactation consultant who assisted mothers in both groups with initiating breastfeeding. In Carfoot 2005, the midwife usually provided breastfeeding assistance, but if she was unavailable, the research assistant often provided help with breastfeeding. More definitive results might have been obtained if the control groups received only usual hospital care.
Infant physiological/behavioral outcomes
The between-group differences in SCRIP scores and maintenance of physiological parameters in late preterm infants is certainly clinically significant, especially given the fact that SSC was compared with a servo-controlled incubator. The clinical significance of some of the other physiological outcomes for healthy full term infants is debatable. Full term infants in the SSC group were less than one degree warmer than control infants.Their heart rate was three BPM slower and their respiratory rate was three breaths less per minute, on average. However, their blood glucose was 10.56 mg/dL higher, a significant finding.The results suggest that early SSC is a safe intervention for healthy infants and that it may increase cardio-respiratory stability, thermal stability, and blood glucose in late preterm infants. Lagercrantz 1986 and Lagercrantz 1996 found that newborn infants experience a catecholamine surge after vaginal birth, caused by compression of the fetal head and intermittent hypoxia during contractions. This response is felt to aid in adaptation to the extrauterine environment immediately postbirth by causing an increase in infant level of alertness, lung compliance, blood glucose, body temperature, and shunting of blood to the vital organs. However, this response may become maladaptive if allowed to continue.These findings correlate accurately with findings predicted from mammalian research on separation in the newborn period. The neurobehavioral stabilization achieved in SSC correlates in mammalian studies with a parasympathetically mediated allostasis, the purpose of which is growth and development. The stabilization achieved in the separated state is mediated by a sympathetically driven defense program, whose purpose is primarily to survive the period of separation. In so far as the differences observed corroborate the findings from mammalian research, they can be considered clinically significant.
The large between-group difference in the amount of crying is certainly clinically significant. Anderson 1989 proposed an evidence-based rationale that maternal-infant separation is associated with excessive infant crying and can be harmful because crying re-establishes portions of the fetal circulation. Each cry cycle causes a bolus of desaturated venous blood to shunt through the foramen ovale into the systemic circulation instead of the lungs, creating hypoxemia. This may result in delayed closure of the foramen ovale or explain the approximately 20% incidence in apparently normal adults of a permanently patent foramen ovale (estimates in numerous recent studies range from 15% to 35% (Del Sette 1998). Anderson 1989 further proposed that crying wastes calories meant for growth, and causes increased and fluctuating cerebral blood flow, cerebral blood flow velocity, and intracranial pressure, thereby increasing the risk of intraventricular hemorrhage in preterm infants. Consequences for healthy full term infants are unknown, but may be similar and correlated with gestational age.
The results of this analysis indicate that SSC may affect maternal attachment behaviors, although the results are mixed. A dose-response relationship may exist as well.
Bystrova 2003 found significant between group differences on two of the eight subscales of the PCERA at 12 months postbirth. The effects of rooming-in on these outcomes did not compensate for a short (120 minutes) period of separation.
These findings would make sense from the perspective of programming (Lucas 2005) and early evolution, where human mothers would be expected to form a rapid attachment to their infants to protect them from predators and to provide the high level of parental care necessary for such physiologically immature newborns. However, it is important to document how many infants in the SSC group breastfed and how effectively they nursed. Breastfeeding during SSC stimulates the secretion of hormones such as oxytocin that promote maternal attachment and prolactin which promotes lactation and, at least in rodents, maternal behavior. Breastfeeding has been considered an integral part of the intervention in Kangaroo Mother Care research in low- and middle-income countries. In this review, breastfeeding has been considered an outcome and SSC the habitat that elicits this outcome. However, mothers would logically nurse their infants soon after birth in early human evolution. Early and effective breastfeeding while in SSC may increase the strength of this intervention with respect to maternal attachment behaviors.
Overall completeness and applicability of evidence
The available evidence does address the review question, but seldom abides by any clear definition of acceptable public health breastfeeding outcomes. Only Anderson 2003; Moore 2005; Punthmatharith 2001;and Shiau 1997 used breastfeeding status (Labbok 1990) to measure the degree of breastfeeding exclusivity. In all the other studies, breastfeeding was considered a dichotomous variable. The infant was either breastfeeding (yes/no) or exclusively breastfeeding (yes/no). Further, the actual intervention in terms of timing and duration of SSC was highly variable, and at times very short. Despite this, the evidence is fairly consistent in supporting the effect of SSC in so far as the findings are numerous and pooled findings were consistently in favor of SSC and show moderate effects. However, for many outcomes findings were from individual studies: the variety of outcomes measured and the lack of consistency in the way outcomes were measured meant that meta-analysis was not appropriate.
The high levels of heterogeneity between studies could possibly reflect bias with selective outcome reporting, with data reported on the basis of post-hoc observations rather than predefined public health outcomes. Another possible source of bias concerns the quality of breastfeeding support provided, and whether this was controlled for adequately between groups. In some instances, co-interventions were added to SSC that make it difficult to disentangle the effects of SSC from the other interventions.
The variability in outcomes reported, instruments used, context, and timing made it difficult to combine many of the attachment outcomes for meta-analysis. Because of these methodological limitations, the overall quality of the evidence is again considered moderate.
Quality of the evidence
The presently available evidence has a number of limitations.
(1) Design limitations
All studies were randomized controlled trials. However, the methodological quality of trials was mixed. Overall, the quality of reporting on study methods was poor. For the majority of trials we did not have sufficient information on the methods used to carry out randomization to allow us to assess whether findings were at high risk of bias. A particular problem in all of the included studies was the lack of blinding. SSC cannot be implemented masked, but the assessment of physiologic changes or outcomes can often be carried out by individuals masked to allocation but overall it is very difficult to judge the impact of lack of blinding or only partial blinding on findings. It is possible that differences in the care women received in SSC and control groups were not confined to whether or not they had early SSC. In some studies the staff providing care to the two groups were different, and in these cases it is possible that the overall care experience for women in different arms of trials was not the same, and it may be that aspects of care other than, or in addition to, SSC led to reported differences between groups. This may have been compounded by the fact that in some studies the same staff delivering interventions also measured outcomes. Outcomes such as observed maternal and infant behavior may have been susceptible to detection bias. The impact of lack of blinding may have been less for some of the outcomes measured, for example, some infant physiological outcomes however, even outcomes such as infant temperature may be affected by bias in staff collecting outcome data.
(2) Outcome variability
Meta-analysis was limited in this review, due to the numerous outcomes and the limited number of randomized trials that could be included for each outcome. Although many of the studies measured broadly similar outcomes, the outcomes were too dissimilar to be included in a meta-analysis. In some studies, means were reported without standard deviations, or exact P values, or both. The context, the instruments used, and the timing of the measurement of attachment and temperature outcomes varied greatly among studies. Breastfeeding was measured as a dichotomous variable in some studies or as an interval level measure of breastfeeding exclusivity in four. Modality for measurement of temperature outcomes varied between studies. These contextual and measurement differences should be noted when considering the results of the review.
Potential biases in the review process
We are aware that the review process may be affected by bias; and we attempted to minimize bias in a number of ways. At least two review authors independently assessed study eligibility, carried out data extraction, and assessed risk of bias. However, some aspects of the review process involve subjective judgements: assessing risk of bias in included studies, for example, is not an exact science, and it is possible that a different review team may have reached different conclusions about the quality of the evidence. We have attempted to explain our decisions regarding study quality in the 'Risk of bias' tables. We have also provided details about the participants and interventions in individual studies and we would encourage readers to interpret results in the light of the information set out in the Characteristics of included studies tables.
Agreements and disagreements with other studies or reviews
The findings are in general agreement with results from other studies mentioned in this review notably Bramson 2010. This large hospital-based study (n = 21,842) demonstrated a clear dose-response effect on exclusive breastfeeding at hospital discharge. The data from this review, although suggestive, are inadequate to demonstrate a dose-response effect. Although the modality and timing of measurement of infant temperature varied between studies, this review found minimal increases in temperature with SSC although the results were often not statistically significant. These results support those obtained Mori 2010 who found a mean increase of 0.22 degrees C. in a meta-analysis of 21 studies of infant temperature pre SSC compared with during the intervention. Mori 2010 found an increase in infant heart rate of 2.04 BPM in a meta-analysis of 12 studies of preterm infants pre versus during SSC. The process of transferring a preterm infant from a radiant warmer or isolette to SSC and back again can be somewhat stressful and may account for these findings. The length of SSC in some of the included studies was very short, 15 minutes in one study, 30 minutes in another. This review found a decrease in infant heart rate of 3.05 in between group comparisons of early SSC with usual care. However, the findings were not clinically significant. We assess the methodological quality of the evidence as moderate because these studies have the same limitations as those with breastfeeding outcomes; small samples, varied contexts and heterogenous outcomes.
Implications for practice
Breastfeeding outcomes: this review does provide evidence to support current practices as recommended by the UNICEF endorsed Baby Friendly Hospital Initiative, in which SSC is encouraged for the first hour after birth. There is, however, inadequate evidence with respect to details such as timing of initiation, dose of skin-to-skin contact (SSC) and technique. This review does not address subsequent ongoing SSC as an intervention to support breastfeeding. It is, however, noteworthy that an intervention practiced for a short time at birth should have measurable breastfeeding effects one to four months postbirth.
Infant outcomes: the significant increase in blood glucose, and maintenance of infant temperature in the neutral thermal range are both clinically important, and lend support to current American Academy of Pediatrics recommendations for the use of SSC in the first hour after birth (American Academy of Pediatrics 2005). Clearly there is a relationship between improved breastfeeding and higher blood glucose. In terms of evolutionary biology, and mammalian studies, this higher value may in fact be the norm, and a lowering may reflect the autonomic nervous system evoking a separation distress response, consuming excess calories (Christensson 1995). This is further supported by the significantly increased crying seen in separation versus SSC (three studies). The decreased crying is in itself clinically important for other reasons as described in the background (Ludington-Hoe 2002). Late-preterm infants are at increased risk for hypoglycemia and hypothermia which can worsen any symptoms of respiratory distress (Raju 2006).
The SCRIP score attempts to provide a composite measure of cardiorespiratory stability. Only one study reported this, with significant benefit in favor of SSC, providing further support for the use of early SSC. While differences in particular cardiorespiratory outcomes are evident, these are open to different interpretations, and mean little without a sense of trend and direction in terms of stabilization and physiological self-regulation.
Although a number of the infant physiological outcomes, (except SCRIP scores, blood glucose, infant crying, and maintenance of physiological parameters), demonstrated little or no clinically significant differences with or without SSC, no negative short- or long-term effects were found. Based on the available evidence, SSC appears to have some clinical benefit, especially for temperature and cardio-respiratory stability in late preterm infants.
Attachment outcomes: despite the variability in dose and timing of the intervention, there is at least a small effect on several dimensions of maternal neurobehavior in relation to her infant. This is consistent with evolutionary biology theory, in which infant survival depended on an immediate care-giving imperative. There is no benefit shown in any study from infants being separated.
The main results of the meta-analysis, and from the single studies, indicate that SSC appears to have a positive effect on breastfeeding one to four months postbirth, blood glucose, infant crying and on infant temperature stability. These benefits of early SSC can be discussed with mothers and their partners during prenatal visits and mothers can be encouraged to incorporate early SSC into their birth plans.The timing of the intervention may be important, because most infants are very alert in the first two hours postbirth and, if undisturbed and unmedicated, will self-attach to the nipple, and do so correctly, at approximately 55 minutes postbirth. However, Widstrom 2011 noted that it may take some infants up to 45 minutes to latch after they reach the mother's nipple.The temperature of a healthy newly delivered infant will remain in a safe range, provided ventral-to-ventral SSC is uninterrupted; the infant is thoroughly dried and covered across the back with a prewarmed blanket; and the head is covered with a dry cap that is replaced if it becomes damp. These practices need to be incorporated into hospital routines along with the stipulation that mothers and newborn infants should not be left alone and unattended by medical personnel in the delivery or recovery room during this transitional period (Dageville 2008).
Implications for research
Interpreting the findings of this review was hampered by the large number of outcomes reported in included studies and inconsistency in the way outcomes were measured. We have reported results for 31 pre-specified outcomes, however, the included studies reported many other outcomes, measured in different ways, reported at different (and sometimes multiple) time points. This lack of consistency was a particular problem for outcomes relating to mother and infant behavior and mother-infant interaction.
Breastfeeding outcomes: suggestions for improvement of clinical trials examining early skin-to-skin contact (SSC) and breastfeeding outcomes include the following. The mother's prenatal breastfeeding intention (how long she planned to nurse her infant) was not controlled in any study except Punthmatharith 2001 and Moore 2005. Only Carfoot 2004; Carfoot 2005; Gouchon 2010; Khadivzadeh 2008 and Moore 2005 evaluated the success of the first breastfeeding in both the SSC and control groups using a published breastfeeding observation instrument, the IBFAT. A valid measure of effective suckling at a single feeding remains elusive (Riordan 1997) and is needed to identify problems in time to minimize breastfeeding attrition; this would be a major contribution to the field. It remains difficult to disentangle the effects of early SSC from the effects of assistance provided by an experienced nurse with the first breastfeeding. The protractility of the mother's nipples is a potential confounding factor that could influence breastfeeding outcomes Dewey 2003 and should be measured in future studies that evaluate breastfeeding and infant suckling patterns. The use of SSC to facilitate breastfeeding for mothers who are having breastfeeding difficulties requires further evaluation. Future research should use standardized public health measures, such as the Index of Breastfeeding Status (IBS) rather than, or in addition to, measuring breastfeeding as a dichotomous variable.
Infant outcomes: rigorous and validated composite measures of physiological benefit are not yet available in the literature. Until better and validated scores or measures for stabilization are presented, we recommend the use of the SCRIP score as a primary outcome measure in studies on early SSC. There is also a clear need for dose-response studies: the intervention of SSC is applied in small doses in many studies. As described in the introduction, in animal studies, the dose of separation is accurately measured and correlated with harmful effects (Poletto 2006; Ziabreva 2003). Further, it is possible that the physiological benefit demonstrated is only as lasting as the actual intervention; therefore, clinical benefit may require prolonged or continuous use of SSC.
Attachment outcomes: improvement is needed in examining maternal attachment behaviors. These studies are all weakened by the lack of consistency in the measurement of these variables. Each research team appeared to have its own ideas about how to operationally define attachment behavior. More recent studies (Anderson 2003; Bystrova 2003) have begun to use more rigorously validated instruments such as the NCAST feeding and teaching scales and the PCERA. No information was provided in the research reports about how many infants successfully breastfed in either group using a validated measurement instrument such as the IBFAT.
Future investigations are recommended because the methodological quality of the included studies is marginally adequate, the characteristics of the SSC and control conditions are diverse, and many outcome measures are difficult to combine. Only four studies (Anderson 2003; Bergman 2004; Chwo 1999; Syfrett 1996) examined the effects of early SSC on late preterm infants who were judged healthy enough to remain on the postpartum ward. The effects of this intervention may be different in this more vulnerable population, and more research is definitely indicated. More research needs to be conducted on the effects of early SSC on mothers who deliver by cesarean birth. To facilitate meta-analysis of the data, future research in this area should involve outcome measures consistent with the best measures used in previous studies or measures developed recently to increase methodological rigor (Anderson 2004b; Labbok 1990). The CONSORT guidelines (Moher 2001; Moher 2010) should be used to document the flow of participants through all clinical trials. Studies should make explicit SSC initiation time, frequency and duration to investigate a possible dose-response relationship.
To improve the methodological quality and reporting in similar clinical trials would be relatively easy. Investigators can provide more details in research reports regarding the method and timing of random assignment, allocation concealment scheme, measures used to control for selection bias, context, timing, and modality of outcome measurements, and means and standard deviations for the interval or ratio level outcome variables examined. However, control for provider and patient performance bias may continue to be problematic, because Institutional Review Boards now require investigators to disclose the purpose of their study to potential participants so they can be informed when they consent to random assignment. Labor and delivery room staff often ask for group assignment of women before delivery so that they will know how to manage the infant immediately postbirth. Outcome assessors should be blinded, however, if at all possible (Polit 2011). Speaking more generally, recommendations by Thomson 1984 provide guidelines for well-controlled clinical trials that remain important to this day.
We thank Dr Busakorn Punthmatharith for her contributions during the earliest phases of the literature review; Dr Mark W Lipsey for his assistance with the categorization of outcome measures for meta-analysis; and Dr Joseph Hepworth for his statistical assistance with the original review. We would also like to thank Anna Fangrath and Lindsay Irish for the English translation of Kastner 2005 and Dr Sheau-Huey Chiu and Danni Li for the translation of Huang 2006. Dr Moore would like to thank Angela Aaron, Ashley Arnold, Candice Bruce, Erin DeBruyn, Monique Deterville, Erin Empting, Kristen McGriff, Jaclyn Miller, Shaunna Parker, Lindsay Piper, Carrie Schrimsher, Lucie Slapnicka, Anna Storvick and Melissa Young, former Women's Health Nurse Practitioner students at Vanderbilt University School of Nursing, for their assistance with reviewing the articles for the updated review, and with completing the Characteristics of included studies tables under the careful supervision and with the guidance of Dr Moore.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Appendix 1. The International Network for Kangaroo Mother Care
The International Network maintains a bibliography of all the research articles published on Kangaroo Mother Care. The bibliography is available from Dr Susan Ludington - Susan.ludington@.case.edu
Appendix 2. Methods used to assess trials included in previous versions of this review
Each study that we identified as a result of the search strategy was evaluated independently for inclusion in the review by two review authors. We rejected trials without a concurrent control group (e.g. those with historical controls). We included relatively high quality quasi-randomized studies in the review. If the assignment to groups appeared to create equivalent groups, then the study was included even if a truly random process was not used for group assignment. For example, if women were alternately assigned to treatment and control groups and there was no reason to think that this should result in nonequivalent groups, that study was included. On the other hand, if assignment to groups was based on woman or provider preference, the study was excluded. Studies conducted by each of the three authors were reviewed for inclusion by the other two authors and a consensus was reached regarding inclusion of these studies in the review. Methods used for generation of the randomization sequence were described for each trial.
Each identified trial was assessed for methodological quality with respect to (1) selection bias, (2) attrition bias, and (3) performance bias. We assigned a quality score for each trial, using the following criteria.
(1) Selection bias (allocation concealment)
(A) Adequate concealment of allocation: centralized randomization, sequentially-numbered, sealed opaque envelopes, computerized minimization technique;
(B) unclear whether adequate concealment of allocation: sealed envelopes but not sequentially numbered or opaque, a trial in which description suggests adequate concealment but other features suspicious, e.g. markedly different treatment and control groups, stated random but unable to obtain further details;
(C) inadequate concealment of allocation: any allocation procedure transparent before assignment, such as open list of random-number tables, use of case record numbers, dates of birth or days of the week.
(2) Attrition bias
We assessed completeness to follow up using the following criteria: complete follow up of all study participants/reasons given for attrition/NSD between participants who terminated their involvement in the study and those who remained enrolled (yes/no/unclear).
(3) Performance bias
We assessed blinding using the following criteria:
(A) blinding of participants (yes/no/unclear);
(B) blinding of caregiver (yes/no/unclear);
(C) blinding of outcome assessment (yes/no/unclear).
We designed a form to extract data. Several review authors extracted data and assessed the methodological quality of each study independently and compared results. Disagreements about study inclusion and methodological quality were resolved through discussion until a consensus was reached. We reviewed the inclusion criteria and therapeutic interventions for each trial to see how they differed between trials. We examined the outcomes in each trial to see how comparable they were between studies. We contacted investigators (if possible) to obtain information about any missing data. For categorical data, we made 2 x 2 tables from each trial for each important outcome, and used odds ratios with 95% CI in the meta-analysis. For continuous variables, we calculated weighted mean differences with 95% CI. We used standardized mean differences to combine trials that used different scales to measure the same outcome. We used fixed-effect meta-analysis for combining data in the absence of significant heterogeneity. We used random-effects meta-analysis for trials with significant heterogeneity identified by using the I² statistic. We were unable to explore heterogeneity using subgroup analysis or sensitivity analysis because there were not enough clinical trials included for the heterogeneous outcomes.
Last assessed as up-to-date: 15 March 2012.
Protocol first published: Issue 1, 2002
Review first published: Issue 2, 2003
Contributions of authors
For this update, Dr Elizabeth Moore wrote the first draft of the review and revised subsequent drafts in response to extensive feedback. Dr Gene Anderson and Dr Nils Bergman commented on the first draft of the updated review and contributed to the writing of the final draft. Therese Dowswell contributed to study assessment, analysis and drafting text.
Declarations of interest
Three of the review authors have been active trialists in this area and have personal contact with many groups in this field, including the International Network for Kangaroo Mother Care based in Trieste, Italy; Bogota, Colombia; and Cleveland, Ohio. Dr Bergman has received reimbursement for lectures that he has conducted on Kangaroo Mother Care and from the sale of KMC related products.
Sources of support
- None, Not specified.
- Therese Dowswell is supported by the NIHR NHS Cochrane Collaboration Programme grant scheme award for NHS-prioritised centrally-managed, pregnancy and childbirth systematic reviews: CPGS 10/4001/02, UK.
Differences between protocol and review
The protocol has been updated. We have modified outcomes and used updated methods. Quasi- randomized controlled trials are no longer part of the inclusion criteria so Anisfeld 1983 has now been excluded.
Medical Subject Headings (MeSH)
MeSH check words
Female; Humans; Infant; Infant, Newborn
* Indicates the major publication for the study