Skin-to-skin contact (SSC) has many benefits for the physical and emotional wellbeing of both newborns and their mothers (Boundy et al 2016; Moore et al, 2016;). Although SSC is a well-established practice following vaginal birth, it is not widely implemented during a caesarean section (CS) (Phillips, 2013). Among the reasons for not establishing perioperative SSC is the increased risk of mothers and newborns becoming hypothermic during a CS, which may act as a barrier to perioperative SSC as a result of maternal–newborn separation for warming of one or both parties (Mangan and Mosher, 2012).
Hypothermia during a CS may impact on the health of mothers and their newborns if no actions are taken to prevent their temperature reduction (Mank et al, 2016). Maternal hypothermia after a CS could cause shivering, delayed wound healing and increased risk of wound infection and haemorrhage (Melling et al, 2001), while neonatal hypothermia can lead to respiratory embarrassment, apnoea, hypoxaemia, carbon dioxide retention, metabolic acidosis, hypoglycaemia and decreased oxygen delivery to the tissues (Bajwa, 2014). Given its potential adverse effects, hypothermia should be prevented and diagnosed and managed promptly.
A possible solution to this issue is to actively warm mothers during a CS (Sultan et al, 2015). Active warming during a CS involves increasing a woman's temperature with the use of warming devices, to prevent and manage hypothermia. The rationale behind this is that providing perioperative active warming to mothers will maintain their temperature within normal levels and, when placing their naked newborn on their bare chest for SSC, newborns will be less likely to experience a reduction in their temperature via conduction; therefore, the frequency of neonatal hypothermia could be reduced.
This systematic review investigates the randomised controlled trial (RCT) evidence on the effect of perioperative maternal active warming versus no warming on the frequency of neonatal hypothermia for newborns who have at-birth SSC during a CS. This systematic review offers an insight as to the evidence that exists on this specific topic: if maternal active warming during a CS is effective in preventing neonatal hypothermia during SSC, if a particular perioperative maternal active warming method is most effective, compared to others, and if active warming is safe for both mothers and newborns. The conduct of the review was guided by the Cochrane Handbook on Systemic Reviews (Higgins et al, 2011).
Methods
Selection criteria
RCTs published in any language that included pregnant women ≥18 years old who were undergoing elective CS at term (from 37 completed weeks of pregnancy) under regional anaesthesia, and who initiated in birth SSC were included in this systematic review. The included studies evaluated active warming (eg forced air warming devices, warmed IV fluids etc) versus no active warming (eg room temperature IV fluids etc) interventions.
The primary outcome measures included neonatal and maternal hypothermia, while the secondary outcome measures included neonatal and maternal core temperatures, room temperature (before, during and after SSC, measured by any method, device or location), duration of SSC, including total time of SSC from the start to end time, umbilical cord blood analysis (umbilical and arterial pH and base excess, as defined by Freeman et al (2012)), the Apgar scores at 1 and 5 minutes, maternal satisfaction (as measured by trial authors), the timing of initiation of first feed, type and length of feed, neonatal jitteriness and neonatal blood sugar level 2 hours after birth.
Search strategy and selection of included studies
Databases searched from inception to June 2020 included EMBASE, Scopus, Web of Science, CINAHL, PubMed and Maternity and Infant Care. Two authors (AV and MMc) independently screened citations by titles, abstracts and full texts for trial eligibility. Any disagreements were resolved by discussion or involvement of a third assessor (MB).
Assessment of risk of bias, quality assessment of the evidence
Three reviewers (AV, MMc and LN) independently assessed risk of bias of included studies using the Cochrane Risk of Bias assessment tool (Higgins et al, 2011). The tool evaluates seven criteria considered to influence bias in trials: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other sources of bias. Judgements are made (ie high, low or unclear risk of bias) for each of the seven criteria, followed by an overall risk of bias judgement for each included study (Appendix 1).
The quality of the evidence was individually assessed by two reviewers (AVR and MMC) using the GRADE approach, as outlined in the GRADE handbook. The GRADEpro software (Evidence Prime, 2015) was used to import data from Review Manager 5.3 (RevMan, 2014) to assess the quality of evidence arising from the included studies that contributed to the meta-analysis of the specified outcomes (Appendix 2). This assessment was based on five criteria: assesment tool, inconsistency, indirectness, imprecision and publication bias, in order to assess the quality of the body of evidence for each outcome. The evidence could be downgraded from ‘high quality’ by one level for serious (or by two levels for very serious) limitations, depending on assessments of these five criteria. Any disagreements were resolved by discussing with MB.
Data analysis
Data were entered into RevMan software (2014) and analysed using a random effects model. The fixed effects model was also used to analyse the data, to ensure robustness of the model chosen and susceptibility to outliers. For continuous outcomes, for example core temperature, standard mean difference were used for studies using different outcome measurement scales. Dichotomous outcomes, for example maternal hypothermia results, are expressed as risk ratios with 95% confidence intervals. Where it was not feasible to undertake a meta-analysis of review outcome, a narrative analysis was provided.
Heterogeneity across included studies was analysed using the I2 test and Chi2 test, as recommended by Higgins (2011). I2 values were classed as follows: 0–40%=might not be important, 30–60%=moderate heterogeneity and 50–90%=substantial heterogeneity. Sensitivity analyses were planned to explore the effect of trial quality on the results of the review. All of the included studies had unclear and/or high risk of bias; therefore, no sensitivity analysis was conducted. A subgroup analysis was undertaken, comparing the only active warming method common to two out of the three included studies, which was warm IV fluids versus no active warming.
Results
The search of the predetermined databases (Appendix 3) revealed a total of 19 905 records, out of which 18 040 were duplicates. A total of 1 842 were excluded at title and abstract-screening level, as they were clearly not relevant to the review. Of the 23 papers that were assessed at full text level, three RCTs (n=286) met the inclusion criteria (Appendix 4). The excluded studies did not perform SSC and/or did not measure neonatal outcomes.
Primary outcomes
The primary outcome data are displayed in Figure 1.
Figure 1. Active warming versus no active warming on frequency of hypothermia
Neonatal hypothermia
Neonatal hypothermia in the operation theatre was reported by Horn et al (2014) and Vilinsky et al (2014). The occurrence of neonatal hypothermia was significantly lower in neonates of women receiving active warming as compared to those receiving no active warming (two studies, 60 participants, relative risk=0.14; 95% confidence interval=0.03–0.75). I2 was 28%, indicating not important heterogeneity in this analysis.
Maternal hypothermia
Maternal hypothermia in the operation theatre was reported in all three studies after the initiation of SSC. The occurrence of maternal hypothermia did not differ between mothers receiving active warming and those receiving no active warming (286 participants, relative risk=0.35; 95% confidence interval=0.11–1.17). I2 was 56% indicating moderate heterogeneity in this analysis.
Maternal hypothermia in the post-anaesthesia care unit
Maternal hypothermia in the post-anaesthesia care unit (PACU) was reported by Paris et al (2014) and Vilinsky et al (2016), while SSC was still being performed. The occurrence of maternal hypothermia was not significantly different between mothers receiving active warming and those receiving no active warming (two studies, 246 participants, relative risk=0.47; 95% confidence interval=0.09–2.40). I2 is 64% indicating moderate heterogeneity in this analysis.
Secondary outcomes
Neonatal core temperature before skin-to-skin contact
It was not feasible to undertake a meta-analysis of this outcome becuase of the lack of data across the three studies. Specifically, Horn et al (2014) did not report the mean and standard deviation values for each group. Vilinsky et al (2016) did not measure temperatures at that time, and Paris et al (2014) measured the neonatal birth temperature once only but it was unclear as to when this was conducted in relation to SSC.
Neonatal core temperature during skin-to-skin contact
Neonatal core temperatures were measured by Horn et al (2014) and Vilinsky et al (2016) in the first 5 minutes of SSC. However, the data (mean, standard deviation) documented by Horn et al (2014) were not identifiable in the article. Vilinsky et al (2016) found that the mean neonatal core temperatures was 36.8°C for the intervention group and 36.55°C for the control group, however, no standard deviations were documented; therefore, no meta-analysis was conducted for this outcome.
Neonatal core temperature at the end of skin-to-skin contact
At the end of SSC, neonatal core temperatures were measured by Horn et al (2014) and Vilinsky et al (2016). There was no significant difference in the neonatal temperatures at the end of SSC between those receiving active warming and no active warming (60 participants, standardised mean difference=1.03; 95% confidence interval=-1.58–3.64). I2 is 95%, indicating considerable heterogeneity in this analysis (Figure 2).
Figure 2. Active warming versus no active warming on core temperatures
Maternal core temperature before skin-to-skin contact
Maternal core temperature before SSC was measured in all three studies and meta-analysis found no difference in maternal temperature measurements between active warming and no active warming participants (286 participants, standardised mean difference=-0.01; 95% confidence interval=-0.70–0.69). I2 was 77%, indicating substantial statistical heterogeneity in this analysis (Figure 2).
Maternal core temperature during skin-to-skin contact
All three studies reported maternal core temperature during SSC. The analysis found a significant difference in maternal temperature during SSC in those receiving active warming versus no active warming (286 participants, standardised mean difference=0.75, 95% confidence interval=0.50–1.01). I2 was 0%, indicating no statistical heterogeneity (Figure 2).
Maternal core temperature after skin-to-skin contact
Pooling of data from the three studies showed no difference between active warming and no active warming in maternal temperatures at the end of SSC (286 participants, standardised mean difference=0.64; 95% confidence interval=-0.28–1.56). I2 was 85%, indicating substantial statistical heterogeneity (Figure 2).
Room temperature before, during and after skin-to-skin contact
The ambient operating room temperature was reported in all three studies; however, none of the studies reported with precision the time that the temperature was measured. Therefore, pooling of data was not appropriate.
Duration of skin-to-skin contact including total time
Horn et al (2014) reported that each newborn received a total of 20 minutes SSC in both groups, without reporting any standard deviation. Vilinsky et al (2016) reported an average of 81.3 minutes of SSC in the intervention group and an average of 82 minutes of SSC for the newborns in the control group, without reporting any standard deviation. In correspondence with Paris et al (2014), the lead author mentioned that SSC was performed in all participants, but the duration of SSC was not documented in the article or the correspondence. As a result of missing data, no meta-analysis was performed.
Other secondary outcomes
Umbilical cord blood analysis was only measured by Paris et al (2014); findings suggest that arterial pH was 7.32 (standard deviation=0.8) for the control group, 7.33 (standard deviation=0.7) for the warm IV fluid group and 7.31 (standard deviation=0.6) for the warmed underbody pad group. Similarly, the venous pH was 7.37 (standard deviation=0.6) for the control group, 7.39 (standard deviation=0.5) for the warm IV fluid group and 7.37 (standard deviation=0.6) for the warmed underbody pad group. Apgar scores were reported by Horn et al (2014) and Paris et al (2014). Horn et al (2014) reported a mean Apgar score of 9 in the first minutes and 10 in the fifth minute.
In contrast, Paris et al (2014) reported the low first minute and fifth minute Apgar scores. The low first minute Apgar score was 8 (standard deviation=10.5), 2 (standard deviation=2.7) and 4 (standard deviation=5.2) for the control, warm IV fluid and warmed underbody pad groups, respectively. The low 5-minute Apgar score was 1 (standard deviation=1.3) for the control group, 0 for the warm IV fluid group and 2 (standard deviation=2.6) for the warmed underbody pad group. As a result of the lack of data, no meta-analysis was performed on these outcomes.
Non-reported secondary outcomes
The following outcomes were not reported: maternal satisfaction, type/timing and duration of newborn feeding, neonatal jitteriness and neonatal blood sugar level 2 hours after birth.
Subgroup analysis
A subgroup analysis was performed (Figure 3), evaluating studies that used warm IV fluids as an intervention versus no active warming (Paris et al 2014; Vilinsky et al 2016). A meta-analysis found a significant difference in maternal hypothermia in the operating theatre between warm IV fluids and no active warming participants (169 participants, relative risk=0.57, 95% confidence interval=0.41–0.79). I2 was 0%, meaning there was no statistical heterogeneity. However, no difference was found in maternal hypothermia in the PACU between participants receiving warm IV fluids and those receiving no active warming (169 participants, standardised mean difference=0.47, 95% confidence interval=-0.08–2.77) (Figure 3). I2 was 52%, meaning there was a moderate statistical heterogeneity.
Figure 3. Subgroup analysis maternal hypothermia (warm IV fluids only)
There was a significant difference in the maternal core temperature before SSC (169 participants, standardised mean difference=0.40, 95% confidence interbal=0.10–0.71; I2=0%) and dur ing SSC (169 participants, standardised mean difference=0.80, 95% confidence interval=0.04–1.57; I2=58%) in participants receiving warm IV fluids when compared to those receiving no active warming (Figure 4). Maternal core temperature after SSC did not differ between warm IV fluids and no active warming participants (169 participants, standardised mean difference=0.57, 95% confidence interval=-0.94–2.43) (Figure 4). I2 was 88%, which represents substantial statistical heterogeneity.
Figure 4. Subgroup analysis on maternal core temperatures (warm IV fluids only)
Quality of evidence and potential biases in the review process
This review was based on a predetermined registered protocol (Prospero registration number: CRD42016039003) and a search strategy that had no language restrictions. The quality assessment of the included studies was robust and based on the Cochrane guidelines and standards. Quality of evidence was determined using the GRADE approach. To minimise the risk of bias in the review, two reviewers conducted each stage of the review independently, and a third reviewer reviewed the systematic review during each stage to ensure the study's rigor.
The review findings were limited by the variation of different warming interventions, the instruments used for measuring maternal and neonatal temperatures and the body locations used for measurements of the participant's core temperatures. It was not possible to explore a number of the secondary outcomes (eg maternal satisfaction, the type, timing and duration of newborn feeding, neonatal jitteriness and neonatal blood sugar level 2 hours after birth), as they were not the focus of the included studies. As a result of the high risk of bias (Appendix 1), the small sample size of the studies and the significant clinical and statistical heterogeneity, the quality of data is considered by the authors to be very low.
Overall, the grade of evidence suggests a very low certainty with regard to the review findings, meaning the authors have very little confidence in the effect estimate (Appendix 2). The true effect is likely to be substantially different from the estimate of effect and so the findings of the studies should be interpreted with caution. Overall, evidence from the included studies is of very low quality and insufficient to guide clinicians and nursing/midwifery staff working in operating theatre departments.
Discussion
This is the first systematic review of neonatal outcomes of RCTs comparing perioperative active warming versus no active warming methods in mothers performing at birth SSC during/after CS.
The findings suggest that maternal active warming during elective CS is more effective when compared with no active maternal warming for reducing the frequency of neonatal hypothermia during SSC. The same effect was seen in relation to maintaining maternal core temperature within normal levels during SSC. However, these effects were not evident after SSC. A possible explanation for this is the methodological differences between the three studies. For example, maternal and neonatal temperatures were recorded at different time frames. In all three sudies, it is unclear if the maternal temperature measurements at the end of SSC were performed after the discontinuation of SSC or while SSC was still performed. If the maternal temperature measurements at the end of SSC were performed after the discontinuation of SSC, this may suggest that the measurements were not actually linked to SSC but rather the location of the mother. Additionally, maternal active warming was performed and discontinued at different time frames in the three included studies.
Key issues arising from this review include insufficient quality of studies, small number of relevant studies, variation in methodological design, and insufficient evidence on the longitudinal measurement of the intervention effect on both maternal and neonatal temperatures. Consequently, there is insufficient evidence to permit conclusions about the effects of maternal active warming on the frequency of neonatal hypothermia during/after a CS, while at birth SSC is performed.
In the current literature, only two systematic reviews were found that explored the effect of maternal active warming during CS. Munday et al (2014) conducted a systematic review of 12 RCTs, and investigated the effectiveness of warming interventions for women undergoing SSC during CS. Overall, this SR suggested that maternal administration of warm IV fluids during CS was effective at maintaining maternal temperatures within normal levels and preventing maternal shivering. Alternative warming devices, such as forced air warming and under-body carbon polymer mattresses, were also found to be effective at preventing maternal hypothermia. Additionally, this review found that preoperative warming devices improved neonatal temperatures at birth, but warm IV fluids during CS did not improve neonatal temperatures, and the effectiveness of any warming intervention on umbilical pH remains unclear. However, this study reviewed the findings of neonatal temperatures only once (immediately after birth), while no SSC was performed and no additional neonatal temperature measurements were reviewed during the newborn stay in the operating theatre and the PACU. Additionally, the only other neonatal outcomes reviewed were Apgar Scores and umbilical pH.
A meta-analysis by Sultan et al (2015) of 13 RCTs (789 participants) on the effect of maternal warming during CS (not focusing on SSC) on maternal and neonatal outcomes concluded that maternal active warming reduced the frequency of maternal hypothermia, increased the maternal end of surgery temperatures, reduced maternal shivering, improved maternal thermal comfort and increased the umbilical artery pH. This review found no difference in neonatal temperature at birth or in Apgar Scores between the intervention and comparator groups. No additional neonatal outcomes were included in this meta-analysis, including the impact of SSC on maternal and neonatal outcomes.
Although the current literature suggests that SSC can reduce the occurrence of neonatal temperature drop and maintain neonatal normothermia after vaginal birth (Yokoyama et al, 2009; Keshavarz et al 2010), this is not the case for newborns having SSC after a CS (Cobb et al, 2016). The present systematic review suggests that women performing at birth SSC during a CS were prone to dropping their core temperature during their operation unless active warming was applied. This may have a direct effect on the occurrence of neonatal hypothermia in the operating theatre for the group of women who did not receive an active warming intervention.
As a result of the small number of studies investigating the frequency of neonatal hypothermia during CS while at birth SSC is performed, the authors suggest that additional research is required for definitive confirmation of the true effects of perioperative maternal active warming (warmed IV fluids) on neonatal temperatures.
Conclusions
This systematic review and meta-analysis of three trials identified a potential relationship between perioperative active warming of women undergoing CS and a reduction of neonatal hypothermia and maintenance of maternal core temperature within normal levels, during SSC. No difference was found between maternal hypothermia in the operating theatre or the PACU, maternal core temperatures before and after SSC, or neonatal temperatures at the end of SSC. Future, high-quality RCTs are required to add to the evidence of the effects of various perioperative maternal active warming methods on neonatal outcomes.
Key points
- This systematic review focuses on the available evidence for skin-to-skin contact after a caesarean section, on the effects of perioperative warming of women on maternal and neonatal outcomes.
- Active warming of women resulted in significantly less occurrence of neonatal hypothermia during skin-to-skin contact.
- There is no difference in maternal hypothermia at the end of the operation.
CPD reflective questions
- What alternative active warming methods can be safely used during the emergency lower segment caesarean section to promote skin-to-skin contact?
- Which active warming methods are most cost effective?
- What other obstacles are evident in everyday practice that could prevent the promotion of perioperative skin-to-skin contact?
- What are the theatre staff perceptions on using perioperative active warming methods in order to promote perioperative skin-to-skin contact?