Supplemental oxygen for caesarean section during regional anaesthesia

Abstract

Background

Supplementary oxygen is routinely administered to low‐risk pregnant women during an elective caesarean section under regional anaesthesia; however, maternal and foetal outcomes have not been well established. This is an update of a review first published in 2013.

Objectives

The primary objective was to determine whether supplementary oxygen given to low‐risk term pregnant women undergoing elective caesarean section under regional anaesthesia can prevent maternal and neonatal desaturation. The secondary objective was to compare the mean values of maternal and neonatal blood gas levels between mothers who received supplementary oxygen and those who did not (control group).

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL; 2014, issue 11), MEDLINE (1948 to November 2014) and EMBASE (1980 to November 2014). The original search was first performed in February 2012. We reran the search in CENTRAL, MEDLINE, EMBASE in February 2016. One potential new study of interest was added to the list of ‘Studies awaiting Classification' and will be incorporated into the formal review findings during the next review update.

Selection criteria

We included randomized controlled trials (RCTs) of low‐risk pregnant women undergoing an elective caesarean section under regional anaesthesia and compared outcomes with, and without, oxygen supplementation.

Data collection and analysis

Two review authors independently extracted data, assessed methodological quality and performed subgroup and sensitivity analyses.

Main results

We found one new included study in this updated version. In total, our updated review includes 11 trials (with 753 participants). The low quality of evidence showed no significant differences in average Apgar scores at one minute (N = six trials, 519 participants; 95% confidence (CI) ‐0.16 to 0.31, P = 0.53) and at five minutes (N = six trials, 519 participants; 95% CI ‐0.06 to 0.06, P = 0.98). None of the 11 trials reported maternal desaturation. The very low quality of evidence showed that in comparison to room air, women in labour receiving supplementary oxygen had higher maternal oxygen saturation (N = three trials, 209 participants), maternal PaO₂ (oxygen pressure in the blood; N = six trials, 241 participants), UaPO₂ (foetal umbilical arterial blood; N = eight trials, 504 participants; 95% CI 1.8 to 4.9, P < 0.0001) and UvPO₂ (foetal umbilical venous blood; N = 10 trials, 683 participants). There was high heterogeneity among these outcomes. A subgroup analysis showed no significant difference in UaPO₂ between the two intervention groups in low‐risk studies, whereas the high‐risk studies showed a benefit for the neonatal oxygen group.

Authors' conclusions

Overall, we found no convincing evidence that giving supplementary oxygen to healthy term pregnant women during elective caesarean section under regional anaesthesia is either beneficial or harmful for either the mother or the foetus' short‐term clinical outcome as assessed by Apgar scores. Although, there were significant higher maternal and neonatal blood gas values and markers of free radicals when extra oxygen was given, the results should be interpreted with caution due to the low grade quality of the evidence.

Plain language summary

Comparing supplemental oxygen with room air for low‐risk pregnant women undergoing an elective caesarean section under regional anaesthesia

Review question

We reviewed the evidence about the effect of giving extra oxygen to pregnant women during planned caesarean section under epidural or spinal anaesthesia. (Epidural anaesthesia is when a drug is injected into the epidural space of the spinal cord; spinal anaesthesia is when a local anaesthetic is injected into the subarachnoid space).

Background

Oxygen was routinely given to pregnant women to provide an extra supply for the foetus in order for the foetus to cope with any unplanned loss of oxygen during or after the birth. Previous studies have shown that giving extra oxygen provided better oxygenation for the mother and foetus in terms of oxygen saturation (a measure of how much oxygen the blood is carrying), PaO₂ (oxygen pressure in the blood) and pH (a measure of acidity or alkalinity). However, clear evidence of foetal clinical outcomes have not been obtained.

Search date

The evidence is current to November 2014.We reran the search in CENTRAL, MEDLINE, EMBASE in February 2016. We found one potential new study of interest which was added to the list of ‘Studies awaiting Classification'. This study will be incorporated into the formal review findings when we next update the review.

Study characteristics

This updated Cochrane review included 11 studies involving 753 participants. The studies compared maternal (mother) and neonatal (foetal) outcomes when pregnant women received extra oxygen versus room air. Oxygen was given to the women in different ways (at any flow rate or concentration via any oxygen delivery device).

Results

Overall, the results of this updated review reach the same conclusions as the original published review. None of the 11 included trials reported maternal desaturation. No differences were noted in routine measures of foetal wellbeing (Apgar scores) when mothers who received extra oxygen were compared with those who did not. The pregnant women receiving extra oxygen in comparison with room air had significantly higher oxygen saturation (three trials) and partial pressure of oxygen in arterial blood (five trials), as well as a significantly higher partial pressure of oxygen in both the umbilical artery and the umbilical vein (eight and 11 trials, respectively). Two trials reported higher markers of free radicals (perhaps indicating stress from excess oxygen) in mothers and foetuses when extra oxygen was given, but this is of no clinical significance Overall, we found no convincing evidence that giving oxygen in this situation is either beneficial or harmful for either the mother or the foetus.

Quality of evidence

None of the 11 studies focused on maternal changes in oxygen saturation (defined as maternal saturation less than 90%). We graded the quality of evidence as low for the primary outcome (Apgar scores), and very low for the secondary outcomes (maternal oxygen saturation; partial pressure of oxygen in arterial blood, the umbilical artery and the umbilical vein). The reasons for our grading were risk of bias and inconsistency of the results.

Summary of findings

for the main comparison.

Oxygen supplement compared with room air for low‐risk term pregnant women undergoing elective caesarean section under regional anaesthesia
Patient or population: low‐risk term pregnant women undergoing elective caesarean section under regional anaesthesia Settings: operating room Intervention: oxygen supplement Comparison: room air
Outcomes	Illustrative comparative risks* (95% CI)		Effect estimates (mean difference (95% CI))	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Room air (weighted mean ± SD)	Oxygen supplement (weighted mean ± SD)
Apgar scores at 1 minute	8.8 ± 0.8	8.8 ± 0.9	Apgar scores at 1 minute in the intervention group were 0.07 point higher (0.16 lower to 0.31 points higher)	519 (6 studies)	++00 low2
Apgar scores at 5 minutes	9.7 ± 0.4	9.7 ± 0.4	Apgar scores at 5 minutes in the intervention group were 0.00 point higher (0.06 lower to 0.06 points higher)	519 (6 studies)	++00 low3
Proportion of maternal participants with desaturation1	None of the 11 studies focused on maternal changes in oxygen saturation
Mean maternal arterial oxygen tension (mm Hg) (PaO2)	‐	‐	We did not pool the data4	241 (6 studies)	+000 very low5
Mean maternal oxygen saturation (%)	‐	‐	We did not pool the data6	209 (3 studies)	+000 very low7
Mean oxygen tension (mm Hg) in umbilical artery (UaPO2)	16.0 ± 6.9	18.8 ± 6.7	Mean oxygen tension in umbilical artery in the intervention group was 3.3 mm Hg higher (1.8 to 4.9 mm Hg higher)	504 (8 studies)	+000 very low8
Mean oxygen tension (mm Hg) in umbilical vein (UvPO2)	‐	‐	We did not pool the data9	683 (10 studies)	+000 very low10
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval;
GRADE Working Group grades of evidence: High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

¹ We defined maternal participants with desaturation as maternal SpO₂ less than 90%.

² We downgraded the result from high to low quality due to inconsistency of the results (high heterogeneity; I² statistic = 79%) and four out of six trials having high risk of bias.

³ We downgraded the result from high to low quality due to 50% of the studies having high risk of bias.

⁴ There was high heterogeneity (I² statistic = 93%).

⁵ We downgraded the result from high to very low quality due to unexplained high heterogeneity and five out of six trials having high risk of bias.

⁶ There was high heterogeneity (I² statistic = 88%).

⁷We downgraded the result from high to very low quality due to unexplained high heterogeneity and all three trials being at high risk of bias.

⁸ We downgraded the result from high to very low quality due to substantial heterogeneity (I² statistic = 72%) and seven out of eight trials having high risk of bias.

⁹There was high heterogeneity (I² statistic = 88%).

¹⁰We downgraded the result from high to very low quality due to unexplained high heterogeneity (I² statistic = 88%) and eight out of 10 trials having high risk of bias.

Background

Description of the condition

During caesarean section, the primary responsibility of anaesthetists and obstetricians is the safety and wellbeing of both mother and baby. Spinal anaesthesia has been recommended for a caesarean section (Riley 1995), but may lead to impaired maternal respiratory function and hypotension, resulting in impaired maternal‐foetal gas exchange (Kelly 1996). Other factors, such as unanticipated prolonged uterine incision‐to‐delivery (UD) intervals may also compromise placental circulation and cause a reduction in foetal oxygen content (Jordan 2002). However, the optimal intrauterine foetal arterial oxygen content before delivery during a routine caesarean section is not known.

Description of the intervention

Many pregnant women are empirically given supplemental oxygen intraoperatively (Stone 2002), at least until the delivery of the baby. This is done theoretically to provide hyperoxygenation to the foetus to enable it to better tolerate any unforeseen intranatal and postnatal oxygen deprivation.

Several factors may affect the transfer of oxygen from mother to baby (Parer 2002). Foetal haemoglobin has a higher oxygen affinity than that of the mother to promote extraction of oxygen from maternal oxygenated haemoglobin across the placenta. The normal range for neonatal haemoglobin is high compared with that for maternal blood (approximately 14 to 24 g/dL). An increase in oxygen delivered to the mother may not give rise to a clinically relevant increase in the foetal blood oxygen content (Bassel 1995). Blood carries oxygen to the foetus largely through oxygen binding with haemoglobin. If the saturation of haemoglobin (SaO₂) with oxygen in the blood is greater than 97%, increasing oxygen tension (PaO₂) to increase the amount of oxygen dissolved in the blood will make only a minimal difference (Kelly 1996). Healthy women about to give birth frequently have SaO₂ levels greater than 96% while breathing air (Blackburn 2003). Therefore, it might be expected that oxygen administration in routine practice would be unlikely to improve delivery of oxygen to the baby (Bassel 1995).

In many units, supplemental oxygen is commonly administered to women through a face mask (Cogliano 2002; Crosby 1992). However, some women complain that when a face mask is applied, they feel claustrophobic, have difficulty getting enough air and are unable to communicate. Cogliano 2002 found that nasal cannulae were more acceptable to women. Oxygen is typically delivered at two to five litres per minute, resulting in an inspired oxygen concentration of approximately 30% to 40% through the nasal cannula (Shapiro 1991).

The benefit of administering supplemental oxygen to healthy, low‐risk pregnant women during an elective caesarean section under regional anaesthesia is controversial (Backe 2002). The partial pressure of oxygen in foetal umbilical arterial (UaPO₂) and umbilical venous (UvPO₂) blood increases when maternal‐inspired oxygen levels are increased (FIO₂). Foetal oxygen partial pressure approaches a plateau when maternal arterial oxygen partial pressure (PaO₂) is increased beyond 300 mm Hg during a caesarean section performed with the patient under general anaesthesia (Baraka 1970); however, Ramanathan 1982 reported that UvPO₂ was improved when the mother's fraction of inspired oxygen (FIO₂) was between 0.47 and 1 during caesarean section performed with the mother under epidural anaesthesia.

In contrast, Kelly 1996 found that administration of 35% oxygen did not alter maternal PaO₂ nor foetal UvPO₂ during a caesarean section under spinal anaesthesia. Although 60% supplemental oxygen modestly increases maternal PaO₂ and foetal UvPO₂, Apgar scores were similar in neonates born to hyperoxic (the condition of having higher than normal level of oxygen) or normoxic (the condition of having a normal level of oxygen) mothers (Adenekan 2010; Khaw 2002). However, Apgar scores are not a specific marker for asphyxia and can demonstrate only moderate to severe hypoxic brain injury (Hogan 2007). Nevertheless, one‐minute Apgar scores greater than seven generally predict good neonatal outcomes. Markers of free radical activity may represent another indicator for unfavourable neonatal outcomes. Khaw 2002 demonstrated that increased free radical activity in both mother and baby was associated with exposure times as short as 10 minutes to FIO₂ of 0.6.

How the intervention might work

Administration of oxygen to the mother during caesarean section may increase the oxygen reserve of the foetus, resulting in protection from the adverse consequences of a prolonged interval from the interruption of oxygen supply from the umbilical artery to the establishment of effective ventilation in the newborn baby.

A prolonged uterine incision‐to‐delivery (UD) interval greater than 180 seconds has been associated with worsening neonatal acid‐base status and lower Apgar scores (Datta 1981). More recent data show that with prolonged UD intervals, neonatal acid‐base status, oxygenation and Apgar scores were no worse in cases where mothers were not given supplemental oxygen; specifically, administration of 60% oxygen to women about to give birth with prolonged UD intervals did not increase umbilical venous oxygen content (Khaw 2004).

Why it is important to do this review

Although oxygen supplementation for otherwise healthy women undergoing uncomplicated caesarean section is a simple and apparently safe procedure, in some settings medical oxygen may be in short supply. Under these circumstances, the unnecessary use of oxygen for some patients may jeopardize the health outcomes of other patients if oxygen supplies are inadequate to meet their needs. In addition, oxygen is a potentially toxic substance and could conceivably present a risk to both foetus and neonate.

For these reasons, it is important to establish whether health benefits may be derived from the use of oxygen for this purpose. This review evaluated the effect of supplementary oxygen given to low‐risk term pregnant women undergoing elective caesarean section under regional anaesthesia.

Objectives

The primary objective was to determine whether supplementary oxygen given to low‐risk term pregnant women undergoing elective caesarean section under regional anaesthesia can prevent maternal and neonatal desaturation. The secondary objective was to compare the mean values of maternal and neonatal blood gas levels between mothers who received supplementary oxygen and those who did not (control group).

Methods

Criteria for considering studies for this review

Types of studies

We included all randomized controlled trials (RCTs) examining maternal and foetal outcomes in women undergoing elective or scheduled caesarean section, under a regional anaesthesia, with or without oxygen supplementation.

Types of participants

We included healthy women of any age with low‐risk term pregnancies, who underwent an elective or scheduled caesarean section under regional anaesthesia.

Types of interventions

Oxygen supplementation to the mother, which may be administered at any flow rate or concentration via any oxygen delivery device, and for any duration, during an elective or scheduled caesarean section under regional anaesthesia.

Types of outcome measures

We included studies if any one of the following clinical outcomes were reported.

Primary outcomes

• Apgar scores at one and five minutes.

• Proportion of maternal participants with desaturation, defined as maternal SpO₂ less than 90%.

Secondary outcomes

To compare mean values of the following variables between participants who received supplementary oxygen and those who did not (control group).

• Mean maternal arterial oxygen tension (mm Hg) (PaO₂).

• Mean maternal oxygen saturation (%).

• Mean oxygen tension (mm Hg) in umbilical artery (UaPO₂).

• Mean oxygen tension (mm Hg) in umbilical vein (UvPO₂).

• Mean pH in umbilical artery (UapH).

• Mean carbon dioxide tension (mm Hg) in umbilical artery (UaPCO₂).

• Maternal plasma concentration of markers of oxygen free radical activity; 8‐isoprostane (pg/mL), malondialdehyde (MDA) (µmol/L).

• Neonatal plasma concentration of markers of oxygen free radical activity; 8‐isoprostane (pg/mL), MDA (µmol/L).

• Any neonatal adverse events requiring intervention (e.g. admission to neonatal intensive care unit, intubation, etc).

Search methods for identification of studies

The search was performed using the standard strategy of the Cochrane Anaesthesia, Critical and Emergency Care Group.

Electronic searches

In this updated review, we searched the Cochrane Central Register of Controlled Trials (CENTRAL; 2014, Issue 11); MEDLINE (1948 to November 2014) and EMBASE (1980 to November 2014). In our earlier review we searched up to February 2012 (Chatmongkolchart 2013).

We reran the search in CENTRAL, MEDLINE, EMBASE to February 2016. We found one potential new study of interest which was added to the list of ‘Studies awaiting Classification' and will be incorporated into the formal review findings when we next update the review.

We used Cochrane's optimally sensitive strategies to identify RCTs for MEDLINE and EMBASE searches (Higgins 2011). We combined these with subject headings and text words.

Our MEDLINE search can be found in Appendix 1. We modified the MEDLINE search strategy to search CENTRAL (Appendix 2) and EMBASE (Appendix 3).

We did not impose any language restrictions.

Searching other resources

We searched the reference lists of review articles, relevant trials, textbooks and abstracts of scientific meetings to identify additional RCTs.

In addition, we made efforts to identify potential RCTs by searching the following data sources.

• The registers of clinical trials (International Clinical Trials Registry Platform; registers compiled by Current Science).

• Earlier printed subject indexes of Index Medicus and Excerpta Medica for relevant studies published before the onset of their respective electronic databases, MEDLINE and EMBASE.

• Other medical databases (Current Contents, Science Citation Index).

• "Grey literature" (theses, internal reports, non‐peer reviewed journals) and databases (System for Information on Grey Literature in Europe (SIGLE) and ZETOC).

• References cited in identified RCTs.

Data collection and analysis

SC entered the data into Review Manager 5 software (RevMan 2014), and SP checked data entry.

Selection of studies

We (SC and SP) independently selected and identified trials eligible for inclusion in the review on the basis of titles and abstracts. After reviewing the abstracts, we retrieved potentially relevant studies in full to evaluate them for inclusion. We resolved differences by discussion.

We (SC and SP) independently extracted data using a standardized data extraction form (Moher 2001; see Appendix 4). Where information in the relevant studies was insufficient, we contacted the study authors for further information. We resolved any discrepancies or disagreements in results or data interpretation through re‐examination of the trial reports. When we were unable to reach a consensus, we consulted with Dr Panthila Rujirojindakul (see Acknowledgements).

We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Moher 2009), and the 'Characteristics of excluded studies' table.

Data extraction and management

We (SC and SP) independently extracted data from published selected studies onto data extraction forms and entered it into the Cochrane software program, Review Manager 5 (RevMan 2014). The lead author (SC) checked the accuracy of the entered data.

We performed analysis of data using Review Manager 5 (RevMan 2014).

Assessment of risk of bias in included studies

We (SC and SP) independently assessed each of the selected studies for methodological quality. We assessed the quality of reporting of each trial based on Cochrane's domain‐based evaluation tool for assessing the risk of bias in included studies, as follows (Higgins 2011).

• Random sequence generation (selection bias).

• Allocation concealment (selection bias).

• Blinding of participants and personnel (performance bias).

• Blinding of outcome assessment (detection bias).

• Incomplete outcome data (attrition bias).

• Selective reporting (reporting bias).

• Other bias.

We evaluated the methodological quality of the included studies using the methods described in Chapter eight of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Based on these criteria, we subdivided the studies into the following categories.

• All quality criteria met: low risk of bias.

• One or more of the quality criteria met only in part: unclear risk of bias.

• One or more quality criteria not met: high risk of bias.

We used this classification as the basis of a sensitivity analysis. Additionally, we explored the influence of individual quality criteria in a sensitivity analysis.

Measures of treatment effect

For dichotomous data, we performed a meta‐analysis using pooled risk ratio (RR) with 95% confidence intervals (CIs). We calculated a mean difference (MD) with 95% CIs for continuous data, which were measured on the same scale. We calculated continuous data measured on different scales as a standardized mean difference (SMD) with 95% CIs.

Unit of analysis issues

All included trials used individual participants as the unit of analysis. We combined, into a single group, trials that included multiple intervention groups, using the formula described in Table 7.7.a of Higgins 2011. For subgroup analysis, if multiple intervention groups were included, we split the sample size of the control group (Higgins 2011).

Dealing with missing data

In the event of missing data in any of the included trials, we first attempted to obtain the missing data from the authors. If unavailable, we performed a sensitivity analysis through imputation as described in the Sensitivity analysis section.

Assessment of heterogeneity

We evaluated the fixed‐effect model for statistical heterogeneity using the Chi² test, the Q statistic and I² statistic to quantify the degree of inconsistency across studies. If the Chi² test P value was < 0.1, or if the I² statistical value was greater than 50%, we took this to indicate substantial heterogeneity (Higgins 2002; Higgins 2003), and used a random‐effects model. If the I² statistic was higher than 75%, we did not report the results of the pooling, rather we reported and discussed individual trial results only. We assessed possible sources of heterogeneity by performing subgroup analyses as described in Subgroup analysis and investigation of heterogeneity.

Data synthesis

We included only RCTs in this meta‐analysis. We calculated pooled estimates of the difference in means using the fixed‐effect model or the random‐effects model, depending on the degree of heterogeneity. If heterogeneity was greater than 50%, we used the random‐effects model to calculate pooled estimates.

Subgroup analysis and investigation of heterogeneity

We planned to investigate the effects of different oxygen concentrations, different oxygen flow rates and different oxygen devices on outcomes where appropriate.

Sensitivity analysis

If appropriate, we planned sensitivity analyses for both missing data and study quality. For missing data, we performed 'best‐case' and 'worst‐case' analyses. The 'best‐case' analysis assumed that all missing data reflected favourable outcomes in the oxygen group and unfavourable outcomes in the control group. The 'worst‐case' analysis assumed the reverse. For assessment of study quality, we planned to compare the overall pooled analysis with an analysis in which the lower quality studies had been removed.

Summary of findings

We used the principles of the GRADE system to assess the quality of the body of evidence in our review (Guyatt 2008). We appraised the quality of evidence based on risk of bias, directness of the evidence, inconsistency (heterogeneity) of the data, precision of the effect estimates and risk of publication bias using GRADEproGDT software (GRADEproGDT 2015).

We created Table 1 and included the following outcomes in that table.

• Apgar scores at one and five minutes.

• Proportion of maternal participants with desaturation.

• Mean maternal arterial oxygen tension (mm Hg) (PaO₂).

• Mean maternal oxygen saturation (%).

• Mean oxygen tension (mm Hg) in umbilical artery (UaPO₂).

• Mean oxygen tension (mm Hg) in umbilical vein (UvPO₂).

Results

Description of studies

We included RCTs conducted on healthy women of any age with low‐risk term pregnancies, who underwent an elective or scheduled caesarean section under regional anaesthesia. The intervention groups received supplementary oxygen, at any flow rate or concentration, via any oxygen delivery device, and for any duration, during caesarean section. The control groups received room air.

Results of the search

In the previous version of this review, we searched the databases to February 2012 (Chatmongkolchart 2013).

In this updated review, we reran the searches to November 2014. As a result of our updated search, we retrieved a total of 1083 citations. After reviewing all titles and abstracts, we identified and retrieved 32 references in full text for the updated review.

We excluded 21 studies, leaving 11 trials for inclusion in this review (see Excluded studies and Figure 1).

1.

Study flow diagram. We reran the search in CENTRAL, MEDLINE, EMBASE to February 2016 and found one study of potential interest. This study was added to ‘Studies awaiting Classification' and will be incorporated into the formal review findings during the review update.

We reran the search in CENTRAL, MEDLINE, EMBASE in February 2016. One potential new study of interest was added to the list of ‘Studies awaiting Classification' and will be incorporated into the formal review findings during the next review update.

Included studies

We included in this updated review 11 trials that enrolled a total of 753 participants (Adenekan 2010; Castro 2009; Cogliano 2002; Gunaydin 2011; Khaw 2002; Khaw 2004; Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). The sample size of the individual trials varied between 20 and 204 participants. In one trial (Khaw 2004), outcomes were compared between participants who had a normal UD interval (< 180 seconds) and those who had a long UD interval (> 180 seconds). We included in this review only participants who had a normal UD interval.

We highlighted the differences in populations studied and interventions applied. Full details of participants, interventions and outcomes for each trial are provided in the Characteristics of included studies table.

Participants were scheduled for an elective caesarean section under spinal anaesthesia in eight trials (Adenekan 2010; Castro 2009; Cogliano 2002; Gunaydin 2011; Khaw 2002; Khaw 2004; Palacio 2008; Singh 2006), and under epidural anaesthesia in three (Ramanathan 1982; Young 1980; Yu 1992). All information on the anaesthetic techniques used is given in the Characteristics of included studies table and in Appendix 5.

The 11 included trials compared oxygen in a variety of concentrations and delivery devices (Appendix 6). Inspired concentrations of supplementary oxygen in the study groups were 35% (Singh 2006); 40% (Cogliano 2002; Khaw 2004; Palacio 2008); 47% (Ramanathan 1982); 60% (Castro 2009; Khaw 2002; Khaw 2004; Singh 2006); 74% (Ramanathan 1982); and 100% (Ramanathan 1982). Four trials confirmed the delivered FIO₂ with an oxygen analyser (Castro 2009; Khaw 2002; Khaw 2004; Ramanathan 1982). Oxygen in the other six trials was given at various flow rates between 2 L/min and 10 L/min (Adenekan 2010; Cogliano 2002; Gunaydin 2011; Palacio 2008; Young 1980; Yu 1992). Adenekan 2010 studied the oxygen flow of 4 L/min delivered from a Datex Ohmeda anaesthetic machine via the breathing circuit. Cogliano 2002 studied the oxygen flow of 2 L/min via a nasal cannula and 4 L/min via a face mask. Gunaydin 2011 studied the oxygen flow of 5 L/min via either a nasal cannula or a facial mask. Palacio 2008 studied the oxygen flow of 6 L/min via a Ventimask. Young 1980 studied pure oxygen at a 'high flow', which we assumed to mean greater than 60%. Yu 1992 studied the oxygen flow of 6 L/min and 10 L/min via a face mask. Therefore, the oxygen concentration actually achieved in these trials has been estimated rather than directly measured (Shapiro 1991).

The duration of oxygen exposure was poorly reported. Khaw 2002 recorded a median duration of supplementary gas exposure of 53.2 minutes (range 33 to 150 minutes) in the intervention group and 52.7 minutes (range 35 to 70 minutes) in the control group. Ramanathan 1982 reported only the starting time of oxygen administration, and Young 1980 reported only the time at which oxygen administration was ceased, with the duration of oxygen exposure of approximately 10 minutes or greater. Adenekan 2010 administered supplementary oxygen after induction of spinal block till the end of surgery. None of the other trials reported duration of exposure (Castro 2009; Cogliano 2002; Gunaydin 2011; Khaw 2004; Palacio 2008; Singh 2006; Yu 1992).

Four trials reported the UD interval (Khaw 2002; Khaw 2004; Ramanathan 1982; Yu 1992); three trials reported the mean ± SD (Khaw 2004; Ramanathan 1982; Yu 1992), and one trial reported the median and range (Khaw 2002). The UD interval in Yu 1992 was 79 ± 31 seconds in the control group and 67.9 ± 23.89 and 89.0 ± 45.65 seconds in the two intervention groups. Ramanathan 1982 reported a UD interval of 64 ± 6 seconds in the control group and 68 ± 7, 72 ± 11 and 69 ± 9 seconds in the three intervention groups. Khaw 2002 reported a median UD interval of 68 (range 52 to 75 seconds) in the control group and 69 (range 55 to 85 seconds) in the intervention group. Khaw 2004 compared outcomes between participants who had normal (< 180 seconds) and long (> 180 seconds) UD intervals. When these data were pooled with data from the other trials included in this review, only participants who had a normal UD interval were included. The UD interval was 102 ± 38 seconds in the control group and 102 ± 38 and 102 ± 35 seconds in the intervention groups. One study excluded participants with foetal extraction times longer than three minutes (Castro 2009) .

Excluded studies

In our original review we excluded 20 studies for the reasons detailed in Characteristics of excluded studies. In this updated version we excluded one new study (Siriussawakul 2014). In total we excluded 21 studies.

No control group was included in three of the studies (Backe 2007; Crosby 1992; Halpern 1990). We excluded seven studies because they enrolled participants outside the inclusion criteria for this review (Bogod 1988; Khaw 2009; Nesterenko 2012; Perreault 1990; Perreault 1992; Rorke 1968; Siriussawakul 2014). We excluded four studies because the intervention was not relevant (Hollmen 1978; Okudarira 2005; Petropoulos 2003; Tonni 2007). We excluded the remaining studies for the following reasons: a meta‐analysis (Reynolds 2005); not randomized (Haruta 1984); no reporting of number of participants (Kelly 1996); participants were included in a subsequent study (Backe 2002; Kelly 1995); and not enough information to be retrieved for analysis (Kelly 1997; Salwanis 2012).

Studies awaiting classification

There is one study awaiting classification which was found during the February 2016 search (Yalcin 2013). This study will be incorporated into the review process when we next update the review. For further details of this study see the Characteristics of studies awaiting classification table.

Ongoing studies

There are no ongoing studies

Risk of bias in included studies

We used Cochrane's domain‐based evaluation table provided in RevMan 2014 to assess the validity and the quality of included trials (see the Characteristics of included studies table). Nine of the 11 included trials had a high risk of bias (Adenekan 2010; Castro 2009; Cogliano 2002; Gunaydin 2011; Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). We assessed only two trials as having a low risk of bias (Khaw 2002; Khaw 2004). (See Figure 2; Figure 3).

2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

3.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Allocation to the treatment group was performed by randomized assignment in all trials. Randomization was clearly adequate in five trials ( Castro 2009; Gunaydin 2011; Khaw 2002; Khaw 2004; Singh 2006), and was unclear in four trials (Adenekan 2010; Cogliano 2002; Palacio 2008; Ramanathan 1982). There was no randomization description in one trial (Yu 1992). Forced randomization with every six participants was used in one trial (Young 1980). The allocation concealment schemes were adequate in five trials (Adenekan 2010; Castro 2009; Khaw 2002; Khaw 2004; Young 1980). The remaining trials may be affected by selection bias because of the absence of adequate allocation concealment (Cogliano 2002; Gunaydin 2011; Palacio 2008; Ramanathan 1982; Singh 2006; Yu 1992).

Blinding

Participants were adequately blinded to the intervention in one trial (Young 1980). Blinding was not adequate because of differences in oxygen delivery devices between study groups in three trials (Adenekan 2010; Cogliano 2002; Gunaydin 2011). Whether participants were unaware of group allocation was not discussed, although a gas mixture was delivered to control and study groups using the same delivery system and oxygen devices (Khaw 2002; Khaw 2004; Ramanathan 1982). However, the participants who were unblinded had no influence on the outcome measured in these three trials. The subject of blinding was not discussed in four trials (Castro 2009; Palacio 2008; Singh 2006; Yu 1992).

In two trials, therapists were blinded to the assigned FIO₂ (Khaw 2004; Young 1980). One trial was adequately blinded to the investigators (Khaw 2004). Another trial blinded the obstetricians but not the investigators (Young 1980).

Assessors who assessed the Apgar scores were unaware of group allocation in five trials (Cogliano 2002; Gunaydin 2011; Khaw 2002; Khaw 2004; Yu 1992). Investigators who performed the blood analysis were blinded to group allocation in two trials (Khaw 2002; Khaw 2004).

Incomplete outcome data

Explicit criteria for participant withdrawal were stated in four trials (Castro 2009; Cogliano 2002; Khaw 2004; Yu 1992). Fourteen participants were excluded in two trials (Khaw 2004; Palacio 2008); six because of an insufficient umbilical blood sample (Khaw 2004); one because of vomiting and not wearing a face mask during the uterine incision (Khaw 2004); one because of complaints of discomfort from wearing the mask (Khaw 2004); two because of coagulation of the samples (Palacio 2008); and four because of supplementary oxygen administration in the study group (Palacio 2008). Investigators did not use intention‐to‐treat methodology.

Selective reporting

All prespecified outcomes were reported in three trials (Adenekan 2010; Khaw 2002; Khaw 2004). Outcomes were not prespecified in two trials (Cogliano 2002; Young 1980). Castro 2009 did not report the amount of FIO₂ the intervention group actually received. More than one reported outcome was not prespecified in two trials (Ramanathan 1982; Yu 1992). Some of the prespecified outcomes were not reported in three trials (Gunaydin 2011; Palacio 2008; Singh 2006).

Other potential sources of bias

Several factors may affect maternal‐foetal gas exchange, such as duration of oxygen exposure; UD interval, maternal hypotension and vasopressor consumption were poorly reported. Duration of oxygen exposure was adequately reported in only one trial (Khaw 2002). The UD interval was reported in four trials (Khaw 2002; Khaw 2004; Ramanathan 1982; Yu 1992). Hypotension was reported in three trials (Khaw 2002; Singh 2006; Yu 1992), and vasopressor consumption was reported in three trials (Gunaydin 2011; Khaw 2004; Singh 2006).

Withdrawal criteria may be a source of other bias. Five trials had maternal desaturation withdrawal criteria using a maternal oxygen saturation reading from SpO₂ (Castro 2009; Cogliano 2002; Khaw 2002; Khaw 2004; Palacio 2008). Two trials stated that mothers who had desaturation would be excluded from the analysis, but none actually met that criteria (Khaw 2004; Palacio 2008). One trial stated that participants in the control group who had a pulse oximetry level below 92% would not be withdrawn, but instead their oxygen concentration level would be increased to 28% (Khaw 2002).

Desaturation withdrawal criteria were different from trial to trial: 92% (Cogliano 2002); 93% (Castro 2009); 94% (Palacio 2008); and 95% (Khaw 2004). Palacio 2008 also reported withdrawal of participants if desaturation occurred more than 4 points from baseline. The other six trials did not state the maternal desaturation withdrawal criteria (Adenekan 2010; Gunaydin 2011; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992) .

A priori sample size calculation was not stated in seven trials (Castro 2009; Gunaydin 2011; Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992).

Effects of interventions

See: Table 1

See: Table 1.

Note on subgroup analyses. We had planned to investigate the effects of different oxygen concentrations, different oxygen flow rates, and different oxygen devices on outcomes where appropriate. Because of the wide variation in oxygen delivery methods used, subgroup analysis was possible only for different oxygen concentrations, and only after some assumptions were made. For trials where the delivered percentage of oxygen was not specified (Adenekan 2010; Cogliano 2002; Gunaydin 2011; Young 1980; Yu 1992), we calculated the estimated oxygen concentration from the device and flow rates, as described in Shapiro 1991. We then used a post hoc value of 60% oxygen delivered as the cut point to divide those studies using a high concentration of oxygen (60% and above) from those using lower concentrations (less than 60%).

While using this methodology, five trials investigated low oxygen concentration (Adenekan 2010; Cogliano 2002; Gunaydin 2011; Palacio 2008; Yu 1992), three trials investigated high oxygen concentration (Castro 2009; Khaw 2002; Young 1980), and three trials investigated both low and high oxygen concentrations (Khaw 2004; Ramanathan 1982; Singh 2006). Three of the four trials (Cogliano 2002; Gunaydin 2011; Yu 1992), while using a low oxygen concentration, had two intervention groups; we combined the results of those groups using the formula shown in Table 7.7.a in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). For trials using both low and high oxygen concentrations (Khaw 2004; Ramanathan 1982; Singh 2006), we split the control group when performing the subgroup analysis. For the trial that had three supplementary oxygen groups (Ramanathan 1982), 47%, 74% and 100%, we combined the results of participants given 74% and 100% oxygen in the high oxygen concentration group.

We noted differences in the units of blood gas values. Three trials reported the values in kilopascal (kPa) (Cogliano 2002; Khaw 2002; Khaw 2004), the remaining trials reported the values in millimetre mercury (mm Hg) (Castro 2009; Gunaydin 2011; Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). To pool the data, we converted kPa into mm Hg by multiplying by a factor of 7.5.

Primary outcomes

1. Apgar scores at one and five minutes

All studies except Castro 2009 studied Apgar scores at one and five minutes. Six studies reported the values of Apgar scores at one and five minutes (Adenekan 2010; Gunaydin 2011; Khaw 2002; Khaw 2004; Palacio 2008; Young 1980). Five studies reported the number of neonates having Apgar scores less than seven at one and five minutes (Cogliano 2002; Khaw 2002; Khaw 2004; Ramanathan 1982; Yu 1992) . Singh 2006 stated only that no differences in Apgar scores were seen in all three groups at one and five minutes (P = 0.76, 0.53, respectively); therefore we did not pool the data from this study. Four trials reported Apgar scores as mean values with standard deviations (Adenekan 2010; Gunaydin 2011; Palacio 2008; Young 1980), and two trials reported the scores as median values with ranges (Khaw 2002; Khaw 2004). To pool the data to obtain average Apgar scores at one and five minutes, we calculated mean and standard deviation from median and range using the formulae provided in Hozo 2005.

Mean Apgar scores at one minute. The pooled data suggest that a non‐significant increase in Apgar score of 0.07 favoured the oxygen group (MD 0.07, 95% CI ‐0.16 to 0.31, P = 0.53, 519 participants). However, heterogeneity was high (I² statistic = 70%). Therefore, we performed a subgroup analysis (Analysis 1.1). Results showed that mean Apgar scores were not significantly different in low (less than 60%) (MD ‐0.03, 95% CI ‐0.19 to 0.13, P = 0.7, I² statistic = 3%) versus high (60% or greater) oxygen groups (MD 0.18, 95% CI ‐0.26 to 0.62, P = 0.41, I² statistic = 81%). The I² statistic was reduced from 79% to 70%. However, the heterogeneity of the high oxygen group was 81%. Of the three trials in this group (Khaw 2002; Khaw 2004; Young 1980), results in only one favoured the oxygen group (MD 0.50, 95% CI 0.26 to 0.74, P < 0.0001) (Young 1980). We downgraded the results from high to low GRADE due to inconsistency of the results (high heterogeneity; I² statistic = 79%), and four out of six trials had high risk of bias (Adenekan 2010; Gunaydin 2011; Palacio 2008; Young 1980).

1.1. Analysis.

Comparison 1 Primary outcomes, Outcome 1 Apgar scores at one minute.

Mean Apgar score at five minutes. We used a fixed‐effect model to pool the data and found no significant differences in mean Apgar scores at five minutes (MD ‐0.00, 95% CI ‐0.06 to 0.06, P = 0.98, I² statistic = 0%, 519 participants; Analysis 1.2). We downgraded the result from high to low GRADE due to 50% of the trials being at high risk of bias.

1.2. Analysis.

Comparison 1 Primary outcomes, Outcome 2 Apgar scores at five minutes.

2. Proportion of maternal participants with desaturation, defined as SpO₂ less than 90%

None of the trials included in this review reported this outcome.

Secondary outcomes

1. Mean maternal arterial oxygen tension (mm Hg) (PaO₂)

Six trials enrolling a total of 241 participants reported maternal PaO₂ (Castro 2009; Khaw 2002; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). We noted significant heterogeneity (I² statistic = 93%) (Analysis 2.1). We performed a subgroup analysis, separating trials with low (less than 60%) and high (60% or greater) inspired oxygen. The results (not shown) could not explain the heterogeneity. Therefore, we did not pool the data. In each trial, results favoured the oxygen group, that is, the mean maternal PaO₂ was higher in pregnant women receiving oxygen. The MD varied from 91.2 mm Hg (95% CI 76.81 to 105.61 mm Hg, P < 0.00001) (Singh 2006) to 227.0 mm Hg (95% CI 195.34 to 258.66 mm Hg, P < 0.00001) (Ramanathan 1982). We downgraded the results from high to very low GRADE due to unexplained high heterogeneity and five out of six trials having high risk of bias.

2.1. Analysis.

Comparison 2 Secondary outcomes, Outcome 1 Mean maternal arterial oxygen tension (PaO₂).

2. Mean maternal oxygen saturation (%)

Five trials enrolling a total of 209 participants reported maternal oxygen saturation (Adenekan 2010; Gunaydin 2011; Palacio 2008; Ramanathan 1982; Yu 1992). We pooled data from three of these five trials; we excluded Gunaydin 2011 because this trial reported the maternal oxygen saturation only for participants receiving oxygen via a face mask at three and five minutes after spinal block (the maternal oxygen saturation of the nasal cannula group was omitted) and we excluded Adenekan 2010 as this trial reported the least arterial oxygen saturation during surgery and the recovery. Pooled maternal oxygen saturation in the analysis was obtained at different times: at the time of uterine incision (Ramanathan 1982), and at the time of delivery (Palacio 2008; Yu 1992).

Heterogeneity was high (I² statistic = 88%); therefore we did not pool the data. Individual data favoured the oxygen group. MD ranged from 1.1% (95% CI 0.8% to 1.4%, P < 0.00001) (Yu 1992) to 2.8% (95% CI 2.0 to 3.5%, P < 0.00001) (Ramanathan 1982) (Analysis 2.2). We downgraded the result from high to very low quality of evidence due to unexplained high heterogeneity, and all three trials being at high risk of bias.

2.2. Analysis.

Comparison 2 Secondary outcomes, Outcome 2 Mean maternal oxygen saturation.

3. Mean oxygen tension (mm Hg) in umbilical artery (UaPO₂)

Eight trials enrolling a total of 504 participants reported UaPO₂ (Cogliano 2002; Gunaydin 2011; Khaw 2002; Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). We used a random‐effects model to pool the data because substantial heterogeneity (I² statistic = 72%) was noted among the studies. Supplementary oxygen significantly increased UaPO₂ with a MD of 3.3 mm Hg (95% CI 1.8 to 4.9, P < 0.0001) (Analysis 2.3). We downgraded the results from high to very low GRADE due to substantial heterogeneity (I² statistic = 72%), and seven out of the eight trials having high risk of bias.

2.3. Analysis.

Comparison 2 Secondary outcomes, Outcome 3 Mean oxygen tension in umbilical artery (UaPO₂).

We performed a subgroup analysis to assess the impact of oxygen concentration on UaPO₂. Pooled data showed a significant increase in UaPO₂ with a MD of 2.5 mm Hg (95% CI 1.2 to 3.8, P = 0.0002) in the low oxygen concentration group and a significant increase in UaPO₂ with a MD of 4.8 mm Hg (95% CI 1.5 to 8.2, P = 0.005) in the high oxygen concentration group (Figure 4).

4.

Forest plot of comparison: 2 Secondary outcomes, outcome: 2.3 Mean oxygen tension in umbilical artery (UaPO₂) (mm Hg).

A sensitivity analysis including low risk of bias trials showed no significant difference in UaPO₂ (MD 0.5, 95% CI ‐1.2 to 2.2, P = 0.57) (Khaw 2002; Palacio 2008). We used a fixed‐effect model to pool the data because no heterogeneity was noted among the trials (I² statistic = 0%).

4. Mean oxygen tension (mm Hg) in umbilical vein (UvPO₂)

Ten trials enrolling a total of 683 participants reported the UvPO₂ (Castro 2009; Cogliano 2002; Gunaydin 2011; Khaw 2002; Khaw 2004; Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). We did not pool the data because heterogeneity was significant (I² statistic = 88%) (Analysis 2.4). Five trials showed no significant difference in UvPO₂ between babies of the pregnant women receiving supplementary oxygen and babies of those not receiving it (Castro 2009; Cogliano 2002; Gunaydin 2011; Khaw 2002; Khaw 2004). The other trials showed significantly higher UvPO₂ of babies in the supplementary oxygen group (Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). We downgraded the results from high to very low quality of evidence due to unexplained high heterogeneity (I² statistic = 88%), and eight out of 10 trials having high risk of bias.

2.4. Analysis.

Comparison 2 Secondary outcomes, Outcome 4 Mean oxygen tension in umbilical vein (UvPO₂).

A subgroup analysis showed a significant increase in UvPO₂ with a MD of 4.0 mm Hg (95% CI 1.7 to 6.2 mm Hg, P = 0.0007) in the low oxygen concentration group and a MD of 8.2 mm Hg (95% CI 3.6 to 12.8 mm Hg, P = 0.0005) in the high oxygen concentration group. The subgroup analysis also showed significant heterogeneity in both low (I² statistic = 76%) and high oxygen concentration groups ( I² statistic = 88%).

A sensitivity analysis including low risk of bias trials showed a significant difference in UvPO₂ values (higher with oxygen; MD 2.82 mm Hg, 95% CI 1.23 to 4.42 mm Hg, P = 0.0005) (Khaw 2002; Khaw 2004; Palacio 2008). Heterogeneity was low (I² statistic = 23%).

5. Mean pH in umbilical artery (UapH)

Eight studies enrolling a total of 623 participants reported UapH (Cogliano 2002; Gunaydin 2011; Khaw 2002; Khaw 2004; Palacio 2008; Singh 2006; Young 1980; Yu 1992). We pooled the data using a fixed‐effect model because we did not note any heterogeneity (I² statistic = 0%). Results showed no significant differences in UapH (MD 0.00, 95% CI ‐0.01 to 0.00, P = 0.26) between the two groups (Analysis 2.5).

2.5. Analysis.

Comparison 2 Secondary outcomes, Outcome 5 Mean pH in umbilical artery (UapH).

6. Mean carbon dioxide tension (mm Hg) in umbilical artery (UaPCO₂)

Eight studies enrolling a total of 504 participants reported UaPCO₂ (Cogliano 2002; Gunaydin 2011; Khaw 2002; Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). We pooled data using a random‐effects model because we found moderate heterogeneity(I² statistic = 52%). Results showed no significant differences between groups in UaPCO₂ (MD 1.46 mm Hg, 95% CI ‐0.51 to 3.42, P = 0.15) (Analysis 2.6).

2.6. Analysis.

Comparison 2 Secondary outcomes, Outcome 6 Mean carbon dioxide tension in umbilical artery (UaPCO₂).

7. Mean maternal plasma concentration of markers of oxygen free radical activity; 8‐isoprostane (pg/mL), malondialdehyde (MDA) (µmol/L)

Markers of oxygen free radical activity of mothers were reported in two trials, which enrolled a total of 104 participants (Khaw 2002; Singh 2006). Khaw 2002 studied the effect of giving 60% oxygen, and Singh 2006 studied the effect of giving 35% and 60% oxygen, compared with room air. The pooled data, using a fixed‐effect model, showed significantly higher MDA (MD ‐0.24 µmol/L, 95% CI ‐0.37 to ‐0.12, P = 0.0002, I² statistic = 0%) (Analysis 2.7) and isoprostane (MD ‐64.27 pg/mL, 95% CI ‐76.82 to ‐51.72, P < 0.00001, I² statistic = 44%) in maternal plasma among those receiving supplementary oxygen (Analysis 2.8).

2.7. Analysis.

Comparison 2 Secondary outcomes, Outcome 7 Mean maternal MDA.

2.8. Analysis.

Comparison 2 Secondary outcomes, Outcome 8 Mean maternal isoprostane.

8. Mean neonatal plasma concentration of markers of oxygen free radical activity; 8‐isoprostane (pg/mL), malondialdehyde (MDA) (µmol/L)

Markers of oxygen free radical activity in neonates were reported in two studies, which enrolled a total of 104 participants (Khaw 2002; Singh 2006). The pooled data showed a significant increase in MDA (MD ‐0.09 µmol/L, 95% CI ‐0.13 to ‐0.04, P = 0.0002, I² statistic =11%) (Analysis 2.9) and isoprostane (MD ‐67.74 pg/mL, 95% CI ‐82.53 to ‐52.96, P < 0.0001, I² statistic = 10% ) (Analysis 2.10) in the umbilical artery, as well as MDA (MD ‐0.30 µmol/L, 95% CI ‐0.37 to ‐0.22, P < 0.0001, I² statistic = 0%) (Analysis 2.11) and isoprostane (MD ‐186.02 pg/mL, 95% CI ‐343.48 to ‐28.56, P = 0.02, I² statistic = 97%) (Analysis 2.12) in the umbilical vein.

2.9. Analysis.

Comparison 2 Secondary outcomes, Outcome 9 Umbilical artery MDA.

2.10. Analysis.

Comparison 2 Secondary outcomes, Outcome 10 Umbilical artery isoprostane.

2.11. Analysis.

Comparison 2 Secondary outcomes, Outcome 11 Umbilical vein MDA.

2.12. Analysis.

Comparison 2 Secondary outcomes, Outcome 12 Umbilical vein isoprostane.

9. Any requirement for neonatal intervention (e.g. admission to neonatal intensive care unit, intubation, etc.)

There were no reports of this outcome.

Discussion

The conclusions of our updated review remain the same as the original review (Chatmongkolchart 2013). The first primary outcome, Apgar scores, was studied in 10 out of the 11 included trials (Adenekan 2010; Cogliano 2002; Gunaydin 2011; Khaw 2002; Khaw 2004; Palacio 2008; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992). Because of differences in unit of measurement, we only pooled six of the included trials (519 participants) (Adenekan 2010; Gunaydin 2011; Khaw 2002; Khaw 2004; Palacio 2008; Young 1980). None of the trials included in this review reported the second primary outcome, the proportion of maternal participants with desaturation. Data were available for eight out of nine secondary outcomes. Quality of the evidence was low for the primary outcomes and very low for the secondary outcomes. The reasons for our grading were that only two of the 11 included trials were at low risk of bias across all domains, and there were unexplained high levels of heterogeneity (Khaw 2002; Khaw 2004).

Summary of main results

Impact on neonatal outcomes

In comparison to breathing room air, giving supplementary oxygen to healthy parturients scheduled for caesarean section under regional anaesthesia had no significant impact on neonatal outcome (assessed by Apgar scores). However, oxygen tension in umbilical vein and artery were significant higher in the supplementary oxygen group.

Impact on maternal outcomes

None of the 11 trials included in this review reported the proportion of maternal participants with desaturation (defined as SpO₂ less than 90%).

Parturients receiving supplementary oxygen had higher mean maternal arterial oxygen tension and saturation than parturients who received no supplementary oxygen. However, very low quality of evidence, due to risk of bias and inconsistency of the effect estimates, indicates that this finding should be interpreted with caution.

Impact on adverse events

A significant increase in the markers of oxygen free radicals was seen in both mothers and neonates after supplemental oxygen administration. However, this result used pooled data from two trials only (Khaw 2002; Singh 2006). None of the 11 trials included in this review reported neonatal adverse events requiring intervention, such as admission to neonatal intensive care units, intubation, etc.

Overall completeness and applicability of evidence

We are confident that our search strategy obtained all available updated studies.

This review found no data related to one of our primary outcomes (maternal desaturation) but did confirm that Apgar scores are not significantly improved at one or five minutes after delivery when supplemental oxygen is administered. Thus, although the data are incomplete for important clinical outcomes, no evidence suggested clear benefit derived from oxygen administration.

The results of this review are applicable to pregnant women who have uncomplicated pregnancies with a normal UD interval. It is not known whether oxygen supplementation would be of benefit to pregnant women with complicated caesarean sections or a long UD interval.

Quality of the evidence

We found the overall quality of evidence being low for the one primary outcome that was reported, and very low for the secondary outcomes, due to risk of bias and inconsistency of the results.

There was substantial heterogeneity across the studied effects. It is plausible that heterogeneity may be due to differences in intervention (different oxygen concentration, flow rates, delivery devices, and duration of administration); and variations in blood gases measurement (methods of analysis and calibration, time delay between drawing of samples and measurement).

Because of the wide variation in oxygen delivery methods used in the included trials, we performed only a subgroup analysis using oxygen concentration with a cutoff value of 60% to assess the impact of the intervention. Moreover, the desaturation withdrawal criteria may have an impact on the results because six of the trials did not mention the withdrawal criteria (Adenekan 2010; Gunaydin 2011; Ramanathan 1982; Singh 2006; Young 1980; Yu 1992), and in one trial (Khaw 2004), the investigators stated that participants would be withdrawn and excluded from the analysis if desaturation occurred.

Potential biases in the review process

We adhered closely to our protocol which outlined our procedures for minimizing bias in the review (Chatmongkolchart 2006): these included independent screening for trial inclusion, data extraction and assessment of risk of bias by two review authors. With assistance from the Cochrane Anaesthesia, Critical and Emergency Group’s Trials Search Co‐ordinator, we conducted a thorough search strategy, and believe we have identified all relevant trials.

Agreements and disagreements with other studies or reviews

This review showed no convincing evidence that giving supplementary oxygen given to healthy term pregnant women during elective caesarean section under regional anaesthesia is either beneficial or harmful for either the mother or the foetus clinically. No differences were noted in foetus' short‐term clinical outcome (Apgar scores) when mothers who received extra oxygen were compared with those who did not.

Our review may different from other studies or reviews in neonatal welfare assessment. We chose Apgar scores because it was used routinely. Other studies may use neonatal acid‐base status or other neurological assessment. In studying of oxidative stress, we included only participants who were undergoing elective caesarean section, but other studies may also include participants who were undergoing emergency caesarean section.

Authors' conclusions

Implications for practice.

There is low GRADE evidence that administration of supplementary oxygen in healthy, uncomplicated term pregnant women undergoing an elective caesarean section, under regional anaesthesia yields no difference in Apgar scores, even though oxygen levels for both mother and foetus were higher. There are insufficient studies to determine the clinical relevance of adverse effects of maternal hyperoxygenation to pregnant women or neonates.

There is very low GRADE evidence, due to risk of bias and inconsistency of the effect estimates, that findings should be interpreted with caution.

Implications for research.

The low and very low quality of evidence and the inconsistency in trials requires careful planning of future RCTs, which should aim to standardize several issues, including methodological quality, the use of less variable oxygen delivery equipment, confirmation of the oxygen concentration delivered to participants, and long‐term clinical endpoints.

Further research is required to identify one of the primary outcomes (the proportion of maternal desaturation) and clarify the clinical effect of maternal hyperoxia on the foetus, and any association between oxygen levels and clinical outcomes, such as neurological disorders and growth retardation.

What's new

Date	Event	Description
16 March 2016	New citation required but conclusions have not changed	This is an update of a review last published in Issue 6, 2013 of the Cochrane Library that included 10 RCTs (Chatmongkolchart 2013). In this updated review, we found one new trial that met our inclusion criteria (Adenekan 2010). In general, our updated review reaches the same conclusions as the original review.
16 March 2016	New search has been performed	In the previous version, we searched the databases to February 2012 (Chatmongkolchart 2013). In this updated review, we reran the searches to February 2016 .

History

Protocol first published: Issue 3, 2006
 Review first published: Issue 6, 2013

Date	Event	Description
19 May 2008	Amended	Converted to new review format.

Appendices

Appendix 1. Search strategy for MEDLINE (Ovid SP)

1. exp Oxygen‐Inhalation‐Therapy/ or Oxygen/ or Oxygen‐Consumption/ or (oxygen* adj3 supplement*).mp. or oxygen*.ti,ab. 
 2. exp Cesarean‐Section/ or (c?section* or c?esar?an or kesarevo or kaiserschnitt).mp. or ((abdominal or operativ*) adj5 (birth or delivery)).mp. 
 3. ((randomized controlled trial or controlled clinical trial).pt. or randomized.ab. or placebo.ab. or drug therapy.fs. or randomly.ab. or trial.ab. or groups.ab.) not (animals not (humans and animals)).sh. 
 4. 1 and 2 and 3

Appendix 2. Search strategy for CENTRAL, the Cochrane Library

#1. MeSH descriptor Oxygen Inhalation Therapy explode all trees
 #2. MeSH descriptor Oxygen, this term only
 #3. MeSH descriptor Oxygen Consumption explode all trees
 #4. oxygen* near supplement*
 #5. oxygen*:ti,ab
 #6. (#1 OR #2 OR #3 OR #4 OR #5)
 #7. MeSH descriptor Cesarean Section explode all trees
 #8. c?section* or c?esar?an or kesarevo or kaiserschnitt
 #9. (abdominal or operativ*) and (birth or delivery)
 #10. (#7 OR #8 OR #9)
 #11. (#6 AND #10)

Appendix 3. Search strategy for EMBASE (Ovid SP)

1. exp oxygen therapy/ or oxygen/ or oxygen consumption/ or (oxygen* adj3 supplement*).mp. or oxygen*.ti,ab. 
 2. exp caesarean section/ or (c?section* or c?esar?an or kesarevo or kaiserschnitt).mp. or ((abdominal or operativ*) adj5 (birth or delivery)).mp. 
 3. ((((singl* or doubl* or tripl*) adj3 blind) or crossover).ti,ab. or multicenter.ab. or placebo.sh. or controlled study.ab. or random*.ti,ab. or trial*.ti,ab.) not (animal not (human and animal)).sh. 
 4. 1 and 2 and 3

Appendix 4. Data extraction form

1)  Title……………………………………………………………………………………………...

2)  Study ID…………...……………………………………………………………………………

3)  Authors…………………………………………………………………………………………

4)  Publication status……………………………………………………………………………..

5)  Date of extraction……………………………………………………………………………..

6)  Reviewer’s name………………………………………………………………………………

7)  Risk of bias table: (Check one: low risk or high risk or unclear)

Domain	Low risk	High risk	Unclear	Description or quote
7.1) Was the allocation sequence adequately generated?				Describe the method used to generate the allocation sequence
7.2) Was allocation adequately concealed?				State concealment approach
7.3) Blinding				Describe all measures used
(a) Those receiving care
(b) Those providing care
(c) Outcome assessors
7.4) Were incomplete outcome data adequately addressed?				State whether attrition and exclusions were reported, reasons for attrition/exclusions
7.5) Are reports of the study free of suggestion of selective outcome reporting?
7.6) Other sources of bias

8)  Participants: If only two groups were compared, the third column will be NA

Participants	Study group 1 or control	Study group 2	Study group 3
Number of women randomized (n)	(value)	(value)	(value)

9)  Interventions: If only two groups were compared, the third column will be NA

Interventions	Study group 1 or control	Study group 2	Study group 3
Type of oxygen delivery device
Flow rate of oxygen (L/min)	(value)	(value)	(value)
Concentration of oxygen (%)	(value)	(value)	(value)
Duration of supplementary oxygen (min)	(value)	(value)	(value)

10) Outcomes: (Intervention vs control) (if yes, fill value; if no, fill NA)

10.1 Maternal outcomes	Study group 1 or control	Study group 2	Study group 3
10.1.1 ABG	Yes or NA	Yes or NA	Yes or NA
pH	(value or NA)	(value or NA)	(value or NA)
PO2	(value or NA)	(value or NA)	(value or NA)
PCO2	(value or NA)	(value or NA)	(value or NA)
10.1.2 Maternal oxygen saturation	(value or NA)	(value or NA)	(value or NA)
10.1.3 Plasma concentration of oxygen free radical activity	Yes or NA	Yes or NA	Yes or NA
Isoprostane	(value or NA)	(value or NA)	(value or NA)
Malondialdehyde	(value or NA)	(value or NA)	(value or NA)
Hydroperoxides	(value or NA)	(value or NA)	(value or NA)
Purine metabolites	(value or NA)	(value or NA)	(value or NA)
10.1.4 Maternal acceptance of devices	Favour or not favour	Favour or not favour	Favour or not favour

10.2 Foetal outcomes	Study group 1 or control	Study group 2	Study group 3
10.2.1 Umbilical cord blood
UapH      (umbilical arterial pH)	(value or NA)	(value or NA)	(value or NA)
UaPO2      (umbilical arterial partial pressure of oxygen)	(value or NA)	(value or NA)	(value or NA)
UaPCO2      (umbilical arterial partial pressure of carbon dioxide)
UvpH      (umbilical venous pH)	(value or NA)	(value or NA)	(value or NA)
UvPO2      (umbilical venous partial pressure of oxygen)	(value or NA)	(value or NA)	(value or NA)
UvPCO2      (umbilical venous partial pressure of carbon dioxide)
10.2.2 Apgar scores (values)	(value or NA)	(value or NA)	(value or NA)
10.2.3 Neonatal plasma concentration of markers of oxygen free radicals	(value or NA)	(value or NA)	(value or NA)
Isoprostane	(value or NA)	(value or NA)	(value or NA)
Malondialdehyde	(value or NA)	(value or NA)	(value or NA)
Hydroperoxides	(value or NA)	(value or NA)	(value or NA)

11) Other comment on the study (co‐interventions, potential confounders)

12) Check one

 [ ] Include in the review

  [ ] Exclude from the review

13) Reasons for excluding from the review

Footnotes

ABG: arterial blood gas
 min: minute
 NA: not applicable
 pH: potential of hydrogen ion
 vs: versus

Appendix 5. Anaesthetic technique

Study ID	Premedication	Preload	Anaesthetic technique	Anaesthetic agent
Adenekan 2010	NA	NA	Spinal block	2.2‐2.8 mL of 0.5% hyperbaric bupivacaine
Castro 2009	No	LRS 15 mL/kg	Spinal block	10 mg 0.5% hyperbaric bupivacaine with 60 µg diamorphine
Cogliano 2002	Ranitidine 150 mg oral 0.3 M Sodium citrate 30 mL oral	Not stated	Spinal block	2.5 to 2.75 mL of 0.5% hyperbaric bupivacaine with 0.3 mg diamorphine
Gunaydin 2011	Metoclopramide 10 mg IV and Ranitidine 50 mg IV 30 min before the operation	LRS 15 mL/kg	Spinal block	12 mg of 0.5% hyperbaric bupivacaine with fentanyl 10 µg and morphine 100 µg
Khaw 2002	Ranitidine 150 mg oral	LRS 20 mL/kg	Spinal block	2 mL of 0.5% hyperbaric bupivacaine with 15 µg fentanyl
Khaw 2004	Ranitidine 150 mg oral	LRS 20 mL/kg	Spinal block	Not stated
Palacio 2008	Not stated	6% starch hydroxyethyl 10 mL/kg	Spinal block	0.06 mg/cm height of 0.5% isobaric bupivacaine with 20 µg fentanyl
Ramanathan 1982	Not stated	Hartman's solution 1200 mL	Epidural block	18 ± 3 mL of 0.75% bupivacaine
Singh 2006	Not stated	LRS 15 mL/kg	Spinal block	2 mL of o.5% hyperbaric bupivacaine
Young 1980	No	LRS 1000 mL	Epidural block	Not stated
Yu 1992	Not stated	LRS 1500‐2000 mL	Epidural block	18‐20 mL of 2% xylocaine with adrenaline 1:200,000

Footnotes

IV: intravenous
 LRS: lactated Ringer's solution
 mL/kg: millilitre per kilogram
 N/A: not applicable

Appendix 6. Fraction of inspired oxygen (FiO₂) of studied group

Study ID	Study group	Device	Expected FIO2	Confirmed FIO2	Duration of oxygen exposure (minutes)
Adenekan 2010	Room air Oxygen flow 4 L/min	Room air Facemask	21% < 60%	No No	After induction of spinal anaesthesia until the end of surgery
Castro 2009	21% 60%	Mask with reservoir	21% 60%	No No	N/A
Cogliano 2002	Air (flow not indicated) Oxygen flow 2 L/min Oxygen flow 4 L/min	Face mask Nasal cannula Face mask	21% 28% 40% (manufacturer's data)	No No No	N/A
Gunaydin 2011	Room air Oxygen flow 5 L/min Oxygen flow 5 L/min	Breathe room air Nasal cannula Face mask	21% 32% (stated in the study) 30% (stated in the study)	No No No	Oxygen was administered immediately after spinal block
Khaw 2002	21% oxygen 60% oxygen	Venturi face mask Venturi face mask	21% 60%	Yes Yes	52.7 (35‐70) 53.2 (33‐150)
Khaw 2004	21% oxygen 40% oxygen 60% oxygen	Venturi face mask Venturi face mask Venturi face mask	21% 40% 60%	Yes Yes Yes	N/A
Palacio 2008	21% oxygen Oxygen flow 6 L/min	N/A Face mask	21% 40%	N/A	N/A
Ramanathan 1982	21% 47% 74% 100%	Face mask Face mask Face mask Face mask	21% 47% 74% 100%	Yes Yes Yes Yes	Oxygen was administered immediately after the epidural injection
Singh 2006	21% 35% 60%	Breathe room air Ventimask Ventimask	21% 35% 60%	No No No	Throughout the surgery
Young 1980	Room air Pure oxygen at high flow	Face mask Face mask	21% More than 40%	No No	10 minutes or greater (after anaesthesia was established until delivery)
Yu 1992	Room air Oxygen flow 6 L/min Oxygen flow 10 L/min	Breathe room air Face mask Face mask	21% 35% (stated in the study) 55% (stated in the study)	No No No	Oxygen mask given immediately after the epidural injection

Footnotes

L: litre
 min: minute
 N/A: not applicable

Appendix 7. Differences between protocol and review

	Protocol	Review
Objectives	Our primary objective is to compare maternal and foetal outcomes between women with low‐risk term pregnancies undergoing elective or scheduled caesarean section, under regional anaesthesia, with or without oxygen supplementation. Our intention is to identify adverse outcomes that might be prevented by the use of supplemental oxygen Our secondary objectives are to compare maternal and foetal outcomes between women who are given oxygen at: Different flow rates of administration Different concentrations of oxygen Different oxygen delivery devices	The primary objective was to determine whether supplementary oxygen given to low‐risk term pregnant women undergoing elective caesarean section under regional anaesthesia can prevent maternal and neonatal desaturation The secondary objective was to compare the mean values of maternal and neonatal blood gas levels between mothers who received supplementary oxygen and those who did not (control group)
Types of outcome measures	We will consider trials if any of the following clinical outcomes are reported: Our main outcome of interest is the prevention of maternal and neonatal hypoxaemia. We will define maternal hypoxaemia as SpO2 less than 90% corresponding to an arterial oxygen tension less than 7.9 kPa or that requires treatment	We included studies if any of the following clinical outcomes were reported: Primary outcomes: Apgar scores at one and five minutes Proportion of maternal participant desaturation, defined as maternal SpO2 less than 90%
	Our additional outcome measures include the following: Maternal outcomes: Arterial blood gas measurements that focus on maternal PaO2, PaCO2, and pH Maternal plasma concentrations of markers of oxygen free radical activity (e.g. 8‐isoprostane, malondialdehyde (MDA), hydroperoxide (OHP), purine metabolites, etc.) Maternal comfort and acceptance of the devices for delivering supplementary oxygen Maternal oxygen saturation Cardiac dysrhythmia defined as any rhythm requiring intervention Any other outcomes (e.g. for nausea and vomiting during surgery, etc.) Foetal outcomes: Umbilical cord blood: UvPO2: umbilical venous partial pressure of oxygen; UvpH: umbilical venous pH UaPO2: umbilical arterial partial pressure of oxygen; UapH: umbilical arterial pH Neonatal plasma concentration of markers of oxygen free radical activity (e.g. 8‐isoprostane, malondialdehyde (MDA), hydroperoxide (OHP), purine metabolites, etc.) Apgar score less than seven at age five minutes and ten minutes Acidosis as defined by cord or neonatal blood with a pH less than 7.20 Any requirement for neonatal intervention (e.g. admission to neonatal intensive care unit, intubation, etc.)	Secondary outcomes:   To compare mean values of the following variables between participants who received supplementary oxygen and those who did not (control group): Mean arterial oxygen tension (mm Hg) for maternal subjects Mean maternal oxygen saturation (%) Mean oxygen tension (mm Hg) in umbilical artery (UaPO2) Mean oxygen tension (mm Hg) in umbilical vein (UvPO2) Mean pH in umbilical artery (UapH) Mean carbon dioxide tension (mm Hg) in umbilical artery (UaPCO2) Maternal plasma concentration of markers of oxygen free radical activity; 8‐isoprostane (pg/mL), malondialdehyde (MDA) (µmol/L) Neonatal plasma concentration of markers of oxygen free radical activity; 8‐isoprostane (pg/mL), malondialdehyde (MDA) (µmol/L) Any requirement for neonatal intervention (e.g. admission to neonatal intensive care unit, intubation, etc.)
Searching other resources	In addition, we will make efforts to identify potential RCTs by searching the following data sources: The registers of clinical trials (The International Register of Clinical Trials Register; registers compiled by Current Science) Earlier printed subject indexes of Index Medicus and Excerpta Medica for relevant studies published before the onset of their respective electronic databases, MEDLINE and EMBASE Other medical databases (Current Contents, Science Citation Index) "Grey literature" (theses, internal reports, non－peer‐reviewed journals) and databases (System for Information on Grey Literature in Europe (SIGLE) and ZETOC) References cited in identified RCTs Other unpublished sources known to be experts in the speciality (to be sought by personal communication)	In addition, we will make efforts to identify potential RCTs by searching the following data sources: The registers of clinical trials (The International Register of Clinical Trials Register; registers compiled by Current Science) Earlier printed subject indexes of Index Medicus and Excerpta Medica for relevant studies published before the onset of their respective electronic databases, MEDLINE and EMBASE Other medical databases (Current Contents, Science Citation Index) "Grey literature" (theses, internal reports, non－peer‐reviewed journals) and databases (System for Information on Grey Literature in Europe (SIGLE) and ZETOC) References cited in identified RCTs
Assessment of risk of bias in included studies	Two authors (SC and TU) will independently assess each of the selected studies for methodological quality. We will assess the quality of reporting of each trial based on the quality criteria, specifying, in particular, the following factors (Moher 2001): Minimization of selection bias: (a) Was the randomization procedure adequate? (b) Was the allocation concealment adequate? Minimization of performance bias: Were the participants and people administering the treatment blinded to the intervention? Minimization of attrition bias: (a) Were withdrawals and dropouts completely described? (b) Was analysis by intention‐to‐treat? Minimization of detection bias: Were outcome assessors blinded to the intervention? We will evaluate the methodological quality of the included studies, using the methods described in Chapter eight of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2005) Based on these criteria, we will subdivide the studies into the following three categories: All quality criteria met: low risk of bias One or more of the quality criteria met only in part: moderate risk of bias One or more criteria not met: high risk of bias We will use this classification as the basis of a sensitivity analysis. Additionally, we will explore the influence of individual quality criteria in a sensitivity analysis. We will incorporate the results of quality assessment in the data analysis. If meta‐analysis of results is possible, we will combine the results sequentially in order of their assessed validity ('cumulative meta‐analysis')	Two authors (SC and SP) independently assessed each of the selected studies for methodological quality. We assessed the quality of reporting of each trial based on Cochrane's domain‐based evaluation tool for assessing the risk of bias in included studies (Higgins 2011): Random sequence generation (selection bias) Allocation concealment (selection bias) Blinding of participants and personnel (performance bias) Blinding of outcome assessment (detection bias) Incomplete outcome data (attrition bias) Selective reporting (reporting bias) Other bias We evaluated the methodological quality of the included studies, using the methods described in Chapter eight of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) Based on these criteria, we subdivided the studies into the following three categories: All quality criteria met: low risk of bias One or more of the quality criteria met only in part: unclear risk of bias One or more criteria not met: high risk of bias We used this classification as the basis of a sensitivity analysis. Additionally, we explored the influence of individual quality criteria in a sensitivity analysis
Subgroup analysis and investigation of heterogeneity	We plan to investigate the effects of different types of devices and different concentrations of oxygen on outcomes	We planned to investigate the effects of different oxygen concentrations, different oxygen flow rates and different oxygen devices on outcomes where appropriate
Sensitivity analysis	We will perform sensitivity analysis to explore the influence of the following factors on effect size: Repeating the analysis, excluding unpublished studies (if there are any) Repeating the analysis, taking account of study quality, as specified above Repeating the analysis, excluding studies using the following filters: diagnostic criteria, language of publication, source of funding (industry versus other) and country Repeating the analysis, both with and without outlying trials We will test the robustness of the results by repeating the analysis using different measures of effect size (risk difference, odds ratio, etc.) and different statistical models (fixed‐effect and random‐effects models)	If appropriate, we planned sensitivity analysis by study quality to investigate the effect of separating high from low risk of bias studies, and for missing data using a best‐case/worst‐case imputation

Footnotes

RCT: randomized control trial
 SIGLE: System for Information on Grey Literature in Europe
 ZETOC: Z Electronic Table of Contents

Data and analyses

Comparison 1. Primary outcomes.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Apgar scores at one minute	6	519	Mean Difference (IV, Random, 95% CI)	0.07 [‐0.16, 0.31]
1.1 Less than 60%	4	356	Mean Difference (IV, Random, 95% CI)	‐0.03 [‐0.19, 0.13]
1.2 At least 60%	3	163	Mean Difference (IV, Random, 95% CI)	0.18 [‐0.26, 0.62]
2 Apgar scores at five minutes	6	519	Mean Difference (IV, Fixed, 95% CI)	‐0.00 [‐0.06, 0.06]

Comparison 2. Secondary outcomes.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mean maternal arterial oxygen tension (PaO2)	6	241	Mean Difference (IV, Random, 95% CI)	141.79 [109.31, 174.27]
2 Mean maternal oxygen saturation	3	209	Mean Difference (IV, Random, 95% CI)	1.58 [0.84, 2.33]
3 Mean oxygen tension in umbilical artery (UaPO2)	8	504	Mean Difference (IV, Random, 95% CI)	3.34 [1.76, 4.91]
3.1 Inspired oxygen less than 60%	6	373	Mean Difference (IV, Random, 95% CI)	2.48 [1.18, 3.78]
3.2 Inspired oxygen at least 60%	4	131	Mean Difference (IV, Random, 95% CI)	4.82 [1.48, 8.16]
4 Mean oxygen tension in umbilical vein (UvPO2)	10		Mean Difference (IV, Random, 95% CI)	Subtotals only
4.1 Inspired oxygen less than 60%	7	445	Mean Difference (IV, Random, 95% CI)	3.95 [1.68, 6.22]
4.2 Inspired oxygen at least 60%	6	238	Mean Difference (IV, Random, 95% CI)	8.19 [3.58, 12.80]
5 Mean pH in umbilical artery (UapH)	8	623	Mean Difference (IV, Fixed, 95% CI)	‐0.00 [‐0.01, 0.00]
6 Mean carbon dioxide tension in umbilical artery (UaPCO2)	8	504	Mean Difference (IV, Random, 95% CI)	1.46 [‐0.51, 3.42]
7 Mean maternal MDA	2	104	Mean Difference (IV, Fixed, 95% CI)	‐0.24 [‐0.37, ‐0.12]
8 Mean maternal isoprostane	2	104	Mean Difference (IV, Fixed, 95% CI)	‐64.27 [‐76.82, ‐51.72]
9 Umbilical artery MDA	2	104	Mean Difference (IV, Fixed, 95% CI)	‐0.09 [‐0.13, ‐0.04]
10 Umbilical artery isoprostane	2	104	Mean Difference (IV, Fixed, 95% CI)	‐67.74 [‐82.53, ‐52.96]
11 Umbilical vein MDA	2	104	Mean Difference (IV, Fixed, 95% CI)	‐0.30 [‐0.37, ‐0.22]
12 Umbilical vein isoprostane	2	104	Mean Difference (IV, Random, 95% CI)	‐186.02 [‐343.48, ‐28.56]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Adenekan 2010.

Methods	Simple randomized trial
Participants	n = 70 Excluded: Women requiring inhalational or intravenous general anaesthesia after subarachnoid block Women with uncontrolled medical diseases Women required emergency caesarean section due to foetal distress
Interventions	Two groups: group 1 received oxygen by face mask 4 L/min, group 2 received room air Anaesthetic technique: Premedication: no Preload: not described Hypotension treated: not described Spinal anaesthesia with a 25‐G or 26‐G spinal needle with 2.2‐2.8 mL of 0.5% hyperbaric bupivacaine
Outcomes	Maternal arterial oxygen saturation (SaO2) prespinal and one minute after induction of spinal anaesthesia Apgar scores at 1, 5 min
Notes	Sample size calculation stated (based on 90% power to detect a difference between two means of 2%, alpha 5%, total 70 participants FIO2 confirmed: no Withdrawal criteria: no
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Simple randomized
Allocation concealment (selection bias)	Low risk	Sealed envelope
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of those receiving care: not described Blinding of those providing care: not described "The study group received supplemental oxygen at 4 L/min from Datex Ohmeda anaesthetic machine via the breathing circuit and a loosely fitting facemask after induction of spinal anaesthesia till the end of surgery and the control group breathed room air throughout the procedure" Comment: probably not done
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of outcome assessors: not described
Incomplete outcome data (attrition bias) All outcomes	High risk	Attrition and exclusions data absent Comment: all participants probably were analysed. None were excluded due to maternal desaturation "The least recorded SaO2 in the study and control groups were 91% and 92% respectively"
Selective reporting (reporting bias)	Low risk	All prespecified outcomes reported
Other bias	High risk	The actual FIO2 of the intervention group was not confirmed. Data were entered as FIO2 of less than 0.6 for analysis The additional supplemental oxygen were not described when maternal desaturation occurred. "The least recorded SaO2 in the study and control groups were 91% and 92% respectively" Sample size calculation stated: "Sample size of seventy was arrived at using the equation for Studies Comparing Two Group Means and a significance level of 0.05, power of 0.9 and minimum expected differences between the two means of 2%"

Castro 2009.

Methods	RCT
Participants	n = 20 Excluded: Maternal diseases with repercussions in foetal oxygenation (pre‐eclampsia, hypertension, pulmonary disease, cardiac disease with shunt, and diabetes mellitus types 1 and 2) Presence of intraoperative intercurrents that compromised foetal oxygenation (hypertension not compensated by the therapeutic manoeuvres described, foetal extraction time > 3 minutes)
Interventions	Two groups: group 1 received FIO2 at least 0.6 via mask with reservoir, group 2 received room air Anaesthetic technique: Premedication: no Preload with lactated Ringer's solution 15 mL/kg Hypotension treated with ephedrine 5‐10 mg Spinal anaesthesia with a pencil‐tip 27‐G spinal needle with 10 mg 0.5% hyperbaric bupivacaine with 60 μg morphine
Outcomes	Maternal ABG (pH, PO2, PCO2, bicarbonate) before and 10 minutes after administration of oxygen Foetal UV (pH, PO2, PCO2, bicarbonate) analysis at the time of delivery Correlation between inspired oxygen, maternal partial oxygen pressure and foetal partial oxygen pressure
Notes	FIO2 confirmed: yes Withdrawal criteria: maternal SpO2 < 92%
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The FIO2 to be administered was determined randomly using a table sealed in an envelope"
Allocation concealment (selection bias)	Low risk	Sealed envelope
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of those receiving care: not described Blinding of those providing care: not described "It was determined by the mixer of the anaesthesia device and the reading was confirmed by the oximeter of the anaesthesia equipment. Oxygen was delivered through a mask with reservoir" Comment: probably not done. Outcomes are not likely to be influenced by lack of blinding of those who provide care
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of outcome assessors: not stated
Incomplete outcome data (attrition bias) All outcomes	High risk	Attrition and exclusions data absent In the methods: the author stated, "Reductions in SpO2 below 92% were treated immediately with supplementary oxygen and the patient was removed from the study". The author did not mention this in the results
Selective reporting (reporting bias)	Unclear risk	"...one (study) received supplementary oxygen using an inspired oxygen fraction (FIO2) equal or greater than 0.6....It was determined by the mixer of the anaesthesia device and the reading was confirmed by the oximeter of the anaesthesia equipment" Comment: The FIO2 the intervention group actually received was not reported. The data were entered as FIO2 of 0.6 for analysis in this review
Other bias	High risk	The FIO2 of the intervention group actually received was not reported. The data were entered as FIO2 of 0.6 for analysis in this review. Sample size calculation not stated

Cogliano 2002.

Methods	RCT
Participants	n = 69 Excluded: age < 18, height < 152 cm and weight > 85 kg at the time of the first antenatal visit, having a history of foetal compromise, having significant coexisting medical conditions, not having English as their first language
Interventions	Three groups: group 1 received oxygen by face mask 4 L/min (40%, manufacturer's data), group 2 received air by face mask and group 3 received oxygen by nasal cannula 2 L/min Anaesthetic technique: Premedication: ranitidine 150 mg orally on the morning of surgery, sodium citrate 0.3 M 30 mL before administration of spinal anaesthesia Preload: not described Spinal anaesthesia: 2.5‐2.75 mL of hyperbaric bupivacaine 0.5% with 0.3 mg diamorphine
Outcomes	Foetal UV and UA (pH, PO2, PCO2) and differences between UvPO2 and UaPO2 at the time of delivery Number of foetuses with Apgar scores < 7 at 1, 5 min Patient's acceptance of oxygen delivery devices regarding comfort, communication, nausea and smell of the gas delivery system
Notes	Theoretically delivered oxygen: Group 1 = 40% Group 2 = 21% Group 3 = 28% FIO2 confirmed: no Withdrawal criteria: maternal blood pressure could not be easily maintained above 100 mm Hg or within 20% of the baseline reading with intermittent doses of ephedrine/or phenylephrine, maternal SpO2 < 93%
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"69 women were recruited and randomly allocated into one of three study groups" Comment: no randomization description
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding of those receiving care: incomplete blinding Blinding of those providing care: incomplete blinding "The administration of oxygen or air by face mask was double‐blinded using a purpose‐built delivery system that was set up before entering theatre by an operating department practitioner. The delivery system consisted of oxygen and medical air supplies combining at a common gas outlet to which the face mask was connected" "...either oxygen or air to be delivered from the common gas outlet without the attending anaesthetist or patient knowing which gas was being delivered" Comment: The participants receiving air or oxygen via face mask can be blinded, the participants receiving oxygen from nasal cannula could not be blinded
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding of outcome assessors: yes (who assessed Apgar scores) "A midwife blinded to group allocation assessed Apgar scores at 1 to 5 min"
Incomplete outcome data (attrition bias) All outcomes	Low risk	"No women had a SpO2 below 94% in any of the groups"
Selective reporting (reporting bias)	Unclear risk	Outcomes were not prespecified
Other bias	High risk	The FIO2 of the intervention group actually received was not reported. The data in a nasal cannula and face mask group were entered as FIO2 of 0.28 and 0.4, respectively, for analysis. Sample size calculation stated (based on 80% power to detect a difference in umbilical venous partial pressure (UvPO2) of 0.5 kPa between the three groups, alpha 5%, 23 participants in each group)

Gunaydin 2011.

Methods	RCT
Participants	n = 90
Interventions	3 groups: group 1 breathed room air, group 2 received nasal cannula 5 L/min and group 3 received face mask 5 L/min Anaesthetic technique: Premedication: metoclopramide 10 mg IV and ranitidine 50 mg IV 30 min before the operation Preload: lactated Ringer's solution 15 mL/kg Spinal anaesthesia: 12 mg of 0.5% hyperbaric bupivacaine with fentanyl 10 microgram and morphine 100 microgram
Outcomes	Foetal UV and UA (pH, PO2, PCO2, base excess) analysis at the time of delivery Foetal Apgar score at 1, 5 min Maternal and neonatal regional cerebral oxygen saturation
Notes	FiO2 confirmed: no Withdrawal criteria: no
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"... parturients were randomly allocated into three groups according to computer‐generated numbers..."
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of those receiving care: not described Blinding of those providing care: no "A junior anaesthesiology resident was asked to record NIRS data, while the staff anaesthesiologist responsible for the study was taking care of the patient" Comment: Outcomes are not likely to be influenced by lack of blinding of those who provide care
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding of outcome assessors: yes "The paediatrician was blinded to the study protocol (either absence or presence of maternal oxygen administration via face mask or nasal cannula)"
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Recruitment and attrition data absent
Selective reporting (reporting bias)	High risk	In the methods: "All maternal parameters were recorded and documented at 0 (baseline control), 1, 3, 5 and 8 min consecutively after administration of the spinal block until delivery (time of umbilical cord clamping)" In the results: Only the maternal SpO2 values at 3 and 5 min after spinal block were reported, and only the mean SpO2 of the facial mask group compared with the control group was reported; the mean SpO2 of the nasal cannula group was omitted "The mean maternal SpO2 recorded 3 and 5 min after administration of spinal block was significantly higher in Group FM (99.5 ± 0.7% and 99.4 ± 0.9%, respectively) than in group RA (98.0 ± 2.2% and 98.4 ± 1.1%, respectively) (P value < 0.001) (data not shown)"
Other bias	Unclear risk	No data to support the following statement: "...supplementary oxygen provided through a face mask at a flow rate of 5 L/min, which was estimated to correspond roughly to the fraction on inspired oxygen (FIO2) ≅ 0.3 ..... oxygen delivery by nasal cannula (FIO2 ≅ 0.32)" Sample size calculation stated: "Power analysis showed that a sample size of 30 parturients in each group would yield 90% power to detect a 10% change in the rSO2 (regional cerebral oxygen saturation) with a type I error of 0.05"

Khaw 2002.

Methods	RCT
Participants	n = 44
Interventions	Two groups: group 1 received air (21%), and group 2 received oxygen (60%) via high‐flow Venturi‐type face mask or air or oxygen delivered from the anaesthetic machine Anaesthetic technique: Premedication: ranitidine 150 mg orally the night before and on the morning of surgery Preload: lactated Ringer's solution 20 mL/kg Spinal anaesthesia: 0.5% hyperbaric bupivacaine 2 mL and fentanyl 15 microgram
Outcomes	Maternal ABG (pH, PO2, PCO2, base excess) before start of anaesthesia and at the time of delivery Foetal UV and UA (pH, PO2, PCO2, base excess) analysis at the time of delivery Foetal Apgar score at 1, 5 min Number of foetuses with Apgar scores < 7 at 1, 5 min Maternal and umbilical lipid peroxide concentrations (isoprostane, malondialdehyde (MDA) and hydroperoxide (OHP)) at 10 min of exposure to oxygen
Notes	FiO2 confirmed: yes Withdrawal criteria: no
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Drawing shuffled coded opaque envelopes “Patients were then randomized, by drawing shuffled coded opaque envelopes”
Allocation concealment (selection bias)	Low risk	Opaque envelope
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Blinding of those receiving care: yes Blinding of those providing care: not described "Air or oxygen was supplied from the anaesthetic machine to a high‐flow Venturi‐type face mask to provide the assigned FIO2" Comment: Outcomes are not likely to be influenced by lack of blinding of those who provide care
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding of outcome assessors: yes (paediatrician who assessed Apgar scores, investigator who performed all blood analyses) "After delivery, a blinded paediatrician assessed Apgar scores" "The investigator who performed all the blood analyses [was] blinded to the FIO2 and did not participate in patient care"
Incomplete outcome data (attrition bias) All outcomes	Low risk	"All participants completed the study"
Selective reporting (reporting bias)	Low risk	All prespecified outcomes reported
Other bias	Low risk	Appears to be free of other sources of bias. In methods, the authors stated, "Our contingency for participants in the air group who developed a pulse oximetry reading of < 94% was to increase the FIO2 to 28%". The authors stated in the results, "All participants completed the study". They did not mention how many participants in the control group have to increase FIO2 to 28%. We received additional information from the authors that no one in the room air group received additional oxygen. Sample size calculation stated: "based on 80% power to detect a difference of 0.05 micromol/L of MDA or OHP, alpha 5%, 22 participants in each group"

Khaw 2004.

Methods	RCT: Comparing between UD < 180 seconds and UD > 180 seconds
Participants	n = 204
Interventions	Three groups: group 1 received room air, group 2 received 40% oxygen and group 3 received 60% oxygen via high‐flow Venturi‐type face mask Anaesthetic technique: Premedication: ranitidine 150 mg orally the night before and on the morning of surgery Preload: lactated Ringer's solution 20 mL/kg Spinal anaesthesia: not described
Outcomes	Foetal UV (PO2, saturation and O2 content) and UapH analysis at the time of delivery Number of foetuses developing foetal acidosis (pH < 7.2) Apgar score at 1, 5 min Number of foetuses with Apgar scores < 7 at 1, 5 min
Notes	FIO2 confirmed: yes Withdrawal criteria: maternal SpO2 < 95%
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Drawing of shuffled opaque sealed envelopes “Patients were allocated randomly, by drawing of shuffled opaque sealed envelopes”
Allocation concealment (selection bias)	Low risk	Sealed envelope
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Blinding of those receiving care: not stated Blinding of those providing care: yes "To maintain blinding of the investigators, an assistant connected the tubing from the flowmeters via an opaque relay box" "Identical management according to a set of predetermined protocols was provided by an anaesthetist who was blinded to the FIO2"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding of outcome assessors: yes (a paediatrician who assessed Apgar scores and the investigator who performed all blood analyses) "A paediatrician who was unaware of group allocation attended each delivery and assessed Apgar scores" "The investigator performing all the blood analyses was blinded to patient allocation and did not participate in patient care"
Incomplete outcome data (attrition bias) All outcomes	Low risk	Recruitment and attrition data presented 212 participants were recruited. 8 participants were excluded: 6 because insufficient umbilical cord blood was obtained, 1 vomited and did not wear the face mask during uterine incision and 1 complained of discomfort from wearing face mask. Intention‐to‐treat analysis was not used
Selective reporting (reporting bias)	Low risk	All prespecified outcomes reported
Other bias	Low risk	Appears to be free of other sources of bias. Sample size calculation stated: "Based on 80% power to detect a 2.6 mL/dL (20%) increase in oxygen content, alpha 5%, 13 participants required for each group"

Palacio 2008.

Methods	RCT
Participants	n = 124 Excluded: maternal illness, uncontrolled pregnancy, gestational age of less than 37 weeks, antenatal diagnosis of foetal malformation, caesarean with risk, suspicion or proof of loss of foetal wellbeing
Interventions	Two groups: group 1 received room air, group 2 received oxygen 6 L/min via face mask (concentration of 40%) Anaesthetic technique: Premedication: not stated Preload: 6% starch hydroxyethyl 10 mL/kg Spinal anaesthesia: 0.06 mg/cm height of 0.5% isobaric bupivacaine with 20 microgram Fentanyl
Outcomes	Maternal oxygen saturation Foetal UV and UA (pH, PO2, PCO2, bicarbonate, base excess, lactate) Number of foetuses with Apgar scores < 7 at 1, 5 min
Notes	FiO2 confirmed: no Withdrawal criteria: desaturation in more than 4 points from baseline , SpO2 < 94%, maternal anxiety or dyspnoea, then supplementary oxygen administered and participants excluded from the study. Six withdrawals were reported: 2 due to coagulation of the samples and 4 due to supplementary oxygen. Investigators did not use intention‐to‐treat methodology
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomization description
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of those receiving care: not described Blinding of those providing care: not described Comment: Outcomes are not likely to be influenced by lack of blinding of those who provided care
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of outcome assessors: not described
Incomplete outcome data (attrition bias) All outcomes	Low risk	Six withdrawals were reported:  2 due to coagulation of the samples and 4 due to supplementary oxygen administration
Selective reporting (reporting bias)	High risk	The neonatal resuscitation was graded in the methodology but in the results stated only no difference in the type of resuscitation. In the methodology stated that: "the type practiced the neonate resuscitation was recorded by the paediatrician. (I: drying and aspiration of oropharyngeal II : mask ventilation III : endotracheal intubation and ventilation with ambu , IV : external cardiac massage and ventilation coordinated with V Drug Administration)" In the result stated that: "No differences in the type of resuscitation practiced the newborn were observed" and "There is also no differences in weight , Apgar , or resuscitation of the newborn" .
Other bias	Unclear risk	Investigators did not use intention‐to‐treat methodology Sample size calculation not stated Excluding 6 are: 2 coagulation of blood gas samples and 4 supplemental oxygen administration, (1 anxiety accompanied by dyspnoea and 3 specifying arterial hypotension administration of ephedrine)

Ramanathan 1982.

Methods	RCT
Participants	n = 40
Interventions	Four groups: control group received FIO2 0.21, group 2 received FIO2 0.47, group 3 received FIO2 0.74 and group 4 received FIO2 1.0 via anaesthesia face mask. The gas mixture was diluted with nitrogen and was delivered through the anaesthetic machine with gas flow of 10 L/min Anaesthetic technique: Premedication: not described Preload: 1200 mL of Hartman's solution Epidural anaesthesia: 18 ± 3 mL of 0.75% bupivacaine
Outcomes	Maternal ABG (pH, PO2, PCO2, base excess, saturation, oxygen content) at the time of delivery Foetal UV and UA (pH, PO2, PCO2, base excess, saturation, oxygen content) analysis at the time of delivery Number of foetuses with Apgar scores < 7 at 1, 5 min
Notes	FIO2 confirmed: yes Withdrawal criteria: no
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"Participants were randomly assigned to one of the following FIO2 groups: 0.21, 0.47, 0.74 or 1.0 (10 participants per group)" Comment: no randomization description
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Blinding of those receiving care: probably blinded Blinding of those providing care: not stated "The gas mixture was delivered from an anaesthesia machine at a fresh gas flow rate of 10 L/min through a circle absorber system. When FIO2 levels of 0.21, 0.47 and 0.74 were studied, the machine oxygen was diluted by nitrogen delivered to the common gas outlet from a calibrated flowmeter. The FIO2 was measured within the face mask....." Comment: participants probably blinded by using face mask for all intervention, staff and research personnel probably unblinded to the intervention. However, outcomes are not likely to be influenced by lack of blinding of those who provide care
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of outcome assessors: not stated
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Recruitment and attrition data absent
Selective reporting (reporting bias)	High risk	"Our study assesses the effects of changes in maternal PaO2 levels ranging between 70 and 490 torr on foetal PO2 levels during caesarean sections performed under epidural anaesthesia" Comment: More than one reported outcome was not prespecified
Other bias	High risk	Some mistakes in the data shown in Table 1 of the study. The author reported SO2 from the foetal artery in the group receiving FIO2 0.21 equal to 26.4 ± 26.4. No comparison of participant characteristics between control and intervention groups. Withdrawal criteria not stated. Sample size calculation not stated

Singh 2006.

Methods	RCT
Participants	n = 60
Interventions	Three groups: group 1 received FIO2 0.21 (breathe atmospheric air), group 2 received FIO2 0.35 and group 3 received FIO2 0.6 via Ventimask Anaesthetic technique: Premedication Preload: lactated Ringer's solution15 mL/kg Spinal anaesthesia: 2 mL of 0.5% hyperbaric bupivacaine
Outcomes	Maternal ABG (pH, PO2, PCO2, base excess) after spinal anaesthesia and at the time of delivery Foetal UV and UA (pH, PO2, PCO2, base excess) analysis at the time of delivery Apgar scores at 1, 5 min Maternal and umbilical lipid peroxide concentrations (8‐isoprostane,malondialdehyde (MDA) and uric acid) The number of episodes of hypotension, bradycardia and desaturation in pregnant women
Notes	FiO2 confirmed: no Withdrawal criteria: no
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was done by draw of chits labelled as 21% oxygen, 35% oxygen or 60% oxygen"
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of those receiving care: no (the control group breathed atmospheric air while the intervention groups received oxygen via Ventimask) Blinding of those providing care: no Comment: participants, staff and research personnel probably unblinded to the intervention. Outcomes are not likely to be influenced by lack of blinding of those who provide care
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of outcome assessors: not stated
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Recruitment and attrition data absent
Selective reporting (reporting bias)	High risk	It was not clear what is meant: "Their effect on foetal outcome following elective caesarean was also assessed". "This study was conducted to find out the effects of different inspired oxygen concentrations (i.e. 21% (air group), 35% and 60%) on maternal and foetal oxygenation and free radical formation. Their effects on foetal outcome following elective caesarean were also assessed" In results: The Apgar score was reported only: "The difference in Apgar scores at 1 minute and 5 minutes in all three groups was statistically insignificant (P = 0.763; 0.533 for 1 minute and 5 minutes, respectively)"
Other bias	High risk	The FIO2 was not confirmed. Sample size calculation was not stated

Young 1980.

Methods	RCT
Participants	n = 32 All were < 35 years Weighed between 45 and 90 kg All had intact membranes No complicating maternal medical or obstetric conditions (e.g. diabetes, anaemia, hypertension, respiratory or renal disease, premature labour, pre‐eclampsia, antepartum bleeding) Foetuses were considered of appropriate size for gestational age
Interventions	Two groups: a control group received room air via face mask, a study group received pure oxygen via oxygen face mask Anaesthetic technique: Premedication: no Preload: lactated Ringer's solution 1 L Epidural anaesthesia: not described
Outcomes	Maternal ABG (pH, PO2, PCO2, HCO3, base excess) before induction of epidural anaesthesia and at the time of delivery Foetal UV and UA (pH, PO2, PCO2, HCO3, base excess, saturation) analysis at the time of delivery Apgar scores at 1 and 5 min
Notes	Theoretically delivered oxygen: Group 1 = 21% Group 2 = over 60% FIO2 confirmed: no Withdrawal criteria: no
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"Forced randomization was carried out so that in each six participants, three would go into either experimental group"
Allocation concealment (selection bias)	Low risk	"The study team determined group assignment by opening a sealed envelope at the time the experiment was commenced"
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Blinding of those receiving care: yes Blinding of those providing care: incomplete blinding (only obstetrician was blinded; blinding of the study team was not described) "Neither the patient nor her obstetrician knew which gas mixture was to be administered" Comment: Outcomes are not likely to be influenced by lack of blinding of those who provide care
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of outcome assessors: not stated ."Apgar scores at 1 and 5 minutes were assigned by the senior neonatal resident"
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Outcomes were not prespecified. Recruitment and attrition data absent
Selective reporting (reporting bias)	Unclear risk	Outcomes were not prespecified
Other bias	High risk	The actual FIO2 of the intervention group was not reported. Data were entered as FIO2 of 1.0 for analysis. Sample size calculation not stated

Yu 1992.

Methods	RCT
Participants	n = 45
Interventions	Group 1 received room air, group 2 received oxygen 6 L/min and group 3 received oxygen 10 L/min via simple face mask Anaesthetic technique: Premedication: no Preload: lactated Ringer's solution 1,500‐2,000 mL Epidural anaesthesia: 18‐20 mL of 2% xylocaine with adrenaline 1:200,000 via epidural catheter
Outcomes	Maternal ABG (pH, PO2, PCO2, HCO3, base excess, saturation) at the time of delivery Foetal UV and UA (pH, PO2, PCO2, HCO3, base excess, saturation) analysis at the time of delivery Number of foetuses with Apgar scores < 7 at 1, 5 min
Notes	Theoretically delivered oxygen: Group 1 = 21% Group 2 = 40% Group 3 = 60% FIO2 confirmed: no Withdrawal criteria: unsatisfactory epidural anaesthesia that necessitated adjuvant drug administration
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"They were allocated randomly to the following 3 groups; breathing room air (group 1, control, breathing through a simple face mask with oxygen inflow of 6 L/min or (group 2) breathing through a simple face mask with oxygen inflow of 10 L/min" Comment: no randomization description and in the results stated: "Forty‐five participants were divided equally into 3 groups in the study"
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding of those receiving care: not stated Blinding of those providing care: not stated Comment: participants, staff and research personnel probably unblinded to the intervention, since group 1 breathing room air, groups 2 and 3 breathing through a simple face mask
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding of outcome assessors: yes (paediatrician who assessed Apgar scores) "The 1‐ and 5‐minute Apgar scores were awarded by a paediatrician unaware of the maternal inspired oxygen concentration"
Incomplete outcome data (attrition bias) All outcomes	High risk	More than one reported outcome was not prespecified. Recruitment and attrition data absent
Selective reporting (reporting bias)	High risk	"We were interested in knowing (1) whether such a method of delivering oxygen to the mother could improve foetal oxygenation in our current practice and (2) whether foetal PO2 could be increased by an increment of oxygen flow to the simple face mask" Comment: More than one reported outcome was not prespecified
Other bias	High risk	No data support the following statement: "In our study, a simple O2 mask was used to deliver oxygen to the patient......The oxygen percentages delivered through the simple mask can approach 35% to 55% when oxygen flow rates of 6 L to 10 L/min are used for adults" The actual FIO2 of the intervention group was not reported. The data were entered as FIO2 of 0.4 (for oxygen 6 L/min) and 0.6 (for oxygen 10 L/min) for analysis. Sample size calculation not stated

ABG: arterial blood gases
 FIO_2: fraction of inspired oxygen
 FM: face mask
 G:gauge
 IV:intravenous
 kg: kilogram

kPa: kiloPascal
 L: litre
 MDA: malondialdehyde
 mg: milligram;
 min: minute

mL: millilitre

mm Hg: millimetre mercury
 n: number (of participants)
 NA: not applicable
 O_2: oxygen
 OHP: hydroperoxide

PCO₂: partial pressure of carbon dioxide
 pH: potential of hydrogen (measurement of acidity)

PO₂: partial pressure of oxygen
 RA: regional anaesthesia
 RCT: randomized control trial
 SaO₂: saturation of haemoglobin

SpO₂: saturation of haemoglobin with oxygen as measured by pulse oximetry
 UA: umbilical artery
 UD: uterine incision‐to‐delivery
 UV: umbilical vein

UaPO₂: umbilical artery oxygen tension

UvPO₂: umbilical vein oxygen tension

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Backe 2002	Included in the subsequent study published in 2007 (Backe 2007)
Backe 2007	The inspired oxygen concentration of a control group varied between 21% and 25%
Bogod 1988	Participants not relevant (general anaesthesia)
Crosby 1992	No control group
Halpern 1990	No control group
Haruta 1984	Not randomized
Hollmen 1978	Intervention not relevant (compares the effects of epidural and general anaesthesia)
Kelly 1995	Included in the subsequent study published in 1996 (Kelly 1996)
Kelly 1996	No reporting of number of participants in each group
Kelly 1997	Not enough detail to be included in the review (abstract for annual meeting of the European Academy of Anesthesiology 1996)
Khaw 2009	Participants not relevant (emergency caesarean section)
Nesterenko 2012	Participants include normal delivery
Okudarira 2005	Intervention not relevant (evaluate the effects of spinal hypotension)
Perreault 1990	Participants not relevant (general anaesthesia)
Perreault 1992	Participants not relevant (general anaesthesia)
Petropoulos 2003	Intervention not relevant (evaluate the effects of various types of anaesthesia)
Reynolds 2005	Not RCT (a meta‐analysis); evaluate effects of types of anaesthesia
Rorke 1968	Participants not relevant (general anaesthesia)
Salwanis 2012	Not enough detail to be included in the review (abstract for 15th WFSA World Congress of Anaesthesiologists 2012)
Siriussawakul 2014	Participants not relevant (spinal block + sedation)
Tonni 2007	Intervention not relevant (evaluate the effects of various types of anaesthesia)

RCT: randomized control trial
 WSFA: World Federation Of Societies of Anaesthesiologists

Characteristics of studies awaiting assessment [ordered by study ID]

Yalcin 2013.

Methods	Randomized controlled trial
Participants	term parturients undergoing elective caesarean section under spinal anesthesia
Interventions	21% vs 40% oxygen
Outcomes	maternal and neonatal oxidative stress, Apgar score 1 and 5 minutes
Notes

Differences between protocol and review

Differences between the published protocol (Chatmongkolchart 2006) and the review are as follows (see details in Appendix 7).

• We changed the contributing authors from Sunisa Chatmongkolchart, Bee B Lee and Thida Uakridathikarn to Sunisa Chatmongkolchart and Sumidtra Prathep.

• We included only published studies.

• We modified the assessment of methodological quality of included studies from four domains based on Moher 2001 to six domains according to Cochrane's domain‐based evaluation tool for Assessment of risk of bias in included studies (Higgins 2011).

• We changed the judgement 'Risk of bias' tool from yes/no to low risk/high risk (Assessment of risk of bias in included studies).

• We performed a sensitivity analysis only on study quality. We did not take the following factors into account: unpublished studies, diagnostic criteria, language of publication, source of funding (industry versus other), country and outlying trials.

• We added the sensitivity analysis of missing data for maternal desaturation.

• We performed the subgroup analysis only on different oxygen concentrations using a cutoff value of 60%; we introduced this cutoff value post hoc.

Contributions of authors

Conceiving the review: Sunisa Chatmongkolchart (SC), Sumidtra Prathep (SP)

Designing the review: SC

Co‐ordinating the review: SC

Undertaking manual searches: none

Screening search results: SC, SP

Organizing retrieval of papers: SC

Screening retrieved papers against inclusion criteria: SC, SP

Appraising quality of papers: SC, SP

Abstracting data from papers: SC, SP

Writing to authors of papers for additional information: SC

Providing additional data about papers: SC 

Obtaining and screening data on unpublished studies: none

Managing data for the review: SC

Entering data into Review Manager (RevMan 2014): SC

Performing double entry of data: (data entered by person one: SC; data entered by person two: SP)

Interpreting data: SC, SP

Making statistical inferences: SC

Writing the review: SC

Serving as guarantor for the review (one author): SC

Person responsible for reading and checking review before submission: SC, SP

Sources of support

Internal sources

• No sources of support supplied

External sources

• Thai Cochrane Network, Thailand.

Declarations of interest

Sunisa Chatmongkolchart: none known

Sumidtra Prathep: none known

New search for studies and content updated (no change to conclusions)