Prophylactic management of postpartum haemorrhage in the third stage of labour: an overview of systematic reviews

Background Postpartum haemorrhage is a direct cause of maternal death worldwide and usually occurs during the third stage of labour. Most women receive some type of prophylactic management, which may include pharmacological or non-pharmacological interventions. The objective of this study was to summarize systematic reviews that assessed the effects of postpartum haemorrhage prophylactic management during the third stage of labour. Methods We applied the guidelines for conducting an overview of reviews from the Cochrane Handbook for Systematic Reviews of Interventions. We searched MEDLINE, EMBASE, and the Cochrane Database of Systematic Reviews to identify all relevant systematic reviews of randomized controlled trials of prophylactic management of postpartum haemorrhage in the third stage of labour compared with no treatment, placebo, or another management technique. Two review authors independently extracted data and assessed methodological quality using a measurement tool to assess reviews and quality of evidence using the Grades of Recommendation, Assessment, Development, and Evaluation for primary outcomes, summarizing results narratively. Results We identified 29 systematic reviews: 18 Cochrane and 11 non-Cochrane. Cochrane systematic reviews were high quality, while the quality of non-Cochrane systematic reviews varied. The following techniques suggested effective, third-stage interventions to reduce the incidence of severe postpartum haemorrhage: active management of the third stage of labour compared to physiological management, active management compared to expectant management, administration of oxytocin compared to placebo, and use of tranexamic acid compared to placebo. The following third-stage management approaches reduced the need for blood transfusion: active management compared to physiological management, active management compared to expectant management, oral misoprostol compared to placebo, and tranexamic acid compared to placebo. Conclusions No effective prophylactic management techniques were identified for maternal mortality. Most methods of effective prophylactic management of postpartum haemorrhage were supported by evidence; however, they were limited to low- or moderate-quality evidence, and high-quality studies are therefore needed. Outcome measures of the included systematic reviews varied. It is recommended that outcome measures in preventive postpartum haemorrhage intervention trials align with the World Health Organization guidelines. Systematic review registration PROSPERO: CRD42016049220. Electronic supplementary material The online version of this article (10.1186/s13643-018-0817-3) contains supplementary material, which is available to authorized users.


Description of the condition
Maternal mortality is an indicator of women's health [1]. The primary cause of maternal death in both developing and developed countries is postpartum haemorrhage (PPH) [2,3], representing 19.7% of maternal deaths worldwide [4]. PPH is defined as blood loss ≥ 500 mL within the first 24 h after delivery [5]. The prevalence of PPH (blood loss ≥ 500 mL) is approximately 6.0-10.0%, and the prevalence of severe PPH (blood loss ≥ 1000 mL) is approximately 1.8-3.0% in any type of delivery but varies by region in the world [6]. Representing a significant medical threat worldwide, effective strategies for the prevention of PPH are essential to decrease maternal mortality rates.

Description of the intervention
Given that PPH usually occurs during and after the third stage of labour [6], primary guidelines recommend active management of this stage. The main component of effective PPH prophylactic management is the administration of uterotonics [5,10,[18][19][20]. Active management is a set of prophylactic interventions consisting of the following components: administration of uterotonics after delivery, early umbilical cord clamping, controlled cord traction for earlier delivery of the placenta, and in certain cases, uterine massage [5].
In contrast with active management, expectant and physiological managements are "hands-off" techniques involving no administration of prophylactic uterotonic agents and delivery of the placenta only through maternal efforts [5]. The World Health Organization (WHO) guideline for preventing PPH recommend the following interventions: use of uterotonics during the third stage of labour for all births, use of oxytocin (10 IU) as the uterotonic drug, controlled umbilical cord clamping in settings where skilled birth attendants are available, and late cord clamping [5]. Moreover, the National Institute for Health and Care Excellence's guidelines were revised in 2014, including a change in the definition of active management of the third stage of labour. The ideal timing of cord clamping, as one component of active management, was changed from early to late, as evidence indicated that late cord clamping did not negatively impact maternal outcomes and had benefits for the neonate [18].
The effectiveness of some prophylactic management techniques for PPH in the third stage of labour has been evaluated in Cochrane systematic reviews. Active management of the third stage of labour was evaluated in comparison with expectant management [21]. Early cord clamping plus controlled cord traction, as one component of active management, is believed to prevent retained placenta and a prolonged third stage of labour. The effectiveness of the timing of cord clamping [22] and controlled cord traction [23] has been recently evaluated.
Other prophylactic interventions were considered in this overview, including use of uterotonic drugs, use of hemostatic agents and uterine massage. For uterotonic drugs, the following were considered as representative methods of augmenting uterine contractions: oxytocin [24][25][26][27][28][29], prostaglandin [30][31][32], and ergot alkaloid [33]. Oxytocin is a naturally occurring hormone that stimulates uterine contractions [34] and is commonly used as a uterotonic. The half-life of oxytocin is short (4-7 min) [35]; therefore, both repeated doses and continuous infusion are acceptable [20]. Prostaglandin is also a naturally occurring hormone; misoprostol, a prostaglandin E1 analogue, can be used orally, sublingually, vaginally, or rectally [36]. Furthermore, misoprostol has mild side effects, such as shivering and pyrexia [37]. Ergot alkaloids act to contract the myometrium through calcium channel mechanisms; however, this also increases the incidence of side effects such as hypertension [38]. A survey examining the use of prophylactic uterotonic agents in 28 countries noted that 95.3% of deliveries used prophylactic uterotonics for the prevention of PPH, and the most commonly used uterotonic agent was oxytocin [13].

Why it is important to do this review
Many prophylactic management techniques for PPH in the third stage of labour have been evaluated in systematic reviews of randomized controlled trials (RCTs). The extant systematic reviews of evidence from RCTs regarding prophylactic management of PPH in the third stage of labour have never been summarized. Furthermore, many clinical guidelines for preventing PPH [18][19][20] were published and are reflected in these systematic reviews. This overview will allow the many readers (such as clinicians, midwives, policy makers, and consumers) to quickly assess a range of evidence about prophylactic management techniques for PPH and utilize this information for making decisions. Regarding the application to research, through summarizing the effectiveness of interventions by outcomes, this overview will also provide a set of outcome measures that are clinically meaningful that can be applied to future studies.

Objectives
The objective of this overview was to: provide a narrative summary of systematic reviews of RCTs; provide the effectiveness of prophylactic management of PPH during the third stage of labour of any type of delivery (vaginal or caesarean section) in terms of outcome measures including maternal mortality, blood loss greater than 1000 mL, and use of blood transfusion.

Methods
In this overview, we applied the guidelines for conducting an overview of reviews from the Cochrane Handbook for Systematic Reviews of Interventions [39] and adhered to the systematic reporting guidelines of the preferred reporting items for systematic reviews and meta-analysis (PRISMA) statement [40]. The PRISMA checklist is shown in Additional file 1. Our review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42016049220).

Criteria for considering included review
Type of reviews In this overview, we included published systematic reviews of RCTs in which prophylactic management of PPH in the third stage of labour was administered after delivery. We excluded reviews that were not systematic reviews of RCTs and those that were only abstracts. When the identified Cochrane review was an updated review with a previous version, we also excluded the previous version and only included the updated version.

Type of participants
The review subjects were women who delivered vaginally or by caesarean section. We recognize that the risk of PPH for vaginal delivery or caesarean section may vary, but we included both delivery methods because some of the reviews presented their results regardless of the mode of delivery.

Type of intervention and comparisons
We included any prophylactic managements of PPH in the third stage of labour and divided them into the following subgroups.

Pharmacological interventions
Active management of the third stage of labour Oxytocin Prostaglandin Ergot alkaloids Tranexamic acid

Non-pharmacological interventions
Early umbilical cord clamping Controlled cord traction Uterine massage We excluded non-prophylactic management. We compared these interventions with placebo, no treatment, contrasting interventions, or other interventions.

Type of outcomes
We searched for the three critical outcomes proposed in the WHO's recommendations for the prevention and treatment of postpartum haemorrhage guidelines [5].
Maternal mortality Blood loss greater than 1000 mL Use of blood transfusion

Search strategy
A comprehensive search was conducted for relevant reviews published in any language in MEDLINE (via EBSCO, 11 October 2016), EMBASE (1980 to 11 October 2016), Cochrane Database of Systematic Reviews (issue 10 of 12 October 2016), and Database of Abstracts of Reviews of Effect (Cochrane library issue 2 of 4 April 2015), using the search terms "postpartum haemorrhage" and "prevention." Systematic review search filters in clinical evidence [41] were used to search MEDLINE and EMBASE. The search strategy is detailed in Additional file 2.

Selection of systematic reviews
Two authors (YY, KF) independently assessed all potential systematic reviews that resulted from our search strategy for inclusion in the present review. We resolved any disagreement through discussion and/or by consulting the third author (YK).

Data extraction
A predefined form was used for data extraction, which included the following sections: study design, participants, experimental intervention, comparison intervention, outcomes, quality of the review, and pooled effect sizes for main outcome measures in metanalyses. When an included review did not identify the number of outcome events or was not meta-analysed, we verified the primary sources included in the review and then performed meta-analysis ourselves using Review Manager 5.3 [42]. We assessed each study's statistical heterogeneity using I 2 statistics. Where heterogeneity was observed (I 2 > 50%), we conducted a random-effects model for combining data. A fixed-effect model was used if the heterogeneity indicated non-importance (I 2 < 50%). We presented a risk ratio with 95% confidence intervals (CI) about dichotomous data.
Two review authors (YM, SI) independently extracted data from the reviews using the data extraction form. We entered data into the Review Manager software 5.3 [42] and GRADEpro GDT software [43] and verified accuracy.

Quality of evidence in included reviews
Quality of evidence in the included reviews was examined using the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) approach [44] for outcomes. The GRADE approach is a system that evaluates quality of evidence, which is assessed on a 4-point scale ("very low," "low," "moderate," and "high") in five domains: (1) study limitations, (2) inconsistency of results, (3) indirectness of evidence, (4) imprecision, and (5) publication bias [44]. We assessed "study limitation" by using the approach of the Cochrane risk of bias tool [39]. If the original study did not assess the risk of bias, we assessed the risk of bias on low, high, or unclear for: (1) selection biases, (2) performance bias, (3) detection bias, (4) attrition bias, (5) reporting bias, and (6) other potential sources of bias [39]. When we assessed the domain of the "imprecision" of the GRADE approach, we focused on the 95% CI around the difference in effect between intervention and comparison. We concluded that the imprecision was not serious when: (1) the 95% CI did not include the RR of 1.0 and included appreciable benefit or harm (RR of under 0.75 or over 1.25); and (2) the total event number and the optimal information size were enough [45]. We used the GRADEpro GDT software's [43] "summary of findings" tables for each outcome.

Assessment of methodological quality of included reviews
The review authors (YM, KF, YK) independently assessed the quality of evidence and methodological quality of the included reviews using the instrument: a measurement tool to assess reviews (AMSTAR) [46]. The AMSTAR tool [46] consists of 11 criteria for measuring the methodological quality of systematic reviews, which is determined by a questionnaire with 11 items that asks reviewers to answer "yes," "no," "cannot answer," or "not applicable." We resolved discrepancies through discussion.

Data synthesis
We provided a narrative summary of the individual review results for each outcome displayed in tables and figures that included characteristics of each review, AMSTAR ratings and outcomes using GRADE. Although we planned to present the data divided by the mode of delivery, almost all original reviews including vaginal and caesarean section deliveries presented the data together, not separately. Therefore, we did not provide the review result for each outcome divided by the delivery mode. Figure 1 is a flow diagram of the selection process. A total of 291 studies were identified from the database search. After removing duplicates, 171 studies remained. A total of 135 titles and abstracts were excluded because they were not systematic reviews or did not examine prophylactic management of PPH. There were 46 full-text studies remaining; of these, 17 were excluded because they did not include prophylactic management, were not systematic reviews of RCTs, or were only abstracts. The excluded studies list is detailed in Additional file 3. A final sample of 29 studies met the inclusion criteria.

Quality of included reviews
AMSTAR [46] ratings representing the quality of systematic reviews are displayed in Table 2. The methodological quality of 16 Cochrane systematic reviews was high, with scores ranging from 10 to 11. The 10 non-Cochrane systematic review scores varied from 1 to 7. The majority of non-Cochrane reviews did not list the included and excluded studies and/or did not consider the quality of included studies.

Quality of evidence in included reviews
Quality of evidence of the included reviews, as measured using the GRADE approach [44], varied by review and is displayed in Tables 3, 4, and 5. The risk of biased scores or imprecision was serious; therefore, the quality of evidence was low to moderate. For the rating of limitations of GRADE approach [44], we provided the result of assessing the risk of bias of the included reviews which did not provide the risk of bias for the included RCTs [48,[52][53][54] in Additional file 4.

Effect of interventions for maternal mortality
We identified seven reviews that examined maternal mortality: two regarding active management of the third stage of labour [23,47], one examining the use of oxytocin [51], three examining the use of prostaglandins [30,32,52], and one examining nipple stimulation [62]. Table 3 portrays the effects of these management techniques.

Oxytocin versus no treatment
There were no cases reporting maternal deaths related to oxytocin injection. Therefore, the effect of an oxytocin injection (10 IU) in the thigh compared to no treatment could not be estimated [51].

Nipple stimulation versus no treatment
In the comparison of breastfeeding immediately after delivery versus no breastfeeding or nipple stimulation, no significant difference was found between the groups            Serious inconsistency, serious imprecision GRADE working group grades of evidence High quality: We are very confident that the true effect lies close to that of the estimate of the effect Moderate quality: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low quality: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect Very low quality: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect CI Confidence interval, RR risk ratio *The risk in the intervention group (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)   [62].

Tranexamic acid versus placebo
Tranexamic acid significantly reduced the incidence of severe PPH compared to placebo or no treatment ( [61].

Tranexamic acid versus placebo
Tranexamic acid significantly reduced the incidence of blood transfusion compared to placebo: (

Summary of main results
In this overview, we appraised and summarized the evidence from 29 systematic reviews that assessed the effectiveness of prophylactic management of PPH in the third stage of labour, including caesarean sections. We then summarized review results based on three outcomes: (1) maternal mortality, (2) blood loss greater than 1000 mL, and (3) use of blood transfusion.

Maternal mortality
There were no effective interventions which we identified. Active management of the third stage of labour with or without controlled cord traction [23,47], nipple stimulation [62], and prostaglandins [30,32,52] were the only interventions that assessed this outcome.

Blood loss greater than 1000 mL
A few effective interventions were identified: active management of the third stage of labour rather than physiological or expectant management [21,48], administration of oxytocin (5-10 IU) [29], and tranexamic acid versus placebo [56,59,60]. A lack of evidence was identified (i.e., no studies found) for the following interventions: fundal pressure versus controlled cord traction as part of the active management of the third stage of labour [49] and intramuscular versus intravenous oxytocin [26].

Use of blood transfusion
The following were effective interventions resulting in reduced blood transfusions: active management of the third stage of labour rather than physiological management [48], active management of the third stage of labour rather than expectant management [21], oral misoprostol (400-600 μg) compared to placebo [32], and tranexamic acid compared to placebo [58][59][60].

Overall completeness and applicability of evidence
This overview systematically summarized 29 systematic reviews of eight different methods that included pharmacological and non-pharmacological managements; however, this overview had several limitations. We only focused on interventions that compared placebo or contrasting management. In order to make results applicable to practice, broad-based pharmacological interventions compared to other pharmacological interventions should have been included in our main analysis. Maternal mortality was reported in a few reviews [23,30,32,47,52,62] ( Table 3). Because PPH is one of the leading causes of maternal death, a greater number of reviews examining various interventions are needed to apply findings to clinical settings. Other outcomes had large amounts of data from numerous trials, but most reviews had serious flaws due to high heterogeneity or few outcome events (Tables 4 and 5). This review did not perform sub-group analysis for the delivery mode or setting. Study participants in the included reviews had undergone vaginal delivery or caesarean section, yet none of the reviews identified risk to participants or examined labour interventions. In this overview, we did not show participant details, settings, or interventions. As such, clinicians who wish to apply the evidence from this review to clinical settings should do so cautiously.

Quality of the evidence
Using the AMSTAR tool [46], we found that the quality of the Cochrane systematic reviews was high; however, non-Cochrane systematic reviews varied from low to high. Only three of 11 non-Cochrane systematic reviews provided the list of included and excluded studies. Providing the list of all studies which appear to meet the inclusion criteria could reduce the risk of publication bias. Most of the systematic reviews in which the quality of evidence was low needed to include a comprehensive research analysis and should have provided details about publication bias.

Potential biases in the overview process
For this overview, we adopted the method outlined in the Cochrane systematic reviews of interventions [39], which minimized the potential bias introduced at all stages in the review process.

Agreements and disagreements with other studies or reviews
WHO has published guidelines for the prevention and treatment of PPH [5]. This document assessed numerous systematic reviews and RCTs that were included in the present overview. The intrapartum guideline, which was published by the National Institute for Health and Care Excellence, assessed active management of the third stage of labour [18]. Several RCTs were included in the NICE guideline despite the omission of a Cochrane systematic review. However, in this overview, we were able to include a broader spectrum of published systematic reviews.

Conclusions
No effective prophylactic management of maternal mortality was identified. Most methods of effective prophylactic management of PPH were supported by evidence; however, they were limited to low-or moderate-quality evidence. Higher quality studies are therefore needed. Study participants had undergone vaginal delivery or caesarean section, and their risks and presence or absence of labour interventions were unclear. Therefore, when these prophylactic strategies are used, the state of participants and access to medical care should be considered.
The critical outcome measures about prevention and treatment of PPH are proposed in the WHO guidelines [5]. However, the systematic reviews assessed herein had a variety of outcome measures, as did the individual trials, reducing our ability to compare the results. For example, the outcome regarding the proportion of PPH had several cut-points (including ≥ 300, 400, and 500 mL). This made it difficult to pool results and show all the evidence from similar trials. It is therefore recommended that trials examining preventive interventions for PPH use consistent outcome measures and those that are recommended in the WHO guidelines [5].