Facility-based active management of the third stage of labour: assessment of quality in six countries in sub-Saharan Africa

Abstract Objective To assess the quality of facility-based active management of the third stage of labour in Ethiopia, Kenya, Madagascar, Mozambique, Rwanda and the United Republic of Tanzania. Methods Between 2009 and 2012, using a cross-sectional design, 2317 women in 390 health facilities were directly observed during the third stage of labour. Observers recorded the use of uterotonic medicines, controlled cord traction and uterine massage. Facility infrastructure and supplies needed for active management were audited and relevant guidelines reviewed. Findings Most (94%; 2173) of the women observed were given oxytocin (2043) or another uterotonic (130). The frequencies of controlled cord traction and uterine massage and the timing of uterotonic administration showed considerable between-country variation. Of the women given a uterotonic, 1640 (76%) received it within three minutes of the birth. Uterotonics and related supplies were generally available onsite. Although all of the study countries had national policies and/or guidelines that supported the active management of the third stage of labour, the presence of guidelines in facilities varied across countries and only 377 (36%) of 1037 investigated providers had received relevant training in the previous three years. Conclusion In the study countries, quality and coverage of the active management of the third stage of labour were high. However, to improve active management, there needs to be more research on optimizing the timing of uterotonic administration. Training on the use of new clinical guidelines and implementation research on the best methods to update such training are also needed.


Introduction
Haemorrhage is estimated to cause 27.1% of the 287 000 maternal deaths that occur annually. 1 Postpartum haemorrhage can be prevented by the active management of the third stage of labour -an intervention that can reduce maternal blood loss by up to 66% compared with physiological or expectant management. [1][2][3] While the annual numbers of maternal deaths attributable to haemorrhage fell sharply between 1990 and 2013, postpartum haemorrhage continues to be the global leading cause of maternal death. 4 The problem does not appear to be a lack of effective interventions but rather the failure to implement such interventions properly in all settings. 1 Maternal care has traditionally been tracked by two key indicators: the proportion of births attended by skilled birth attendants and antenatal care coverage. 5 However, these two indicators may not reflect the content or quality of the care available. 6 For example, the presence of skilled birth attendants does not guarantee that appropriate interventions are correctly implemented at appropriate times. A recent assessment identified 18 quality-of-care indicators for evaluating facility-based deliveries, including the "proportion of women who are administered uterotonics in the third stage of labour. " 7 Recommendations for specific actions that make up the active management of the third stage of labour have evolved with research. Since 2003, these recommendations have resulted in several attempts to define the essential components of such management (Table 1). In a recent multicentre trial led by the World Health Organization (WHO), it was suggested that use of a uterotonic alone may suffice to prevent postpartum haemorrhage and that "omission of CCT [controlled cord traction] has very little effect on the risk of severe haemorrhage. " 12 In 2012, based on these findings, WHO issued revised recommendations that emphasized the use of a uterotonic, suggested that controlled cord traction should be optional -and only ever implemented by a skilled birth attendant -and did not recommend the use of sustained uterine massage. 11 Delayed cord clamping, which appears to benefit the neonate, is also now recommended. 13,14 There have been few reports on the coverage and quality of the active management of the third stage of labour in developing countries. In a global survey it was found that only 16 (43%) of 37 countries investigated included administration of a uterotonic and/or the active management of the third stage of labour in their national health management information systems. 15 Often, any quality indicators relating to postpartum haemorrhage prevention are monitored non-systematically at subnational level and then only in the context of specific projects. A study done in seven countries in [2005][2006] reported that the active management of the third stage of labour was only implemented correctly in 0.5-32% of the deliveries observed. 16 No study since has had a similar size and scope and used observation to assess such management.
To provide a baseline for future measurement and inform policy and programme interventions, we assessed the quality Objective To assess the quality of facility-based active management of the third stage of labour in Ethiopia, Kenya, Madagascar, Mozambique, Rwanda and the United Republic of Tanzania. Methods Between 2009 and 2012, using a cross-sectional design, 2317 women in 390 health facilities were directly observed during the third stage of labour. Observers recorded the use of uterotonic medicines, controlled cord traction and uterine massage. Facility infrastructure and supplies needed for active management were audited and relevant guidelines reviewed. Findings Most (94%; 2173) of the women observed were given oxytocin (2043) or another uterotonic (130). The frequencies of controlled cord traction and uterine massage and the timing of uterotonic administration showed considerable between-country variation. Of the women given a uterotonic, 1640 (76%) received it within three minutes of the birth. Uterotonics and related supplies were generally available onsite. Although all of the study countries had national policies and/or guidelines that supported the active management of the third stage of labour, the presence of guidelines in facilities varied across countries and only 377 (36%) of 1037 investigated providers had received relevant training in the previous three years. Conclusion In the study countries, quality and coverage of the active management of the third stage of labour were high. However, to improve active management, there needs to be more research on optimizing the timing of uterotonic administration. Training on the use of new clinical guidelines and implementation research on the best methods to update such training are also needed.
Active management of third stage of labour in sub-Saharan Africa Linda Bartlett et al.
and coverage of the active management of the third stage of labour in facilitybased deliveries in six countries in sub-Saharan Africa. We investigated the separate components of such management -focusing on uterotonic provision to reflect the most recent research and guidelines. The relevant national policies -if any -and the availability of the various commodities needed for such management were also assessed.

Study design
With a cross-sectional design, we used direct observation of facility-based labour and delivery to assess quality of care in normal delivery practice and the management of selected complications during active management of the third stage of labour. For each of our six study countries, a routine checklist for the clinical observation of labour and delivery (available from the corresponding author) was adapted from a previous study 16 and partly based on the Managing complications in pregnancy and childbirth: a guide for midwives and doctors manual. 17 There were only minor differences between the six checklists: each was piloted during the training of the data collectors. Lessons from the first two countries where the survey was implemented -i.e. Ethiopia and Kenya -helped refine the tools used elsewhere.
In each study facility, we audited the infrastructure and supplies needed and reviewed whether national policies and/ or practice guidelines supported the active management of the third stage of labour. Providers were interviewed and tested on their knowledge of maternity care. In five of our study countries, data were collected, using customized forms, on smartphones or tablet computers. In Kenya, however, data were recorded on paper.
Our data collectors were midwives and doctors who were currently in clinical practice. Clinical refresher training was offered before the collectors were trained as observers. The latter training included four days in a classroom followed by one or two days of pretesting the data collection form -in all the study countries except Kenya -on smartphones or tablets. In role-play simulations based on the MamaNatalie and NeoNatalie models (Laerdal, Stavanger, Norway), trainees assumed the roles of observer, client and health-care provider and practised using the checklists for uncomplicated and complicated births. At the end of the training, data collectors also visited a nearby non-study facility to practise using the checklist in the field.

Study setting
The data for this study were collected, between 2009 and 2012, in surveys in Ethiopia, Kenya, Madagascar, Mozambique, Rwanda and the United Republic of Tanzania (Table 2). Each survey, which took two to four weeks to complete, was supported by the United States Agency for International Development via the Maternal and Child Health Integrated Program and facilitated by staff at the programme's headquarters in Washington, United States of America, the programme's country office in each study country and the six corresponding ministries of health. At the time of sur vey implementation, the maternal mortality ratio, in deaths per 100 000 live births, ranged from 440 in Madagascar to 790 in the United Republic of Tanzania. In five of our six study countries, approximately 35-55% of women gave birth in facilities and nearly all pregnant women made at least one visit to an antenatal care clinic. Ethiopia had the lowest percentages of facility-based births (10%) and of pregnant women receiving antenatal care at least once (34%). 18

Participants
Women were approached as they arrived at the labour and delivery ward, received a description of the study by the observer and those that consented to participate were followed. There were up to three women per observer and several observers per facility. If a woman who came in had a complication -such as pre-eclampsia -or if she developed a complication during labour, she would be prioritized for observation.  Research Active management of third stage of labour in sub-Saharan Africa Linda Bartlett et al.
Overall, 2689 women consented to observation and 2317 of these women were observed during the third stage of labour and therefore included in our final analysis (Table 2). Although 643 health facilities were visited, the number visited in each study country varied widely -from 19 in Ethiopia to 409 in Kenya (Table 2). Only the 390 visited facilities where labour and delivery were observed were included in the final analysis. The other 253 either did not offer labour and delivery services or had no clients during the observation period.

Study size
All samples, except that of Tanzania, were believed to be nationally representative of facilities with at least moderately high utilization (Table 3). 20 In Kenya, the survey was implemented within a national Service Provision Assessment run by ICF Macro (Calverton, USA). Ethiopia's sample was limited to hospitals with at least five deliveries per day. In Madagascar, the sample included all facilities with at least two deliveries daily. Rwanda's survey was a census of district and referral hospitals and a random selection of district health centres. The two surveys in the United Republic of Tanzania were planned to serve as the baseline and endline of a quality improvement project run by the Maternal and Child Health Integrated Program and only included facilities in project regions.

Variables
At the time that our study was conceived in 2008, the International Federation of Gynaecology and Obstetrics/International Confederation of Midwives' definition of the active management of the third stage of labour was still widely used. This definition includes uterotonic administration within a minute of the birth, controlled cord traction and uterine massage. 8 We collected data on each of these components and also on the components of the relaxed definition 16 that included uterotonic administration within three minutes of the birth. 9,21 The type of uterotonic administered -if any -was recorded. Variables were created based on "yes" or "no" responses to checklist items. Any "do not know" responses were excluded. Analyses of the timing of uterotonic administration were based on observers' recordings of the times. If not recorded, the timing of administration was assumed to have been more than three minutes after the birth. Kenyan observers estimated the timing of administration as at delivery of the anterior shoulder, within a minute of the baby's delivery or after placental delivery.

Statistical analysis
The data for each study country were analysed separately. Post-stratification weights were applied to the observations to account for differences between the numbers of observed and expected deliveries at each facility. Weights were based on the relevant national health management information systems or facility registers. For each study country, descriptive statistics were generated separately for each investigated component of the active management of the third stage of labour and for the combination of all such components.
Facilities were assessed for the presence of at least one non-expired dose of oxytocin, ergometrine or misoprostol Active management of third stage of labour in sub-Saharan Africa Linda Bartlett et al.
that was onsite -i.e. in the delivery room or a neighbouring room. Such drugs were recorded as "not present" if the observer did not personally see a dose.

Ethical considerations
The study protocol was approved by ethical review boards in each country and by the Johns Hopkins Bloomberg School of Public Health, which ruled that the protocol was exempt from review under the United States Code of Federal Regulations, 45 CFR 46.101(b) (5). Informed consent was obtained from all study participants, including facility directors, health workers and patients.

Results
Providers with nurse or midwifery training performed most of the observed deliveries in each study country (Table 4).
In the knowledge test, 440 (42%) of the 1037 providers investigated indicated that, in the previous three years, they had received pre-service or in-service training in delivery care but only 377 (36%) said that they had received training in the active management of the third stage of labour (Table 5).
Data on the availability of a uterotonic in the delivery room were missing for 12 of the 390 facilities included in the final analysis. Of the remaining 378 facilities, 344 (91%) and 329 (87%) had at least one uterotonic and oxytocin available in the delivery room, respectively. Only 41 (75%) of the 55 Tanzanian facilities included in the final analysis had oxytocin available onsite -with more hospitals stocking the drug than health centres (Fig. 1). The syringes and needles needed to administer oxytocin were available in almost all facilities. Availability of ergometrine and misoprostol varied widely. Of the 378 facilities, 166 (44%) -including only four (22%) of the 18 Ethiopian facilities -displayed clinical guidelines for a normal deliver y, that included the provision of active management of the third stage of labour, either on a wall or in another easily visible location.
For routine deliveries, each study country included the active management of the third stage of labour -including all components in the International Federation of Gynaecolog y and Obstetrics/International Confederation of Midwives definition 21 and oxytocin as the preferred uterotonic -in its service delivery guidelines. In each country's essential drug list, oxytocin was registered and indicated for use in the active management of the third stage of labour. All of the relevant national policies noted that any provider who was considered to be a skilled birth attendant was eligible to administer uterotonics.

Individual management components
In the 2317 deliveries observed, uterotonic administration was nearly universal (Table 6). Oxytocin was the most frequently used uterotonic. Among the study countries, Kenya demonstrated the highest frequency of controlled cord traction and uterine massage. Of the 2173 women given a uterotonic at any time, 1640 (76%) received it within three minutes of the birth. However, in only 1124 (52%) of the 2173 women given a uterotonic was it administered within a minute of the birth.
Fifty of the women observed developed postpartum haemorrhage and all but one of these 50 women had been given oxytocin. The other woman had not received any uterotonic.

Research
Active management of third stage of labour in sub-Saharan Africa Linda Bartlett et al.

Discussion
In all six of our study countries, the quality and coverage of the active management of the third stage of labour were high. The practice of at least one component of such active management was nearly universal. Uterotonic administration was the most frequently observed component and is generally considered to be the most important. 11 However, there was wide variation among the study countries in the use of controlled cord traction, uterine massage and the timing of uterotonic administration.
Encouragingly, skilled birth attendants conducted almost all of the observed deliveries, uterotonics and other related supplies were usually present onsite and all of the study countries had national policies or guidelines for the active management of the third stage of labour. However, the surveys revealed a low frequency of provider training in active management during the previous three years and the frequent unavailability in delivery rooms of relevant guidelines.
In our study, almost as many women received a uterotonic more than one minute after the birth as within a minute of the birth. Confusingly, there are many differing recommendations on when a uterotonic should be administered. A review of active versus expectant management for women in the third stage of labour, found six recommendations, including "at the deliver y of the anterior shoulder", "immediately following birth" and "within two minutes of birth". 3 The International Federation of Gynaecolog y and Obstetrics/ Inter nat iona l C onfe der at ion of Midwives definition 8 recommended "within one minute" -whereas the 2007 22 and 2009 10 WHO guidelines recommended "soon after birth of the baby". The most recent -i.e. 2012 -WHO guidelines simply recommended "during the third stage of labour". 11 The need for further information on the optimal timing of uterotonic administration has been identified in almost all of the relevant WHO guidelines, trial reports and Cochrane reviews since 2007. 3,11,12,[22][23][24][25] However, neither in a five-country assessment of the impact of all components of the active management of the third stage of labour 26 nor in an eight-country assessment of such active management with and without controlled cord traction 12 was the timing of uterotonic administration discussed.
C o n f u s i o n o v e r c h a n g i n g definitions and guidelines is a barrier to optimal implementation of the active management of the third stage of labour. Studies from Colombia, Ghana and the United Republic of Tanzania have concluded that the lack of uniformity in definitions may contribute to the creation of barriers to effective dissemination of knowledge, consistent training, and implementation of clinical guidelines in practice. 27,28,9 Many health facilities in low-resource countries are under-staffed so that a single provider may need to manage several deliveries concurrently and may be unable to provide all of the recommended interventions at the recommended times -even when the necessary supplies are available. 29,30 Given the current focus on uterotonic use, future research and guidelines should define the upper and lower time-limits for uterotonic administration to prevent postpartum haemorrhage.
The presence of confusing guidelines, low provision of training and lack of monitoring of content have previously been identified as barriers to optimal implementation of the active management of the third stage of labour. 16 In 2012, it was observed that the providers of active management need improved educational and training opportunities. 15 A multifactorial intervention -using clinical leaders, clear service delivery guidelines, regular reviews and supportive materials -could improve the implementation of active management. 31 The development of appropriate standards and guidelines and clinical audits could promote a so-called culture of quality throughout a country's health facilities and systems. 32 The active management of the third stage of labour in Ethiopia and the United Republic of Tanzania has been assessed in 2005-2006. 16 We also surveyed these two countries in 2010. Comparisons between the data indicate that progress has been made in both countries. However, sampling differences and changing definitions mean that such comparisons have to be handled with care. Since 2005, both countries have developed their first national policies and guidelines for the prevention of postpartum haemorrhage. Over the same period, the percentage of oxytocin use increased from 68% to 98%. The use of direct observationwhich remains rare in the assessment of obstetric quality of care -may be considered a strength of this study. However, it also allows potential bias. Observers' judgments -even if standardized through training and assessed using inter-rater reliability measures -may not be correct. Further, the observer's presence may have stimulated improvements in the performance of the observed provider. 33 The surveys were limited to observing care practices for facility-based deliveries only and do not provide data on home births. In a recent study of uterotonic use after delivery that included both facilities and homes, it was estimated that only 40% of Tanzanian women received a uterotonic 34 -a value much lower than the 99% recorded by us in health facilities. While we used a wide variety of sampling strategies, the surveys were nationally representative and used standardized approaches for the assessment of active management that enabled cross-country comparisons. This study built local capacity to conduct direct observational research and collected baseline data that should be useful in future assessments. Based on these survey tools, a new index has been developed to measure the quality of facility-based labour and delivery care. This should make it quicker and easier to repeat such assessments. 35 Our analysis focuses primarily on the process component of quality of care -i.e. the actual health care given to patients. [36][37][38] Although we present some information on the human and material resources, 37,38 our study was not designed to assess quality of care based on outcomes. 38 A full evaluation of the quality of the active management of the third stage of labour would require assessment of the inputs, processes, outputs and outcomes.
Although we found evidence of progress being made since 2005, there is still room for improvement. As new evidence becomes available and revisions to global guidelines are developed, national policies and guidelines should also be updated. As an organization responsible for setting global standards in health practice, WHO is in the best position to ensure that new guidelines are introduced in countries. National guidelines, in turn, should stimulate appropriate training and the production of updated standard management guidelines that are readily available at the facility level. 39 National health management information systems should include uterotonic provision to enable