Data collection

The maternal-newborn bottleneck analysis tool was developed as part of the Every Newborn Action Plan (ENAP) process to assist countries in identifying context-specific bottlenecks to the scale-up and provision of maternal and newborn health interventions across seven health system building blocks [25]. The tool was utilised during a series of national consultations supported by the global Every Newborn Steering Group between July 1^st and December 31^st 2013 (see Additional file 1). The workshops for each country included participants from national ministries of health, United Nations (UN) agencies, the private sector, non-governmental organisations, professional bodies, academia, bilateral agencies, and other stakeholders. For each workshop, a facilitator, orientated on the tool, coordinated the process and facilitated groups to reach consensus on the specific bottlenecks for each health system building block. This paper, third in the series, is focused on antenatal corticosteroids for the management of preterm birth.

Data analysis methods

We graded bottlenecks for each health system building block using one of the following options: not a bottleneck (=1), minor bottleneck (=2), significant bottleneck (=3), or very major bottleneck (=4). We first present the grading in heat maps according to the very major or significant health system bottlenecks as reported by all 12 countries, then by mortality contexts (neonatal mortality rate (NMR) <30 or NMR ≥30 deaths per 1000 live births) and by region (countries in Africa and countries in Asia). We developed a second heat map showing the specific grading of bottlenecks for each health system building block by individual country. Survey responses were analysed for common specific bottlenecks, defined to be those reported by at least 3 countries. Where the same specific bottleneck was reported under more than one building block, it was categorised in accordance with the relevant survey question, or under the building block where most countries reported the specific bottleneck.

Finally, we categorised context-specific solutions for scaling up under the subcategories of corresponding bottlenecks within each health system building block.

Leadership and governance bottlenecks and solutions

Leadership and governance was graded as a very major or significant bottleneck by 7 of 11 country teams (Figure 2), representing a larger proportion of Asian countries (4 of 5 countries in Asia, compared to 3 of 6 countries in Africa). The most commonly cited specific bottleneck was inadequate guidelines on ACS use, cited by 9 of 11 country teams. Of these, 5 country teams (4 in Asia) reported no clear guidelines on ACS for management of preterm birth, and 4 country teams (all in Africa) reported available guidelines, which were either outdated or not disseminated.

Proposed solutions included development (or update) and dissemination of national guidelines and protocols on prevention and management of preterm labour, including ACS. Updated WHO guidelines will be useful to inform national guidelines.

Health financing bottlenecks and solutions

Health financing was graded as a very major or significant bottleneck by 8 of 11 country teams. The most commonly cited specific bottleneck was insufficient funding, reported by 9 country teams, with 8 indicating a lack of priority or policy as the underlying cause.

Proposed solutions included increased advocacy and leadership, particularly with funding and budget allocation, for preterm birth prevention, management and care including ACS.

Health workforce bottlenecks and solutions

Health workforce was graded as a very major or significant bottleneck by 7 of 11 country teams. All country teams questioned reported a shortage of health workers especially in higher cadres, inadequate training, and inadequate supervision or mentoring. India did not provide an answer to the question on health worker shortages.

Proposed solutions included integration of ACS into pre-service and in-service training, development and dissemination of job aids, and consideration of expanded prescription authority for more cadres of health workers.

Essential medical products and technologies bottlenecks and solutions

Essential medical products and technologies was graded as a very major or significant bottleneck by 9 of 11 country teams. All 11 country teams cited as a specific bottleneck the lack of inclusion of ACS on the national essential medicines list (NEML) for fetal lung maturation, though all included dexamethasone for other indications. Of 8 country teams reporting inadequate procurement or distribution of ACS, 6 cited the lack of policy or NEML listing for fetal lung (or other lack of alignment between supply chain policy and ACS recommendations) as a direct cause of lack of integration into supply chain system. Additionally, no respondent country included ACS in quantification and forecasting, and none reported having an indicator for ACS drug availability.

Proposed solutions included listing of ACS drugs on NEMLs for a fetal lung maturation indication, development and dissemination of guidelines, and software to support needs-based forecasting and procurement.

Health service delivery bottlenecks and solutions

Health service delivery was graded as a very major or significant bottleneck by 9 of 11 country teams. All country teams reported having inadequate or non-existent quality monitoring and improvement systems, and 9 country teams noted a lack of alignment between ACS prescribing authority and health worker cadres providing care for women at risk of preterm birth. Additionally, 7 countries cited delays due to referrals as a specific bottleneck. Table S3 and S4 of the Additional file 2 summarises levels of care where ACS is allowed and health worker cadres permitted to prescribe and/or administer ACS, as reported by country teams.

Proposed solutions included development and dissemination of guidelines across facilities, creation of supervision and mentoring systems, integration into clinical audits and reviews, and strengthened referral systems.

Health information system bottlenecks and solutions

Health information system was identified as the most problematic building block, with all 11 country teams reporting very major or significant bottlenecks. Responses to health information systems questions revealed an almost complete absence of defined and standardised indicators for ACS use, with the exception of Nepal and one province in Pakistan, where country teams reported that ACS use was tracked on partographs. No countries reported having an indicator for ACS drug availability. Moreover, no countries included ACS use in clinical audits or perinatal reviews.

Proposed solutions included creation of indicator(s) for ACS use and integration of ACS data into record-keeping systems and regular reviews.

Community ownership and partnership bottlenecks and solutions

Community engagement was graded as a very major or significant bottleneck by 8 of 11 country teams. The most commonly cited specific bottlenecks were lack of awareness and initiatives targeted at raising awareness of preterm birth risks and management (9 countries), and the absence of an accessible facility or transportation to reach a facility where ACS could be administered (9 countries).

Proposed solutions included leveraging existing community outreach infrastructure, such as community groups, opinion leaders, and media.

Health information systems priority actions

Data on ACS coverage, use, and outcomes including safety were absent in nearly all countries. To enable continuous and quantitative assessment of ACS programmes, appropriate indicators need to be defined, and these data integrated into existing health information systems in order to allow tracking and comparison at facility, district, and national levels. ACS coverage data are one of the priority indicators identified in the Every Newborn Action Plan [28]. Coverage data are lacking and should also aim to assess false positives (women treated whose babies are born after 34 completed weeks).

Use of data for quality improvement would be aided if ACS were systematically included in safe childbirth checklists, partographs, clinical audits, and perinatal death reviews [29]. Data are also needed on process such as logistics and stock out. Two case studies of in-country implementation programmes provide examples of data collection and recordkeeping systems for ACS use, and how audit and feedback mechanisms using these records can be harnessed to improve quality of care for the management of preterm birth (Figure 4).

Health service delivery priority actions

Estimates suggest ACS coverage in LMICs is generally low even in tertiary facilities (Table 1), consistent with the poor grading of health service delivery among respondent country teams. All 11 countries in this analysis are among the 75 Countdown to 2015 countries with an estimate of 41% (weighted average) coverage of preterm births in secondary and tertiary facilities. Eight of 11 countries were also included in the WHO Multicountry Survey estimate of coverage in high-volume facilities; these were DRC (16%), Kenya (32%), Nigeria (30%), Uganda (27%), India (69%), Nepal (20%), Pakistan (63%), and Vietnam (52%). These estimates are likely to be higher than the national average as the survey methodology sampled only large facilities from the capital city and two randomly selected provinces. National population coverage with ACS is likely to be much lower than facility coverage estimates, as a large proportion of births take place outside of high-volume facilities or outside any facility. In the 11 countries, institutional delivery rates (in any facility) ranged from 33% in Bangladesh to 92% in Vietnam, according to data compiled by UNICEF in 2015 [30].

Available evidence supports focusing scale-up efforts in facilities with the capacity for gestational age assessment and diagnosis of women at risk of imminent preterm birth, as well as ongoing support for preterm infants and well-functioning maternity wards [4]. The recent study (ACT) which extended ACS to include non-hospital facilities in six countries (including Kenya, India, and Pakistan) found increased perinatal mortality as well as increased rates of presumed maternal infection [12]. The specific cause(s) of newborn mortality were not ascertainable, but service delivery challenges included limited ability to estimate gestational age and accurately diagnose high risk of preterm birth. As a consequence, ACS was under prescribed for preterm and early preterm babies likely to benefit from treatment and overprescribed to babies born at or near term, when ACS may have increased risks.

With clear evidence for use in hospitals in HICs with adequate neonatal support and reason for caution at lower-level facilities in LIC settings, programmatic scale-up is best focused on higher-level and high-volume facilities, where resources can also be used to produce significant impact cost-effectively and also improve targeting and safety tracking. Increased population coverage will then depend on increased and timely identification and referral of at-risk women, both in lower-level facilities and in homes. Systemic improvements are needed to encourage institutional deliveries with skilled birth attendants and to strengthen referral systems, while guidelines and training at lower levels of care will be critical to building capacity for timely identification. Expanded care such as a pre-referral dose may also improve care by allowing a longer time for ACS to take effect. However, such steps must be considered cautiously by each country based on the capacity of lower-level facilities to provide adequate care to ensure safety.

Guidelines implementation and quality improvement

For any policy to reach every mother and every newborn, clear guidelines and adequate training must reach all relevant cadres of health worker and levels of care. As indicated by the country teams' responses to leadership and governance questions, it is critical both to develop clear guidelines and to ensure active dissemination. An active model of dissemination should ideally integrate systems of supervision, mentoring, and monitoring of quality improvement - all areas reported as inadequate by all 11 country teams.

A study of active dissemination in the United States showed that use of local opinion leaders, technical updates, reminders, interactive small-group learning, and audit and feedback were effective in accelerating facility uptake of ACS. These five elements were also effective in increasing use of a maternal health intervention in LMICs, with a structure reflecting the differing needs of low-resource settings. Figure 5 shows the five-component structure of the interventions in these two RCTs and details their implementation in each setting. Figure 6 outlines a program with similar features currently being studied in a facility in Cambodia. In particular, the audit and feedback component provides one model for continuous quality monitoring and improvement, an essential part of quality implementation of guidelines. This pilot model also provides a further example of the role of the continuous use of outcome data in improving safety and quality of care.

Essential medical products and technologies priority actions

Corticosteroids for ACS treatment are inexpensive (often under 1 USD for a full course) and are often assumed to be widely available. Yet most respondents indicated shortages at the country level, primarily attributed to a lack of supporting policy or effective logistics systems. NEML listing for the fetal lung maturation indication is essential to prioritisation for procurement as well as integration into supply chain, from forecasting to distribution.

Adequate procurement further depends on development and dissemination of guidelines on ACS use, including clarity on the choice of corticosteroid and appropriate regimen. Two corticosteroids, dexamethasone and betamethasone, have been shown to be safe and effective to manage preterm birth. Betamethasone is sometimes preferred in HICs due to limited evidence suggesting better outcomes (44% reduction in RDS and 33% reduction in neonatal mortality versus placebo or no treatment, compared with 20% reduction in RDS and 28% reduction in neonatal mortality for dexamethasone versus placebo or no treatment) and potentially lower risk of maternal infection [4].

However, evidence on dexamethasone still conclusively supports its overall safe and effective use for management of preterm birth. A meta-analysis of 9 studies directly comparing dexamethasone with betamethasone further found no statistically significant differences apart from a greater reduction in intraventricular haemorrhage using dexamethasone (RR 0.44, 95% CI 0.21 to 0.92, 4 studies, 549 infants). A large RCT directly comparing dexamethasone to betamethasone is currently underway [32].

Critically, only dexamethasone is a feasible choice for scale-up in most LMICs. While dexamethasone is widely available from international suppliers for a variety of indications, the formulation of betamethasone supported by most evidence (a suspension of betamethasone acetate in betamethasone phosphate, rather than betamethasone phosphate alone) has been subject to global shortages [33] and costs 25 times as much as dexamethasone per course of ACS [34].

Dexamethasone also faces fewer policy hurdles to increased use for management of preterm birth. With a variety of uses, dexamethasone in the recommended formulation is already registered, listed on the NEML, and included in procurement and supply chain in all 11 countries, for other indications. Dexamethasone is the only ACS listed on the WHO EML with a fetal lung maturation indication [35].

National policies, in line with the current WHO guideline, should therefore focus on promoting appropriate use of dexamethasone through procurement policy, clear guidelines, and integration into forecasting, procurement, and supply chain.

Limitations and implications for further research

Soliciting responses from a wide range of in-country partners and practitioners in maternal and newborn health captured context-specific challenges and generated collaborative solution ideas. The grading process also created consensus around priority bottlenecks and health system building blocks to be addressed. However, these consensus views are subjective. The quality and amount of information also varied depending on the level of knowledge of participants on health system issues and on workshop facilitation. In addition, bottlenecks were reported as perceived bottlenecks relative to the other health system building blocks. National-level assessment may mask regional disparities, particularly between urban and rural areas. This comprehensive questionnaire may have led to respondent fatigue, but this effect is least significant for ACS as the first intervention in the bottleneck analysis tool. For India and Pakistan, subnational responses may not mirror all national-level challenges. However, Pakistan submitted responses from all regions except two tribal territories, and although India returned data from only three of its 28 states, these areas are amongst the poorest states and include populations similar to those of Vietnam, Kenya, and DRC [36].

Further research is needed to establish the effectiveness of the solutions described here, especially in the context of ACS use in LMIC hospital settings where most of the world's births now occur. To date the evidence on ACS is primarily from HIC settings with neonatal intensive care [4] or the ACT trial from low-level settings with less-skilled workers [12]. The evidence base could be greatly advanced in LMIC settings, for instance through a multi-country, cluster-randomised trial in LMIC hospitals to assess the mortality impact and safety of a package optimising gestational age (where first trimester ultrasound is not routine) and clinical assessment of mothers, including risk of preterm birth and possible maternal infection, whilst providing appropriate maternal and newborn care. Given the lack of pharmacokinetic and dynamics data for both dexamethasone and betamethasone, other research questions include optimal dosage regimens. Data are now even more critically important and this gap is urgent to address, possibly for coverage and outcome data that could be tracked in routine health management information systems [29]. Further work is also needed to define signal functions for newborns by level of care. A signal function on ACS could be incorporated into existing monitoring systems, such as the signal functions for basic and comprehensive emergency obstetric care, to ensure consistent tracking of maternal and newborn interventions along the continuum of care [37].

Another important track of research is to improve the accuracy and also feasibility of gestational age assessment, including testing tools to improve accuracy of last menstrual period assessment, use of ultrasound dating in later pregnancy and ways to promote better recording in notes and use of the data by clinicians. Key messages and actions are summarised in Figure 7.