Cost-effectiveness of strategies to improve the utilization and provision of maternal and newborn health care in low-income and lower-middle-income countries: a systematic review

Worldwide, each year 3 million newborns die within the first 28 days of life [1], 2.6 million babies are stillborn [2], and 287,000 women die from complications of pregnancy and childbirth [3]. The vast majority of these deaths occur in Africa and Asia, many could be prevented by improving access to existing interventions [1‐3]. Substantial evidence exists on a wide range of interventions and behaviours, and reviews have identified life-saving maternal and newborn health (MNH) interventions that are not only effective, but also cost-effective and suitable for implementation in resource-constrained settings [4‐6]. Examples include iron supplements to prevent anaemia, tetanus toxoid immunization, magnesium sulphate for eclampsia, uterotronics to prevent and manage post-partum haemorrhage, hygienic cord care, immediate thermal care, exclusive breastfeeding, and management of neonatal sepsis, meningitis and pneumonia [6]. It has been estimated that increased coverage and quality of pre-conception, antenatal, intra-partum, and post-natal interventions by 2025 could avert 71% of newborn deaths [5].

Despite efforts to identify priority interventions, access to life-saving MNH interventions remains inadequate in low and middle income countries [6]. There may be delays making the decision to seek care, difficulties in reaching care, or problems with the range or quality of care available [7]. To achieve higher coverage of MNH interventions, effective and cost-effective strategies need to be identified that address these challenges and lead to improvements in the utilization and provision of MNH care. Demand-side strategies are needed to influence health practices of individuals and communities and promote uptake of preventive and curative MNH care during pregnancy, childbirth and in the post-natal period. These strategies may provide health information and education, or address geographic, financial, or cultural barriers to accessing care. At the same time, supply-side strategies are needed to enhance the capability and performance of front-line health workers, who as the first point-of-contact for women and newborns provide essential care in the community and at primary health facilities. Supply-side strategies may involve training to ensure health workers have the necessary knowledge and skills, and strategies to motivate health workers, improve the working environment and resources available, or strengthen other aspects of the health system.

Several recent reviews have summarized the evidence on the effectiveness of these strategies to improve MNH care in low and middle income countries [8‐14]. The reviews range in scope and setting, and have focused on community-based strategies, integrated primary care, and urban settings. However, given resource constraints, it is important to know not only what strategies are effective at improving coverage of MNH interventions, but also whether the strategies are cost-effective. This has been highlighted as a research priority [12], and the existing reviews contain only limited information on cost-effectiveness [9, 11, 12, 14].

This paper presents a systematic review of the cost-effectiveness of strategies to improve the demand and supply of maternal and newborn health care in low-income countries (LICs) and lower-middle-income countries (LMICs). Our focus is not the interventions and behaviours themselves, but strategies for ensuring they are made available and taken up. The protocol for the systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO) [15].

The title and abstract of 3236 publications were screened and the full text of 140 publications was reviewed (Figure 1). This included three publications identified from reference lists [19, 37, 47]. One publication identified for full-text review could not be retrieved and so was not reviewed. Ninety-two publications were excluded at full-text review. Fourteen of the excluded publications were literature reviews, whose references were manually searched for relevant studies. One publication was excluded because the study was not in a LIC or LMIC. Sixteen publications were excluded because they assessed medical interventions rather than a strategy, and a further 6 publications were excluded because they focused on health topics, such as immunization or HIV, without specific reference to MNH care. Of those remaining, 8 publications were excluded because the cost-effectiveness of the strategy was not measured in a maternal or newborn population, though we did include one article that reported the cost per infant death averted. The latter reported on the distribution of mosquito nets to pregnant women and was included since the deaths averted would primarily be in newborns. A further 11 publications mentioned cost-effectiveness in the abstract but were excluded because they did not contain any cost data. Finally, 36 publications were excluded because they did not report a measure that combined cost and effect. For instance, several studies reported the total cost or a unit cost of implementing the strategy, such as the cost per health worker trained.

Our systematic review identified 48 publications on the cost-effectiveness of strategies to improve the utilization and provision of MNH care in low-income and lower-middle-income countries. The 48 publications reported on 43 separate studies and we judged the reporting of 16 of these studies to be of high quality with respect to the CHEERS reporting criteria.

There was considerable diversity in the strategies used to improve MNH care, and also in the setting, intensity and scale of implementation. However, it was possible to identify some common themes among the strategies, and these were presented in relation to the continuum of care and the level of the health system. This synthesis summarized the cost-effectiveness literature available and highlights the extent to which the evidence focuses on community-based strategies and care during pregnancy. It also emphasizes the lack of evidence on the cost-effectiveness of strategies to improve post-partum care. In addition, it was interesting to note the large proportion of strategies focused on a specific aspect of MNH care. This was particularly noticeable in the facility-based strategies, which included a number of strategies to extend antenatal care and improve specialist care.

It is clear that demand and supply-side strategies can be cost-effective in enhancing the utilization and provision of MNH care and improving health outcomes. Strategies that reported on the cost per life-year saved and cost per DALY averted were cost-effective when compared to GDP per capita. These strategies, in their specific settings, included the use of women’s groups [18, 37, 47, 49, 51]; home-based newborn care using community health workers, volunteers and traditional birth attendants [22, 39, 59, 65]; adding services to routine antenatal care [33]; a facility-based quality improvement initiative to enhance compliance to care standards [52]; and the promotion of breastfeeding in maternity hospitals [36]. However, it should be noted that the results may not be transferable beyond the study setting and using GDP per capita as a threshold does not necessarily ensure that the strategy is affordable. Using GDP per capita as the threshold also approaches cost-effectiveness from a national perspective, and it has been suggested that a global minimum monetary value for a DALY would improve the transparency and efficiency of priority setting for international donors [72].

It is more difficult to interpret the cost-effectiveness results when strategy-specific measures were reported, such as the cost per Caesarean-section, or cost per syphilis case treated. Very few studies use the same measure, though some report estimates from the same setting at different points in time [39, 41, 44‐46]. In many of the studies the costs appear low given the potential health benefits, though the results should be interpreted with care as the choice of measure can also affect conclusions. For example, the cost per woman vaccinated with tetanus toxoid was much lower when the vaccination took place in a campaign rather than during routine antenatal care, however, targeting the vaccine to pregnant women attending antenatal care was found to be more cost-effective when the cost per life saved was estimated [42]. This example highlights the benefits of using health outcomes data, and where that is not available, the value in extrapolating beyond intermediate measures to estimate the number of lives saved using models, such as the Lives Saved Tool (LiST) [73]. Strategy-specific measures also have limited use for priority-setting as decision-makers cannot directly compare strategies that report different cost-effectiveness measures.

The extent to which alternative strategies can be compared was also limited by the how the cost-effectiveness analysis was framed. The reported results may depend on the choice of comparator, the costs included and any assumptions about the effect of the strategy on the quantity and quality of life. For instance, a commentary accompanying the article on training assistant medical officers to perform emergency obstetric care argued the cost-effectiveness results may be an under-estimate since ‘doing nothing’ would be a more realistic comparator than training surgeons given the shortage of medical personnel in Mozambique [50]. The costing perspective is another dimension that can affect study results. For instance, from a health services perspective, delivering home-based care was found to be more expensive than providing services in an outreach clinic or at the health facility, but this perspective does not take into account the direct and indirect costs incurred by households to obtain care [25]. The range of costs included can also have a dramatic impact. For example, when interpreting the results of the study on the management of birth asphyxia by health volunteers it is important to acknowledge the estimate of US$25 per life saved only takes into account the cost of the equipment, and does not include the cost of other supplies or value the time donated by health volunteers [40].

We had hoped to learn the extent to which cost-effectiveness depends on the strength or scale of implementation, or country-specific factors, and each of these considerations is salient for the transferability of study results. Intensive implementation may improve the strategy’s effectiveness, but this is likely to come at a cost. Similarly, the scale of implementation may affect costs or effects, and the potential to generate economies of scale will depend on the characteristics of the strategy and the geographic setting. The context for a study can also influence costs and effects, and the degree to which a strategy can be replicated in another setting. One study reported that there would be economies of scale if the women’s group initiative were rolled out in Bangladesh, and another on the Bamako Initiative reported relatively similar cost-effectiveness in Benin and Guinea [18, 27]. However, drawing conclusions across studies is extremely challenging given the wide-ranging strategies implemented in disparate contexts. The women’s group studies were the most comparable since a common strategy was applied in different countries and with differing intensity (in terms of population coverage), and they were methodologically similar as the studies had several researchers in common. Looking across these studies there was some evidence that greater intensity of implementation produced a larger effect, though implementation and contextual factors may impact on the effect sizes [14]. As others have noted, understanding determinants of differences in cost, and the effect of scale on cost are priorities for future work [14].

Only three studies reported a cost-effectiveness measure that took into account the combined effect of the strategy on both maternal and newborn health. One of the distinctive characteristics of pregnancy and intra-partum care is the potential for the care to benefit both the woman and her child. In addition, intervening during pregnancy can affect subsequent care seeking and health outcomes. Thus, the cost-effectiveness of a strategy may be under-estimated if the health benefits were measured in either women or newborns but not both [74], though this point was not highlighted in any of the publications.

In drawing conclusions we need to bear in mind the quality of the publications. The CHEERS checklist sets a standard for the information that should be reported, and although there were some high quality publications, many fell short of the standard. The publications of low and medium quality were often those in which the cost-effectiveness analysis was presented as supplementary to other study results and a strategy-specific measure was used. In some cases, gaps in reporting made it difficult to assess whether methods were appropriate, what assumptions had been made and how the findings should be interpreted. As many peer-reviewed journals have restrictions on article length, authors should make better use of web appendices or, ideally, report the cost-effectiveness analysis as a separate, stand-alone article. Discussion of the cost-effectiveness results was another aspect where the quality was often limited, and conclusions on the cost-effectiveness of the strategy were often stated with only a cursory consideration of study design, context, and limitations, and without reference to the existing literature.

However, the checklist also has some limitations. It focuses on the quality of the reporting rather than the quality of the economic evaluation conducted. This means articles with a clear description of their methods may rank highly even when there are shortcomings in how the study was framed or the range of costs included. In addition, several of the criteria more readily apply to cost-effectiveness analyses that use individual patient-level data or economic modelling, which are the backbone of cost-effectiveness studies in high income countries, but less frequently used in low and middle income countries. We also acknowledge the quality categorization was based on a simple approach and alternative methodologies could be applied, such as assigning weights to the criteria based on subjective judgment or statistical analysis.

We took a systematic approach to the literature search and selection process, though there is always a risk that a relevant article has been missed, and it was not always straightforward to determine whether an article satisfied the inclusion criteria. For example, we carefully considered whether to include article publication that compared universal HIV screening in pregnant women with screening restricted to areas of high HIV prevalence [38]. We also carefully considered the eligibility of several articles that reported costs in relation to a process indicator, such as the study from Cambodia on midwife-led group discussions with pregnant women that reported the cost per educational interaction [32]. We decided that process indicators would be eligible if they referred to an interaction between women (or newborns) and a front-line worker. Publication bias is also a potential concern, since cost-effectiveness analyses are often only undertaken in field-studies once a positive effect has been demonstrated, though the lack of statistically significant effect does not necessarily preclude a strategy from being cost-effective, nor should it prevent a cost-effectiveness analysis being published [75, 76]. That all the studies in our review had positive findings was striking, and suggests that cost-effectiveness analyses with negative findings may not have been published.

The review highlights a need for future research. Only a minority of studies on strategies to improve the utilization and provision of MNH care consider their cost-effectiveness [8‐14]. Moreover, the evidence available is limited by the lack of high quality reports presenting comparable cost-effectiveness measures. In particular, there are gaps relating to post-partum and post-natal care, and relatively few studies focused on the quality of intra-partum care. Further work on the extent to which implementation, scale and context impact cost-effectiveness would also be useful to understand the degree to which strategies can be replicated elsewhere. Moreover, for future research to generate results that can be transferred beyond the immediate study setting, greater consideration should be given to how the economic evaluation is framed. This includes the use of a comparable cost-effectiveness measure, such as cost per DALY averted, and also a costing that reflects the full cost to everyone of implementing the strategy within the prevailing health system. It is good practice to take household costs into account and to value donated goods and volunteer time. As the initiative to promote facility-births in Burkina Faso demonstrated, the cost of the strategy (as opposed to the intervention alone) can be substantial: there was an eightfold increase in the cost of a facility birth when programme costs were included in the calculations [29]. Thus, estimates of the cost-effectiveness of life-saving MNH interventions that do not take into account demand or supply strategies will substantially under-estimate the resources required to reduce maternal and neonatal mortality.

Demand and supply-side strategies can be cost-effective in enhancing the utilization and provision of MNH care and improving health outcomes, though the evidence available is limited by the lack of high quality studies using comparable cost-effectiveness measures. Direct comparison of alternative strategies was also limited by how the studies were framed, as there was substantial variation in how researchers approached, designed and analyzed cost-effectiveness. Further studies are needed, though existing evidence shows the cost-effectiveness of several strategies implemented in the community to influence health practices and care seeking, and also facility-based initiatives to improve the range and quality of MNH care available.