Triple return on investment: The cost and impact of 13 interventions that could prevent stillbirths and save the lives of mothers and babies in South Africa

Background: The time of labor, birth and the first days of life are the most vulnerable period for mothers and children. Despite significant global advocacy, there is insufficient understanding of the investment required to save additional lives. In particular, stillbirths have been neglected. Over 20 000 stillbirths are recorded annually in South Africa, many of which could be averted. This analysis examines available South Africa specific stillbirth data and evaluates the impact and cost-effectiveness of 13 interventions acknowledged to prevent stillbirths and maternal and newborn mortality. Methods: Multiple data sources were reviewed to evaluate changes in stillbirth rates since 2000. The intervention analysis used the Lives Saved tool (LiST) and the Family Planning module (FamPlan) in Spectrum. LiST was used to determine the number of stillbirths and maternal and neonatal deaths that could be averted by scaling up the interventions to full coverage (99%) in 2030. The impact of family planning was assessed by increasing FamPlan’s default 70% coverage of modern contraception to 75% and 80% coverage. Total and incremental costs were determined in the LiST costing module. Cost-effectiveness measured incremental cost effectiveness ratios per potential life years gained. Results: Significant variability exists in national stillbirth data. Using the international stillbirth definition, the SBR was 17.6 per 1 000 births in 2013. Full coverage of the 13 interventions in 2030 could reduce the SBR by 30% to 12.4 per 1 000 births, leading to an MMR of 132 per 100 000 and an NMR of 7 per 1 000 live births. Increased family planning coverage reduces the number of deaths significantly. The full intervention package, with 80% family planning coverage in 2030, would require US$420 million (US$7.8 per capita) annually, which is less than baseline costs of US$550 million (US$10.2 per capita). All interventions were highly cost-effective. Conclusion: This is the first analysis in South Africa to assess the impact of scaling up interventions to avert stillbirths. Improved coverage of 13 interventions that are already recommended could significantly impact the rates of stillbirth and maternal and neonatal mortality. Family planning should also be prioritized to reduce mortality and overall costs.

Data on stillbirths in South Africa were reviewed to evaluate changes in stillbirth rates over time. Stats SA, DHIS and PPIP data were collated to approximate trends in South Africa between 2000 and 2013. Stats SA data were obtained from a report by the Medical Research Council (MRC) [16]. The data have been adjusted for under-registration. DHIS data were acquired from the South African Health Review [3,17] and PPIP data were compiled from Saving Babies reports [6,18-23]. Additional PPIP data were made available to the authors for 2012 and 2013, which were used to approximate the distribution of stillbirths in health facilities in South Africa. The Lives Saved Tool (LiST) was used to analyze the number of stillbirths and deaths of mothers and newborns that could be averted by scaling up the intervention packages for maternal and neonatal health to full coverage (99%). LiST is a module in Spectrum, a demographic software package, which preloads national data for health status, mortality rates, and coverage of more than 60 interventions and their effectiveness in relation to specific causes of death [24,25]. The modelling methods in LiST have been widely reviewed [26,27]. The effects of scaling up the coverage of specific interventions were modelled to estimate the deaths averted, overall and by each intervention. The analysis was performed using Spectrum version 5.04. The WHO third trimester stillbirth definition (>1 000 g) was used to ensure projections are of international relevance. The modelled interventions would however also be effective in averting second trimester stillbirths (500–1 000 g). Mortality rates were estimated for 2012, or the closest year: third trimester stillbirth rate (SBR) was 17.6 per 1 000 births [6], maternal mortality ratio (MMR) was 269 deaths per 100 000 live births [28] and neonatal mortality rate (NMR) was 12 per 1 000 live births [28]. The causes of maternal [29] and newborn [30] mortality were adapted from the South African Medical Research Council Burden of Disease (BOD) estimates to fit the causal categories provided by LiST (Figure 1). The causes of death categorized in LiST are slightly different from those presented by the MRC. For example, neonatal diarrhoea is not reported separately in the MRC BOD, but rather combined with under-five diarrhoeal deaths. Therefore, we separated these using the default proportions in LiST. LiST broadly defines stillbirths as antepartum or intrapartum. Rates of 35% for intrapartum and 65% for antepartum stillbirths were calculated by dividing the numbers of fresh and macerated stillbirths by the total number of stillbirths, respectively, using data provided by PPIP for 2012 and 2013 [6]. Causes of newborn, child and maternal deaths in South Africa used in the LiST model. Source: Adapted from Medical Research Council burden of disease estimates in 2010. In The Lancet’s Stillbirth Series, Bhutta et al. [31] identified a package of 10 interventions during pregnancy and childbirth to prevent stillbirths globally, including expanded antenatal care packages for improved detection and management of hypertension and diabetes in pregnancy, improved detection of high risk pregnancies and fetal compromise, and induction of labor after 41 weeks gestation. Pattinson et al. [32] include five additional interventions in their global analysis which provide significant health benefits for mothers and newborns. These interventions were adapted for the South African context. Interventions for malaria prevention were excluded as South Africa has succeeded in preventing malaria transmission throughout most of the country following introduction of DDT in 2001 [33]. Folic acid supplementation or fortification was also excluded because South Africa’s folic acid fortification program, implemented in 2003, has achieved significant declines in neural tube defects [34]. Early detection and treatment of HIV in pregnant mothers was added to the model as a maternal intervention, since over 40% of maternal deaths in South Africa are due to AIDS [29]. The availability of obstetric care was also adjusted. The obstetric levels in LiST must sum to cover 100% of deliveries. This includes assisted and unassisted home deliveries and facility deliveries with essential care, basic emergency obstetric care (EmOC) and comprehensive EmOC. At baseline, facility delivery coverage is distributed between the three levels of care. In an ideal scenario, everyone in South Africa needing comprehensive EmOC would have access to it. This intervention has thus been scaled up to 99% and the other obstetric packages have correspondingly been scaled down to maintain 100% delivery coverage. Table 2 presents the final 13 priority interventions, along with their effect estimates and affected fractions. In The Lancet’s Stillbirth Series, the intervention effects were compiled using the literature review criteria developed by the Child Health Epidemiology Group (CHERG) [35] and through Delphi consultation with global experts and practitioners [31,32]. The same effects have been used in this analysis; these values are different from the default effects in version 5.04 of LiST, except where indicated. The effects used are also available online as supplementary documentation for LiST [36]. In keeping with The Lancet’s methods, pathways were added to LiST for the effect of hypertensive disease case management on stillbirths and the effect of labor and delivery management on maternal sepsis, neonatal sepsis and neonatal tetanus [32]. The affected fractions were also derived from the online supplementary material [36]. It has been indicated where default affected fractions were used. South African experts were consulted to determine the effect and affected fraction for the added intervention, early detection and treatment of HIV in pregnant mothers. Effect and affected fraction estimates for interventions to reduce stillbirths and maternal and neonatal deaths AF = Affected fraction; *Default LiST value in Spectrum V5.04. Sources: Adapted from Bhutta [31,32,36]. A one day expert consultation was hosted locally to discuss trends in maternal and newborn interventions in South Africa. Twenty-three participants were invited from the health sector, including clinicians, academics and those in positions at national and district level. Coverage levels for all interventions were reviewed and modified by the expert panel (Table 3). Unchanged default LiST coverage levels have been indicated. A linear increase in coverage was applied to each of the 13 priority interventions from 2014, the baseline coverage year, to reach 99% coverage in 2030. Coverage levels for other maternal and child health interventions were not altered, so that the overall impact of the priority interventions on South Africa’s SBR, MMR and NMR could be isolated. (Ramping up the coverage of all maternal, newborn and child interventions simultaneously in LiST did not have a significant impact on the deaths averted by the 13 interventions). In order to evaluate the impact of increasing the modern contraceptive prevalence rate (CPR) on pregnancies and subsequent stillbirths and maternal and newborn mortality, three scenarios were analyzed for family planning using the FamPlan module available in Spectrum: 70% coverage in 2030 (FamPlan default); scale up to 75% coverage in 2030; and scale up to 80% coverage in 2030. For all three scenarios, a baseline family planning coverage of 65% and a total fertility rate (TFR) of 2.81 were used (default FamPlan values). Baseline coverage estimates for interventions to reduce stillbirths and maternal and neonatal deaths *Default LiST value in Spectrum V5.04. Source: Consultation with expert South African panel. The interventions were costed using the costing module in LiST, using the most recently available data. In the model, the cost of each intervention is based on four components: personnel and labor; drugs and supplies; other recurrent costs; and capital costs. Default international prices were used for drugs and supplies but staff remuneration was added based on public health salary data for South Africa [37,38]. The direct (recurrent) and indirect (capital) costs for hospitalization and outpatient visits were obtained from the WHO-CHOICE (CHOosing Interventions that are Cost Effective) database [39]. Family planning was costed by combining the costs for each contraceptive method, including oral contraceptive pills, condoms, injectables, intrauterine devices, and female and male sterilization. Costs related to infrastructure development are not included in LiST [40]. All costs were adjusted to 2014 US dollars. All per capita estimations used a population size of 54 million, according to the Stats SA mid-year population estimates for 2014 [41]. Cost-effectiveness was determined by combining the effects (additional lives saved) and costs calculated by LiST. The baseline used was 2014 such that effects and incremental costs were calculated based on the difference between estimates in 2014 and 2030. Lives saved were multiplied by life expectancy to estimate potential life years gained. A life expectancy at birth of 60 years was used for stillborns and neonates [41] and a Reproductive-Aged Life Expectancy (RALE) [42] of 27 years was used for mothers, based on SA’s 2011 life tables [43]. Incremental cost-effectiveness ratios (ICERs) were obtained by dividing incremental intervention costs by incremental effects. The WHO criteria on cost-effectiveness were used, which specify that an intervention is highly cost-effective if it averts a year of life lost for less than the national gross domestic product (GDP) per capita, cost-effective if 1 to 3 times the GDP per capita, and not cost-effective if greater than 3 times the GDP per capita [44]. Given SA’s GDP per capita (2012) of US$7 500 [45], interventions were considered highly cost-effective if the ICER was less than US$7 500, cost-effective if between US$7 500 and US$22 500, and not cost-effective if more than US$22 500. Ethical review board approval was not required for this paper as no human subjects were involved and only secondary data were used.

N/A

• Dima Health