A cost-effectiveness analysis of early detection and bundled treatment of postpartum hemorrhage alongside the E-MOTIVE trial | Nature Medicine

Nature Medicine volume 30, pages 2343–2348 (2024)Cite this article

3138 Accesses

1 Citations

89 Altmetric

Metrics details

Timely detection and treatment of postpartum hemorrhage (PPH) are crucial to prevent complications or death. A calibrated blood-collection drape can help provide objective, accurate and early diagnosis of PPH, and a treatment bundle can address delays or inconsistencies in the use of effective interventions. Here we conducted an economic evaluation alongside the E-MOTIVE trial, an international, parallel cluster-randomized trial with a baseline control phase involving 210,132 women undergoing vaginal delivery across 78 secondary-level hospitals in Kenya, Nigeria, South Africa and Tanzania. We aimed to assess the cost-effectiveness of the E-MOTIVE intervention, which included a calibrated blood-collection drape for early detection of PPH and a bundle of first-response treatments (uterine massage, oxytocic drugs, tranexamic acid, intravenous fluids, examination and escalation), compared with usual care. We used multilevel modeling to estimate incremental cost-effectiveness ratios from the perspective of the public healthcare system for outcomes of cost per severe PPH (blood loss ≥1,000 ml) avoided and cost per disability-adjusted life-year averted. Our findings suggest that the use of a calibrated blood-collection drape for early detection of PPH and bundled first-response treatment is cost-effective and should be perceived by decision-makers as a worthwhile use of healthcare budgets. ClinicalTrials.gov identifier: NCT04341662.

Postpartum hemorrhage (PPH), defined as blood loss ≥500 ml from the genital tract after childbirth, is the leading cause of maternal death worldwide, accounting for approximately 27% of maternal deaths1,2. PPH is a major concern in low- and middle-income countries (LMICs), where PPH-associated mortality is disproportionately high3. PPH is associated with considerable economic burden: recent estimates from a study conducted in Kenya, India, Nigeria and Uganda suggest the costs of direct hospital care for patients with PPH can be up to 2.8 times higher than for a birth without PPH4. In addition, the immediate and long-term economic consequences of maternal mortality incurred by households can be substantial5,6,7.

The World Health Organization (WHO) has published and updated several evidence-informed recommendations for the prevention and treatment of PPH8,9. However, adherence to these recommendations in many low-resource settings is limited by numerous challenges. First, PPH is often undetected or detected late; consequently, life-saving treatment is not promptly initiated. The current usual practice of blood-loss assessment is visual estimation, which is widely recognized as inaccurate and typically leads to underestimation of blood loss10. An additional challenge is delayed or inconsistent use of effective interventions for the management of PPH. Treatments for PPH are often administered sequentially; healthcare providers wait to observe the effects of one intervention before administering another intervention11. However, PPH is a time-critical condition, and such delays can result in loss of life. Some cost-effective interventions may not be used at all. Evidence from hospitals in Kenya, Nigeria, South Africa and Tanzania showed that tranexamic acid (TXA), a medication used to prevent the breakdown of blood clots, was administered late and mostly as a last resort for patients requiring surgery due to PPH12. Furthermore, despite the availability of clear recommendations regarding PPH and their wide dissemination, uptake at the point of care remains low13. An underpinning factor to some of the challenges relates to limited resources; therefore, it is imperative to evaluate the resource implications of new interventions for managing PPH.

To address these challenges, the cluster-randomized E-MOTIVE trial was designed to assess a multicomponent intervention for detection and treatment of PPH in patients having vaginal delivery. The E-MOTIVE intervention consisted of a calibrated blood-collection drape—a sterile fold-out sheet placed on the delivery bed enabling blood to be swept into a pouch with measurement lines indicating warning and action points—for early detection of PPH, and the WHO-proposed first-response bundle, which included uterine massage, oxytocic drugs, TXA, intravenous (IV) fluids and a process for examination and escalation (Fig. 1). The clinical effectiveness of the E-MOTIVE intervention has already been reported14. Evidence from the trial supported WHO recommendations for both routine objective measurement of postpartum blood loss for vaginal births, and a standardized and timely approach for managing PPH, comprising objective assessment of blood loss and the bundle, supported by an implementation strategy, for all vaginal births. In this Article, we report the economic evaluation conducted alongside the E-MOTIVE trial, an integral component of the E-MOTIVE project, which aimed to assess the cost-effectiveness of the E-MOTIVE intervention compared with usual care. The economic evaluation, which was carried out from a healthcare system perspective, was based on the outcomes of cost per case of severe PPH prevented (blood loss, ≥1,000 ml) and cost per disability-adjusted life-year (DALY) averted.

The E-MOTIVE intervention included a calibrated blood-collection drape for early detection of PPH and a bundle of first-response treatments (uterine massage, oxytocic drugs, TXA, IV fluids, examination and escalation), supported by an implementation strategy.

A total of 104 secondary-level hospitals were assessed for eligibility for the E-MOTIVE trial. Fourteen were excluded due to prior implementation of an early-detection protocol or treatment bundle for PPH. Ninety hospitals in Kenya, Nigeria, Pakistan, South Africa and Tanzania entered the baseline phase. The independent data monitoring committee recommended completing the trial before randomizing hospitals in Pakistan, as the required sample size was achieved in the other four countries. Two hospitals in Kenya were excluded before randomization as they were unable to carry out source-data verification.

Eighty hospitals in Kenya, Nigeria, South Africa and Tanzania underwent randomization at a 1:1 ratio to receive the E-MOTIVE intervention or continue providing usual care. Two hospitals in Tanzania, one in each group, did not receive the assigned intervention due to participation in a conflicting program. Following randomization, a 2-month transition was implemented to allow hospitals in the E-MOTIVE group to adapt clinical practices for intervention delivery. Data collected during this phase did not contribute to the analysis.

Data for analysis were obtained from 78 secondary-level hospitals (from 14 in Kenya, 38 in Nigeria, 14 in South Africa and 12 in Tanzania), with a total of 210,132 patients (110,473 in the baseline phase and 99,659 in the implementation phase) giving birth vaginally in the hospitals between 2 August 2021 and 3 March 2023. Source-verified data regarding blood loss were available for 206,455 patients (107,733 in the baseline phase and 98,722 in the implementation phase; 98% follow-up) (Fig. 2). The clinical findings of the E-MOTIVE trial have been published in full elsewhere14.

All participating hospitals entered a 7-month baseline period in which they provided usual care for patients having vaginal delivery. After the baseline phase, hospitals were randomly assigned in a 1:1 ratio to receive the E-MOTIVE intervention or to continue providing usual care. Eighty hospitals across Kenya, Nigeria, South Africa and Tanzania underwent randomization. Due to participating in a conflicting program, two hospitals in Tanzania did not receive the assigned intervention. Data for analysis were therefore available from 78 hospitals, with a total of 210,132 patients. Source-verified blood loss data for analysis were available for 206,455 patients.

Severe PPH occurred in 786 of 48,678 patients (1.6%) in the E-MOTIVE group and in 2129 of 50,043 (4.3%) in the usual-care group (adjusted risk difference −2.6%, 95% confidence interval (CI) −3.1% to −2.1%; Table 1). In the E-MOTIVE group, the mean DALYs per patient was 0.00767 (standard deviation (s.d.) 0.394), and in the usual-care group, the mean DALYs per patient was 0.01158 (s.d. 0.454). The adjusted DALY difference between E-MOTIVE and usual care per patient was −0.00266 (95% CI −0.00814 to 0.00287; Table 1).

The resource utilization per group is presented in Supplementary Table 1. Notably, administration of oxytocin, TXA and IV fluids—three core elements of the MOTIVE first-response bundle—was more common in the E-MOTIVE group despite lower rates of PPH (8.5% compared with 16.7% in the usual-care group). This can be explained by the improved detection of PPH facilitated by the use of a calibrated blood-collection drape and consequent triggering of the bundle. The usual-care group experienced higher numbers of blood transfusions, marginally longer hospitalization and greater need for additional treatment interventions. Also, notably more severe PPH cases in the usual-care group necessitated additional time for physician attendance.

Disaggregated mean per-patient costs are presented in Supplementary Table 2. The total unadjusted mean per patient cost was 45.15 USD (s.d. 107.93) in the E-MOTIVE group and 43.19 USD (s.d. 126.84) in the usual-care group (Table 1). The adjusted total cost difference was 0.30 USD (95% CI −2.31 to 2.78; Table 1). The estimated incremental cost-effectiveness ratios (ICERs) (Table 1) are therefore 11.83 USD per case of severe PPH averted and 113.91 USD per DALY averted. The ICER in terms of DALYs is below both the weighted gross domestic product (GDP)-based threshold (2,816 USD) and opportunity-cost based threshold (1,690 USD) (Extended Data Table 1), suggesting the E-MOTIVE intervention is cost-effective. Figure 3 shows the probability of the E-MOTIVE intervention being cost-effective compared with usual care across a range of willingness-to-pay (WTP) thresholds per DALY averted. For thresholds of WTP per DALY averted greater than approximately 1,500 USD, there is >80% probability that the E-MOTIVE intervention is cost-effective (Fig. 3).

The dashed lines show the expected WTP for a DALY averted, as estimated from WHO recommendations (green) and Woods and colleagues (blue).

If the device cost of the calibrated drape is reduced to 1 USD (2023 prices), the E-MOTIVE intervention becomes comparable in cost to usual care, while being more effective (Table 2). Further reductions in the cost of the calibrated drape could potentially result in cost savings. Additional sensitivity analyses to explore the impact of the costing assumptions and the use of multiple imputation (Supplementary Tables 3–5) made no substantial difference to the base-case results; the E-MOTIVE intervention remained cost-effective.

The mean per-patient total costs, DALYs and ICERs from the country-level analyses are presented in Extended Data Table 2. These were estimated using fully pooled, one-country costing models. Briefly, the E-MOTIVE intervention was judged to be cost-effective for each participating country when the ICERs were compared against both country-specific GDP-based WTP thresholds and opportunity-cost-based WTP thresholds (Extended Data Table 1). In South Africa, where the cost of calibrated drapes was lower relative to other resources, the E-MOTIVE intervention was estimated to be less expensive than usual care and, therefore, the dominant intervention based on the point estimates. Accordingly, exploratory analyses (see Supplementary Information, p. 7–11) suggest the budget impact of delivering the E-MOTIVE intervention in these countries would be insubstantial.

This study assessed the cost-effectiveness of early detection of PPH using a calibrated drape and treatment using the WHO first-response treatment bundle, which included uterine massage, oxytocic drugs, TXA, IV fluids and a process for examination and escalation, compared with usual care. The findings suggest that early detection of PPH using a calibrated blood-loss collection drape and treatment with the WHO first-response bundle is cost-effective compared with usual care. Our sensitivity analysis suggested that for WTP values above 1,500 USD per DALY averted there is more than an 80% probability of the E-MOTIVE intervention being cost-effective. Furthermore, deterministic sensitivity analyses showed that potential reductions in the cost of the calibrated blood-collection drape could lead to cost savings, substantially improving the affordability of the E-MOTIVE intervention.

Although a formal quantification of resource use relating to the implementation strategies used to support the E-MOTIVE intervention was not conducted as part of the trial, emerging data suggest the cost of implementation can be effectively absorbed into the existing healthcare system. The post-trial implementation pivot in the four countries indicates that implementing E-MOTIVE does not necessitate additional staffing, and on-site training can be conducted with negligible cost implications. Furthermore, costs related to PPH trolleys or carry cases are minimal and nonrecurrent, while the utilization of audit and feedback solutions and champions does not require additional resources (E-MOTIVE implementation pivot team, personal communication).

The study benefited from a large sample size recruited from 78 hospitals across four countries, broad inclusion criteria to capture all patients with vaginal births in the trial hospitals, and a wide range of primary data. However, the study is not without limitations. Although the analysis considered a range of costs for calibrated blood-collection drapes to account for potential price variations due to increased production, the cost-effectiveness implications of emerging sustainable and climate-friendly alternative devices could not feasibly be assessed15. Also, owing to the pragmatic design of the trial, extensive bottom-up costing of all resource items was not conducted. This naturally increases the uncertainty around the unit cost estimates used in the analysis. However, when feasible, cost estimates were obtained from established sources and other secondary sources based on bottom-up costing. Some assumptions were required to estimate country-specific unit costs when these were not available. All assumptions were agreed upon before any analysis was undertaken, and sensitivity analyses exploring their importance found that they did not substantially impact the cost-effectiveness results.

Furthermore, PPH and associated maternal mortality can involve considerable economic costs to patients, their families and wider society5,6,7. Owing to the pragmatic design of the trial, these costs were not captured. Given that there were fewer cases of severe PPH and less severe PPH in the E-MOTIVE group, and maternal deaths from bleeding, though rare, were in the same direction, it is likely that an analysis from the societal perspective—which considers medical and nonmedical costs not directly linked to the intervention—would produce even more favorable cost-effectiveness estimates for the E-MOTIVE intervention.

In addition, this analysis was conducted alongside a large international, cluster-randomized trial with a baseline control phase that presents complexities with respect to data analysis; for example, randomization took place at the cluster level, but outcomes were measured at the level of the individual. This was addressed using methods to account for the hierarchical nature of the data, and the analysis was adjusted for imbalances in outcomes during the baseline phase across trial groups. In addition, due to the substantial loss of power that would be experienced by analyzing countries in isolation, country-specific cost-effectiveness analyses were not conducted. However, we assessed cost-effectiveness from the perspective of each participating country based on whole trial data. To this end, we conducted fully pooled, one-country costing cost-utility analyses (CUAs) in which clinical data from all participating countries were pooled, and country-specific unit costs and life-expectancy data were applied to all patients in the trial. Although not fully country specific, we believe these estimates provide useful indicative information on cost-effectiveness for decision-makers given the widespread occurrence of visual blood loss estimation, and delayed and inconsistent use of effective PPH interventions, such as TXA, across countries. However, these estimates should be interpreted with caution.

Finally, this analysis does not quantify the potential health equity impacts associated with delivering the E-MOTIVE intervention—information likely to be important to decision-makers. The methods of conventional cost-effectiveness analysis (CEA) focus on efficiency, that is, maximizing population health gain from available resources, rather than reducing health inequities. Although frameworks to robustly incorporate equity concerns into CEA are emerging, the substantial data requirements to conduct such an analysis were not feasible for the present analysis.

In summary, our findings suggest that early detection of PPH and bundled treatment for PPH is cost-effective. Therefore, provision of calibrated blood-collection drapes and use of bundled first-response treatment can be considered a worthwhile use of constrained healthcare budgets, and every effort should be made to adhere to the WHO recommendations.

The E-MOTIVE trial was an international, parallel cluster-randomized trial that included a baseline control phase14. A cluster design was required as the E-MOTIVE intervention was delivered at the hospital level, targeting health care providers. Between August and October 2021, all participating hospitals entered a 7-month baseline period during which they provided usual care for PPH in patients having vaginal delivery. Following this 7-month baseline period, hospitals were randomly assigned, in a 1:1 ratio, to continue providing usual care or to receive the E-MOTIVE intervention for 7 months, with a 2-month ‘transition phase’ to allow hospitals to adapt clinical practices for intervention delivery.

A minimization algorithm generated by an independent statistician was used to ensure balance between the intervention hospitals and usual-care hospitals within each country for key prognostic variables, including the number of vaginal births per hospital, the prevalence of primary-outcome events (for the clinical analysis) during the baseline, the quality of oxytocin and the number of hospitals per country.

We included secondary-level hospitals in Kenya, Nigeria, South Africa and Tanzania. Hospitals in Pakistan were initially included in the baseline phase but could not be included in the randomization process (Fig. 2). Hospitals were eligible for inclusion if they were geographically and administratively distinct from each other, had between 1,000 and 5,000 vaginal births per year, and were able to provide comprehensive obstetrical care with the ability to perform surgery for PPH. Hospitals were excluded if they had already implemented a treatment bundle for PPH. Written permission was granted by each participating hospital for clinical staff to extract anonymized clinical-outcome data for each vaginal birth.

The E-MOTIVE intervention consisted of a blood-collection drape, with calibrated lines to measure blood-loss volume, for early detection of PPH and the WHO-proposed first-response treatment bundle, which included uterine massage, oxytocic drugs, TXA, IV fluids and a process for examination and escalation (Fig. 1). Detailed information on the E-MOTIVE intervention is published elsewhere14.

In usual care, blood loss was estimated visually, with healthcare providers relying on their perceptions to subjectively assess the volume of blood lost. First-response treatment for PPH typically consisted of some or all of the components of the WHO-proposed first-response bundle. These were typically administered sequentially, with oxytocic drugs given as first-line treatment and TXA reserved for refractory bleeding. Established dosage regimens for usual care were applied, consistent with the E-MOTIVE group (Fig. 1).

Noncalibrated drapes, without warning or action lines, were used in the usual-care group hospitals to quantify blood loss for the purpose of the trial.

We estimated cost-effectiveness based on outcomes of severe PPH prevented and DALYs averted.

Severe PPH, defined as blood loss of at least 1,000 ml, was measured at 1 h and, if there was continued bleeding, for up to 2 h postpartum. Blood loss was objectively measured with the use of a blood-collection drape. Calibrated drapes were used in the hospitals in the E-MOTIVE group to enable early and accurate diagnosis of PPH and to obtain data on blood loss. Noncalibrated drapes were used in the hospitals in the usual-care group to obtain data on blood loss. Data on blood loss were source-verified by capturing a photograph of the drape with collected blood inside it, on a digital weighing scale, with the weight visible in the photograph. Only data that had been source-verified were used in the analysis.

This outcome differs from the primary outcome in the E-MOTIVE clinical analysis, which was a composite of severe PPH, laparotomy for bleeding or maternal death from bleeding. Given that composite outcomes are generally inadequate for economic evaluation due to varying component importance, disaggregation is recommended16. However, the infrequency of laparotomies and maternal deaths from bleeding in the trial limited a meaningful cost-effectiveness assessment based on these outcomes.

The DALY is a composite summary measure of disease burden that accounts for both mortality and nonfatal health consequences and is the preferred metric for economic evaluations to support resource allocation decisions in LMICs17. DALYs were estimated on the basis of nonfatal PPH events and maternal death from bleeding for both arms of the trial.

For nonfatal PPH events, years lived with disability were estimated on the basis of the magnitude of the disability and its duration. Disability weights for severe PPH (0.324 (≥1,000 ml blood lost)) and less severe PPH (0.114 (<1,000 ml blood lost)) were drawn from the Global Burden of Disease study18. The duration of disability due to PPH (both severe and less severe) was considered to last for a postpartum period of 6 weeks. Given that the trigger criterion of the E-MOTIVE intervention imposes a benefit on less-severe PPH, it was imperative to include disability for less-severe PPH to ensure relevant effects were captured.

Years of life lost for premature death due to bleeding were calculated using life expectancy of country-specific female populations drawn from Global Burden of Disease abridged life tables19. Years of life lost were calculated using a discount rate of 3%, as recommended for economic evaluations in global health17.

Resource use information was collected prospectively via electronic case report forms and recorded in REDCap (version 10.9.0–13.3.2). Information was collected from the perspective of the healthcare system for calibrated drapes, uterotonic drugs, TXA, IV fluids, duration of hospitalization, intensive care unit (ICU) admission, transfer to a higher-level facility, blood transfusions, postpartum laparotomy, hysterectomy, nonpneumatic anti-shock garments, uterine balloon tamponades and bimanual compression. When necessary, data from an observational study conducted alongside the E-MOTIVE trial and expert clinical opinion from within the research study team supplemented case report form information.

Extended Data Table 3 presents the unit costs used in the analysis. Calibrated blood-collection drape costs were obtained from Excellent Fixable Drapes in India, the manufacturer and supplier of the drapes used in the E-MOTIVE trial. We considered the price at which the drapes are currently being procured, 1.25 USD, in our base-case analysis. Costs of oxytocic drugs and TXA were obtained from a recent publication by the United States Agency for International Development Global Health Supply Chain Program20. Uterotonic drug costs were sourced from the United Nations Populations Fund Product Catalogue, while the TXA costs reported were the United States Agency for International Development wholesale prices. We obtained costs of IV fluids from the International Medical Product Price Guide, a recommended source of medication costs in LMIC settings21. An adjustment of 25% was used to account for shipping and handling charges, as well as internal distribution of traded goods22.

Country-specific unit cost estimates for non-ICU hospitalization in secondary-level hospitals were obtained from the WHO-CHOICE initiative23,24. Country-specific personnel costs were obtained from publicly available records regarding health sector pay, and personal communication with E-MOTIVE country trial management groups25; costs from the latter were based on local government salaries. Conservative estimates of the lowest-grade doctor who could attend a case of severe PPH were used. We used other secondary sources to estimate the cost of blood transfusions, additional treatment interventions, transfer to a higher-level facility and ICU admission4,26,27,28,29,30.

Due to a lack of cost data for postpartum laparotomy, we assumed a unit cost equivalent to 80% of a hysterectomy, based on expert clinical opinion from within the E-MOTIVE study team. Furthermore, we estimated unit costs for bimanual compression based on personnel requirements and procedure duration, and for uterine balloon tamponades in Kenya, Nigeria and Tanzania, we estimated costs considering materials and labor required for an improvised device. For the base case, we did not apply unit costs to activities perceived as a reprioritization of existing staff time, that is, uterine massage and examination, as we assumed no additional resource was required. Additional details on costing assumptions are provided in Extended Data Fig. 1.

To standardize unit costs across countries where data were unavailable, a market basket approach was used, wherein an index table based on WHO-CHOICE estimates (Extended Data Table 4) was used to indicate the relative mean cost of estimates for inpatient and outpatient health service delivery for each country pair in the study22,23,24. The market basket approach is an established costing method for the development of a complete set of country-specific unit cost data in the economic evaluation of multinational trials22. All unit costs were adjusted to 2022 USD using average exchange rates and the average US inflation rate between the price base year used in individual studies and 2022, as recommended when there is a relatively high proportion of imported commodities in economic analyses31. Given the short follow-up period of the trial, costs were not discounted.

The economic evaluation comprised two main analyses: a CEA based on the outcome of cost per case of severe PPH prevented and a CUA based on the outcome of cost per DALY averted. Both were carried out on an intention-to-treat basis and relied on complete case analysis wherein cases without source-verified blood loss data were excluded.

Following recommendations for the economic evaluation of cluster and multinational trials32,33, we used multilevel modeling to estimate the difference in mean costs and outcomes between the E-MOTIVE and usual-care groups. Multilevel modeling accounts for unobserved cluster-specific effects on costs and outcomes and facilitates the estimation of cost-effectiveness across the whole sample34. Consistent with the clinical analysis, we fit generalized linear mixed models incorporating a constrained baseline analysis14. For severe PPH, we used the binomial family and logit link, in addition to robust standard errors, followed by marginal standardization to estimate risk difference. Differences in mean costs and DALYs were estimated using the Gaussian family and identity link, in combination with nonparametric permutation tests given the inherent skewness of such data35. We included fixed effects for allocated exposure to E-MOTIVE, time period, country and covariates used in the randomization method (number of vaginal births per hospital, the proportion of patients with a clinical primary-outcome event at each hospital, and the quality of oxytocin at each hospital during the baseline phase). We adjusted for clustering using random cluster and cluster-by-period effects.

Model estimates of the difference in costs and outcomes were used to derive an incremental cost per case of severe PPH prevented and an incremental cost per DALY averted. For the CUA, we used two thresholds to judge the cost-effectiveness of the E-MOTIVE intervention (Extended Data Table 1): a weighted threshold based on the WHO recommended threshold for a ‘highly cost-effective’ intervention of the countries’ per capita GDP and a weighted threshold based on recently advocated opportunity-cost based thresholds put forward by Woods and colleagues36,37,38, equivalent to 51% GDP per capita for Kenya, Nigeria and Tanzania, and 71% GDP per capita for South Africa.

We conducted sensitivity analyses to quantify the uncertainty relating to key assumptions and sampling variations. To characterize the inherent uncertainty around incremental cost-effectiveness estimates, we used nonparametric clustered bootstrapping with multilevel models to generate 1,000 paired estimates of incremental mean total costs and DALYs. These estimates were used to construct a cost-effectiveness acceptability curve that shows the probability that the E-MOTIVE intervention is cost-effective across a range of WTP threshold values per additional DALY averted39. We also conducted deterministic sensitivity analyses on input parameters for the base-case analysis (Supplementary Information, p. 3). This included varying the device cost of the calibrated drapes to 1 USD, 0.75 USD and 0.50 per unit (2023 prices) respectively, considering potential price decreases with expanded production.

Given that only source-verified blood-loss data were used in the main analysis, we conducted a sensitivity analysis using multiple imputation to assess the effect of missing data. Missing data were imputed under the assumption that data were missing at random, with an allowance for clustering. The multiple imputation was performed using chained equations. Differences between the E-MOTIVE and usual-care groups in terms of risk of severe PPH, means costs and mean DALYs from the seven multiply imputed datasets were obtained using multilevel models in the same manner as the main analysis and pooled using Rubin’s rules.

To provide indicative context for local decision-makers, we estimated the cost-effectiveness of the E-MOTIVE intervention from the perspective of each participating country using four fully pooled, one-country costing CUAs. Clinical outcome and utilization data from all participating countries were pooled, and country-specific unit costs and life-expectancy data were applied to all patients in the trial. The country-level analyses were adjusted analogously to the main analyses. Model estimates of differences in cost and DALYs were used to derive ICERs, which were judged against the country-specific thresholds reported in Extended Data Table 4. We extended these estimates to explore the potential budget impact of implementing the E-MOTIVE intervention (Supplementary Information, p. 7–11).

All analyses were carried out using Stata, version 17.1 (StataCorp).

Ethical approval was granted by the University of Birmingham Science, Technology, Engineering and Mathematics (STEM) ethics committee in the UK (ERN_19-1557); the World Health Organization – Human Reproduction Programme (WHO-HRP) (approval for formative phase) in Switzerland; the Kenyatta National Hospital (KNH) – University of Nairobi (UoN) Ethics and Research Committee (KNH-ERC/A/197), the National Commission for Science, Technology and Innovation (NACOSTI) (NACOSTI/P/21/8330), and the Pharmacy and Poisons Board (PPB) in Kenya (PPB/ECCT/20/06/08/2020(122)); the National Health Research Ethics Committee of Nigeria (NHREC) (NHREC/01/01/2007) and National Agency for Food and Drug Administration and Control (NAFDAC) in Nigeria (NAFDAC/DER/VCTD/E-MOTIVE/2022); the University of the Witwatersrand Human Research Ethics Committee (Medical) (M200241), the Eastern Cape Department of Health – Eastern Cape Health Research Committee (EC_202007_014), the KwaZulu-Natal Department of Health – KZN Health Research Committee (KZ_202008_036) and the University of Cape Town – Human Research Ethics Committee in South Africa (091/2020); the Muhimbili University of Health and Allied Sciences (MUHAS) – Senate Research and Publications Committee (DA.282/298/01.C/) and the National Institute for Medical Research in Tanzania (NIMR/HQ/R.8a/Vol. IX/3510). The trial was registered on ClinicalTrials.gov (NCT04341662) and the Pan African Clinical Trials Registry (PACTR202002791391791). All participants provided written informed consent before participation in intervention training.

Local researchers, including national principal investigators from each participating country, contributed to the E-MOTIVE study design. National principal investigators also led the implementation of the study in their respective countries, supported by a national team of local study coordinators and data managers. Additionally, local research midwives/nurses were also employed at each hospital to facilitate data collection and adherence to study protocols. Moreover, both national principal investigators and local study coordinators are acknowledged as authors of publications arising from the E-MOTIVE study.

This research is locally relevant to each of the participating countries as maternal mortality rates due to PPH are highest in sub-Saharan Africa. Co-design workshops were conducted in each country before implementing the E-MOTIVE intervention, which enabled key local stakeholders to contribute to discussions on adapting implementation strategies to local contexts.

Roles and responsibilities were agreed upon among collaborators ahead of the research. Capacity-building plans for local researchers focused on training research hub staff to conduct a large international, cluster-randomized trial, and on training local research midwives/nurses to facilitate implementation of the E-MOTIVE intervention during client care.

This research would not have been severely restricted or prohibited in the setting of the researchers and does not result in stigmatization, incrimination or discrimination to participants. There is a risk to participants (healthcare providers) if their personal data are not adequately protected. However, the study strictly adhered to applicable data protection regulations in each country, including de-identifying data collected from interviews and surveys before review and conducting on-site monitoring visits to ensure secure storage of participant data.

A central sponsor (University of Birmingham) level risk assessment was put in place during the setup phase of the study. Subsequently, within each country, a separate risk assessment was developed in collaboration with the national coordinating team and finalized before data collection commenced. A central and country-specific monitoring plan and data management plan were also put in place.

Lastly, local and regional research relevant to our study was taken into account in the write-up of this manuscript and the wider E-MOTIVE project.

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Patient data cannot be made publicly available due to privacy concerns. The complete de-identified patient data that support the findings of this study can be obtained from the Chief Investigator of the E-MOTIVE trial, on approval from the E-MOTIVE Trial Data Analysis Sub-Committee. Approval from this committee can be requested by directly contacting the Chief Investigator ([email protected]), with an expected review period of approximately 2–3 months. After approval, researchers will be granted access to perform analyses, ensuring data security and confidentiality, with measures in place to prevent any breach of personal information. Additional data used for the analysis are publicly available and referenced in Methods and Supplementary Information. The parameter values and their sources are reported in Extended Data Table 3 and Supplementary Tables 6–9.

Stata codes are available via GitHub at https://github.com/ewbham/E-MOTIVE.

Kassebaum, N. J. et al. Global, regional, and national levels and causes of maternal mortality during 1990-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 384, 980–1004 (2014).

Article PubMed PubMed Central Google Scholar

Sheldon, W. R. et al. Postpartum haemorrhage management, risks, and maternal outcomes: findings from the World Health Organization Multicountry Survey on Maternal and Newborn Health. BJOG 121, 5–13 (2014).

Article PubMed Google Scholar

Say, L. et al. Global causes of maternal death: a WHO systematic analysis. Lancet Glob. Health 2, e323–e333 (2014).

Article PubMed Google Scholar

Theunissen, F. et al. Cost of hospital care of women with postpartum haemorrhage in India, Kenya, Nigeria and Uganda: a financial case for improved prevention. Reprod. Health 18, 18 (2021).

Article PubMed PubMed Central Google Scholar

Kes, A. et al. The economic burden of maternal mortality on households: evidence from three sub-counties in rural western Kenya. Reprod. Health 12, S3 (2015).

Article PubMed PubMed Central Google Scholar

Molla, M., Mitiku, I., Worku, A. & Yamin, A. Impacts of maternal mortality on living children and families: a qualitative study from Butajira, Ethiopia. Reprod. Health 12, S6 (2015).

Article PubMed PubMed Central Google Scholar

Bazile, J. et al. Intergenerational impacts of maternal mortality: qualitative findings from rural Malawi. Reprod. Health 12, S1 (2015).

Article PubMed PubMed Central Google Scholar

WHO Recommendations for the Prevention and Treatment of Postpartum Haemorrhage (World Health Organization, 2012).

Vogel, J. P., Oladapo, O. T., Dowswell, T. & Gülmezoglu, A. M. Updated WHO recommendation on intravenous tranexamic acid for the treatment of post-partum haemorrhage. Lancet Glob. Health 6, e18–e19 (2018).

Article PubMed Google Scholar

Hancock, A., Weeks, A. D. & Lavender, D. T. Is accurate and reliable blood loss estimation the ‘crucial step’ in early detection of postpartum haemorrhage: an integrative review of the literature. BMC Pregnancy Childbirth 15, 230 (2015).

Article PubMed PubMed Central Google Scholar

Althabe, F. et al. Postpartum hemorrhage care bundles to improve adherence to guidelines: a WHO technical consultation. Int J. Gynaecol. Obstet. 148, 290–299 (2020).

Article PubMed Google Scholar

Forbes, G. et al. Factors influencing postpartum haemorrhage detection and management and the implementation of a new postpartum haemorrhage care bundle (E-MOTIVE) in Kenya, Nigeria, and South Africa. Implement. Sci. 18, 1 (2023).

Article PubMed PubMed Central Google Scholar

Akter, S. et al. Detection and management of postpartum haemorrhage: qualitative evidence on healthcare providers’ knowledge and practices in Kenya, Nigeria, and South Africa. Front. Glob. Women’s Health 3, 1020163 (2022).

Article Google Scholar

Gallos, I. et al. Randomized trial of early detection and treatment of postpartum hemorrhage. N. Engl. J. Med. 389, 11–21 (2023).

Article PubMed Google Scholar

Singata-Madliki, M. & Hofmeyr, G. J. A novel, re-usable ‘Safe birth Tray’ for postpartum blood loss monitoring: a preliminary acceptability assessment. Int. J. Gynaecol. Obstet. 155, 553–555 (2021).

Article PubMed Google Scholar

Ramsey, S. D. et al. Cost-effectiveness analysis alongside clinical trials II-An ISPOR Good Research Practices Task Force report. Value Health 18, 161–172 (2015).

Article PubMed Google Scholar

Wilkinson, T. et al. The international decision support initiative reference case for economic evaluation: an aid to thought. Value Health 19, 921–928 (2016).

Article PubMed Google Scholar

Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2019 (GBD 2019) Disability Weights (Institute for Health Metrics and Evaluation, 2020).

Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2019 (GBD 2019) Life Tables 1950–2019 (Institute for Health Metrics and Evaluation, 2020).

Uses of Medicines for Prevention and Treatment of Post-partum Hemorrhage and Other Obstetric Purposes (United States Agency for International Development, 2022).

International Medical Product Price Guide (Management Sciences for Health, 2015).

Mulligan, J.-A. et al. Unit costs of health care inputs in low and middle income regions. Disease control priorities project working paper. LSHTM Research Online http://researchonline.lshtm.ac.uk/12952/ (LSHTM, 2005).

WHO-CHOICE Estimates of Cost for Inpatient and Outpatient Health Service Delivery (World Health Organization, 2021).

Stenberg, K., Lauer, J. A., Gkountouras, G., Fitzpatrick, C. & Stanciole, A. Econometric estimation of WHO-CHOICE country-specific costs for inpatient and outpatient health service delivery. Cost. Eff. Resour. Alloc. 16, 11 (2018).

Article PubMed PubMed Central Google Scholar

Salary Scales, with Translation Keys, for Employees on Salary Levels 1 to 12 and Those Employees Covered by Occupation Specific Dispensions (OSDs) (Department of Public Service and Administration, 2023).

UNICEF Supply Catalogue Vol. 2023 (UNICEF Supply Division, 2023).

Mvundura, M. et al. Cost-effectiveness of condom uterine balloon tamponade to control severe postpartum hemorrhage in Kenya. Int J. Gynaecol. Obstet. 137, 185–191 (2017).

Article PubMed Google Scholar

Uniform Patient Fee Schedule 2022 Vol. 2023 (National Department of Health, 2022).

Theron, G. B. Management of postpartum hemorrhage with free-flow pressure controlled uterine balloon. Int J. Gynaecol. Obstet. 142, 371–373 (2018).

Article PubMed Google Scholar

Dayananda, K. et al. Selective non-operative management of abdominal stab wounds is a safe and cost effective strategy: a South African experience. Ann. R. Coll. Surg. Engl. 99, 490–496 (2017).

Article PubMed PubMed Central Google Scholar

Kumaranayake, L. The real and the nominal? Making inflationary adjustments to cost and other economic data. Health Policy Plan 15, 230–234 (2000).

Article CAS PubMed Google Scholar

Gomes, M. et al. Developing appropriate methods for cost-effectiveness analysis of cluster randomized trials. Med. Decis. Mak. 32, 350–361 (2012).

Article Google Scholar

Manca, A., Sculpher, M. J. & Goeree, R. The analysis of multinational cost-effectiveness data for reimbursement decisions: a critical appraisal of recent methodological developments. Pharmacoeconomics 28, 1079–1096 (2010).

Article PubMed Google Scholar

Manca, A., Rice, N., Sculpher, M. J. & Briggs, A. H. Assessing generalisability by location in trial-based cost-effectiveness analysis: the use of multilevel models. Health Econ. 14, 471–485 (2005).

Article PubMed Google Scholar

Thompson, J., Davey, C., Hayes, R., Hargreaves, J. & Fielding, K. swpermute: permutation tests for stepped-wedge cluster-randomised trials. Stata J. 19, 803–819 (2019).

Article PubMed PubMed Central Google Scholar

Woods, B., Revill, P., Sculpher, M. & Claxton, K. Country-level cost-effectiveness thresholds: initial estimates and the need for further research. Value Health 19, 929–935 (2016).

Article PubMed PubMed Central Google Scholar

The World Health Report 2002: Reducing Risks, Promoting Healthy Life (World Health Organization, 2002).

GDP per Capita (Current US$) (The World Bank, 2023).

Fenwick, E., Claxton, K. & Sculpher, M. Representing uncertainty: the role of cost-effectiveness acceptability curves. Health Econ. 10, 779–787 (2001).

Article CAS PubMed Google Scholar

Download references

This study was funded by the Bill & Melinda Gates Foundation: E.V.W., A.D., J.T.M., K.-M.M., L.E.B., K.N.S., K.H., L.M., A.C., T.E.R., I.G., H.G., F.A.-A., Z.Q., G.J.H., N.M., S.F., S.M., O.T.O. and I.D.G. were supported by investment grant (INV-001393). The funder of the study had no role in study design, data collection, analysis, interpretation or writing of the report. We thank M. Podesek, I. Horne, F. Althabe, J. Smith, C. Evans, S. Miller, M. Gulmezoglu, J. Okore, A. Ado Wakili, M. Singata-Madliki, E. Muller and A. Mwampashi and all those not otherwise mentioned above who contributed to the E-MOTIVE trial on which this CEA is based on.

College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK

Eleanor V. Williams, Raymond Oppong, Samuel J. Perry, Adam J. Devall, James T. Martin, Kristie-Marie Mammoliti, Leanne E. Beeson, Kulandaipalayam N. Sindhu, Lee Middleton, Karla Hemming, Arri Coomarasamy & Tracy E. Roberts

Melbourne School of Population and Global Health, University of Melbourne, Melbourne, Victoria, Australia

Ilias Goranitis

African Center of Excellence for Population Health and Policy, College of Health Sciences, Bayero University, Kano, Nigeria

Hadiza Galadanci

Department of Obstetrics and Gynecology, Muhimbili University of Health and Allied Sciences, Dar es Salaam, Tanzania

Fadhlun Alwy Al‑beity

Department of Obstetrics and Gynecology, University of Nairobi, Nairobi, Kenya

Zahida Qureshi

Effective Care Research Unit, University of the Witwatersrand, Johannesburg, South Africa

G. Justus Hofmeyr

Department of Obstetrics and Gynecology, University of Botswana, Gaborone, Botswana

G. Justus Hofmeyr

KwaZulu-Natal Department of Health, Pietermaritzburg, South Africa

Neil Moran

Department of Obstetrics and Gynaecology, University of Cape Town, Cape Town, South Africa

Sue Fawcus

Eastern Cape Department of Health, Bhisho, South Africa

Sibongile Mandondo

Department of Obstetrics and Gynaecology, Walter Sisulu University, Mthatha, South Africa

Sibongile Mandondo

UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction, Department of Sexual and Reproductive Health and Research, World Health Organization, Geneva, Switzerland

Olufemi T. Oladapo & Ioannis D. Gallos

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

E.V.W., I.G., R.O., S.J.P. and T.E.R. were responsible for the economic analysis. E.V.W. did the economic analysis and received advice from I.G., R.O., S.J.P. and T.E.R. T.E.R. supervised the economic analysis. A.C., A.J.D., J.T.M., K.H., L.M., I.G., T.E.R. and I.D.G. contributed to the study design and methodology. H.G., F.A.-A., Z.Q., G.J.H., N.M. and S.F. were responsible for the oversight of the study in their respective countries, acting as principal investigators. L.E.B., K.-M.M., K.N.S., A.J.D. and I.D.G. were responsible for trial management and had oversight of data collection. J.T.M., K.H. and L.M. did the statistical analysis for clinical outcomes. All authors contributed to data interpretation. E.V.W. wrote the first draft of the manuscript. All authors contributed to critical revision of the manuscript for important intellectual content and gave final approval.

Correspondence to Tracy E. Roberts.

G.J.H. has consulted for Equalize Health, a not-for-profit health technology company. This did not influence the design, conduct or reporting of the research presented in this manuscript. The other authors declare no competing interests.

Nature Medicine thanks Aduragbemi Banke-Thomas, Brooke Nichols and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Sonia Muliyil, in collaboration with the Nature Medicine team.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Costing assumptions used for economic evaluation.

Supplementary Tables 1 and 2. List of deterministic sensitivity analysis conducted. Supplementary Tables 3–5. Potential budget impact analysis. Supplementary Tables 6–9.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

Williams, E.V., Goranitis, I., Oppong, R. et al. A cost-effectiveness analysis of early detection and bundled treatment of postpartum hemorrhage alongside the E-MOTIVE trial. Nat Med 30, 2343–2348 (2024). https://doi.org/10.1038/s41591-024-03069-5

Download citation

Received: 18 March 2024

Accepted: 16 May 2024

Published: 06 June 2024

Issue Date: August 2024

DOI: https://doi.org/10.1038/s41591-024-03069-5

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative