Computer generated transcript

Warning!
The following transcript was generated automatically from the content and has not been checked or corrected manually.

The new england journal of medicine established in 1812 July 6, 2023 vol. 389 no. 1 Randomized Trial of Early Detection and Treatment of Postpartum Hemorrhage I. Gallos, A. Devall, J. Martin, L. Middleton, L. Beeson, H. Galadanci, F. Alwy Al‑beity, Z. Qureshi, G.J. Hofmeyr, N. Moran, S. Fawcus, L. Sheikh, G. Gwako, A. Osoti, A. Aswat, K.‑M. Mammoliti, K.N. Sindhu, M. Podesek, I. Horne, R. Timms, I. Yunas, J. Okore, M. Singata‑Madliki, E. Arends, A.A. Wakili, A. Mwampashi, S. Nausheen, S. Muhammad, P. Latthe, C. Evans, S. Akter, G. Forbes, D. Lissauer, S. Meher, A. Weeks, A. Shennan, A. Ammerdorffer, E. Williams, T. Roberts, M. Widmer, O.T. Oladapo, F. Lorencatto, M.A. Bohren, S. Miller, F. Althabe, M. Gülmezoglu, J.M. Smith, K. Hemming, and A. Coomarasamy abstract BACKGROUND Delays in the detection or treatment of postpartum hemorrhage can result in com- The authors’ full names, academic de‑ plications or death. A blood-collection drape can help provide objective, accurate, Appendix. Dr. Devall can be contacted the and early diagnosis of postpartum hemorrhage, and delayed or inconsistent use of at a.j.devall@bham.ac.uk or at the Insti‑ effective interventions may be able to be addressed by a treatment bundle. tute of Metabolism and Systems Re‑ search, WHO Collaborating Centre on METHODS Global Women’s Health, College of Med‑ We conducted an international, cluster-randomized trial to assess a multicomponent ical and Dental Sciences, University of clinical intervention for postpartum hemorrhage in patients having vaginal delivery. Birmingham, 4th Fl. E., Heritage Bldg., The intervention included a calibrated blood-collection drape for early detection of B15 2TH, United Kingdom.n, Birmingham postpartum hemorrhage and a bundle of first-response treatments (uterine massage, ThisarticlewaspublishedonMay9,2023, oxytocic drugs, tranexamic acid, intravenous fluids, examination, and escalation), at NEJM.org. supported by an implementation strategy (intervention group). Hospitals in the con- This is the New England Journal of Medi- trol group provided usual care. The primary outcome was a composite of severe cine version of record, which includes all postpartum hemorrhage (blood loss, ≥1000 ml), laparotomy for bleeding, or mater- Journal editing and enhancements. The nal death from bleeding. Key secondary implementation outcomes were the detec- Author Accepted Manuscript, which is the tion of postpartum hemorrhage and adherence to the treatment bundle. author’s version after external peer review available at PubMed Central.e Journal, is RESULTS A total of 80 secondary-level hospitals across Kenya, Nigeria, South Africa, and Tan- N Engl J Med 2023;389:11-21. zania, in which 210,132 patients underwent vaginal delivery, were randomly assigned DOI: 10.1056/NEJMoa2303966 to the intervention group or the usual-care group. Among hospitals and patients with Copyright © 2023 Massachusetts Medical Society. CME data, a primary-outcome event occurred in 1.6% of the patients in the intervention at NEJM.org group, as compared with 4.3% of those in the usual-care group (risk ratio, 0.40; 95% confidence interval [CI], 0.32 to 0.50; P<0.001). Postpartum hemorrhage was detected in 93.1% of the patients in the intervention group and in 51.1% of those in the usual-care group (rate ratio, 1.58; 95% CI, 1.41 to 1.76), and the treatment bundle was used in 91.2% and 19.4%, respectively (rate ratio, 4.94; 95% CI, 3.88 to 6.28). CONCLUSIONS Early detection of postpartum hemorrhage and use of bundled treatment led to a lower risk of the primary outcome, a composite of severe postpartum hemor- rhage, laparotomy for bleeding, or death from bleeding, than usual care among patients having vaginal delivery. (Funded by the Bill and Melinda Gates Founda- tion; E-MOTIVE ClinicalTrials.gov number, NCT04341662.) n engl j med 389;1nejm.org July 6, 2023 11 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. The new england journal ofmedicine ostpartum hemorrhage, defined as strategy for the detection and treatment of post- blood loss of at least 500 ml after birth, partum hemorrhage after vaginal delivery. P accounts for 27% of maternal deaths world- wide.1,2The World Health Organization (WHO) has published and updated several evidence- Methods A Quick Take is available atformed recommendations for the prevention Trial Design and Oversight NEJM.org and treatment of postpartum hemorrhage. 3,De- The E-MOTIVE trial was an international, parallel spite strong efforts to adopt and scale up the use cluster-randomized trial that included a baseline of these recommendations, postpartum hemor- control phase, along with mixed-methods evalu- rhage remains the leading cause of maternal ation. A cluster design was necessary because the complications and death worldwide. Three key intervention was delivered at the hospital level, challenges contribute to this lack of progress. targeting health care providers. The trial was The first challenge is that postpartum hemor- approved by the University of Birmingham, the rhage is often undetected or detected late; thus Ethics Review Committee of the World Health lifesaving treatment is not promptly initiated. In Organization (WHO) (for the formative phase), a large, international, randomized trial of pro- and the relevant ethics and regulatory review com- phylaxis for postpartum hemorrhage in 29,645 mittees in each country. 5 participants, only 53% of the participants in Between August and October 2021, all the whom postpartum hemorrhage developed received participating hospitals entered a 7-month base- a diagnosis and were treated with a uterotonic line period during which they provided usual care drug. The current approach for blood-loss as- for postpartum hemorrhage. After this 7-month sessment at birth is visual estimation, which is baseline period, hospitals were randomly as- widely recognized to be inaccurate and tends to signed, in a 1:1 ratio, in a sequential manner as underestimate blood loss. 6 they approached the end of their assigned base- The second challenge is delayed or inconsis- line phase either to continue providing usual care tent use of interventions for the management of or to receive the trial intervention for 7 months, postpartum hemorrhage. Treatments for post- with an allowance of 2 months for transition in partum hemorrhage are often administered in a order to conduct training and to implement and sequential manner; a health care provider ad- embed the intervention in practice. A minimiza- ministers an intervention and waits to see wheth- tion algorithm that was generated by an inde- er it has had an effect before another intervention pendent statistician was used to ensure balance is administered. However, postpartum hemor- between the intervention hospitals and usual- rhage is a time-critical condition, and delays in care hospitals within each country for key prog- the use of lifesaving interventions can result in nostic variables, including the number of vaginal loss of life. Some effective interventions may not births per hospital (dichotomized with the use be used at all. For example, a survey of hospitals of the median value obtained during the base- in Kenya, Nigeria, South Africa, and Tanzania line phase), the prevalence of primary-outcome showed that tranexamic acid was used late and events during the baseline phase (dichotomized most often as a last resort in women in whom with the use of the median value obtained during surgery for postpartum hemorrhage was indi- the baseline phase), the quality of oxytocin (di- cated.8 chotomized as high or low quality on the basis The third challenge is that despite the avail- of the percent of active ingredient contained in ability of clear recommendations regarding post- the product), and the number of hospitals per partum hemorrhage and their wide dissemina- country (count). During the 7-month implemen- tion, uptake is poor at the point of care. In tation phase, we conducted mixed-methods pro- previous work, we identified several barriers to cess evaluations to assess implementation out- implementation, including limited staffing, lack comes. of relevant knowledge and skills, lack of engage- Trial oversight and monitoring were provided ment from health care providers, and profes- by a trial steering committee and an independent sional attitudes that discouraged task sharing. 9 data monitoring committee. The first two and To address these challenges, we designed a clus- last two authors vouch for the accuracy and com- ter-randomized trial to assess a multicomponent pleteness of the data and for the fidelity of the 12 n engl j med 389;nejm.org July 6, 2023 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. Early Detection and Treatment of PPH E M O T IV E and Trigger Criteria of Uterus ODrugsic TraAcidmic IV Fluids and Escalation Calibrated drape for the Massage until uterus 10 IU IV oxytocin 1 g IV tranexamic acid IV fluids in addition to Ensure bladder is the collection of has contracted or injected or diluted injected or diluted in the infusion should empty, evacuate blood, with trigger for 1 min in 200–500 ml 200 ml crystalloid be given if clinically clots, check for tears 500 ml for the first hr istered over 10-min 10-min periodover citation and willus- examination andl after birth period, plus a main- require a second placenta for com- Observations (blood tenance dose of intravenous access pleteness loss, blood flow, 20 IU IV oxytocin Escalate if bleeding uterine tone) every diluted in 1000 ml does not stop after 15 min documented saline administered first response or on the blood-loss over 4-hr period clinician is unable to monitoring chart (with misoprostol identify or manage Blpulse monitoredd 800 µg if used) cause of bleeding once in the first hr post partum and Implementation Strategies documented on the Audit newsletters: Sharing with all staff monthly rates of detection and bundle use, along with rates of PPH, severe PPH, blood blood-loss moni- transfusion, laparotomy, and death from PPH and giving feedback at monthly departmental meetings toring chart Champions: Midwife and doctor to oversee change, troubleshoot, give feedback on audit newsletters, connect with other champions by means of chats, meetings, and websites for sharing knowledge and lessons learned Trigger Criteria Trolley or carry case: Restocking of all medicines and devices used for treatment of PPH after every use and completion of a Blood loss ≥500 ml stocking checklist at the start of every shift Blood loss ≥300 ml Training: Onsite, simulation-based, and peer-assisted training, lasting from 90 min to an entire workday, facilitated by the use of plus one abnormal provider guides, flipcharts, and job aids displayed in labor wards observation Figure 1. E-MOTIVE Treatment Bundle. Early detection and treatment of postpartum hemorrhage (PPH) involved the use of a blood‑collection drape and the World Health Or‑ ganization first‑response treatment bundle, which together comprise the E‑MOTIVE protocol. Misoprostol may be administered rectally or sublingually. IV denotes intravenous. trial to the protocol, available with the full text of granted by each participating hospital for clini- this article at NEJM.org. Commercial suppliers cal staff employed in that hospital to extract and contractors had no role in the design of the anonymized clinical-outcome data for each vagi- trial; the collection, analysis, or interpretation of nal birth. the data; or the writing of the manuscript. E-MOTIVE Intervention and Usual Care Participating Hospitals The E-MOTIVE intervention consisted of a cali- Hospitals were the randomization unit. We in- brated drape for early detection of postpartum cluded secondary-level hospitals in Kenya, Nigeria, hemorrhage and the WHO first-response treat- South Africa, and Tanzania; hospitals in Pakistan ment bundle, which included uterine massage, were initially included in the baseline phase but oxytocic drugs, tranexamic acid, intravenous could not be included in the randomization pro- fluids, and a process for examination and esca- cess (see below). Hospitals were eligible for in- lation (Fig. 1; and see the Supplementary Ap- clusion if they were geographically and adminis- pendix, available at NEJM.org). Implementation tratively distinct from each other, had between was supported by several components, including 1000 and 5000 vaginal births per year, and were the use of trolleys or carry cases for postpartum able to provide comprehensive obstetrical care hemorrhage; simulation-based, on-site training; with the ability to perform surgery for postpar- local champions (midwives and doctors who tum hemorrhage. We excluded hospitals that had lead and support change in participating hospi- already implemented a bundle for treatment of tals); and audit and feedback of actionable data postpartum hemorrhage. Written permission was to providers. The implementation strategy was n engl j med 389;1 nejm.org July 6, 2023 13 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. Thenew england journal of medicine informed8,9 the findings from our formative had been source-verified were used in the analy- research and refined during multidisciplinary sis for blood-loss outcomes, according to the workshops in each of the participating countries. recommendation of the independent data moni- The intervention was piloted and refined in three toring committee and the trial steering commit- hospitals in each country that did not participate tee after data-reliability concerns were raised at in the main trial. an external pilot site. An end-point review com- The control hospitals, where usual care was mittee whose members were unaware of the trial- provided, estimated blood loss visually and used group assignments assessed case summaries to various interventions for postpartum hemorrhage confirm whether any postpartum laparotomy or in accordance with local or national guidelines. maternal death was due to bleeding. These interventions were often administered se- Prespecified key secondary implementation out- quentially, with oxytocic drugs given as first-line comes were the detection of postpartum hemor- treatment and tranexamic acid reserved for re- rhage (assessed in patients with a diagnosis of fractory bleeding. Uncalibrated drapes, without postpartum hemorrhage recorded by the birth alert or action lines, were used in the control attendants, out of the total number of patients hospitals to quantify blood loss for the purpose who had a postpartum hemorrhage as objectively of the trial. Drapes were manufactured and sup- measured in the blood-collection drape), and plied by Excellent Fixable Drapes in India. adherence to the treatment bundle (assessed in Themedications(oxytocicdrugsandtranexam- patients treated with the bundle after a diagno- ic acid) and intravenous fluids that were used in sis of postpartum hemorrhage recorded by the the trial were obtained by means of existing birth attendants, out of the total number of pa- procurement pathways and sourced from local tients in whom postpartum hemorrhage was stocks at the hospitals. Samples of medications objectively measured). Adherence to the treat- from the participating hospitals were analyzed ment bundle was defined as adherence to at least according to the International Pharmacopoeia three core bundle elements: administration of (oxytocin) and British Pharmacopoeia (tranexamic oxytocic drugs, tranexamic acid, and intrave- acid) standards to ensure that they were of ade- nous fluids. quate quality (see the Supplementary Appendix).0 Other secondary outcomes included the indi- vidual components of the primary composite Outcome Measures outcome, postpartum hemorrhage (defined as The primary outcome was a composite of three blood loss of ≥500 ml), death from any cause, clinical outcomes after vaginal birth: severe post- blood transfusion for any cause, blood transfu- partum hemorrhage, defined as blood loss of at sion for postpartum hemorrhage, blood loss as a least 1000 ml after vaginal birth, measured at continuous variable, uterine tamponade use, in- 1 hour and, if there was continued bleeding, for tensive care unit (ICU) admission or higher-level up to 2 hours post partum; postpartum laparoto- hospital transfer, newborn death, implementation my for bleeding at any time up to discharge from outcomes, and resource-use outcomes. The inde- the hospital; or maternal death from bleeding at pendent data monitoring committee monitored any time up to discharge from the hospital. maternal deaths and ICU admissions as markers Blood loss was objectively measured with the use of serious adverse events. A detailed list of all of a blood-collection drape. Uncalibrated drapes the secondary outcomes is provided in the Sup- were used in the hospitals in the usual-care plementary Appendix. group to obtain data on blood loss; calibrated Statistical Analysis drapes were used in the hospitals in the inter- vention group to enable early and accurate diag- In order for the trial to have 90% power at 5% nosis of postpartum hemorrhage as well as to significance to detect a change from 4% to 3% obtain data on blood loss. Data on blood loss (a 25% relative reduction) in the risk of a primary- were source-verified by capturing a photograph outcome event, with allowance for clustering of the drape with collected blood inside it, posi- and for varying cluster sizes across most realis- tioned on a digital weighing scale, with the tic scenarios, we calculated that at least 72 clus- weight visible in the photograph. Only data that ters would be required (inflated to 80 clusters to 14 n engl j med 389;nejm.org July 6, 2023 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. Early Detection and Treatment of PPH allow for withdrawal of 10% of the clusters). All The widths of the confidence intervals have not the analyses were performed according to a been adjusted for multiplicity, so the intervals modified intention-to-treat principle (including should not be used to infer definitive treatment all the hospitals that underwent randomization effects for secondary outcomes. All the analyses and had available data). A full sample-size justi- were carried out with the use of Stata software, fication is provided in the trial protocol. version 17 (StataCorp). The primary comparison was between the hospitals (clusters) that had been randomly as- signed to the intervention group and those ran- Results domly assigned to the usual-care group. For the Hospital and Patient Characteristics primary outcome, we fitted a generalized linear A total of 104 secondary-level hospitals were as- mixed model incorporating a constrained base- sessed for eligibility. Fourteen hospitals were line analysis. We used the binomial distribu- excluded because they had already implemented tion and logit link, followed by marginal stan- an early-detection protocol or treatment bundle dardization to estimate risk differences and risk for postpartum hemorrhage. A total of 90 hos- ratios. All the analyses were adjusted for cluster- pitals in Kenya, Nigeria, Pakistan, South Africa, ing with the use of random cluster and cluster- and Tanzania started the baseline prerandom- by-period effects and used robust standard errors. ization period (Fig. 2). These facilities were rep- The primary analysis was unadjusted, except for resentative of our target population (Table S2). the factors used in the randomization method The independent data monitoring committee (number of vaginal births per hospital, country, recommended completion of the trial before the proportion of patients with a primary-outcome hospitals in Pakistan could undergo randomiza- event at each hospital, and the quality of oxyto- tion, since the required sample size had been cin at each hospital during the baseline phase). achieved in the other four countries. Two hospi- A sensitivity analysis was additionally adjusted tals in Kenya could not implement the full trial for prespecified clinically important prognostic protocol including source-data verification re- factors at the patient level (age, newborn birth quirements for blood-loss measurements; these weight, parity, multiple pregnancy, and mode of hospitals were excluded before randomization. delivery [spontaneous or instrumentally assisted The remaining hospitals stayed in the baseline delivery]). Finally, we allowed for missing covari- phase and provided usual care for 7 to 8 months. ate data by using multiple imputation and an A total of 80 hospitals in Kenya, Nigeria, evaluation of none-missing-at-random patterns South Africa, and Tanzania underwent random- under a tipping-point analysis (all allowing for ization, with 40 assigned to the intervention clustering and a number of auxiliary covariates). group and the remaining 40 to continue provid- We analyzed the treatment effect on the pri- ing usual care. Two hospitals in Tanzania, 1 in mary outcome in prespecified subgroups (Table S1 each group, did not receive the assigned inter- in the Supplementary Appendix). Summaries of vention because of participation in a conflicting data (numbers and percentages according to program(Fig.2).Afterrandomization,a2-month randomized group) about the primary outcome, transition period was implemented so that the maternal death, and ICU admission were pro- intervention could be embedded into clinical vided to the independent data monitoring com- practice in the intervention sites. Data that were mittee by the trial statistician (who remained un- collected in the trial groups during this transi- aware of the trial-group assignments) once after tion period did not contribute to the analysis. randomization. Because the interim analyses were Data for analysis were available from 78 hospi- performed with the use of the Peto principle, no tals (from 14 in Kenya, 38 in Nigeria, 14 in adjustment was made in the final P value to deter- South Africa, and 12 in Tanzania), with a total mine statistical significance. Because the statisti- of 210,132 patients (110,473 in the baseline cal analysis plan did not include a provision for phase and 99,659 in the implementation phase) correcting for multiplicity for tests of secondary giving birth in the hospitals during the trial pe- or other outcomes, those results are reported as riod (Table 1). Source-verified data regarding point estimates with 95% confidence intervals. blood loss were available for 206,455 patients n engl j med 389;nejm.org July 6, 2023 15 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. The new england journal of medicine postpartum hemorrhage (oxytocin and tranexamic 104 Hospitals were assessed for eligibility acid) were similar in the two groups (Table 1). Outcomes 14 Were excluded because A primary-outcome event occurred in 794 of inclusion criteria were not met 48,678 patients (1.6%) in the intervention group and in 2139 of 50,044 (4.3%) in the usual-care 90 Initiated the baseline phase of the trial group (risk ratio, 0.40; 95% confidence interval [CI], 0.32 to 0.50; P<0.001) (Table 2). Postpartum 10 Were excluded before hemorrhage was detected in 93.1% of the pa- randomization 2 Were unable to carry out tients in the intervention group and in 51.1% of source-data verification those in the usual-care group (rate ratio, 1.58; 8 Were excluded before 95% CI, 1.41 to 1.76), and adherence to the treat- randomization because the required sample size ment bundle was 91.2% and 19.4%, respectively had been achieved (rate ratio, 4.94; 95% CI, 3.88 to 6.28). The risk of the primary outcome in the intervention group 80 Hospitals in Kenya, Nigeria, South Africa, progressively decreased with time after random- and Tanzania underwent randomization ization, from a mean of 3.8% during the base- line prerandomization phase to 1.1% by the last month of the implementation phase (Fig. 3). 40 Were assigned to the intervention 40 Were assigned to the usual-care The median blood loss was 160 ml (interquar- group group tile range, 100 to 280) in the intervention group 39 Received assigned intervention 39 Received assigned intervention and 220 ml (interquartile range, 120 to 380) in 1 Did not receive assigned inter- 1 Did not receive assigned inter- vention owing to participation in vention owing to participation in the usual-care group (Fig. S1). Postpartum hem- a conflicting program a conflicting program orrhage (defined as blood loss of ≥500 ml) was diagnosed in 8.5% of the patients in the inter- 39 Hospitals (with data for 101,104 39 Hospitals (with data for 109,028 vention group and in 16.7% of those in the usual- patients) were included in the patients) were included in the analysis analysis care group (risk ratio, 0.51; 95% CI, 0.44 to 0.60), and severe postpartum hemorrhage (defined as 99,399 Patients had primary-outcome 107,056 Patients had primary-outcome blood loss of ≥1000 ml) in 1.6% and 4.3%, re- data data 50,721 Patients in the baseline phase 57,012 Patients in the baseline phase spectively (risk ratio, 0.39; 95% CI, 0.31 to 0.49). 48tation phases in the implemen- 50,phaseatients in the implementation Postpartum blood transfusion for bleeding was 1705 Patients had missing primary- 1972 Patients had missing primary- outcome data outcome data used in 1.2% of the patients in the intervention 1282 Patients in the baseline phase 1458 Patients in the baseline phase group and in 1.9% of those in the usual-care 423 Patients in the implementation 514 Patients in the implementation group (risk ratio, 0.71; 95% CI, 0.55 to 0.90). phase phase There were 17 maternal deaths in the inter- vention group and 28 deaths in the usual-care Figure 2. Randomization of Hospitals in the Cluster-Randomized Trial. All the participating hospitals entered a 7‑month baseline period in which group (risk ratio, 0.73; 95% CI, 0.40 to 1.31). A total of 12 and 18 of these deaths, respectively, they provided usual care for postpartum hemorrhage. After the baseline period, hospitals were randomly assigned, in a 1:1 ratio, in a sequential were attributed to postpartum bleeding. manner as they approached the end of the baseline phase to either the in‑ There were few cases of laparotomy, compres- tervention group (in which hospitals implemented the E‑MOTIVE protocol) sion sutures, uterine-artery ligation or hysterec- or the usual‑care group (in which hospitals continued to provide usual care). The 80 hospitals that underwent randomization were in Kenya, Nige‑ tomy, a situation that limited meaningful com- parisons between the trial groups. The results for ria, South Africa, and Tanzania. Two hospitals in Tanzania, 1 in each trial group, did not receive the assigned intervention because of participation in all the secondary outcomes are shown in Table 2 a conflicting program and were not included in the analyses. and Tables S3 and S4. Exploratory analyses indicated consistent ef- fects across subgroups (Table S5). Results of sensi- (98% follow-up), and data on laparotomy and tivity analyses, including analyses with full adjust- maternal death were available for all the patients. ment for covariates and with multiple imputation The hospital characteristics, patient characteris- for missing data, were consistent with those of the tics, and the availability of essential drugs for primary analysis (Tables S6, S7, and S8). 16 n engl j med 389;1 nejm.org July 6, 2023 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. Early Detection and Treatment of PPH Table 1. Characteristics of Participating Hospitals and Clinical Characteristics of the Patients.* Intervention Usual Care Characteristic (N=49,101) (N=50,558) Hospital characteristics Median no. of vaginal births per hospital (IQR) 1136 (775–1881) 1263 (787–1854) Median availability of bundle components (IQR) — % of time available Oxytocin 100 (100–100) 100 (100–100) Tranexamic acid 100 (100–100) 100 (100–100) Intravenous fluid 100 (100–100) 100 (100–100) Clinical characteristics of the patients Country — no. (%) Kenya 11,475 (23.4) 9,992 (19.8) Nigeria 17,300 (35.2) 20,909 (41.4) South Africa 9,668 (19.7) 9,030 (17.9) Tanzania 10,658 (21.7) 10,627 (21.0) Median age (IQR) — yr 26 (21–31) 26 (21–30) No. of previous births† Median (IQR) 1 (0–2) 1 (0–3) Distribution — no./total no. (%) 0 17,719/47,575 (37.2) 17,642/48,228 (36.6) 1–4 25,477/47,575 (53.6) 25,805/48,228 (53.5) ≥5 4,379/47,575 (9.2) 4,781/48,228 (9.9) Previous cesarean section — no./total no. (%) 1456/48,911 (3.0) 1281/50,364 (2.5) Postpartum hemorrhage in previous pregnancy — 487/47,869 (1.0) 405/48,925 (0.8) no./total no. (%) Multiple pregnancy — no. (%) 804 (1.6) 960 (1.9) Delivery with forceps or vacuum — no./total no. (%) 358/49,100 (0.7) 278/50,558 (0.5) Birth weight — g 3033±559 3044±552 Median gestational age (IQR) — wk 39 (37–40) 38 (37–39) Gestational age <37 wk — no./total no. (%) 6,877/44,389 (15.5) 8,565/48,844 (17.5) Antepartum hemorrhage — no./total no. (%) 372/48,000 (0.8) 275/48,692 (0.6) Preeclampsia — no./total no. (%) 1,038/48,280 (2.1) 1,182/50,171 (2.4) Labor augmented or induced — no. (%) 6,811 (13.9) 9,323 (18.4) Retained placenta or manual removal of placenta — no. (%) 566 (1.2) 1,072 (2.1) * Plus–minus values are means ±SD. Hospitals in the intervention group implemented a protocol for the early detection and treatment of postpartum hemorrhage that included a calibrated blood‑collection drape and a bundle of first‑response treatments; hospitals in the control group provided usual care. Each trial group in this cluster‑randomized trial includ‑ ed 39 clusters (i.e., hospitals). The numbers in the column heads are the numbers of patients. Percentages may not total 100 because of rounding. IQR denotes interquartile range. †Data on the number of previous births were missing for 1526 patients in the intervention group and for 2330 in the usual‑care group. Discussion my for postpartum hemorrhage, or maternal death from postpartum hemorrhage — after The E-MOTIVE intervention resulted in a 60% vaginal birth across secondary-level hospitals in lower risk of the primary outcome — a compos- Kenya, Nigeria, South Africa, and Tanzania. This ite of severe postpartum hemorrhage, laparoto- benefit was presumably attributable to observed n engl j med 389;1 nejm.org July 6, 2023 17 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. The new england journal of medicine Table 2. Primary Outcomes, Key Secondary Implementation Outcomes, and Secondary Outcomes.* Intervention Usual Care Risk or Rate Ratio Difference Outcome (N=49,101) (N=50,558) (95% CI)† (95% CI)† Primary outcome Composite of severe postpartum hemor‑ 794/48,678 (1.6) 2139/50,044 (4.3) 0.40 (0.32 to 0.50)§ −2.5 (−3.0 to −2.0)§ rhage, laparotomy for bleeding, or maternal death from bleeding — no./total no. (%)‡ Key secondary implementation outcomes Detection of postpartum hemorrhage — 3870/4158 (93.1) 4244/8299 (51.1) 1.58 (1.41 to 1.76) 33.3 (26.9 to 39.8) no./total no. (%)¶ Adherence to treatment bundle — no./ 3791/4158 (91.2) 1623/8351 (19.4) 4.94 (3.88 to 6.28) 70.2 (64.6 to 75.7) total no. (%)‖ Secondary outcomes Postpartumhemorrhage—no./totalno.(%)** 4158/48,678 (8.5) 8351/50,043 (16.7) 0.51 (0.44 to 0.60) −8.2 (−9.7 to −6.6) Severe postpartum hemorrhage — no./ 786/48,678 (1.6) 2129/50,043 (4.3) 0.39 (0.31 to 0.49) −2.6 (−3.1 to −2.0) total no. (%)** Laparotomy for bleeding — no. (%) 12 (<0.1) 7 (<0.1) 1.72 (0.57 to 5.16) 0.01 (−0.02 to 0.04) Maternal death — no. (%) From bleeding 12 (<0.1) 18 (<0.1) 0.80 (0.38 to 1.68) −0.01 (−0.03 to 0.02) From any cause 17 (<0.1) 28 (0.1) 0.73 (0.40 to 1.31) −0.02 (−0.04 to 0.01) Blood transfusion — no. (%) For any cause 1074 (2.2) 1296 (2.6) 0.87 (0.69 to 1.10) −0.4 (−0.9 to 0.2) For bleeding†† 580 (1.2) 944 (1.9) 0.71 (0.55 to 0.90) −0.6 (−1.0 to −0.2) Blood loss at ≤2 hr post partum — ml‡‡ Median (IQR) 160 (100 to 280) 220 (120 to 380) — — Mean 225±229 318±321 — −84 (−103 to −64) * Plus–minus values are means ±SD. Laparotomy related to bleeding and maternal death from bleeding were determined by the end‑point review committee, whose members were unaware of the trial‑group assignments. The widths of the confidence intervals for secondary outcomes have not been adjusted for multiplicity and cannot be used to infer treatment effects. † Rate ratios are reported for the outcomes of detection of postpartum hemorrhage and use of treatment bundle; risk ratios are reported for other outcomes. Differences between percents are presented in percentage points, and differences between mean values are presented in the unit of the values. Analyses were adjusted for the cluster‑level covariates that were used in the randomization (number of vaginal births, prevalence of postpartum hemorrhage, country, and prevalence of primary‑outcome events) and for imbalances during the base‑ line period. Baseline data before implementation of the intervention (involving 110,473 patients in 78 clusters) were disaggregated for the interventional and usual‑care sites for each outcome as follows: for the composite primary outcome, 1931 of 50,721 patients (3.8%) in the intervention group and 2546 of 57,012 (4.5%) in the usual‑care group; for the detection of postpartum hemorrhage, 5097 of 8179 (62.3%) and 4971 of 9717 (51.2%), respectively; for adherence to the treatment bundle, 1682 of 8194 (20.5%) and 1038 of 9779 (10.6%), respectively; for postpartum hemorrhage, 8194 of 50,720 (16.2%) and 9779 of 57,010 (17.2%), respectively; for severe postpartum hemor‑ rhage, 1920 of 50,720 (3.8%) and 2535 of 57,010 (4.4%), respectively; for laparotomy for bleeding, 10 of 52,003 (<0.1%) and 12 of 58,470 (<0.1%), respectively; for maternal death from bleeding, 16 of 52,003 (<0.1%) and 24 of 58,470 (<0.1%), respectively; for maternal death from any cause, 29 of 52,003 (0.1%) and 34 of 58,470 (0.1%), respectively; for blood transfusion for any cause, 1507 of 52,003 (2.9%) and 1700 of 58,470 (2.9%), respectively; for blood transfusion for bleeding, 991 of 52,003 (1.9%) and 1176 of 58,470 (2.0%), respectively; for median blood loss up to 2 hours post partum, 220 ml (IQR, 120 to 380) and 220 ml (IQR, 120 to 380), respectively; and for median blood loss up to 24 hours post partum, 220 ml (IQR, 120 to 380) and 220 ml (IQR, 120 to 380). ‡ The intracluster correlation coefficient for the primary outcome on the latent scale was 0.011 (95% CI, 0.008 to 0.014). The cluster auto‑ correlation for the primary outcome was 0.61. The intracluster correlation coefficient and cluster autocorrelation were estimated by fitting a mixed‑effects linear model to the data with random effect for cluster and for a cluster–period interaction. In the analysis of severe post‑ partum hemorrhage, only women with source‑verified data on blood loss were included. § P<0.001. ¶ The detection of postpartum hemorrhage was defined as the recording of a diagnosis of postpartum hemorrhage by the birth attendant. The denominator is the number of patients with objectively measured postpartum hemorrhage (defined as blood loss of ≥500 ml). ‖ Adherence to the treatment bundle was defined as adherence to three core elements of the bundle: the administration of oxytocic drugs, tranexamic acid, and intravenous fluids. The denominator is the number of patients with objectively measured postpartum hemorrhage. ** Only patients with source‑verified data on blood loss were included in this analysis. †† Blood transfusion for bleeding was defined as blood transfusion in patients with postpartum hemorrhage. ‡‡ For the analysis of blood loss as a continuous variable, mean differences are reported. Outcomes were analyzed by permutation tests, and confidence intervals were constructed with the use of permutation tests, by finding the upper and lower boundaries of the intervention ef‑ fect that would lead to a two‑sided P value at less than the 5% level. 18 n engl j med 389;1 nejm.org July 6, 2023 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. Early Detection and Treatment of PPH improvements in the detection of postpartum 100 hemorrhageandtheuseoftheWHOfirst-response Randomization bundle in the hospitals in the intervention group. 8 Baseline Transition Implementation Findings regarding postpartum hemorrhage s Phase Phase (blood loss, ≥500 ml) were consistent with those n 7 for the primary outcome. The E-MOTIVE proto- t 6 f col allowed for triggering of the treatment bun- o 5 Usual care dle at blood loss of 300 ml or more if there was a n 4 an accompanying abnormality in the vital signs r or clinical observations. This trigger criterion was P 3 commonly used in the hospitals in the interven- 2 Intervention tion group, and this trigger criterion probably underlies the apparent benefit of the interven- 1 tion for less-severe postpartum hemorrhage. 0 We minimized identification and recruitment 0 5 10 15 bias by using broad inclusion criteria to include Month all the patients with vaginal births in the trial hospitals. The analysis approach was adjusted Figure 3. Patients with Primary-Outcome Event during the Baseline, for the slight residual imbalance in the baseline Transition, and Implementation Phases. The primary outcome was a composite of severe postpartum hemorrhage, phase across the trial groups in proportion with the primary outcome. We took care, to the ex- laparotomy for bleeding, or maternal death from bleeding. After the 7‑month baseline period, hospitals had an allowance of 2 months for tran‑ tent possible, to avoid contamination between sition in order to conduct training and to implement and embed the inter‑ the trial groups by ensuring that the trial hospi- vention in practice before beginning the 7‑month implementation phase. tals were geographically dispersed and in differ- I bars indicate the 95% confidence interval. ent administrative areas. The hospitals in the usual-care group continued to provide usual care and had the same access to bundle components estimated, which results in underestimation of and quality-checked medicines as those in the blood loss and delays in the initiation of lifesav- intervention group. ing treatment. A Cochrane review showed that Several limitations of this trial warrant con- use of a calibrated drape improved the detection sideration. First, owing to the pragmatic design, of postpartum blood loss as compared with vi- we did not collect information on some clinical sual estimation (rate ratio, 1.86; 95% CI, 1.11 to outcomes, such as the postnatal hemoglobin level 3.11 [high certainty]) but had no clear effect on and anemia, or on patients’ experience of care. health outcomes. 13 Second, our trial was not powered to assess ma- This large, international trial showed that the ternal death, but findings for this outcome, al- use of a calibrated drape for detection of post- beit uncommon, were in the direction of those partum hemorrhage and a bundle of treatments, for the primary outcome. Third, the trial was supported by a multifaceted implementation strat- conducted in low- and middle-income countries; egy, resulted in a substantially lower risk of the further implementation research is needed in primary outcome, a composite of severe postpar- high-income settings, focusing on process out- tum hemorrhage, laparotomy for bleeding, or comes such as postpartum hemorrhage detec- death from bleeding than usual care. tion and bundle use to ensure broader generaliz- The views and opinions in this article are those of the authors ability. Finally, the uncalibrated drapes that were and do not necessarily reflect those of their respective institu- tions or the World Health Organization. used in the control hospitals for the purpose of Supported by an investment grant (INV-001393) from the Bill gathering trial-outcome data were transparent, and Melinda Gates Foundation. Disclosure forms provided by the authors are available with and therefore providers would have been able to the full text of this article at NEJM.org. see the blood collecting in the drape. To the A data sharing statement provided by the authors is available extent that this situation may have influenced with the full text of this article at NEJM.org. We thank Mark Misiko, Jim Kelly, Polycarp Oyoo, Hilal their actions, it would be expected to attenuate Mukhtar, Stephen Abu, Ahmad Danjuma Suleiman, Muhammed the observed effect of the intervention. Shittu, Muftahu Dahiru Mahmoud, Lolade Arigbede, Baba Mai Blood loss after birth is currently visually Yaki, Faisal Dankishiya, Fatima Abubakar, Adeosun Love Funmi, n engl j med 389;1 nejm.org July 6, 2023 19 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. The new england journal of medicine Rahamtu Yusuf, Sadiq Abubakar, Taiwo Amole, Elani Muller, uaibu, Collins Agbeze Kalu, Ahmad Tijjani Ibrahim, Muhammad Sara Willemse, Fawzia Samuels, Sibongile Doris Khambule, Juma Umar, Nura Mainasara, Zaki Abubakar, Amaka Ngozi Ocheke, Makungu,AndreaPembe,DavidSando,AmaniKikula,Mohamed Nasiru Abdullahi Gada, Adebayo Adedokun, Aminu Magaji, Bakari, Sambusa Masumbuko, Beatrice Erastus Mwilike, Akwina- Fatima Rasheed, Hussaina Adamu, Amy Ivori, Umma Bawa, ta Banda, Christina Myalla, Roy Kaberia, Purity Kagendo, Henry Constance Shehu, Jombo Sunday, Ishaku Musa, Buyiswa Eugin- Atsewa, Mary Otuku, Augustine Khaemba, Hassan Mwachome, ia Pongwana, Khokhela Mba, Annelize Barnard, Gwendolene Halima Kaka, Damaris Wambui, Virginia Wangeci Irungu, Lydia Booys, Tracey Hinkel, Ronel Faro, Liaquat Parker, Nompumelelo Mwihaki Wanjiku, Noreen Nangira Were, Getrude Baya, Janet Seyisana, Monique Mieke Amira, Thea Williams, Thandiwe Ma- Munyanzi, Protus Wafula Nyongesa, Jennifer Jennis Saru, Debra busha, Bongekile Ethel Mbewana, Mangola Daniel Kabongo, Chepchirchir, Flora Tanui, Grace Limo, John Karani Kimani, Dhanabagium Gramoney, Katrin Middleton, Eze Collins, Emile Mary Njuguna, Maureen Adundo Okango, Rita Yumbe, Mary Manefeldt, Tessa McMillan, Jessica Westwood, Mziwohlanga Njoroge, Salome Waweru, Ndungu Muchiri, Nalika Robai, Binta Mdondolo, Sibonile Zitha, Knowledge Chipango, Robert James Garba Bello, Mahmoud Kawu Magashi, Stephen Bature, Binta Bwire, Shoma Guyela, Edward Philemon, Yasintha Elifather, Bi- Mohammed, Lewis Lebara, Nonye Enyidah Esther, Emmanuel jengo Zakaria Mabiba, Helenestina Anatory Chizzery, Consolata Agbogo, Hapsat Farouk, Ismail Saad Ayuba, Fatima Adamu Yu- Reuben Kyengu, Isack Mwongera Kyagari, Uddy Kyula, Happiness nusa, Aisha Shehu Satatima, Hajara Sani, Ibrahim Shehu, Hauwa Braison Munisi, Paschal Gerald Kalinga, Immaculatha Mewama Yakubu, Eigbadon Ambrose Odianosen, Osunde Faith Enifome, Samuel, Peter Nachiwa, Jacqueline Amani, Mbaraka Hussein, Fa- Aisha Nana Adamu, Maryam Muhammad, Asma’u Abdullahi, In- tuma Mono, Lulu Boniventure, Baltazary M. Joseph, and Balima nocent Ogbonnaya Michael, Ajiboye Ruth Tayo, Olowogbayi Tope, Itambu;themembersofthetrialsteeringcommittee(Sabaratnam Khadija Muhammad Kanoma, Rashida Badaru, Saadatu Abuba- Arulkumaran [chair], Deborah Armbruster, Pierre Buekens, and kar Tanimu, Habiba Muhammad Idris, Aminu Ibrahim Haruna, Monica Taljaard); the members of the independent data monitor- Maikudu Sulaiman, Chritiana Omuwa Ikhane, Salima Muham- ing committee (Zarko Alfirevic [chair], Pisake Lumbiganon, and mad, Akpofure Harriet Ese, Ikem Christiana Nwanne, Sunday Andrew Copas); the members of the end-point review committee Pius, Maryam Halilu Mohammed, Olorunfemi Ajibola John, (Harry Gee, Amie Wilson, Raffaele Napolitano, Irshad Ahmed, Rebecca Solomon Elisha, Abigail Gwimi, Ejura Ochala, Ehigha and Margarita Bariou); and all those not otherwise mentioned Enabudoso,LawalOyeneyi,HadizaUsman,MuhammadJamilSh- above who contributed to this trial. Appendix The authors’ full names and academic degrees are as follows: Ioannis Gallos, D.M.S., M.D., Adam Devall, B.Med.Sci., Ph.D., James Martin, Ph.D., Lee Middleton, M.Sc., Leanne Beeson, B.Sc., Hadiza Galadanci, F.R.C.O.G., Fadhlun Alwy Al-beity, M.D., Ph.D., Zahida Qureshi, M.B., B.S., M.Med., G. Justus Hofmeyr, M.B., B.Ch., D.Sc., Neil Moran, B.M., B.Ch., Sue Fawcus, M.B., B.S., Lumaan Sheikh, F.C.P.S., M.R.C.O.G., George Gwako, M.B., Ch.B., Ph.D., Alfred Osoti, M.B., Ch.B., Ph.D., Ashraf Aswat, B.Sc., Kristie-Marie Mam- moliti, M.Sc., Kulandaipalayam N. Sindhu, M.B., B.S., M.D., Marcelina Podesek, M.Sc., Isobelle Horne, B.A., Rebecca Timms, M.Res., Idnan Yunas, M.B., B.Chir., D.C.H., Jenipher Okore, B.Sc., Mandisa Singata-Madliki, Ph.D., Edna Arends, B.A., Aminu A. Wakili, M.B., B.S.,ArdMwampashi,B.A.P.S&P.A.,SidrahNausheen,M.B.,B.S.,ShahMuhammad,M.B.,B.S.,M.P.H.,PallaviLatthe,M.B.,Ch.B.,M.D., Cherrie Evans, Dr.P.H., C.N.M., Shahinoor Akter, Ph.D., Gillian Forbes, Ph.D., David Lissauer, M.B., Ch.B., Ph.D., Shireen Meher, M.B., B.S., M.D., Andrew Weeks, M.B., Ch.B., M.D., Andrew Shennan, M.B., B.S., M.D., Anne Ammerdorffer, Ph.D., Eleanor Williams, M. Sc., M.A., Tracy Roberts, Ph.D., Mariana Widmer, M.Sc., Olufemi T. Oladapo, M.D., M.P.H., Fabiana Lorencatto, Ph.D., Meghan A. Bohren, Ph.D., M.S.P.H., Suellen Miller, Ph.D., M.H.A., Fernando Althabe, M.D., Metin Gülmezoglu, M.D., Ph.D., Jeffrey M. Smith, M.D., M.P.H., Karla Hemming, Ph.D., and Arri Coomarasamy, M.B., Ch.B., M.D. The authors’ affiliations are as follows: the United Nations (UN) Development Program–UN Population Fund–UN Children’s Fund– World Health Organization (WHO)–World Bank Special Program of Research, Development, and Research Training in Human Repro- duction, the Department of Sexual and Reproductive Health and Research, WHO (I.G., M.W., O.T.O., F.A.), and the Concept Founda- tion (A. Ammerdorffer, M.G.) — both in Geneva; the College of Medical and Dental Sciences, University of Birmingham (A.D., J.M., L.M., L.B., A. Aswat, K.-M.M., K.N.S., M.P., I.H., R.T., E.W., T.R., K.H., A.C.), Health Education England (I.Y.), and the Department of Obstetrics and Gynaecology, Birmingham Women’s and Children’s NHS Foundation Trust (P.L., S.M.), Birmingham, the Centre for Behaviour Change, University College London (G.F., F.L.), and the Department of Women and Children’s Health, School of Life Course Sciences, King’s College London (A.S.), London, and the Institute of Life Course and Medical Sciences, University of Liverpool, Liver- pool (D.L., A.W.) — all in the United Kingdom; the African Center of Excellence for Population Health and Policy, College of Health Sciences, Bayero University, Kano, Nigeria (H.G., A.A.W.); the Department of Obstetrics and Gynecology, Muhimbili University of Health and Allied Sciences, Dar es Salaam, Tanzania (F.A.A., A.M.); the Department of Obstetrics and Gynecology, University of Nai- robi, Nairobi, Kenya (Z.Q., G.G., A.O., J.O.); the Department of Obstetrics and Gynecology, University of Botswana, Gaborone (G.J.H.); the KwaZulu–Natal Department of Health, Pietermaritzburg (N.M.), the Department of Obstetrics and Gynaecology, University of Cape Town, Cape Town (S.F., E.A.), and the Effective Care Research Unit, University of the Witwatersrand, Johannesburg (G.J.H., M.S.-M.) — all in South Africa; the Department of Obstetrics and Gynecology, Aga Khan University, Karachi, Pakistan (L.S., S.N., S.M.); the Maternal and Newborn Health Unit, Technical Leadership and Innovation, Jhpiego, and Johns Hopkins University, Baltimore (C.E.); the Gender and Women’s Health Unit, Centre for Health Equity, School of Population and Global Health, University of Melbourne, Mel- bourne, VIC, Australia (S.A., M.A.B.); the Department of Obstetrics and Reproductive Sciences, School of Medicine, University of Cali- fornia, San Francisco, San Francisco (S.M.); and the Maternal, Newborn, and Child Health Team, Bill and Melinda Gates Foundation, Seattle (J.M.S). References 1. Trends in maternal mortality 2000 to tion Division. Geneva: World Health Or- JP, Gülmezoglu AM, Winikoff B. Postpar- 2020: estimates by WHO, UNICEF, UNFPA, ganization, 2023. tum haemorrhage management, risks, World Bank Group and UNDESA/Popula- 2. Sheldon WR, Blum J, Vogel JP, Souza and maternal outcomes: findings from 20 n engl j med 389;1 nejm.org July 6, 2023 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved. Early Detection and Treatment of PPH the World Health Organization multi- postpartum haemorrhage: an integrative 10.Ammerdorffer A, Rushwan S, Timms country survey on maternal and newborn review of the literature. BMC Pregnancy R,etal.Qualityofoxytocinandtranexam- health. BJOG 2014;121:Suppl 1:5-13. Childbirth 2015;15:230. ic acid for the prevention and treatment of 3. Recommendations WHO for the pre- 7. Althabe F, Therrien MNS, Pingray V, postpartum hemorrhage in Kenya, Nige- vention and treatment of postpartum et al. Postpartum hemorrhage care bun- ria, South Africa, and Tanzania. Int J Gyn- haemorrhage. Geneva: World Health Or- dles to improve adherence to guidelines: a aecol Obstet 2022;158:Suppl 1:46-55. ganization, 2012. WHO technical consultation. Int J Gynae- 11.Hooper R, Forbes A, Hemming K, 4. World Health Organization. Updated col Obstet 2020;148:290-9. Takeda A, Beresford L. Analysis of cluster WHO recommendation on tranexamic 8. Forbes G, Akter S, Miller S, et al. Fac-randomised trials with an assessment acid for the treatment of postpartum tors influencing postpartum haemor- of outcome at baseline. BMJ 2018;360: haemorrhage. October 2017 (https://apps rhage detection and management and the k1121. .who.int/iris/bitstream/handle/10665/ implementation of a new postpartum 12.Haybittle JL. Repeated assessment of 259379/WHO-RHR-17.21-eng.pdf; haemorrhage care bundle (E-MOTIVE) in results in clinical trials of cancer treat- sequence=1). Kenya, Nigeria, and South Africa. Imple- ment. Br J Radiol 1971;44:793-7. 5. Widmer M, Piaggio G, Nguyen TMH, ment Sci 2023;18:1. 13.Diaz V, Abalos E, Carroli G. Methods et al. Heat-stable carbetocin versus oxyto-9. Akter S, Forbes G, Miller S, et al. De- for blood loss estimation after vaginal cin to prevent hemorrhage after vaginal tection and management of postpartum birth. Cochrane Database Syst Rev 2018; birth. N Engl J Med 2018;379:743-52. haemorrhage: qualitative evidence on 9(9):CD010980. 6. Hancock A, Weeks AD, Lavender DT. healthcare providers’ knowledge and prac- Copyright © 2023 Massachusetts Medical Society. Is accurate and reliable blood loss estima-tices in Kenya, Nigeria, and South Africa. tion the ‘crucial step’ in early detection FrontGlobWomensHealth2022;3:1020163. “Intention to Treat,” a new podcast drawing on the deep expertise of the New England Journal of Medicine, offers breaking news, enlightening context, and incisive analysis of critical and urgent issues in medicine and health care. Listen to the latest episode, on the primary care crisis, at NEJM.org or wherever you get your podcasts. n engl j med 389;1 nejm.org July 6, 2023 21 The New England Journal of Medicine Downloaded from nejm.org on November 8, 2023. For personal use only. No other uses without permission. Copyright © 2023 Massachusetts Medical Society. All rights reserved.