Abstract
Introduction:
Postpartum hemorrhage (PPH) remains the leading cause of maternal mortality. A new clinical intervention (E-MOTIVE) holds the potential to improve early PPH detection and management. We aimed to develop and pilot implementation strategies to support uptake of this intervention in Kenya, Nigeria, South Africa, and Tanzania.
Methods:
Implementation strategy development: We triangulated findings from qualitative interviews, surveys and a qualitative evidence synthesis to identify current PPH care practices and influences on future intervention implementation. We mapped influences using implementation science frameworks to identify candidate implementation strategies before presenting these at stakeholder consultation and design workshops to discuss feasibility, acceptability, and local adaptations. Piloting: The intervention and implementation strategies were piloted in 12 health facilities (3 per country) over 3 months. Interviews (n=58), case report forms (n=1,269), and direct observations (18 vaginal births, 7 PPHs) were used to assess feasibility, acceptability, and fidelity.
Results:
Implementation strategy development: Key influences included shortages of drugs, supplies, and staff, limited in-service training, and perceived benefits of the intervention (e.g., more accurate PPH detection and reduced PPH mortality). Proposed implementation strategies included a PPH trolley, on-site simulation-based training, champions, and audit and feedback. Country-specific adaptations included merging the E-MOTIVE intervention with national maternal health trainings, adapting local PPH protocols, and PPH trollies depending on staff needs. Piloting: Intervention and implementation strategy fidelity differed within and across countries. Calibrated drapes resulted in earlier and more accurate PPH detection but were not consistently used at the start. Implementation strategies were feasible to deliver; however, some instances of limited use were observed (e.g., PPH trolley and skills practice after training).
Conclusion:
Systematic intervention development, piloting, and process evaluation helped identify initial challenges related to intervention fidelity, which were addressed ahead of a larger-scale effectiveness evaluation. This has helped maximize the internal validity of the trial.
INTRODUCTION
Postpartum hemorrhage (PPH) remains the leading cause of maternal mortality,1 accounting for up to 27% of all maternal deaths, with disproportionate burden in low- and middle-income countries (LMICs).2 PPH is defined as blood loss of 500 ml or more, and severe PPH as blood loss of 1,000 ml or more after birth.3 Etiologies include uterine atony (the uterus not contracting after birth), retained placental tissue, genital tract trauma, and coagulopathies.4 Severe morbidity and mortality associated with PPH are preventable when detected early and treated by prompt evidence-based clinical interventions, as recommended by World Health Organization (WHO) guidelines.4,5 While trends in maternal mortality rates suggest a global decrease in PPH between 2000 and 20173 and stagnation between 2016 to 2020,2 delays in PPH detection, ineffective management, or a combination continue to be substantial challenges.6 Existing research has identified a range of factors influencing PPH detection and management in LMICs, including limited knowledge about current clinical guidelines and few training opportunities for health workers.7–9 Other contextual barriers include shortages of staff, drugs, and equipment and inadequate referral pathways, combined with poor transportation infrastructure.7–9
Building on these “missed opportunities” to improve PPH detection and management and increase adherence to evidence-based recommendations for PPH treatment,4,10, a new PPH first response clinical care bundle was proposed by a WHO Technical Consultation.11 A clinical care bundle is a set of evidence-based interventions designed to be administered together to improve adherence to evidence-based guidelines.12 The University of Birmingham, along with global collaborators, researchers, clinicians, and project and program developers, added an early detection component to the bundle, resulting in the E-MOTIVE intervention (Figure 1).13 E-MOTIVE stands for early detection of PPH using a calibrated drape, followed rapidly by the MOTIVE bundle (in no particular order): massage of uterus, oxytocic drugs administration, tranexamic acid administration, intravenous (IV) fluids administration, examination for identifying and managing the source of bleeding, and escalation to more advanced care, if bleeding continues despite treatment.11,13
The introduction of health care innovations, including care bundles, requires health workers to change their clinical practice behaviors by either modifying current practice, adopting a new practice, or discontinuing a previous practice. Existing evidence reports a wide range of barriers and enablers influencing clinical care bundle uptake.14,15 Previous research identified likely barriers and enablers of recommended PPH care: delayed detection, a lack of knowledge and training in PPH management, and shortages of staff and drugs.7,11 Therefore, implementing the new E-MOTIVE intervention is likely to necessitate implementation strategies to encourage uptake and sustainability.
Implementing the new E-MOTIVE intervention is likely to necessitate implementation strategies to encourage uptake and sustainability.
Theories and frameworks from behavioral and implementation sciences can aid in identifying barriers and enablers of behaviors that need to be targeted in the development of implementation strategies.16,17 Theory-based interventions are more likely to be effective or provide explanations for why an intervention has not worked.18 Therefore, applying behavioral change theory can facilitate exploration of the influences on current PPH treatment practices and “what it would take” to implement the new E-MOTIVE intervention in practice. However, to date, there has been limited theory-based research exploring factors influencing PPH detection and management in LMICs. This study applies a recognized approach to developing and evaluating behavior change interventions, the Behavior Change Wheel (BCW), which is a synthesis of 19 theoretical frameworks to guide the development of implementation strategies.19 The BCW uses the Theoretical Domains Framework (TDF)20,21 and the COM-B model (Capability, Opportunity, Motivation–Behavior) to guide intervention development by first conducting a behavioral analysis to identify what needs to change and then mapping to BCW intervention types (e.g., training, modeling, physical restructuring) to bring about the desired change.19
This study is part of the larger E-MOTIVE research program (ClinicalTrials.gov: NCT04341662), which aimed to design, develop, and pilot theory-based implementation strategies to support the E-MOTIVE intervention before conducting a large-scale cluster-randomized trial with process and economic evaluations in 4 countries (Kenya, Nigeria, South Africa, and Tanzania).22 This research comprised 3 workstreams that map onto the Medical Research Council framework for complex intervention development and evaluation.23 As part of the E-MOTIVE research program, we have already conducted mixed-methods formative research to understand factors influencing pre-intervention PPH detection and management and to prospectively identify health workers’ perceived barriers and enablers to implementing the E-MOTIVE intervention.24,25 The cluster-randomized clinical trial has since been completed and reported elsewhere.26 The trial results demonstrated a 60% relative reduction in the primary composite adverse PPH outcome (severe PPH 1,000 ml or more, laparotomy for bleeding, or maternal death from bleeding; risk ratio, 0.40, 95% CI=0.32, 0.50).26 The parallel process evaluation has been conducted and will explore and report on the fidelity, acceptability, and feasibility of the E-MOTIVE intervention in the cluster-randomized trial context. This article specifically reports on the stage of research that was conducted before the trial, which involved triangulation of the formative research, stakeholder involvement, and then piloting and evaluating the E-MOTIVE intervention and implementation strategies in a limited number of facilities before the larger clinical trial to inform any necessary intervention refinement ahead of larger-scale evaluation.
METHODS
Design
The study followed a 2-phase, 5-step systematic approach using behavioral and implementation science frameworks. A full protocol has been published.22 We report the methods in 2 phases: (1) implementation strategy development and (2) piloting and process evaluation of the E-MOTIVE intervention and implementation strategies. A mixed-methods approach involving qualitative interviews, surveys, observations, and stakeholder workshops was used in both phases. Figure 2 outlines the timeline of the phases and the steps taken in each phase. The research conducted in step 6 will be reported elsewhere.
Phase 1: Implementation Strategy Development
Triangulating Data Sources
Triangulation data sources were qualitative interviews with health care professionals (n=45; 12 hospitals),24,25 a previously published qualitative study in Tanzania,7 an online survey with health care professionals (n=972; 91 hospitals), and a qualitative evidence synthesis (43 studies).27,28 Hospitals participating in the formative interview and survey studies met the inclusion criteria for the E-MOTIVE trial26 and were secondary-level hospitals in Kenya, Nigeria, South Africa, and Tanzania, had 1,000 to 5,000 vaginal births per year, and were able to provide comprehensive obstetrical care, with the ability to perform surgery for PPH. Within hospitals meeting these inclusion criteria, we used maximum variation sampling to recruit a diverse sample of hospitals based on size and geographical location. We also used maximum variation sampling within hospitals to recruit a diverse participant sample representing the different health care professional cadres who detect and manage PPH, including doctors, nurses, and midwives. The aim of triangulation was to integrate findings across these 3 formative research activities to arrive at a distilled set of key findings presenting current PPH practices and potential influences on the E-MOTIVE intervention and implementation strategies.29,30 First, the interviews, survey, and qualitative evidence synthesis findings were deductively mapped to the TDF.20,21 This integrates constructs from 33 behavior change theories into 14 domains representing the broad range of influences on behavior, including individual (i.e., knowledge, emotions, and perceived capability), sociocultural (e.g., social and professional role, and identity), and environmental (e.g., environmental context and resources) influences.20,21 TDF has been widely used to identify and synthesize influences on implementation and health care provider behavior.31 Following recommended triangulation methods,29,30 we created a matrix structured around findings from each data source and country. Five research team members independently rated each finding across data sources using convergence criteria29,30: “agreement” (i.e., finding is identified); “partial agreement” (i.e., partially covered finding); “disagreement” (i.e., opposed finding) and “not found” (i.e., finding is not present). This analysis was first conducted for Nigeria and then extended to Kenya, South Africa, and Tanzania. We then compared convergence and divergence of data across countries, resulting in key influences being grouped into main findings.
Designing Implementation Strategies
The next step was to generate recommendations for potential implementation strategies to address the key findings identified through triangulation. This step was guided by behavior and implementation science frameworks whereby TDF domains were mapped to the COM-B model and BCW.19 The research team consulted matrices pairing the 2 frameworks to identify potential intervention types,32,33 discussed how these could potentially be operationalized in the context of E-MOTIVE, and generated descriptions of candidate strategies, including diagrams and pictures.
Refining Proposed Implementation Strategies
We then hosted 2-day stakeholder consultation and design workshops separately for each country (4 workshops in total) to actively seek the knowledge, experiences, and perceptions of local stakeholders to refine proposed implementation strategies and to prospectively improve feasibility and acceptability. We invited a range of stakeholders, including senior management (head of obstetrics and matron-in-charge) and labor and delivery ward staff (consultant obstetricians, doctors, midwives, and nurses). All workshops were held virtually on Zoom due to COVID-19 restrictions in place at the time. During the workshops, the research team first presented the findings of the formative research and then described the E-MOTIVE intervention and implementation strategies. Then, we facilitated discussion about feasibility and acceptability, how best to design and deliver each strategy, and whether any adaptation to each country’s local context was needed. Feasibility was defined as the practicality of use, and acceptability was defined as “a multifaceted construct that reflects the extent to which people delivering or receiving a health care intervention consider it to be appropriate.”34 The workshops were audio-recorded and analyzed to identify necessary refinements and delivery preferences for countries and cadres, which were made ahead of piloting.
Phase 2: Piloting and Evaluating the E-MOTIVE Intervention and Implementation Strategies
This phase aimed to field-test and pilot the E-MOTIVE intervention and implementation strategies, evaluate key implementation outcomes, and identify what needed to be refined and/or omitted. The pilot phase was conducted in 3 hospitals in each country (n=12) over a 3-month period (Clinicaltrials.gov: NCT04341662). The pilot aimed to replicate the cluster-randomized trial26 by purposively sampling hospitals to ensure representativeness of hospitals participating in the trial. Staff in participating hospitals were initially trained in E-MOTIVE by external trainers (Jhpiego), and the implementation strategies (e.g., champions and PPH trolleys/kits) were put into place. Hospitals received calibrated drapes during the pilot to enable early detection of PPH but also to pilot objective trial outcome measures (i.e., weighing drapes to measure blood loss). The weighing of drapes for trial outcome assessment was the responsibility of research midwives employed by the study at each site. Drugs were not supplied as part of the pilot.
Toward the end of the pilot period, we conducted a process evaluation consisting of direct clinical observations of women with vaginal birth and PPH and qualitative interviews. Country research teams also completed case report forms (CRFs) to collect data on the number of vaginal births, number of women with PPH, adherence to the MOTIVE bundle for women with PPH, and individual MOTIVE components delivered. Individual CRFs were collected for every woman giving birth at study hospitals. The CRFs were completed by research staff at each hospital and entered directly from the medical record to the REDCap electronic data capture system. Research staff at each hospital also collected drug stock-outs on a monthly facility-based CRF and entered this into the REDCap electronic data capture system. Hospitals taking part in the pilot met the inclusion criteria for the cluster-randomized trial but were excluded from participating in the trial to maintain internal validity and reduce the risk of contamination.
The key implementation outcomes evaluated were fidelity, feasibility, and acceptability. Fidelity was defined as the extent to which the E-MOTIVE intervention and implementation strategies were delivered as intended (i.e., adherence to the study protocol). Published criteria for interpreting fidelity suggest lower than 50% adherence represents “low” fidelity, 51%–79% “medium” fidelity, and 80%–100% “high” fidelity.35 We used a mixed-methods approach, comprising interviews with health workers; direct observations of health workers supporting women with vaginal birth and detecting and treating PPH; and CRFs of vaginal births, PPH cases, E-MOTIVE adherence, and drug stock-outs from patient and facility records to assess these implementation outcomes. Interviews assessed feasibility, including questions about the barriers and enablers to implementation using the TDF.36 To assess acceptability, interview questions were structured around the Theoretical Framework of Acceptability.34 Fidelity was self-reported in interviews, complemented through objective direct observations using a structured data collection form and woman-level data using the aforementioned CRFs. We used maximum variation sampling to recruit interview participants of different cadres, genders, and with varying years of experience. In-country research teams recruited interview participants, conducted the interviews, and completed the clinical observations over a 2-day period.
Ethical Approval
This study was approved by the University of Birmingham (reference number: ERN_19-1557), the WHO Research Project Review Panel, and WHO Research Ethics Review Committee (reference number ERC.0003486) for the formative phase and the following relevant ethics and regulatory review committees in each country: University of Melbourne Medicine and Dentistry Human Ethics Sub-Committee (1956004) in Australia; National Health Research Ethics Committee of Nigeria (reference number: NHREC/01/01/2007)-07/04/2020) in Nigeria; University of Nairobi KNH-UON ERC (P25/01/2020) and the Pharmacy and Poisons Board PPB/ECCT/20/06/06/2020(116), National Commission for Science, Technology and Innovation Nacosti P/21/8437 in Kenya; Eastern Cape Department of Health (EC_202007_015), University of Cape Town Human Research Ethics Committee (reference number: 091/2020), Health Province of KwaZulu-Natal (reference number: KZ_202008_036), and University of the Witwatersrand Human Research Ethics Committee-Medical (reference number:M200241) in South Africa; and Muhimbili University of Health and Allied Sciences (reference number: DA.282/298/01.C/) and the National Institute for Medical Research (Reference number: NIMR/HQ/R.8a/Vol IX/3501) in Tanzania.
RESULTS
We report the findings of the triangulation, mapping to implementation strategies, and stakeholder consultation and design workshops (phase 1), followed by the findings of the process evaluation of the pilot (phase 2).
Phase 1: Development of Implementation Strategies
Triangulation
Table 1 shows the convergence of 11 key findings across data sources and countries (Supplement 1 shows the full set of findings). Across countries, findings show there was agreement across all data sources that most MOTIVE bundle components (uterine massage, oxytocic drugs, IV fluids, and examination/escalation) were already routinely used individually to treat PPH but not used in a bundled approach. Tranexamic acid, which has been recommended for PPH treatment,10 was not routinely used. There was agreement across all countries that introducing a new blood loss collection and measurement tool (i.e., calibrated drape) would need training, and there were some concerns about how this would replace existing blood loss estimation methods (e.g., visual estimation using blood-soaked linens or collecting blood in kidney dishes). However, generally, it was perceived that use of a calibrated drape would help detect PPH earlier. It was reported that the use of the MOTIVE bundle would require sufficient staff for teamwork and that the MOTIVE bundle would likely be effective at improving PPH treatment and reducing PPH morbidities and mortalities. There was also agreement that implementing the MOTIVE bundle would depend on having consistent stocks of drugs and equipment, as well as adequate staffing, as it was perceived that 1 person alone would not be able to deliver the MOTIVE bundle.
Across countries, findings show there was agreement across all data sources that most MOTIVE bundle components were already routinely used individually to treat PPH but not used in a bundled approach.
Across countries, some differences were noted around the division of PPH treatment by different cadres (Kenya and Nigeria), in the use of tranexamic acid to treat PPH (Nigeria and Tanzania), and availability of calibrated drapes to measure blood loss (Kenya, Nigeria, Tanzania, and to less extent South Africa). There were also differences in receiving in-service training and existing skillsets among health workers. In the qualitative evidence synthesis, the provision of in-service training was reported as unsatisfactory, whereas participants in all countries receiving regular training (except Nigeria) reported being appropriately skilled at detecting and managing PPH. For the MOTIVE components, areas of disagreement were noted between countries about midwives not being trained to administer some components of the bundle, such as tranexamic acid (Kenya and Nigeria). Interview participants in all countries said the MOTIVE bundle as described would not be challenging to implement, whereas surveyed participants who did not have the bundle explained to them said rapidly administering multiple interventions (proxy for the MOTIVE bundle) might be challenging.
Mapping to Implementation Strategies
A summary of the mapping of the triangulated findings to the BCW to identify the components of the proposed implementation strategies is presented in Table 2. The proposed implementation strategies represent those agreed with stakeholders during the stakeholder consultation workshops to be feasible to deliver within the scope and context of the E-MOTIVE trial. The E-MOTIVE implementation strategy comprised 4 components:
Simulation-based team-based training to develop facility trainers (delivered by Jhpiego) and implemented on-site by facility trainers with follow-up practice drills on the calibrated drape and MOTIVE bundle use to encourage and enable staff to develop the capability to work and communicate in a coordinated and efficient manner as a team when detecting and managing a PPH.
Introduction of 2 local champions (1 midwife and 1 doctor) to promote E-MOTIVE uptake and provide reminders, encouragement, and social support to staff implementing E-MOTIVE.
Availability of a PPH trolley or carry case with all necessary drugs and equipment.
Provision of audit and feedback to all staff.25
The champions and audit and feedback strategies provided opportunities to deliver nonmaterial incentives and social encouragement for implementing E-MOTIVE by recognizing and praising instances of good practice.
Stakeholder Consultation and Design Workshops
Four interdisciplinary workshops (1 per country) were facilitated with a total of 63 participants: 28 nurses and midwives, 13 doctors, 16 consultants/specialists, and 6 clinical management staff. Most participants across all countries agreed in principle that the E-MOTIVE intervention and implementation strategies were implementable in their contexts. However, there were some challenges identified around adopting new ways of working and uptake of some aspects of individual implementation strategies. Table 3 presents the E-MOTIVE intervention and implementation strategies, workshop findings, and country-specific adaptions. All stakeholder decisions are available in Supplement 2.
In summary, the 300 ml and 500 ml trigger lines on the calibrated drape were acceptable, and the clinical responses for each trigger line were reported as practical and feasible by all countries. The concept of the MOTIVE bundle was acceptable, although local adaptations were required where there was no protocol for midwives to administer tranexamic acid because midwives do not usually have prescribing or administrating authority for IV medications. The education and training strategies were both acceptable if merged with existing training (e.g., Essential Steps in the Managing Obstetric Emergencies in South Africa).37,38 Some adaptations were likely; for example, having as many as 8 different follow-up practice drills could be burdensome for sites with shortages of staff (Nigeria and Tanzania). Having a PPH trolley was acceptable in all countries except for storing oxytocin, which requires refrigeration. A PPH carry case would be more feasible for smaller labor wards where space was limited, and in South Africa, the PPH carry case replaced the existing PPH box. Also, participants reported that it was more practical to use existing drug and equipment checklists instead of developing new trolley-specific checklists and for each site to assign 1 person to check and restock after each shift rather than collective responsibility. Introducing a champion was acceptable, provided there was a clear remit of roles and responsibilities (e.g., champion handbook). The audit and feedback approach needed simplification because the dashboard format proposed at the workshops was difficult to understand and interpret; this was agreed upon by all countries.
Phase 2: Piloting and Process Evaluation
Table 4 presents the pilot results per country. In the pilot, there were a total of 10,052 vaginal births that resulted in 1,269 PPHs. Across all countries, the implementation of all MOTIVE components for women with PPH (as recorded in the CRFs) was 41.1% (522/1269), representing low fidelity. However, there was substantial variation in fidelity across countries: 10.7% (55/512; low fidelity) in South Africa, 57.3% (215/375; medium fidelity) in Nigeria, 62.1% (187/301, medium fidelity) in Kenya, and 80.2% (65/81, high fidelity) in Tanzania.
There was substantial variation in fidelity across countries, ranging from 10.7% in South Africa to 80.2% in Tanzania.
Process evaluation interviews were conducted with 58 health workers at the pilot study hospitals (Table 5). Eighteen women with vaginal birth and 7 women with PPH were observed at each of the 8 hospitals (Kenya=2; Nigeria=3 and South Africa=3).
Table 6 summarizes the interview and observation findings according to the implementation outcomes of interest: fidelity, acceptability, and feasibility, starting with the E-MOTIVE intervention and then each implementation strategy in turn. Supplement 3 presents a full description of identified TDF domain barriers and enablers influencing the feasibility and acceptability of the E-MOTIVE intervention and implementation strategies.
Fidelity of E-MOTIVE Intervention
In contrast to the CRF data on low adherence to the MOTIVE bundle components, overall fidelity of the calibrated drape and the MOTIVE bundle in self-reported interviews and observations was medium to high.
The drape was placed with the funnel open after the birth in 66.7% of the women (12/18) and before the birth with the funnel rolled up in 33.3% (6/18) by a midwife in 61.1% (11/18) or by an E-MOTIVE research midwife in 38.9% (7/18). Monitoring of blood loss was regularly checked in Kenya and South Africa but less often in Nigeria. Across countries, there was different timing of sweeping blood into the funnel of the drape (after birth, after placenta delivery, drape removal, and not done) across all countries. There were some inconsistencies in what materials were included in the drape funnel (blood only or blood and blood-soaked pads/gauze). There were substantial variations in how staff responded when the blood loss reached 300 ml (warning line) and 500 ml (action line) when compared with the E-MOTIVE protocol. For 4 of the 7 women with PPH (57.1%), the MOTIVE bundle was correctly triggered at 500 ml of blood loss. Once triggered, all MOTIVE components were administered (median time from starting first component to starting last component was 6 minutes [range 3–25 minutes]). These findings from the pilot sites were critical in refining the use of the calibrated drape before the E-MOTIVE intervention was introduced into the cluster-randomized trial.
In contrast, self-reported fidelity to the E-MOTIVE intervention was much higher, with health workers reporting always using the calibrated drape. At 300 ml blood loss, health workers said that they would call for help and prepare for a possible PPH. All health workers reported triggering PPH treatment at 500 ml blood loss and adhering to the bundled approach of giving all treatments rapidly. No health workers reported “waiting to see” if 1 component worked before giving the next component.
Therefore, from the direct observations and interviews, it is possible to infer fidelity to the calibrated drape use was medium to high, and it was feasible to administer the MOTIVE components within 15 minutes (per study protocol).
Acceptance and Factors Influencing Feasibility of E-MOTIVE Intervention
Overall, staff reported that the E-MOTIVE intervention was feasible despite some challenges. Across all countries, enablers to delivering the E-MOTIVE intervention fell within the TDF domains: environmental context and resources (having an adequate supply of calibrated drapes and sufficient staff on a shift to give treatments in quick succession), beliefs about capabilities (calibrated drapes allowed quicker and easier and more accurate blood loss measurement), beliefs about consequences (improved PPH detection), and social/professional role and identity (evidence to support midwives’ diagnosis versus a higher-level professional’s estimate of blood loss). Other enablers fell within the TDF domains: social influence (supportive managers to ensure adequate drug stocks), emotion (reduced stress), social professional role and identity (increased scope of midwives’ practice to use all bundle components), social influence (from working within a team who support each other and communicate better), and goals (importance of bundle to effectively reduce PPH-related maternal mortalities).
Barriers fell within the TDF domains: environmental context and resources (long-term supply of calibrated drapes and lack of available delivery beds to maintain calibrated drape use for 1 hour after birth, in contexts where a woman is moved from delivery room to postnatal room, lack of staff to give treatments in quick succession); knowledge (lack of understanding about bundled approach); skills (ability to place the calibrated drape and administer tranexamic acid); beliefs about capabilities (to correctly read the measurement lines); and social influences (women do not want the calibrated drape to be kept in place for up to 2 hours after the birth and/or are traumatized by having a PPH so refuse potential painful treatment). Other barriers were social/professional role and identity (midwives’ hesitancy to administer tranexamic acid, which requires new skills), memory, attention and decision-making (forgetting to give all treatments), and intentions (reluctant to adopt a care bundle when existing PPH treatment is considered effective).
Fidelity, Acceptance, and Factors Influencing Feasibility of Implementation Strategies
Training: The fidelity of initial training was high; however, fidelity of follow-up practice sessions was low to moderate. This was evidenced by all staff receiving training from on-site trainers (from training logbook). Twenty-two simulation practice sessions were held to observe potential challenges or missed opportunities with drape use and MOTIVE administration. The main areas where the staff participating in practice sessions did not adhere to the training were not retrieving the PPH trolley (Kenya) and not rapidly running IV oxytocin (Kenya and Nigeria). In interviews, some health workers in South Africa remarked that simulations did not always reflect the realities of PPH treatment, for example, when there were competing emergencies on the ward.
Acceptability and feasibility of training were high, evidenced by staff welcoming more training on PPH. Enablers fell within the TDF domains: skills (training was adequate), beliefs about capabilities (building confidence) and memory, and attention and decision-making (remembering to deliver E-MOTIVE intervention). The TDF domain of environmental context and resources might be a barrier whereby the training of new staff makes time demands on the system to provide ongoing training on E-MOTIVE intervention use.
PPH trolley/carry case: Although PPH trolleys were available, fidelity of their use was low. This was evidenced by all sites having easy access to a PPH “kit” of supplies, which was stored in either a trolley (Nigeria and Tanzania) or a carry case (Kenya and South Africa). In the interviews, health workers reported that the PPH trolley was used, and it was regularly checked and restocked on a daily, weekly, or after use basis. However, this contrasted with more objective observations, where it was observed that the PPH trolley was not frequently used (14.3%, 1/7).
Despite the low fidelity of the PPH trolley, acceptability of having a PPH trolley was reported as high. Reasons provided for not regularly using the PPH trolley included the trolley not being adequately stocked or beliefs that retrieving it was not a good use of health workers’ limited time. Additional influences on trolley use fit within the domains of environmental context and resources (keeping drugs and equipment together saves time unless space on the ward was limited and if the trolley was in good working order and there was sufficient staff to retrieve it), beliefs about consequences (improving PPH response time), knowledge (all ward staff knew where to find PPH items), beliefs about consequences (trolley could improve PPH treatment), and intention (no plans to use trolley because it was inadequately stocked).
Champions: Fidelity, acceptance, and feasibility of having champions in place was high. This was evidenced by interview participants reporting having 1–3 champions per hospital. Some health workers were unsure about the clinical role of the champion and a few health workers were not aware of having a champion at their site. For those who were aware of the champions’ roles, having champions was considered appropriate. Interviewed participants highlighted some challenges with having champions, which included having limited capacity to carry out roles due to competing clinical tasks, appropriate division of responsibility and labor between 2 or more champions, and difficulties in champions working on different shifts from other staff, therefore, unavailable for some staff. However, champions were reported to give support, ensure supplies of drapes and stock of drugs, provide training to new staff, and manage any ill feelings or conflicts among staff. Enablers to engagement with champions fit within domains of social/professional role and identify (designated on-site staff fulfilling champion’s role), beliefs about consequences (helpful and supportive), and social influence (importance of having a “focal” person who can talk to staff). Barriers also fit within social influence (misunderstandings among staff and champions receiving financial incentives) and beliefs about capabilities (champions’ capacity due to other competing clinical responsibilities).
Audit and feedback: Fidelity, acceptance, and feasibility of engaging in audit and feedback were low. This was evidenced by most interview participants being unaware of audit and feedback strategies, suggesting inconsistent fidelity across countries. Doctors were more likely to participate in feedback sessions than midwives (Kenya and South Africa). The TDF enablers for staff engaging with audit and feedback were beliefs about consequences (beneficial because it identified where improvements are needed), goals (important to set targets), and environmental content and resources (audit and feedback posters displayed in all hospitals). A key barrier related to beliefs about capabilities (perceived accuracy of the audit and feedback [i.e., staff felt actual drug use was higher or lower than expected] and drug use related to poor documentation skills) and memory, attention and decision-making (drugs were usually locked away [e.g., tranexamic acid] or if key holder going off duty forgets to leave keys for next shift).
DISCUSSION
This article describes the systematic approach taken to identify areas of suboptimal PPH detection and management to develop implementation strategies to support delivery of the E-MOTIVE intervention. Using the BCW, the approach involved different country stakeholders and pilot testing of the implementation strategy to identify implementation issues. This approach provided an opportunity to identify what was working well and where to make further intervention modifications to improve implementation before the E-MOTIVE cluster-randomized trial.
This article describes the systematic approach taken to identify areas of suboptimal PPH detection and management to develop implementation strategies to support delivery of the E-MOTIVE intervention.
The pilot shows that the initial implementation of the E-MOTIVE intervention and implementation strategies in the pilot hospitals (which were distinct from the main E-MOTIVE trial hospitals) varied within hospitals and across countries. All components of the bundle were delivered; however, the new calibrated drape was not always consistently and appropriately used by staff, despite health workers acknowledging the reliability of the calibrated drape to measure blood loss quickly and accurately. This suggests switching to the calibrated drape requires time for staff to adapt to new ways of working and success depends not only on health workers understanding the benefits of accurate blood loss measurement. The overall use of oxytocin in women with PPH increased to 80% or higher in all countries except Tanzania, which was already 100% at baseline. Overall use of tranexamic acid in women with PPH also increased in all countries. While improvements in drug use were feasible, there was still an implementation gap in the fidelity of the MOTIVE components. Although fidelity varied across countries, given the relatively short duration of the pilot phase, the limited number of hospitals participating in the pilot, and relatively small number of women with PPH, it is not possible to draw firm conclusions about the relationship between extent of fidelity and PPH outcomes. This is something that will be explored as part of a larger-scale process evaluation conducted alongside the E-MOTVE cluster-randomized trial, with a larger number of participating facilities and over a longer time frame. The short time frame in the present pilot study may also, in part, explain the fact that drug stock-outs were not a reported barrier. The PPH trolley and kit introduced as part of the implementation strategy may have also facilitated the coordination and restocking of necessary medications and supplies.
While the piloting and process evaluation of the E-MOTIVE intervention and implementation strategies highlighted numerous enablers that indicate good feasibility and acceptability of E-MOTIVE, it also identified some barriers that highlight implications for refining E-MOTIVE ahead of the cluster-randomized trial. The pilot has shown that differences in E-MOTIVE intervention implementation could be explained by staff’s familiarity with the MOTIVE components from current PPH treatment practices compared to the more significant changes in PPH detection using the new calibrated drape. In response, the E-MOTIVE training and skill practices for using the calibrated drape needed to ensure the setting and equipment reflect the clinical reality. The interviews and observations showed the research midwives performing some aspect of E-MOTIVE implementation beyond their intended role of placing and weighing the calibrated drapes. Although this represents a deviation in fidelity or intended practice, it is potentially explained by the fact that research midwives may have felt ethical or moral obligations to assist in an emergency, such as PPH, even if support was not required. To try and address this ahead of the trial, Jhpiego (providers of train the trainer activity) updated the materials to stress that the research midwives’ role is to collect data, support training and practice, and help with managing the PPH trolley but not provide routine care. The intention that administration of the bundle should only be done by clinical staff was further emphasized in the E-MOTIVE training. The intervention was to be delivered by regular facility staff who conduct births, and nurses and midwives can implement the entire intervention when a doctor is not present. Alternatively, an additional recommendation for future studies or implementation would be to employ nonclinical research staff to assist with data collection and training to minimize the opportunity and likelihood of them supporting with delivery of clinical care and, in turn, impacting fidelity.
Some staff also reported knowledge barriers related to the drape and limited self-efficacy in interpreting calibration lines. Therefore, other changes included refining training materials to explain and reinforce the importance of calibrations to trigger the bundle, carrying out calibration reading practice, and making modifications to treatment provision (duration of massage and IV fluids with oxytocin to be infused from 1L to 500 ml). We found that the uptake of the supporting implementation strategies also varied across hospitals from all countries. While all health workers received training, there were differences in the organization of follow-up practice sessions, varying use of the PPH trolley, presence of and engagement with champions, and health workers’ receiving audit and feedback on MOTIVE bundle usage. In response, the number of follow-up practice sessions was reduced from 8 to 5, a new PPH trolley checklist was introduced to ensure supplies were restocked after use or daily (supported by research midwives if needed), and new training monitoring materials were introduced to ensure quality and consistency of implementation.
While initial uptake was not always as intended, the country partners worked to increase feasibility and, consequently, fidelity over the pilot period. For example, increased efforts were made to explain the role of the local champions (a new concept in some study sites) and ensure they were appointed as “the face” of any implementation efforts to help overcome resistance to change, promote consistent use of intervention, and promote uptake of implementation strategies. Specifically, efforts were made to encourage the uptake of audit and feedback, which was limited possibly because the existing culture was to only give feedback to senior staff, not all staff. Similar uptake issues were also identified for implementation strategies aimed at improving the clinical care bundle fidelity, such as “Sepsis Six.”39
This study highlights the importance of piloting interventions before large-scale evaluation of effectiveness and targeting different influencing factors on implementation. Our findings identify early challenges to implementing the E-MOTIVE intervention, particularly the introduction of the calibrated drape and some limited uptake of the PPH trolley/carry case, skills practice after training, champions, and audit and feedback implementation strategies. The generalizable lessons learned for use in other trials were to not rush from intervention development to trial, to pilot the intervention and implementation strategies, and to evaluate via a process evaluation first to identify areas for refinement ahead of larger-scale trials. This is important to help maximize intervention fidelity and, in turn, the likelihood of the trial being able to accurately evaluate if the intervention (E-MOTIVE + implementation strategies) is effective or not, reducing potential waste of resources, time, and efforts of testing at scale, if not working as optimally in the pilot. Our study highlighted a “know-do” implementation gap based on differences between real-world direct observations and reported use of E-MOTIVE and uptake of implementation strategies by interview participants. Therefore, when introducing new or modified ways of working, such as implementing a new care bundle and adopting implementation strategies, it is essential to be aware of potential issues with fidelity, acceptability, and feasibility.
Our study highlighted a “know-do” implementation gap based on differences between real-world direct observations and reported use of E-MOTIVE and uptake of implementation strategies by interview participants.
Strengths and Limitations
The strengths of the study are that it was multidisciplinary collaborative research using rigorous data collection methods, and it adhered to Medical Research Council guidelines to involve stakeholders to achieve consensus through discussion of supporting implementation strategies. Other strengths included the use of behavior change theory to develop potentially more effective supporting implementation strategies, identification of any previously missed opportunities to improve PPH detection and management, and addressing any implementation issues of the PPH intervention and implementation strategies before the start of the cluster-randomized trial. Methodological limitations of the research were probable loss of interpersonal and group dynamics from using Zoom for workshops, inability to address some contextual issues (staff shortages and lack of beds), and the sustainably of stocking medicines and supplies. Direct observations limited to dayshifts may have missed better or worse performance. The repeated data collection may have resulted in participant fatigue if interviewed more than once and possible stress from being observed.
We acknowledge differences in fidelity were identified across the qualitative interviews, direct observations, and CRFs findings. Fidelity in self-reported interviews was high, in contrast to observations, which were medium, and CRFs, which were low. This difference can potentially be explained by the data collection biases associated with each method. Interviews are subject to social desirability and recruitment bias, and direct observations are subject to the Hawthorne effect,40 in which participants modify their behavior in response to their awareness of being observed. Moreover, low fidelity, as reported in the CRFs, may have been influenced by incomplete patient records. These discrepancies were anticipated in advance, hence, why a mixed methods approach was used to achieve a more balanced picture of implementation outcomes.
Additional limitations include, first, the focus of the E-MOTIVE program on PPH following births occurring in secondary-level hospitals meeting the trial criteria.26 The proportion of hospital-based births varies across countries, with births also occurring in lower-level hospitals and community settings (i.e., at home or in nonhospital health facilities). Therefore, the E-MOTIVE program of research and associated clinical and implementation interventions addresses a subset of PPH and associated morbidity and mortality, and consideration of adaptation of the drape, bundle, and implementation strategies for births in other settings is a next key step. The qualitative evidence synthesis28 that was conducted as part of the E-MOTIVE formative research to inform intervention development reports on the perceptions and experiences of the prevention, detection, and management of PPH following births in different settings, particularly from the perspective of women, community members, and traditional birth attendants. The qualitative evidence synthesis maps identified barriers and enablers in other settings to potential interventions and summarizes recommendations for implementation.28 Second, the reported triangulation of the findings from the formative research highlighted wider system-level barriers related to staffing levels and infrastructure (e.g., insufficient number of beds and electricity), which were beyond the scope of what was feasible and sustainable to address within the context of E-MOTIVE. Nonetheless, these highlight that ensuring adequate skilled workforce, infrastructure, supplies, and medications are important implementation considerations for policymakers looking to implement and sustain E-MOTIVE beyond the trial, should it be shown to be effective.
CONCLUSION
From an implementation perspective, delivering the E-MOTIVE intervention and implementation strategies was mostly acceptable and feasible across 4 study countries during the pilot phase. The E-MOTIVE intervention, coupled with the described implementation strategies, has the potential to detect PPH earlier and more accurately and to improve PPH management, which may result in reduced PPH-related morbidity and mortality. Potential threats to fidelity and acceptability identified in the pilot phase can be addressed by refining the E-MOTIVE intervention and implementation strategies ahead of the cluster-randomized trial.
Acknowledgments
We would like to acknowledge the time given by participants to attend the workshops, provide clinical training, and take part in the interviews and observations. We thank Hannatu Abdullahi, Isabella Atieno, Jennyfer Oluyemisi Don-Aki, Susheela Engelbrecht, Chrisostom Lipingu, Comfort Ene Okpe, Gaudiosa M. Tibaijuka Idris Usman Takai, Stephen Bature, Nonye-Enyidah Esther, Fatima Abubakar Rasheed, Samuel Adelaiye, Hussaina Adamu, A.M.Y Ivori, Umma Suleiman Bawa, Ehigha Enabudoso, Ahmad Tijjani Ibrahim, Muhammad Umar, Muhammad Kabir Abdullahi, Hauwa Muhammad Baba, Nasir Abubakar, Lawal Oyeneyin, Victor Ohenhen, NuraAbdullahi Mainasara, Zaki Abubakar, Aisha Nana Adamu, Constance E. Shehu, Oyaromade Abidemi, Kikelemo Adesina, Amaka Ngozi Ocheke, Hadiza AbdullahiUsman, Danladi Abubakar, Aminu Gambo, Faruq Umar Faruq, Aminu Ibrahim Haruna, Salima Muhammad, Nasiru Abdullahi Gada, Nwachi Ayo, Maisaratu Bakari, Jombo Sunday, Muhammad Jamil Shuaibu, Hadiza Idris, Adebayo Adedokun, Musa Ishaku, Collins Agbeze Kalu, Aminu Magaji. Juma Makungu, Beatrice Mwilike, Mwanaidi Mlaguzi, Abdalla Bihoga, Patricia Atugonzia Osward, Editha Fundisha, Henry Mwigani, Robert Masagasi, Jim Kelly Mugambi, Mark Misiko, Saida Ramadhan, Collins Kariuki, Victoria Muviku, Joyce Kingori, John Karanja, Jeniffer Kamwaro, Doris Mbithi, Christine Muteria, Mark Sigei, Raheli Mukwana, Edna Arends, Elani Muller, Nontozamo Tsetse, Mandy Manus, Bongeka Mhlauli, Heloise le Riche, Shereen Arontjies, Masamkele Mpongo, Nontle Ntshulana Charles Benjamin Gaunt, Nomveliso Bilisho, Sibongile Doris Khambule, Anele Khumalo, Nokuthula Mthethwa, Nombuyiselo Dlamini, Farida Ahmed, Fowzia Samuels, Dezerie Sanders, Violet Hourn, Linda Makaluza, Farida Ahmed, Jeanette Meiring, Sisanda Malgas, Mandy Manus, Phumla Mlotana, Nothemba Mbatsha, Nonkululo Dangazele, Aphelele Precious Novukela, Lulama Nompandana, Thandeka Khumalo, and Sara Willemse.
Funding
This research is funded by the Bill & Melinda Gates Foundation through a grant to the University of Birmingham (Grant Number: INV-001393).
Disclaimer
The funders were not involved in the study design, data collection, data analysis, or preparation of the article.
Author contributions
GF, SA, SM, MAB, FL: conceptualization, methodology, and formal analysis. GF, SA, SM, HG, ZQ, FAA, GJH, NM, SF, MSM, AAW, TGA, BMM, FD, AAA, ASM, JE, EN, AO, GG, JO, AK, EM, AM, CE, KMM, MAB, FL: investigation and data curation. GF, SA, SM, HG, ZQ, FAA, GJH, NM, SF, MS, AAW, TGA, BMM, FD, AAA, ASM, JE, EN, AO, GG, JO, AK, EM, AM, KMM, CE, AD, AC, IG, OTO, MAB, FL: project administration. GF: writing–original draft. All authors: writing–review and editing. All authors reviewed and approved the final version of the article.
Competing interests
SM’s University (University of California San Francisco) holds the Trademark license for the ‘LifeWrap’ Nonpneumatic Anti-Shock Garment (NASG), a first-aid device used in refractory postpartum hemorrhage. The manufacturer of the LifeWrap NASG pays the University of California a royalty for the use of the name. GJH has an interest in the MaternaWell Tray for blood loss monitoring.
Notes
Peer Reviewed
First Published Online: September 12, 2024.
Cite this article as: Forbes G, Akter S, Miller S, et al. Development and piloting of implementation strategies to support delivery of a clinical intervention for postpartum hemorrhage in four sub-Saharan Africa countries. Glob Health Sci Pract. 2024;12(5). https://doi.org/10.9745/GHSP-D-23-00387
- Received: September 12, 2023.
- Accepted: August 13, 2024.
- © Forbes et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly cited. To view a copy of the license, visit https://creativecommons.org/licenses/by/4.0/. When linking to this article, please use the following permanent link: https://doi.org/10.9745/GHSP-D-23-00387