Factors influencing postpartum haemorrhage detection and management and the implementation of a new postpartum haemorrhage care bundle (E-MOTIVE) in Kenya, Nigeria, and South Africa

Background Postpartum haemorrhage (PPH) is the leading cause of global maternal deaths, accounting for 30–50% of maternal deaths in sub-Saharan Africa. Most PPH-related deaths are preventable with timely detection and initiation of care, which may be facilitated by using a clinical care bundle. We explore influences on current PPH detection and management and on the future implementation of a new PPH bundle (E-MOTIVE) in low-resource, high-burden settings. Methods Semi-structured qualitative interviews based on the Theoretical Domains Framework were conducted with 45 healthcare providers across nine hospitals in Nigeria, Kenya and South Africa, to identify barriers and enablers to current PPH detection and management and future implementation of a new PPH care bundle. Data were analysed using thematic and framework analysis. The Behaviour Change Wheel was used to identify potential interventions to address identified barriers and enablers. Results Influences on current PPH detection and management fell under 12 domains: Environmental Context and Resources (drug and staff shortages), Skills (limited in-service training), Knowledge (variable understanding of the recommended practice), Behaviour Regulation (limited quality improvement culture), Beliefs about Consequences (drawbacks from inaccurate detection), Emotion (stress from the unpredictability of PPH), Social Influence (teamwork), Memory, Attention and Decision-making (limited guideline use), Social/Professional Role and Identity (role clarity), Beliefs about Capabilities (confidence in managing PPH), Reinforcement (disciplinary procedures) and Goals (PPH as a priority). Influences on bundle uptake included: Beliefs about Consequences (perceived benefits of new blood loss measurement tool), Environmental Context and Resources (high cost of drugs and new tools), Memory, Attention and Decision-making (concerns about whether bundle fits current practice), Knowledge (not understanding ‘bundled’ approach), Social Influence (acceptance by women and staff) and Intention (limited acceptance of ‘bundled' approach over existing practice). These influences were consistent across countries. Proposed interventions included: Education, Training, Modelling (core and new skills), Enablement (monitoring uptake), Persuasion (leadership role) and Environmental Restructuring (PPH emergency trolley/kit). Conclusions A wide range of individual, socio-cultural and environmental barriers and enablers to improving PPH detection and management exist in these settings. We identified a range of interventions that could improve PPH care and the implementation of new care bundles in this context. Trial registration ClinicalTrials.gov: NCT04341662 Supplementary Information The online version contains supplementary material available at 10.1186/s13012-022-01253-0.

• This formative research is the first to involve healthcare providers in high-burden settings using a theoretical approach to identify the potential implementation issues of a new clinical care bundle to improve postpartum haemorrhage detection and management • This study illustrates the application of implementation science in low-resource, high-burden settings to address a maternal health priority and implementation gap (i.e., translating postpartum haemorrhage guidelines into practice) • This research demonstrates the benefits of considering implementation issues from the outset of project planning in order to maximise the internal validity of the trial, to ensure fidelity of bundle uptake and to optimise evaluation efforts in the future.

Background
Almost 300,000 maternal deaths occur each year, most of which are preventable with timely access to quality maternity care when complications arise [1]. Postpartum haemorrhage (PPH) is the leading cause of maternal death, accounting for 27% of all maternal deaths worldwide and approximately 30% to 50% of maternal deaths in sub-Saharan Africa [2]. PPH is categorised as a blood loss of ≥ 500mls and severe PPH is a blood loss of ≥ 1000mls after childbirth [3]. It is typically caused by uterine atony (inadequate contraction of the uterus after birth), as well as uterine rupture, retained placental tissue, genital tract trauma, and maternal bleeding disorders [3]. PPH is largely preventable; when it does occur, early detection and prompt evidence-based management can avoid the most severe outcomes [3,4]. Global progress in reducing PPH mortality has been slow [5], and there is an urgent need to address preventable PPH mortality and morbidity, particularly in sub-Saharan Africa, where the burden is highest. The World Health Organization (WHO) first published guidelines for the prevention and treatment of PPH in 2012 and since has regularly updated its recommendations [3,[6][7][8][9].
These recommendations include 13 clinical interventions depending on the PPH type (minor, moderate or severe), and the circumstances in which the PPH occurs (e.g., mode of birth, birth setting, timing of PPH); however, evidence of guideline implementation in practice is limited, particularly in low-and middle-income countries (LMICs) because of recognised implementation gaps in current PPH practice [10]. Also, research has shown that PPH detection is low in LMICs [11], leading to missed opportunities in providing life-saving care. In these settings, 'visual estimation' of blood loss is typically used to detect PPH (e.g., looking at the blood pooling on the bed under the woman or at materials used to absorb or wipe away blood); however, this approach is often unreliable and inaccurate in estimating blood loss, especially when PPH occurs [12].
Identifying effective ways of encouraging healthcare providers to change their clinical practice and follow evidence-based recommendations remains challenging [13], and the dissemination of evidence-based recommendations alone does not ensure implementation by healthcare providers [13,14]. Failure to implement recommendations into practice can be affected by the challenges of changing well-established individual clinical behaviour, changing hospital protocols and healthcare systems [15]. One potential approach to facilitating the implementation of multi-component guideline recommendations is the introduction of a 'care bundle' . A care bundle is a small set of evidence-based interventions (three to five) which are to be administered concurrently or in quick succession to every person with a given diagnosis. When compliance is high and all components are applied, care bundles can change clinical practice and improve patient outcomes [11,16].

The E-MOTIVE Research Programme
As a result, a new care bundle ('E-MOTIVE') for improving the first response to PPH was developed based on the evidence from the WHO's 2012 and 2017 recommendations, following a WHO Technical Consultation on PPH Bundle Development [11]. The E-MOTIVE first response to PPH bundle stands for Early detection of PPH using a calibrated drape, followed rapidly by (in no particular order): Massage of uterus, administration of Oxytocic drugs, administration of Tranexamic acid, administration of IntraVenous fluids, Examination for identifying and managing the source of bleeding, and Escalation to more advanced care, if bleeding despite first response treatment.
Implementing the E-MOTIVE bundle will necessitate a change in clinical practice behaviours. Therefore, to introduce the E-MOTIVE bundle more effectively and change behaviour, we need to first understand what factors are influencing current PPH practice (i.e., PPH detection and management) and identify a priori potential barriers and enablers to bundle uptake. This combined knowledge will help put strategies in place to hopefully target barriers and enablers of current practice that could facilitate or hinder implementation and what additional support is needed to maximise behaviour change and bundle uptake from the start, i.e. behavioural analysis prior to implementation [17]. This is particularly complex in the context of PPH, which involves two overarching behaviours comprised of multiple sub-actions: (1) PPH detection (e.g. measuring blood loss, monitoring of vital signs) and (2) PPH management (examining the woman, fetching necessary drugs, equipment and assistance, and administering drugs). Furthermore, these behaviours are often performed by multiple care providers and cadres at different time points.
Existing research has shown factors that influence the adherence to recommendations related to PPH detection and management in LMICs, include inaccurate estimation of blood loss after birth, limited skills to detect and to manage a PPH appropriately, and wider health system challenges, such as health workforce shortages, limited resources and training, and unreliable drug supplies [18][19][20]. However, existing studies did not draw on theory to guide their investigation unlike this research [19,20]. A potential limitation of not using theory is not exploring a wide range of potential factors influencing the targeted behaviour and which implementation interventions are needed and likely to be more effective at changing behaviour. Considerable research has shown the advantages of applying behavioural theories, models and frameworks to identify and address barriers and enablers, and the uptake of evidence-based clinical care. A widely used theoretical approach for identifying what needs to change is the Theoretical Domains Framework (TDF) [21,22]. The TDF synthesises 33 behaviour change theories into 14 domains representing cognitive, affective, social and environmental influences on behaviour. These 14 domains have also been further condensed into three over-arching minimum constructs necessary for behaviour change: Capability, Opportunity, and Motivation-Behaviour (the COM-B model) [23]. A strength of the TDF and COM-B model is that they have been mapped onto two frameworks of intervention strategies. The first is the Behaviour Change Wheel which specifies 9 broad behaviour change intervention types (i.e. education, training, incentivisation, modelling, coercion, persuasion, enablement, environmental restructuring and restriction) alongside policy categories to aid implementation (e.g. guidelines, service provision, legislation) [23] (see Fig. 1). The second, is the Behaviour Change Technique (BCT) taxonomy v1, which specifies 93 BCTs, defined as the 'observable, replicable, and irreducible components of an intervention designed to alter or redirect causal processes that regulate behaviour' (i.e. the 'active ingredients;' e.g. goal-setting, feedback on behaviour, problem-solving) [24]. There are published matrices which pair domains from the TDF and COM-B with intervention strategies in the Behaviour Change Wheel and BCT taxonomy, to suggest which types of intervention strategies are likely to be more relevant and effective in addressing different types of influences on behaviour [23,25]. This in turn facilitates more systematic progress from initial behavioural diagnosis of barriers and enablers to selection and design of targeted, and likely more effective, implementation interventions [26]. These interlinked frameworks have been applied to identify factors influencing clinical practice and design interventions to improve implementation across a range of clinical settings including maternal care and implementation of clinical care bundles [27,28].
In LMICs, beyond reported delays in PPH detection and inconsistent, non-evidence-based PPH management, the application of theory-based approaches to address the challenges associated with PPH detection and management-related behaviour are limited. Particularly, in sub-Saharan countries such as Nigeria, Kenya and South Africa, where there is evidence of higher proportions of maternal deaths from PPH [15][16][17].
Ahead of a cluster randomised trial to test the effectiveness of the E-MOTIVE bundle for the first response to PPH (Clini calTr ials. gov: NCT04341662), we aimed to identify the determinants of delayed detection and inappropriate management of PPH [27]. As part of the E-MOTIVE formative research, this study aimed to apply the TDF and the Behaviour Change Wheel to: • Identify barriers and enablers influencing current PPH detection and management • Identify barriers and enablers likely to facilitate or hinder the implementation of the E-MOTIVE care bundle • Generate recommendations for potential implementation intervention types to address identified barriers and enablers and improve implementation of the E-MOTIVE bundle.

Design
Semi-structured qualitative interviews were conducted with healthcare providers currently working in maternity care hospitals in Nigeria, Kenya and South Africa. The full protocol has been published [29]. This study is reported according to the consolidated criteria for reporting qualitative research [30] (See Additional File 1).

Participants
A total of 45 participants were recruited (n=5 participants per hospital; n=9 hospitals in total: 3 hospitals per country, 15 participants per country). Maximum variation sampling was used to achieve a wide range of roles including midwives, nurses, doctors, consultants and administrative/management staff of varying levels of experience and diverse settings based on hospital size and location.

Materials
The semi-structured guide (Additional File 2) had three broad sections exploring influences on (1) current PPH detection, (2) current PPH management and (3) potential implementation of the EMOTIVE bundle. Each section included a mixture of open questions followed by prompts structured around the domains of the TDF. For section 3, the concept of a clinical care bundle was explained, and the E-MOTIVE bundle was described, and photographs of the drape and a pictograph of the bundle elements (See Fig. 2) were shown. The topic guide was piloted in each country with midwives, nurses and doctors and refined to enhance clarity, and incorporate locally used language and definition of terms. Table 1 presents examples of TDF-based questions and follow-up prompts for each section of the interview guide.

Procedures
Country research teams contacted eligible healthcare providers at their place of work and provided study information sheets. After they agreed to participate, the providers signed consent forms before the interviews commenced. All interviews were conducted by experienced researchers (n=7: four doctors trained in maternal care, two midwives and one social scientist) from the country research teams. The interviews were conducted either face-to-face, by telephone or by Zoom due to COVID-19-related physical distancing restrictions. All interviews were audio-recorded and transcribed verbatim by the country research teams.

Analysis
An inductive thematic and deductive framework analysis was conducted [31,32] following three steps: • Familiarisation of the data by the research team by listening to interview audios and reading transcripts • Inductive thematic analysis: a coding matrix based on the interview guide and the research questions was used to collate the data. Data were first analysed to develop theme labels within three overarching, pre-defined categories: PPH detection, PPH management and bundle implementation. This involved identifying meaningful and important seg-ments of text to generate descriptive summaries and grouping of similar summaries together under an inductively generated theme label. The descriptive summaries were phrased in terms of 'belief statements' representing a perceived barrier, enabler or mixed (i.e. both a barrier and an enabler) to current PPH practice or bundle implementation. Two researchers (GF, SA) separately produced descriptive summaries and regularly checked the summaries against theme labels for consistency. Any interpretative discrepancies were discussed until agreement was reached. Clinical accuracy of descriptive summaries and appropriateness of current practice compared to evidence-based practice was checked by SM, a midwife with expertise in PPH detection and management in LMICs. Researchers (GF, SA) checked for any contextual misunderstandings with country research interviewers • Deductive framework analysis: the generated belief statements were then coded to the TDF domains they were judged to best represent. For instance, 'a lack of accurate and objective blood measurement of blood loss leads to delays in detecting PPH' was a barrier coded to beliefs about consequences. The mapping of belief statements to relevant TDF domains was conducted by GF and for reliability purposes, checked by FL, LA and MAB. This analysis was first done for each country. The themes generated from the analysis of transcripts from the first country (Nigeria) provided a starting point for the analysis of subsequent countries (Kenya and South Africa). Findings were then compared across countries to look for similarities and differences

Mapping to implementation interventions
Potential strategies to address the identified barriers and enablers and facilitate implementation were identified by consulting published matrices that pair TDF domains with intervention types specified in the Behaviour Change Wheel and BCTs specified in the BCT taxonomy [24,26,33]. Examples of potential ways to operationalise these intervention strategies in the study context were generated from online discussions by the international, multidisciplinary research team and incountry teams. Table 2 provides the sociodemographic characteristics of participants and a description of the hospitals. Fortyfive healthcare providers, consisting of 14 midwives, (one male; 13 females), 6 nurses (all females), 17 doctors (9 males; 8 females), and 8 managerial staff (4 males; 4 females) participated in the study. Their overall years of experience ranged from one year to 30 years (median=8 years) and time working at their hospital from 3 months to 26 years (median=4 years). All participants currently worked in a hospital in a clinical role (n=37) or in a managerial role (n=8). The managerial staff was senior clinical staff members, with responsibility for managing and overseeing a team of clinical staff responsible for PPH detection and management (i.e. in-charge midwifery or nurse matrons or head of obstetrics). The locations of the hospitals were 2 rural and 7 urban and the burden of PPH cases was categorised as either low (less than 50 cases per year), middle (between 51 and 99 cases per year) or high (more than 100 cases per year). The median duration of the interviews was 67 min (range 38-107 min).

Findings
Overall, similar influences on bundle uptake were reported from participants across countries. In the narrative summary of the results below, we highlight any country-specific influences (i.e. reported by participants from one country only) or influences for which participants from different countries had contrasting views. The findings are presented in three sections corresponding to the stated research questions. Contextual differences and similarities across the 3-country settings are discussed throughout.

Research question (RQ)1: influences on current PPH detection and management
A total of 55 belief statements representing perceived barriers and enablers to current PPH detection and management were identified across 12 of the 14 TDF domains. No belief statements mapped to the TDF domains of optimism and intention. Table 3 presents a sub-sample of barriers and enablers identified within each TDF domain, for each country, with supporting quotes. These are summarised descriptively below, highlighting any cross-country differences and similarities within each domain. The full set of identified barriers and enablers within each domain is in Additional File 3.

Environmental context and resources
The influence of environmental context and resources was mixed across all countries. Provision of PPH care was facilitated by all hospitals being organised into separate wards or rooms for maternity care and by having guidelines, protocols, posters and/or charts related to PPH care displayed in some maternity wards across all countries. A common barrier was inconsistent supplies of drugs, particularly in Nigeria where women or their families often had to procure tranexamic acid themselves. Other barriers were delays in transporting women referred from lower-level hospitals to higher-level hospitals because of a lack of ambulances (all countries), unreliable blood supplies (only in Nigeria and South Africa), staff shortages to manage PPH emergencies (all countries) and lack of tools and equipment to measure blood loss (only in Nigeria and Kenya).

Skills
The influence of skills related to PPH detection and management was also mixed across all countries. Training on PPH received since midwifery, nursing or medical school was inconsistent, with some healthcare providers having received in-service training whilst others had no additional training since pre-service certification. Continuous training was seen as an enabler to keeping updated with new approaches and was recommended by participants in Kenya and South Africa. Participants reported more training for inexperienced or new staff was required in Kenya and Nigeria.

Knowledge
Participants across countries reported differing knowledge of definitions of PPH, different knowledge of guidelines, differing awareness of the signs and symptoms of a PPH and different steps constituting appropriate PPH management. These findings were often in contradiction to WHO guidelines. Participants reported recognising PPH as an obstetric emergency and felt that they were knowledgeable at appropriately detecting and managing PPH.

Behaviour regulation
For all countries, a key enabler was receiving feedback on current practice because it helps identify areas for improvement and in Kenya and South Africa existing Rural --2 2

Table 3 Summary of key barriers and enablers of PPH detection and management in Kenya, Nigeria and South Africa
Key findings grouped by TDF Domain

Environmental Context and Resources
Health hospitals organised into separate wards or rooms for maternity care Regular continuing medical education (CME), including case reviews, will reduce PPH cases   Delayed treatment due to reliance on vital signs to detect a PPH 10 M E M "Especially from the signs and symptoms the mother starts experiencing headaches, light headedness and also dizziness and from the vitals when you check the blood pressure" (Doctor, Kenya) "The extent of blood loss is the first thing that we pay attention to, and then the patient's vital status" (Doctor, Nigeria)

Social/Professional Role and Identity
Good levels of role clarity 34 E E E "My role is that after I detect that the woman has PPH, I will call the attention of my colleague and later on the doctor on call" (Midwife, Nigeria) There is a clear division of tasks based on profession 33 E E B "The person doing the delivery, the midwife will examine and then they would call, usually call the doctor who take over examination or just double check" (Midwife, South Africa) The team managing a PPH should be multi-disciplinary 10 X E X "Midwives even the sub-staff can help but if there are many doctors on ground they can come and help in managing the patient. For instance, one setting the line, one taking the sample for blood for grouping and crossmatching or some applying anti-shock garment, so it is multi-disciplinary activity" (Doctor, Nigeria)

Beliefs about Capabilities
Varied levels of confidence in ability to detect and manage PPH

Beliefs about consequences
Despite PPH being the largest contributor to maternal mortality and described as a 'common' and 'big issue' , participants in this study did not always perceive it to be as serious a concern compared to other maternal health issues. For all countries, negative consequences of current practice were delays in PPH detection because of a lack of an appropriate and objective measure of blood loss and late referrals from other hospitals to effectively manage PPH.

Social/professional role and identity
Midwives and nurses were most often the first responders to PPH, responsible for deciding when to call for help after observing a significant blood loss or a sudden onset of bleeding after birth in all countries. Midwives and doctors performed examinations to determine the cause of the blood loss and took vital signs to detect and confirm a PPH. Therefore, an important enabler was a good level of role clarity (individual and team) in all countries, division of tasks based on clinical role (in Nigeria and Kenya) and having a multi-disciplinary team managing a PPH (only reported in Kenya).

Emotion
Healthcare workers, particularly first responders, reported an association between some negative emotions such as panic and stress, they felt when managing a PPH with the unpredictability and potential for maternal mortality from PPH.

Social influence
The impact of social influence was mixed across countries. An enabler for some participants was teamwork, which they described could speed up and improve the management of PPH; therefore, it is necessary to appropriately manage PPH. In contrast, some participants believed PPH had to be detected and managed alone in some hospitals often because of staff shortages that were reported across all countries.

Memory, attention and decision-making
Memory, attention and decision-making influences were also mixed. In all countries, more attention was given to using guidelines to inform clinical care if international guidelines were adapted into hospital-based clinical guidelines. A common barrier in South Africa was that more consideration was given to vital signs (observing a decreasing blood pressure and an increasing pulse rate) than blood loss to detect a PPH, which may delay treatment. Participants often reported detecting PPH using visual blood loss estimation methods, such as looking at the amount of blood on linen or on the floor, using kidney dishes to collect blood or counting the number of bloodsoaked linens or swabs. Decisions made during this period of diagnosis were sometimes unlikely to include any immediate treatment to manage the PPH. After diagnosing a PPH, management sometimes followed a 'wait and see' approach, where clinicians would implement one treatment measure at a time and wait to see if the woman responded to it. The treatments mentioned as the primary or first-line treatments included uterine massage, oxytocic drugs (e.g. oxytocin and misoprostol) and IV fluids. Some of the reported current care conflicted with the evidence-based practice for PPH detection and management, for example not administering the dosage of oxytocin recommended by WHO guidelines.

Beliefs about capabilities
Beliefs about capabilities were a mixed influence, with healthcare providers reporting varying levels of confidence in detecting and managing PPH. In Nigeria, specific barriers were the need to cope with the rapid onset of PPH and abilities to detect and manage PPH with a woman not booked for birth in the hospital, because of having no details of the medical history of potential highrisk factors for PPH (compared to a woman booked to give birth in the hospital, who therefore has a medical history obtained).

Reinforcement
There were varying approaches to disciplinary procedures for the mismanagement of PPH across hospitals and countries. In Nigeria and Kenya, disciplinary procedures were considered a barrier to PPH detection and management; therefore, disciplinary actions were not required or recommended with more training being preferrable. One participant in South Africa expressed fears of punishment (i.e. disciplinary action taken) for failing to manage a PPH appropriately.

Goals
Across all countries, an enabler was that PPH was perceived to be of high importance because it can cause maternal deaths if not detected and managed appropriately.

RQ2: Potential influences on the future use of the E-MOTIVE bundle
Thirty-eight belief statements representing barriers and enablers to implementing the E-MOTIVE bundle were identified across nine TDF domains. No beliefs were mapped to the TDF domains of Behaviour Regulation, Optimism, Reinforcement, Goals and Emotion. Table 4 shows a full list of belief statements across TDF domains, classified as a barrier, an enabler or mixed across the three countries, alongside supporting quotes.

Beliefs about consequences
Collectively, the feedback received about introducing the E-MOTIVE bundle to clinical practice was positive across all countries. Many participants said that using the bundle had the potential to improve current PPH management and to reduce PPH morbidity and mortality in their hospitals or country. The perceived benefits of using the bundle were the accurate measurement of blood loss which could increase the effectiveness of PPH detection and consequently improve PPH management and reduce PPH mortality and morbidity. The bundled approach of using massage and oxytocin together was thought to be advantageous, although not including misoprostol as part of the bundle was a potential disadvantage because misoprostol is typically used in conjunction with oxytocin, especially where the latter was thought to have poor quality. Potential barriers were how the drape may limit women to only recumbent birth positions (lying flat on their back) and that counting of swabs placed in the drape may be time-consuming.

Environmental context and resources
The influence of environmental context and resources was mixed. An enabler was having a plentiful supply of oxytocin and IV fluids and having a large enough team to deliver the bundle as intended. A key barrier to bundle implementation was the limited availability of tranexamic acid in the maternity ward and in the wider hospital. Other barriers were an inadequate supply of and anticipated high ongoing cost of calibrated drapes, having poor quality oxytocin, bed type not suitable for using the calibrated drape, hospital prone to power cuts that could impede internal examinations and staff shortages.

Memory, attention and decision-making
Memory, attention and decision-making was an enabler because components of the bundle were already used independently, such as examinations to detect the cause of bleeding, uterine massage as the first-line treatment for PPH and oxytocin currently administered simultaneously with the uterine massage. An important barrier was that tranexamic acid was not routinely used (unless as a last resort when oxytocin had not worked).
Despite an overall positive evaluation of the bundle, and a recognition that the majority of bundle elements were already used in current practice, there were concerns raised about how collecting blood loss in a calibrated drape would fit in with current practice.

Beliefs about capabilities
Confidence in performing different components of the bundle varied. Across all countries, participants reported feeling confident in their capability to give oxytocin, IV fluids and uterine massage, but there was varying confidence in examining the genital tract and this also varied across cadre, e.g. doctors were often more confident than midwives. Participants also did not feel confident in using a bundled approach and in giving tranexamic acid, specifically in Nigeria and Kenya where tranexamic acid was not routinely used for PPH management.

Skills and knowledge
Both skills and knowledge were reported to be a barrier to implementing the bundle, particularly a lack of education and skills around how to use components of the bundle that are not routinely used in current practice (e.g., a calibrated drape, tranexamic acid). A further barrier was a lack of understanding of the concept of a clinical care bundle and a need for further training on using a bundle approach.

Social/professional role and identity
There were mixed views about whether certain professional groups would be allowed to perform certain components of the bundle, for instance, whether in addition to doctors, midwives or nurses could perform genital tract examinations. A potential enabler of the overall use of the bundle was assigning roles and a need to work as a team.

Social influences
Social influence was a barrier to the massage and examination components of the bundle. Some women refuse to have an internal examination and staff have concerns about causing discomfort and pain to a woman when applying a uterine massage.

Intention
The influence of intention is a potential barrier, with reported concerns over buy-in (i.e. failure to see benefits) to the bundle approach over existing practice, because

Key findings grouped by TDF Domain E-MOTIVE Component Freq (n=45) Nigeria Kenya South Africa Example Quote(s)
Accurately measuring blood loss improves detection of PPH Calibrated Drape 45 E E E "One of the biggest challenges estimating getting a close to a good estimate of the blood loss so if you are able to get something that gives that then I think it would be a very good thing" (Midwife, Kenya) Uterine massage is effective in managing PPH when used in conjunction with Oxytocics/ uterotonics Massage 16 E E E "I start massaging I will now inject oxytocin but immediately you start massaging you will see the uterus starting to contract…. definitely it comes with reduction in blood loss" (Doctor, Nigeria) Oxytocin is routinely administered as it is effective at preventing and managing PPH E E E "…. we have delays in getting blood so initially we have to make do with IV fluids until the blood is available to maintain haemodynamic status" (Doctor, Nigeria) Using the bundle will increase the effectiveness and improve current management of PPH All components 34 E E E "…it [the bundle] is going to improve the services rendered and then it will really reduce the morbidity and mortality associated with this condition" (Doctor, Nigeria) Using the bundle will reduce PPH mortality and morbidity "I think it is better to use one after the other, than using all at once. The reason for the bleeding, could be trauma, tears, so when you repair a trauma the bleeding stops" (Midwife, Kenya)

Social/Professional Role and Identity
Mixed beliefs that both doctors and midwives or nurses can do examinations

Key findings grouped by TDF Domain E-MOTIVE Component Freq (n=45) Nigeria Kenya South Africa Example Quote(s)
Potential lack of buy-in of using the bundle over existing practice All components 3 B B B "We massage the uterus, administer oxytocic, tranexamic acid, and IV and also examine the vaginal track to remove any retained tissues. We do not refer to it as bundle, but we do all these in the EMOTIVE bundle" (Midwife, South Africa) the majority of bundle elements are already being practised-albeit independently rather than bundled.
Overall, similar influences likely to facilitate or hinder the implementation of the E-MOTIVE care bundle were consistently found across all countries. Table 5 presents a mapping of identified barriers and enablers to potential intervention strategies to address these and improve current practice and maximise future bundle implementation. We proposed 5 implementation interventions to address the identified barriers and enablers: Calibrated Drape, E-MOTIVE training, E-MOTIVE champions, PPH emergency trolley/kit and Audit and Feedback. These interventions focus on increasing the competencies and motivation of staff to adopt E-MOTIVE and to address existing contextual and socio-cultural factors which could hinder the uptake and optimal delivery of E-MOTIVE. One strategy which is part of the E-MOTIVE bundle is a new blood collection measurement tool (i.e., a calibrated drape) that targets delays in PPH detection and inaccurate measurement of blood loss. Therefore, this requires specific training on how to use this new calibrated drape appropriately and includes broader education on the purpose of clinical care bundles, such as E-MOTIVE, and on recommended PPH management to address the identified areas of sub-optimal PPH care. Other barriers addressed by training include administering tranexamic acid, which is rarely used to treat mild PPH in Nigeria and Kenya and only prescribed by doctors in South Africa, inadequate team communication skills, breaking down any hierarchical clinical practices and reassuring midwives of their enhanced role in administering tranexamic acid where there is no specific local hospital protocol specifying that nursing staff can administer tranexamic acid.

RQ3: Mapping of identified barriers and enablers to implementation interventions
Other interventions operate on a day-to-day basis to target any resistance to adopting E-MOTIVE by providing on-site leadership (e.g., E-MOTIVE champions) support to encourage and to demonstrate using the bundle. Also, to address the stress and anxiety often felt by staff when coping with a PPH, in particular in sourcing all necessary drugs and equipment by introducing a regularly stocked PPH-specific emergency trolley/kit to keep everything readily available in one place. At the hospital level, healthcare providers were encouraged to routinely use E-MOTIVE by monitoring uptake and subsequently give feedback to identify areas for further improvements (i.e., Audit and Feedback).

Discussion
This study aimed to understand what factors influenced PPH detection and management, a global health priority with demonstrated evidence of guideline implementation challenges in current practice, and to identify a priori what it would take to implement a newly developed clinical care bundle that aims to increase adherence to evidence-based practice. The E-MOTIVE project's overall goal is to reduce PPH-related maternal morbidity and mortality through improved early detection and management of PPH.
We identified a wide range of influences on PPH detection and management that could affect the future delivery of the new E-MOTIVE bundle. Drawing on the TDF, the thirteen domains to address during implementation were Skills; Knowledge; Behaviour Regulation; Memory, Attention and Decision-making; Environmental Context and Resources; Social Influences; Beliefs about Consequences; Beliefs about Capabilities; Social/Professional Role and Identity; Emotion; Reinforcement, Intention and Goals. There were multiple barriers including divergence in participants' knowledge, insufficient skills training and poor teamwork, based on professional hierarchies. Contextual factors were a lack of supplies and staff shortages. These are consistent with the findings from other research investigating what influences PPH care [18][19][20].
Despite the heterogeneous sample, there was consensus across some beliefs and there were areas of disagreement where beliefs that were barriers in one country were enablers in another. At an individual level, similar beliefs seemed to influence current PPH care practices. There was agreement that improvements in PPH detection were necessary, suggesting greater acceptance of implementing the calibrated drape element of the bundle. There were more discordant beliefs about the benefits of giving tranexamic acid, which is perhaps not surprising given that conclusive evidence of safety and effectiveness has only emerged in the past five years [34]. There were also persistent concerns expressed around the sustainability of the bundle and drapes, along with ongoing stock-outs of essential supplies and drugs and inadequate training.
This study aimed to inform and to support the implementation of the new E-MOTIVE bundle which will be tested in a clinical trial. Whilst it is widely evidenced that clinical bundles can standardise practice and patient care [35], there is also evidence that changing any clinical behaviours, including PPH-related detection and management, is likely to come with implementation challenges [14]. Therefore, in advance of the main trial, this formative research investigated what is likely to influence the implementation of the bundle and identify which implementation interventions could promote the use of the bundle in the contexts of Kenya, Nigeria, and South Africa. In addition to the previously reported influences on current PPH care, the TDF domain of intention was  identified as a potential barrier to bundle implementation with participants stating that current management was deemed to be adequate, indicating that the implementation of the bundle might not give added value. The Behaviour Change Wheel mapping enabled us to narrow the target of potential drivers of implementation and which types of implementation interventions would be best suited to address the identified barriers and enablers. The mapping to five intervention types indicated that the more relevant intervention types to improve PPH care were 'education' and 'training' to increase the understanding of using a bundle approach to PPH detection and management care and providing any new skills. Similarly, 'persuasion' through the communication of the benefits for healthcare providers by staff in a leadership role and 'environmental restructuring' including 'enablement' and 'modelling' to address the known barriers of delivering the bundle as intended. These intervention types were also identified as potential interventions to improve the implementation of other clinical care bundles, such as the 'Sepsis Six' [21].

Strengths and limitations
This study focuses on factors influencing behaviours related to PPH detection and management and how they could pose challenges for the future implementation of a new clinical bundle. The inclusion of first respondersmidwives, nurses, doctors and managers with varying experiences-did provide more insight into the challenges of coping with PPH and the future uptake of E-MOTIVE. However, the clinical position of the researchers conducting the interviews may have impacted the responses given about actual practice; for example, if a doctor interviewed a junior midwife, she might give an answer based on what she thought the doctor would want to hear, rather than on reflecting actual clinical practices (social desirability bias). Efforts to minimise this influence included training interviewers on appropriate probing and to avoid judgmental language including continuous monitoring of responses and feedback throughout the data collection period by the formative research team.
A number of barriers were identified, and many of which were structural-and resource-related, such as lack of electricity, staff and number of hospital beds. While we recognise the importance of these barriers, they are unfortunately outside the scope of the trial interventions to feasibly and sustainably address them. In the main trial process evaluation, we plan to document these contextual challenges and ensure that they are accounted for in future efforts to scale-up PPH detection and management.
Approaches to changing clinical behaviours tend to be pragmatic; therefore, demonstrating how improvements were achieved is more difficult in the absence of any theory-based development or conceptualisation. The major strengths of this study are the systematic and theoretical approach to specifying the types of interventions required using implementation and behavioural science methodologies. These behaviour change frameworks have been used across many clinical settings for identifying implementation problems and informing implementation interventions, but not yet in the PPH field. This approach is in keeping with the importance placed on early consideration of implementation to increase the development of an intervention (such as E-MOTIVE), as advocated by the United Kingdom Medical Research Council [36].
An issue that emerged from the study was how challenging participants found it to conceptualise using a 'bundled' approach of clinical care, as this was a new and hypothetical approach for most participants. To enhance understanding about the E-MOTIVE bundle, participants were shown pictures of the new calibrated drape and a pictograph of the bundle, and the bundled approach was described by the researcher conducting the interview. Despite this, the findings from anticipated, hypothetical use of the bundle may differ from those discovered from actual use. However, participants noted a number of potential issues based on current practice and what they would see as likely issues to using the bundle hypothetically, and what would help to implement it in their specific contexts. This hypothetical bundle implementation enables us to think about actual implementation from the outset of project development, rather than waiting to address potential implementation gaps or failures at the trial stage. The approach gives the bundle and interventions the best chance of succeeding and maximising the internal validity of the implementation trial by ensuring the bundle is delivered with fidelity and giving us the optimal opportunity of evaluating it. Barriers and enablers to the actual use of the E-MOTIVE bundle will be assessed as part of a process evaluation of the E-MOTIVE trial.

Conclusions
To our knowledge, this is the first study to conduct theory-based formative research with healthcare providers of a new clinical bundle to inform the design of implementation interventions aiming to improve the detection and management of PPH in low-resource, high-burden settings. It provides additional insights into what implementation intervention might enhance efforts to create the desired changes in PPH detection and management behaviours to maximise bundle uptake. Despite, many participants expressing preliminary acceptance of adopting a bundled approach, some changes are necessary to promote and increase bundle uptake by addressing the identified challenges. These changes include training on using the E-MOTIVE bundle elements all together (rather than a wait-and-see sequential delivery), training new skills for midwives to administer IV tranexamic acid, promoting more teamwork and communication and ensuring an adequate of stock of the necessary consumables, equipment and drugs to deliver the bundle. These findings underscore a critical need to develop strategies to address implementation challenges whereby our formative research may maximise the uptake and appropriate use of the new E-MOTIVE bundle by basing the implementation intervention on the barriers from current practice and challenges identified by the potential users of the bundle.
The next step is further research during the E-MOTIVE clinical trial to investigate the acceptability and usability of these implementation interventions with healthcare providers and stakeholders in Nigeria, Kenya and South Africa.