Caring for providers to improve patient experience (CPIPE): intervention development process

A growing body of research has documented disrespectful, abusive, and neglectful treatment of women in facilities during childbirth, as well as the drivers of such mistreatment. Yet, little research exists on effective interventions to improve Person-Centred Maternal Care (PCMC)—care that is respectful and responsive to individual women’s preferences, needs, and values. We sought to extend knowledge on interventions to improve PCMC, with a focus on two factors–provider stress and implicit bias–that are driving poor PCMC and contributing to disparities in PCMC. In this paper we describe the process towards the development of the intervention. The intervention design was an iterative process informed by existing literature, behaviour change theory, formative research, and continuous feedback in consultation with key stakeholders. The intervention strategies were informed by the Social Cognitive Theory, Trauma Informed System framework, and the Ecological Perspective. This process resulted in the ‘Caring for Providers to Improve Patient Experience (CPIPE)’ intervention, which has 5 components: provider training, peer support, mentorship, embedded champions, and leadership engagement. The training includes didactic and interactive content on PCMC, stress, burnout, dealing with difficult situations, and bias, with some content integrated into emergency obstetric and neonatal care (EmONC) simulations to enable providers apply concepts in the context of managing an emergency. The other components create an enabling environment for ongoing individual behavior and facility culture change. The pilot study is being implemented in Migori County, Kenya. The CPIPE intervention is an innovative theory and evidence-based intervention that addresses key drivers of poor PCMC and centers the unique needs of vulnerable women as well as that of providers. This intervention will advance the evidence base for interventions to improve PCMC and has great potential to improve equity in PCMC and maternal and neonatal health.

The target population for this project is health care workers in maternity units as well as other support units in Migori County, Kenya. This site has previously been described [9,65,66]. To summarise: Migori County, located in western Kenya, has eight sub-counties. The county population is approximately one million, with an estimated 40,000 births annually. The estimated maternal mortality ratio is high at 673 deaths per 100,000 live births compared to 495/100,000 nationally. There is one county referral hospital, seven sub-county hospitals, and several health centres, dispensaries, and faith-based and private health facilities. The health care worker patient ratios are 32 nurses, 19 clinical officers, and 4 doctors per 100,000 people. Based on the most recent Kenya demographic and health survey, 53% of births in the county occurred in health facilities, compared to the national average of 61%. The need for this intervention was informed by findings of our initial research in Migori, which focused on understanding the extent of PCMC as well as the barriers and facilitators to providing it. Thus, this served as the initial formative research for the intervention design. Data collection for this initial period starting in 2016 included surveys and in-depth interviews with over 1000 women who had recently given birth and 49 providers in the county. This was followed by a second phase of data collection in 2019 with about 100 providers focused on better understanding the role of provider stress and bias in PCMC. The methods and findings from this initial formative work in Migori County are published, including quantitative and qualitative interviews with women highlighting gaps in PCMC [9,44,65,66], as well as qualitative and quantitative interviews with providers to understand drivers of poor PCMC [19,20,25], and to assess extent of provider stress and bias [30,31,67]. We summarise the relevant findings from phase 2 to provide context for the intervention components. First, the work highlighted that provider stress, burnout, and bias as well as difficult situations during childbirth were key drivers of poor PCMC [19,20,25,66]. Other provider-level drivers included inadequate knowledge and skill on various aspects of PCMC, perceived lack of time, forgetfulness, self-protection, and assumptions about women’s knowledge and expectations [19,20,25,66]. Further, using an adapted Implicit Association Test (IAT) and situationally specific vignettes to assess implicit and explicit SES biases, respectively, we showed that providers had implicit biases in their associations between difficult patient characteristics and patient SES attributes. Providers also had explicit biases and incorrect assumptions about poorer women and care expectations (e.g. poor women did not expect providers to introduce themselves to them and would not sue if something went wrong, or that once a woman is present at a facility, she has consented to all subsequent care). Data highlighted that differential treatment was linked to women’s appearances, assumptions about who is more likely to understand or be cooperative, women’s ability to advocate for themselves or hold providers accountable, ability to pay for services in a timely manner, and situational factors related to provider stress and burnout [67]. These factors interacted in complex ways to produce PCMC disparities. Many of the drivers of poor PCMC and sources of bias are modifiable and could be addressed in trainings; yet, only 22% of providers surveyed in our study sites in Migori County reported participating in a training to improve patient-provider interactions [67]. Further, 96% of providers surveyed were experiencing moderate to high stress, while 85% were experiencing low to high burnout [31]. When asked what causes them the most stress at work, most providers (55%) stated high workload [68]. Lack of supplies or equipment was the second most common leading cause of stress (15%), followed by poor salaries (8%). Other sources of stress were frequent staff turnover, personal/family problems, incompetence of other providers, attitudes of superiors, colleagues, and patients, and death of a patient. Within the last year, 34%, 37%, and 55% of providers reported they had been treated in a way that was disrespectful or humiliating by their superiors, colleagues, and patients respectively. Close to half of providers (42%) reported they had ever lost a baby or mother during pregnancy or childbirth; with 57% reporting this had happened in the last year. Seven per cent had thoughts of suicide. Yet, few providers have had training on how to cope or deal with stress – over 80% of the providers had never received stress management training – although almost all (98%) reported they would like such a training. Similarly, 88% reported that they had no readily available access to psychological and emotional support, from a counsellor, psychologist, psychiatrist, or other mental health professional, although almost all (94%) reported they would like to have access to such support. Eighty-four per cent (84%) had no access to workplace peer support, although 92% wanted such support. Thirty-six per cent had no mentor in the country, although all wanted to have a mentor in the county. These findings underscored the need for an integrated multi-level intervention that addressed several of the drivers of poor PCMC, as well as supported providers. Drawing on the existing literature on interventions to improve PCMC [27,57], reduce stress and burnout [69–71], and mitigate the effects of bias [72–74], we first designed the CPIPE intervention to target key drivers of poor PCMC using social and behavioural science theories – Ecological Perspective, Social Cognitive Theory (SCT), and Trauma Informed System (TIS) frameworks–to inform the intervention strategies. The ecological perspective recognises that behaviour is influenced at multiple levels and underscores the importance of enabling environments for individual behaviour change [75]. We therefore identified intervention components at different levels of influence (Figure 1), to create an enabling environment for individual behaviour change efforts. SCT describes a dynamic process in which personal factors, environmental factors, and human behaviour exert influence upon each other (reciprocal determinism). SCT posits that if individuals have a sense of self-efficacy, they can change their behaviour even when faced with obstacles. Also, people are more likely to change if they believe the activity has benefits to them (outcome expectancy); and if there are tangible goals, positive role models (observational learning), and reinforcement [76]. These SCT constructs informed the intervention strategies (Figure 2). For example, the intervention will include a training component, which emphasises benefits of stress reduction and coping strategies for the provider, as well as for women and their babies, to decrease resistance and increase their engagement. The training will also use simulation, a technique of choice in training professional teams, to evoke real-life scenarios in an interactive fashion to increase self-efficacy [77], and adult learning techniques such as self-directed inquiry [78]. Finally, the TIS framework recognises stress as a source of trauma to the system, which if not addressed leads to numbing, reactivity, and depersonalisation [79,80]. TIS supports ‘reflection in place of reaction, curiosity in lieu of numbing, self-care instead of self-sacrifice, and collective impact rather than siloed structures’ [80]. This philosophy informed the overall intervention approach. Ecological approach. Conceptual framework. For the training content, we proposed that the CPIPE intervention includes both didactic and interactive sessions addressing key drivers of poor PCMC. Given that the sources of stress are difficult to avoid in the life of a provider in a low resource setting, our approach focuses on the factors that influence the stress response – which is providers’ perceptions of the stressors, coping strategies, and their general wellbeing [32]. For implicit bias, a first step towards mitigation is recognising and creating structures to minimise it [50]. Additionally, people have to be concerned about the effects of bias to be motivated to identify and learn to replace biased responses with responses more consistent with their goals [72]. Further, existing psychological theories posit that people are inherently complex, with multiple and often contradictory patterns of selves. Thus, it is possible to reduce the effects of people’s bias through activities that elevate the alternative selves and goals that people endorse, without actually removing their deep-seated biases – referred to as sidelining implicit bias [81,82]. We therefore utilised content from our prior research to increase bias awareness and mitigation, as well as an overall focus on the caring provider whose goal is to provide PCMC – with particular attention for the needs of the most vulnerable whom they may unintentionally mistreat. To create an opportunity for providers to practice the training content in the context of providing clinical care, we planned to integrate the content on PCMC, stress, bias, and difficult situations into an emergency obstetric and neonatal care (EmONC) simulation and training developed by PRONTO International. The PRONTO training kit, called a PRONTOPack™, includes a hybrid birth simulator called PartoPants™ (a modified pair of surgical scrubs with all anatomical landmarks necessary for delivery) worn by a patient actress, which creates a highly realistic experience that centres the patient [83]. During each simulation, one provider plays the role of the patient, enabling them to engage with the patient experience. This also creates the opportunity to practice providing PCMC while managing an obstetric emergency. Simulations are followed by a debrief, which provides the opportunity for discussion and feedback about the patient and provider experience [83]. This drew on our prior work in Ghana where we successfully integrated PCMC into PRONTO’s simulation and training curriculum and showed that this was acceptable to providers and was an effective means to improving PCMC in SSA [84,85]. For the mentorship, peer support, embedded champions, and leadership engagement components, we decided to collect additional data in phase 3 (described below) to assess the relevance, feasibility, and acceptability of these components and to tailor them to the context. In addition, we sought feedback on the training content and approach as well as its composition, duration, and location. This phase of research involved in-depth interviews with 21 stakeholders in the county, including nurses, midwives, clinical officers, support staff, community members, and county/sub county health management teams. The participant list was collaboratively developed in consultation with the study team, who had prior experience and in-depth knowledge of the context in which this intervention will be implemented. The interviews were conducted remotely over phone from January to March 2021 due to restrictions during the early phase of the pandemic. Semi-structured guides were used by two interviewers, who were core members of the study in the previous phase, to explore perceptions regarding the intervention strategies, as well as preferences for structure, content, format, modality, frequency, location, etc. Question formats included describing the planned intervention to respondents and asking for their perceptions and concerns about it. We also asked open-ended questions about their preferences for how the various components should be implemented. Interviews were audio-recorded transcribed and coded by the study team utilising a collaborative deductive coding framework. Codes were then queried, and excerpts analysed using a thematic approach. Additionally, at the beginning of this phase of the work, we established a local advisory board to advise on the project to ensure the intervention was relevant and feasible to the context and to develop local ownership – a critical component to the sustainability of the project. We summarise the results of this research which informed the final intervention plan below, organised by the five intervention components. Representative quotes are presented in Table 1. Representative quotes for themes from formative research on intervention plan. In general, respondents approved of all the training topics proposed to them and many noted how a training on the topic would be helpful to them. For instance, for PCMC almost all respondents highly recommended the training as something that would improve their performance at work. One respondent however noted some providers might be reluctant to disclose their mistreatment of patients in a training due to fear of reprimand (e.g. job termination). Thus, it was important to provide a safe space where participants can talk about their experiences without fear of judgement or punishment. For implicit bias, most respondents affirmed the presence of bias in the way providers treat patients and felt that the training was relevant. It was, however, acknowledged that although implicit bias is prevalent and the training needed, how the content is delivered might affect how participants receive it. Similarly for stress and coping mechanisms, all respondents acknowledged that providers in the maternity unit undergo stress due to several reasons, and providers will appreciate a training on dealing with stress and will participate effectively. Most respondents held a favourable view of integrating the new content with the PRONTO simulation training, having participated in prior PRONTO trainings. However, a couple of respondents raised concerns about feasibility of integration, specifically related to time constraints and the size of PRONTO’s package – noting that a smaller package could be more effective. Regarding the composition of people at each training, the views were mixed. Some thought the mixing of cadres was a good idea because it reflected their working conditions and provided an opportunity to foster teambuilding skills and promote teamwork. Others, however, thought the mixing of cadres was not a good idea because of different levels of training, knowledge, and understanding between providers and support staff. Others noted prejudice and the feeling of inferiority in junior staff and support staff would discourage them from fully engaging in the training. When asked how long a reasonable period was to take providers out of work for training, the average suggested duration of training was between 2–4 days (ranging from one day to two weeks). Some respondents recommended that the training be held off-site (i.e. hotel or conference centre) to give them a break from the usual work stressors and limit distractions, while others preferred facility-based training due to convenience and increased access to the trainings. Suggested approaches to enhance engagement included making the trainings interactive, such as using scenarios, simulations, dramatisation, and team-building activities. Additionally, they recommended carving out time for participants to have tea together to discuss and ‘share and build each other up.’ Leadership/management involvement and support as well as follow-up trainings were also considered facilitators of engagement. Most respondents greatly appreciated the strategy of peer support groups. They noted that although there were no such groups for health care providers, it has worked well for people living with HIV and for family planning programmes. Respondents cited that in these groups’ providers can debrief after having difficult work experiences. It was also noted that peer support groups will enable providers to bring up issues they all experience, listen to each other and freely discuss how they are dealing with the issues raised among themselves, without supervisors. Like the training, views regarding composition of peer support groups were mixed: Some respondents thought mixing different cadres of providers will create the opportunity to discuss shared experiences. Most however preferred cadre-specific groups – noting that some providers (such as support staff) might not feel comfortable enough to openly share their experiences in the presence of other cadres. Most of the respondents preferred in-person peer support group meetings (as opposed to an online platform) where providers can meet each other, express themselves easily, and have assurance of confidentiality. Meeting frequencies mentioned ranged from weekly to quarterly. Using WhatsApp groups for peer support was acceptable to a few, but others noted they were already on several WhatsApp groups, and discussions on those groups are not confidential. It was also noted that some of the providers do not have access to smart phones. All the respondents appreciated the need for a mentorship programme, as this will bring together providers for knowledge exchange. Respondents recommended that the programme comes up with a key area on what the providers need to be mentored on based on gaps we have identified. It was noted that there are some existing mentorship programmes within the county (focused mostly on clinical topics) hence the need to come up with different focus areas for this programme to avoid duplication. Respondents however had mixed reactions about how mentors and mentees should be paired. Some respondents liked the idea of paring senior and junior staff as mentors and mentees respectively. But others noted that using years of experience may not be effective, given many of the leaders in top positions in the health department are junior in terms of years of experience, but are more qualified. Some also suggested that junior staff may be able to mentor senior staff in areas they have expertise in, as senior staff can have limited knowledge on some topics. It was thus recommended that, what needs to be considered is expertise and competence, and not just seniority. All respondents acknowledged that using embedded champions is good for sustainability. Some respondents recommended that the best way to go about selecting the champions is to have two per facility, so that when one is transferred, the other person can continue to train the rest. Some respondents noted that champions should be people who add value to the project; and are vocal, passionate, and dedicated. Some recommended we identify the most active person during training in a facility as the embedded champion, while others noted all provider cadres should be considered to avoid biases and wrangles in the facilities. Suggested ways to motivate embedded champions included having periodic meetings, supportive supervision by the project team, providing some form of motivation, having continuous assessment and designated roles for the champions, involving facility leaders, and forming WhatsApp groups for champions. One respondent suggested that the champions should be invited for existing continuous medical education (CME) meetings in the referral hospital to learn both soft and technical skills. This was noted as critical for ownership and sustainability. Thus, involving the county leaders in the health system and forming a community advisory board (CAB) to represent all stakeholders, including at the community level, was laudable. Some respondents noted that many projects tend to work with county coordinators and side-line the facility in charge. It was thus commendable that the project involved the facility in charge, as this was important. It was also noted that involving both county and facility leadership in the project would enable them to get to know the challenges that providers encounter. Suggested ways to maintain leadership engagement included inviting leaders to the trainings, identifying a focal person for the project at the facility level to facilitate project activities, and having periodic meetings with the leaders to update them and discuss planned project activities. It was also suggested that the project team connects with county officials to discuss how the project activities can be integrated into the annual county work plan. Following analysis of the formative research data, we presented the findings to the CAB and made several decisions regarding the final intervention plan and approach to implementation. This included an agreement to proceed with all the intervention components as proposed (Table 2), with careful attention to how they were presented. For example, we focused on creating a safe space where providers can be comfortable speaking up and engaging without fear of reprisal. We also continued with the plan to integrate the proposed topics into the EmONC simulation training, but reduced the number of simulations to only two during the initial training, which will also include didactic and interactive sessions on topics listed in Table 2 in the form of case-based learning, teamwork activities, hands-on practice, and reflective sessions. While acknowledging the concerns of those who thought it might not a good idea to combine different cadres of staff, we decided to go ahead and combine them for the training since most approved of this, and this was a way to improve teamwork and communication and promote interprofessional development. We also decided to limit the initial training days to two days, followed by monthly refreshers over the 6-month intervention period. All maternity providers in the intervention facilities will be invited to participate in the training to facilitate a person-centred facility culture change. To help manage patient needs and keep number of participants to a reasonable level, given protocols for the COVID-19 pandemic, facility leads will divide providers in their facility into two groups who will attend alternate 2-day training sessions. The initial training was to be held offsite, while the monthly refreshers will be held onsite in each facility on days and times identified as appropriate at the facility (e.g. during designated CME times). The initial training days and times were selected in consultation with the CAB and county and facility leadership, who were all invited to participate in the training. Other recommendations such as pre and post assessments for training, providing pamphlets, interactive activities, tea breaks, etc., were incorporated into the training plan. CPIPE intervention strategies. For peer support, we agreed these will comprise groups of people of similar cadres (nurses/midwives, support staff, doctors/clinical officers) within each facility. A leader will be selected from each group within the facilities to coordinate and lead in-person meetings of the group each month. WhatsApp will only be used for sharing resources. Based on discussions about mentorship during the formative research, we fielded a short survey to all providers in the intervention facilities to better understand their mentorship needs. The survey also included a list of people who were willing to serve as mentors, and potential mentees were asked to identify those they would like to have as mentors (to ensure mentees were paired with mentors they were comfortable with). Provision was also made for peer mentorship. For embedded champions, each facility will nominate two embedded champions following the initial training, who will serve as the central persons to facilitate monthly refreshers and coordinate peer support activities in their facilities. For leadership engagement, leadership at all levels will be included in the CAB, and CAB meetings will be organised quarterly for continuous leadership engagement. Per the earlier findings on gaps in individual psychological support, we reached out to inquire about the presence of clinical psychologists or mental health counsellors in the county who could provide individual counselling to providers and identified two. The two psychologists agreed to better meet the counselling needs of providers and shared their phone numbers to be given to intervention participants who could reach out to them as needed. The intervention is ongoing in the county. In subsequent manuscripts, we will discuss the implementation of the intervention, lessons learnt, participants perceptions of the intervention, and preliminary effectiveness on various outcomes.

N/A

• afulani, patience a.
• oboke, edwina n.
• ogolla, beryl a.
• getahun, monica
• kinyua, joyceline
• oluoch, iscar
• odour, james
• ongeri, linnet