Exploring the sustainability of perinatal audit in four district hospitals in the Western Cape, South Africa: a multiple case study approach

Perinatal audit in South Africa

All public health hospitals conducting deliveries in the Western Cape Province in South Africa have been implementing perinatal audit for over 15 years using the PPIP.23 Given the long history, hospitals in the Western Cape will be a conducive environment to understanding sustained practice, considering microlevel and mesolevel factors.

Study design

A multiple case study design was applied to understand the ‘how’ or ‘why’ of sustained implementation.24 We used a multiple holistic design whereby the subdistrict was considered as a unitary whole, allowing for comparison across settings to gain insights on factors influencing sustained implementation of perinatal audit.

Sampling

Data collection

Data collection tools included a key informant interview guide and a meeting observation guide (online supplemental file 3). The interview guide focused on individual perceptions about the perinatal audit process, factors needed for implementation and team dynamics related to implementation. The meeting observation guide considered who was in attendance, information presented, and behaviours and interactions of participants.25 Fieldwork and data collection took place from October 2019 to March 2020, ranging from half of a day to 5 days per site. MK conducted the fieldwork and sent a summary report of preliminary findings and reflection to the research team within 1 week of visiting the site. Key informant interviews were in English and ranged from 20 min to 1 hour. All interviews were conducted individually with the exception of case D, which were done in two groups. Non-participant observations occurred at seven meetings: two provincial PPIP meetings, three subdistrict perinatal review meetings (M&M meetings), one monitoring and evaluation (M&E) meeting, and one other staff meeting.

Data management and analysis

Interviews were recorded and transcribed. Transcripts, observation and reflection notes were compiled and analysed using Atlas.ti V.9 by MK with oversight from AG. Thematic analysis was used applying an analysis framework derived from Carl May’s extended normalisation process theory,26 an implementation theory used to consider broader social systems in which interventions are implemented (online supplemental file 4). Undertaking an iterative process, we developed the coding framework by analysing the data from case A using the dimensions and constructs of the extended normalisation process theory. With these findings, we identified emerging themes and gaps in the analysis framework. The codebook was revised to include descriptive factors as well as tailored to suit the intervention and related results. This revised codebook was tested and refined using the same case study as well as another (case C) before being applied to all of the data. A report was developed for each case study by MK and received inputs from all authors.

Rigour, positionality and ethics

Measures were taken to ensure rigour of the case study approach,24 27 such as engagement with stakeholders prior to data collection, voluntary participation of participants, seeking peer and expert feedback, audit trail with clear mapping of the research process and triangulation of data sources. A feedback report was shared with subdistrict managers to verify results with the stakeholders. Permission to take photographs of documents and training materials was given from subdistrict health administrators, with commitment not to include sensitive information and identifiers. The lead author did not know any of the participants prior to the study but was able to develop trust with them through stakeholder engagement, including spending a few days in the subdistricts. This engagement helped to contextualise and interpret the data. Though not involved in the data collection process, other authors (A-MB, NR and RP) may have been known or familiar to some of the participants, given their involvement in the national and provincial perinatal audit processes.

Patient and public involvement

Patients or the public were not involved in the design, conduct, reporting or dissemination plans of this study. A short report disseminating the findings was shared with subdistrict managers, district health managers, and regional and provincial stakeholders. An in-person feedback session occurred in one subdistrict with study participants.

Capability

May posits that routine implementation depends on its workability and integration into everyday practice.26 Participants described perinatal audit activities as embedded into everyday workflows. In all subdistricts, the managers viewed data capturing of information about perinatal deaths as part of their routine data collection and reporting system. However, PPIP was more embedded in the information system in region 2 because the responsibility of the data capturing and analysis using the PPIP software was the responsibility of the information officer, not the clinical staff, as in region 1. The information officers reported that they would collect the PPIP data from the maternity ward at the same time as collecting data for the routine information system. Clinical staff regularly attended M&M meetings, and managers expected and monitored their participation. Participants more involved in the process, such as meeting facilitators or data capturers, reported that they had adequate time to complete the related work and considered it part of their jobs.

Perinatal audit activities were linked to ongoing processes, such as quality improvement. For example, managers tailored trainings or quality-related interventions to the identified issues during the M&Ms. Participants agreed that the ‘response’ component of perinatal audit was taken forward as part of their routine work. Discussion and action points from the M&M meetings were shared at regular management or team (ward or clinical) meetings in order to implement actions:

We'll go through the old minutes with the next Monday [bimonthly management meeting]. And they'll ask you “Did you sort that out?” and we need to give feedback on that. –Clinical manager

New staff orientations formally and informally integrated perinatal audit. Few participants reported undergoing official training on the components of the intervention, with the exception of those involved with data capture using PPIP, most of whom had received some formal training. The district health team and the regional PPIP coordinators embedded staff capacity development efforts to improve perinatal audit activities (eg, data collection) into ongoing trainings. For the most part, training was unofficial and embedded within general orientation and learning their roles on the job:

With the new operational manager of the maternity ward, we will like spot teach what M&M is and what you are supposed to record on that M&M [form] but not official training. –Maternity ward operational manager

The integration of perinatal audit into other subnational level mechanisms and activities further supported sustained practice. For example, the regional PPIP coordinators scheduled their monthly clinical outreach visit to the district hospital (to conduct routine specialist procedures) on the same day as the related M&M meeting. District teams provided materials to support implementation, such as the PPIP software and a standard operating procedure (SOP) template for how to conduct M&M meetings. In all subdistricts, the ‘ideal hospital initiative’ was being implemented, requiring a minimum number of 10 M&M meetings per year, including some perinatal-focused meetings.28 Costs related to implementation of perinatal audit were integrated into existing budgets. Participants did not view activities for perinatal audit as stand-alone but rather as an integral part of clinical governance.

Contribution

Another dimension of the theory suggests that routine implementation depends on people’s contributions through investing meaning, commitment, effort and appraisal.26 For perinatal audit, participants had a collective understanding of the purpose but not the process and how the different components linked to each other. For example, PHC nurses did not include the surveillance system (PPIP) when asked to describe perinatal audit because they did not engage with that component.

Participants highly valued the intervention for improving service delivery, learning skills and debriefing about cases as a team. For improving service delivery, participants gave examples of change due to perinatal audit, such as additional trainings, human resource changes, the development of SOPs and acquiring additional resources. For example, one hospital permanently assigned a medical officer to the maternity ward in response to issues raised during the audit process. Another subdistrict used the audit process to advocate for additional midwives in the maternity ward. The midwives and nurses indicated that the perinatal audit process improved data capture and collection since they knew that information would be reviewed and discussed at the perinatal review meeting.

It helps us as a staff - as midwives - to be accurate with the writing of the notes because most of the time when there is an emergency, you are busy. Sometimes you forget to record … So it [perinatal death audit] helps us to improve our skills as well. –Midwife

For learning skills, participants viewed the M&M meetings as an opportunity to gain clinical skills from the referral hospital specialist (eg, obstetrician or paediatrician).

It’s almost like getting a refresher every month of at least one to three topics in obstetrics that he [outreach specialist] does. –Clinical manager

For debriefing, participants reflected that the meetings were an opportunity to collectively and openly debrief about a difficult case. Any death can be traumatic for the staff, and debriefing can help those involved understand what happened.

You need feedback on what has happened. It doesn’t help if you’ve nursed the patient and baby is gone or mom’s gone and you don’t have any feedback on what happened. –Midwife

It is [valuable] because at what other platform are we gonna discuss? One is one death too many you know. –Nursing manager

Though the team dynamics varied between subdistricts, overall participants used the review process as an opportunity to navigate professional hierarchies, hold each other accountable, improve communication, and build and nurture their relationships. An established cohesive team environment led to participants wanting to contribute to the process as part of the camaraderie felt between staff. The team approach to implementation ensured accountability and representation by multiple cadres (doctors, maternity staff, information and subdistrict management):

It’s not only a doctor driven thing. It’s a nursing and a doctor driven thing… We as the nursing staff - any category of the nursing staff - can give inputs to it [M&M]. –Nursing manager

Everybody’s got a voice there from the juniors to doctors to the sisters and I think we make everybody’s opinion count. –Clinical manager

Participants trusted the process because the review meetings were well facilitated and occurred in a safe, non-blame environment conducive to learning. The M&M meetings did not exceed the scheduled 1 hour, requiring careful preparation of cases and strategic facilitation (box 2). While only one subdistrict presented a code of conduct at the start of the meeting (online supplemental file 5), all participants believed others understood the purpose and rules of the M&M meetings. Some of the nurses and midwives still felt blamed by management and doctors during the review meetings but indicated it gets better over time. When anonymity was not maintained during M&M meetings, it was only because those involved in a case would indicate that it was their case in order to explain better what had happened, signalling they trusted the process and wanted to debrief. Of the meetings observed, the facilitator never first disclosed who was involved in the case.

By seeing how it works over time, participants knew what to expect and did not fear participation. One subdistrict experienced initial resistance to perinatal audit and found the following measures improved the process and led to sustained practice: (1) clear instructions on how to conduct meetings, (2) local adaption of the process to suit their needs, and (3) improved facilitation of the review meetings to ensure ownership and a blame-free environment by having the clinical manager lead the meeting, along with the doctor who was involved in the case.

Potential

A third dimension of the theory posits that implementation depends on people’s commitment to operationalising the intervention. The potential for sustained practice of perinatal audit came from the individual and collective commitment by staff to deliver high-quality maternity care. At an individual level, most of the participants were very dedicated to their jobs and demonstrated pride and confidence in their clinical practice. The maternity ward staff were described as especially passionate about their work and competent:

The sisters that work in maternity they’re excellent with what they do and they’re committed. We trust that they know their job. They can manage everything. –Nursing manager

Overall commitment to high-quality service delivery was reflected by individual motivation to achieve good results and the belief that perinatal audit would help. The enthusiasm of a few committed individuals to implement the process drove others to engage and even lifted the level of commitment to perform, especially among the facilitators (box 2). In general, participants felt open to learning and new approaches.

Participants were motivated when they saw that their subdistrict statistics were among the best in the region. Subdistricts in region 2 provided continuous professional development points to doctors who attended the meetings, though this was not an incentive on its own. Financial incentives were not offered nor did people feel it was necessary (eg, per diems, tea and coffee). Collectively, there was buy-in because people saw that their engagement would yield positive change and was worth their time and their staff’s time:

You only get buy-in if people see why. If people get positive things out of it to see why am I doing this and not feel threatened and can see the learning opportunity… –Medical manager

Low staff turnover also enabled shared commitment to work together and improve the health system collectively. Many of the participants were from the subdistrict or had been there a long time, with no intention of leaving, which facilitated the motivation to improve the health system:

If your people know that this is now their hospital where they’re going to be for a few years. You try to implement things that make your life easier… But if [not], people didn’t really care about improving the system because tomorrow they’re going. –Family physician

The overall implementation process at subdistrict level was a shared task among multiple players who were committed to their role. These individuals acted as informal teams, each having a different responsibility linked to the audit process and holding each other accountable to ensuring the tasks would get done. The common characteristics of these informal teams included open and constant communication, trust in each other, and dedication to quality improvement and expectation of excellence among actors. The multidisciplinary nature of the process demonstrated shared commitment among all actors engaged. Some key actors were consistently absent from the observed processes, notably emergency medical services (EMS) and district health management. Three of the subdistricts reported that an EMS representative would normally attend the perinatal meeting or would attend if asked, but even in these settings, some participants expressed frustration when they did not attend, given their important role in the referral process. Direct engagement from the district health management team was limited in the perinatal audit process. Subdistricts in region 2 reported that the district comprehensive health manager would sometimes attend; subdistricts in region 1 reported no engagement from the district office. These subdistrict managers indicated that information related to perinatal audit would be reported to them in other meetings as relevant.

Capacity

The final dimension of the theory considers that implementation depends on people’s capacity to co-operate and co-ordinate their actions. Across the case studies, participants described working in an environment that supports the implementation of perinatal audit. These subdistricts have well-functioning hospitals with highly competent staff, at the management and clinical levels. Resources were already in place to implement perinatal audit, that is, staff capacity, data capturing forms, computers and available space (meeting room). Some participants reported budget constraints to implement actions identified through the audit process, that is, human resources and equipment procurement. For example, all subdistricts, except for case B, reported not having enough staff in the maternity wards. Subdistrict managers responsible for addressing these challenges considered these challenges as part of the broader budget management and constraints, as demonstrated by this quote:

Most of the time what comes up in these perinatal reviews is the number of staff. But I must look at the budget… a professional nurse in maternity ward is expensive… My hands are tied because this is like my budget. How am I going to cut it? –Nursing Manager

The subdistricts demonstrated professional work environments with clear and regular communication. Communication channels between the district hospital and the regional referral hospital included a range of mediums, for example, phone, email and WhatsApp. Participants felt there was open communication between team members and health system levels, which made it easier to share information. For example, participants indicated they could call the regional hospital and speak directly to a specialist (eg, obstetrician or neonatologist) and get guidance over the phone about how to manage a case. This type of open communication strengthened trust and joint responsibility between health system levels and contributed to a healthy organisational culture conducive for implementation of perinatal audits.

All case studies demonstrated strong data use for decision making more generally with maternity-related statistics visible in the hospital and regular M&E meetings at subdistrict and district levels to inform health system planning. PPIP data use for decision making varied between the regions. Region 1 did not use the PPIP data for local decision making, whereas region 2 had a strong system of using the data and information from PPIP, as demonstrated by this quote:

The M&E - where we have all the role players together - it’s great because using the [PPIP] data then we can say “this is the issue with the transport from [city]. This is the issue with the CPAP (continuous positive airway pressure) at [case D] hospital” and immediately you can address more things. –Regional PPIP coordinator

The overall management was excellent, as observed and reported. The top-level managers (medical manager, clinical manager and matron) demonstrated strong managerial skills, which fed into good management of others on their team (other managers and operational ward managers). The managers interviewed were supportive and protective of their staff. Managers reported one-on-one meetings with those involved in a perinatal death prior to M&M meetings in order to demonstrate support and identify issues before the group meeting. By working alongside their staff, managers were visible and able to mentor staff, including in perinatal audit-related activities, such as how to correctly complete the PPIP forms and apply learning from the M&M meetings. The regional PPIP coordinators also had strong management skills and served as mentors to the staff in subdistricts, with an aim to grow champions to strengthen implementation.

I think it’s just lead by example, be open, be a good example why, and just care. Care for your patients, care for your staff. –PHC operational manager

It’s like a tree. So you start with the stem and a couple of branches and you’re adding leaves all the time. So - like the other day when I went to [district hospital] when [doctor] presented the PPIP data himself. It’s not that the tree is suddenly full of leaves, but it’s a slow process of adding people and getting them enthusiastic. –Regional PPIP coordinator

Service delivery lens: inputs needed for implementation

Our study validates the need for tangible system inputs, such as focal points and regular meetings,5 6 29–33 and shows that for sustainability, integrating these into routine practice and systems gives people the capability to sustain implementation. Organisational incentives (refreshments, per diems and continuous professional development points) did not appear to contribute to sustained participation in our study, though incentives have been identified in other settings.6 34

Training and supervision can also promote sustainability.6 7 9 22 35 This study shows that training was mostly informal and integrated, especially after initial introduction. The long history and scale of the Perinatal Education Programme in South Africa, which includes perinatal audit as part of the curriculum, may also have contributed to sustained practice,36 37 though it was not specifically identified or explored in this study. Low staff turnover and continuous supervision by the regional PPIP coordinators helped maintain skills and knowledge. As with other studies, participants believed they had the skills needed to fulfil their responsibilities related to perinatal audit.38 39 Nonetheless, few people had a full a grasp on all of the steps in the audit cycle and how they linked. A clear explanation of the components of MPDSR and a list of competencies required for implementation remains elusive in the global literature.6 40 The lack of a common understanding of MPDSR implementation, as reflected in global literature and the users of the intervention in this study,6 may impede our ability to demonstrate effectiveness and sustainability of the intervention process, which is not a problem unique to MPDSR.41

While no standard minimum requirements of human and material resources for MPDSR implementation have been identified,6 sufficient and well-managed human and material resources may contribute to sustainability.9 34 42 43 In our study, the belief that there were sufficient resources to respond to identified actions may have reflected the Western Cape Province’s rich and unique experience of health-system transformation and a relatively well-functioning overall health system.44 Future research may want to analyse budgets and expenditures relating to perinatal audit to validate these beliefs.

Societal lens: interactions between those involved

External influences can affect the perceived legitimacy of MPDSR.6 Our study finds that the expectation of reporting and engagement from the regional PPIP coordinators, along with other accountability mechanisms, that is, the ‘ideal hospital’, gave legitimacy to participate in perinatal audit, as found in other South African studies.21–23 The clear and intentional linkages to the routine information system in region 2 added another layer of accountability, as shown in studies from India45 and Malawi.46 The integration of perinatal audit into other processes embedded activities into the broader frame of clinical governance rather than as a stand-alone activity, further supporting the presumption that MPSDR should be implemented along with other clinical governance practices.6 7 47

The belief that the intervention achieves its desired outcome, also called ‘value proposition’, promotes sustainability.20 48 While many studies have shown positive outcomes from MPDSR,6 few have also linked this to buy-in and sustainability.6 20 49–52 Our study shows that seeing the benefits of engaging over time enabled people to buy in and become more committed to perinatal audit, ultimately improving implementation.6

Individual motivation to implement MPDSR is critical for sustained practice.8 26 Our study confirms what others have found regarding intrinsic motivation related to MPDSR,6 such as passion for maternity care and commitment to improving the quality of service delivery. Additionally, we learnt that users valued the opportunity to debrief as a team after difficult cases in a safe and trusted space. As for extrinsic motivation, people appreciated the opportunity to learn clinical skills through MPDSR.6 Individual motivation, buy-in to MPDSR and general commitment to their jobs and quality improvement are linked.21 51 53 Our study finds that clear tools, supportive supervision and continuous oversight from subnational actors improved individual confidence in implementation, and this aligns with the quality improvement literature.35

Sustainability is supported when people have a common understanding about an intervention; they value it, trust it, and use it to help build and nurture relationships.8 54 The nature and quality of teams, including the hierarchies, mentorship, teamwork, facilitation and management, all played a role in MPDSR implementation.6 Multidisciplinary team engagement is widely acknowledged as an enabler,6 22 55 but less studied are the small informal teams that are core to implementation.6 Our study confirms the importance of such teams for sustainability, especially when they operate in an environment with clear communication channels and mutual respect.21 56 The importance and value of investing and strengthening these informal teams require more attention for those seeking to strengthen MPDSR implementation.40

Systems lens: things that trigger change

Local adaption of an intervention is a core element of sustainability.39 57–59 The contextual and local adaptation of the MPDSR process has been well documented in South Africa and other LMIC settings,6 21 60 is promoted by the WHO,40 and aligns with broader quality improvement approaches.35 61 Our study shows variability in implementation processes between sites, and that subdistricts continuously tailored the process to the capacity, interests, and needs of the actors involved.

Implementation culture profoundly influences MPDSR and its sustainability,6 22 55 62 63 and multiple frameworks seek to support how to overcome the blame culture specifically.51 63 Our study validates elements that prevent blame in MPDSR, such as strong leadership, codes of conduct, participation, openness, professionalism and self-reflection.7 21 38 51 60 64 65 Strong leaders or ‘champions’ are a critical factor in MPDSR sustainability,6 20 21 and our study goes further to identify traits and motivations of these individuals (Box 2), which align with common aspects found in good leaders or managers.66 67 Strong, predictable and open communication systems, along with effective management, enable a work culture conducive to learning, innovation and accountability linked to perinatal aduit.8 39 61 The observed positive implementation culture of perinatal audit in this study took time to nurture and also was part of a wider effort to strengthen quality improvement, self-reflection and joint responsibility.23 44 68 The Western Cape Department of Health’s governance approach of collaboration, integration and multisectoral engagement may have influenced the implementation of the perinatal audit programme and enabled it to benefit from and contribute to the broader health system.44 68 69

By focusing on explanatory factors, our study provides a deeper understanding into how and why MPDSR routinely works in the Western Cape. While our study confirms findings from other studies in South Africa,21–23 65 the focus at the mesolevel and microlevel allowed for more understanding on how individuals and teams perceive perinatal audit implementation.

Application of the extended normalisation process theory

The adapted version of the extended normalisation process theory provided a structure to understand and explain an environment with sustained implementation.26 May argues that his theory is useful for understanding complex implementation processes in the ‘real-world’ environment where they are implemented. The MPDSR process is complex, with many parts or steps, different actors at multiple levels, and the interaction between people, teams and the health system.11 By applying a health policy and systems approach, we were able to unpack the contextual factors and underlying mechanisms that might render MPDSR to be sustained.15 35 70 Using theory, we explored issues such as trust, credibility and hierarchies shaped by the power relations between stakeholders even when the implementation process slightly varied between cases.

For the maternal and child health community, this study demonstrates the value of using theory as a means to understand complex implementation.12 15–19 Most studies in maternal and child health fall under the service delivery lens, measuring the tangible markers of an intervention.10 Our study confirms that factors enabling sustained practice of MPDSR require investments in the societal and systems lenses, or intangible elements of the health system, and this will require qualitative research approaches.6

Limitations

This study collected information on perinatal mortality audit, which is a sensitive topic, given the nature of exploring adverse incidents by reporting data on deaths as well as reviewing the situation surrounding the death. Participants may not have shared their actual understanding of the process or experience or may have changed their behaviour during the observed review meetings. Through individual interviews, this study included the perspectives of front-line health workers, subdistrict health management and regional actors involved in the PPIP process. District management staff were not available for interviews (scheduling conflicts), and not all of clinical staff and subdistrict managers were included. Data collection stopped at the end of March 2020 due to the COVID-19 pandemic and related restrictions. This unfortunately prevented further data collection, including observation of additional meetings, and timely validation meetings with the subdistricts.

To ensure rigour and trustworthiness, triangulation of the different data sources was used to verify and validate information including field notes, observations and follow-up interviews with specific people. There was possible interpretive bias of the lead researcher (MK) due to issues of reflexivity and specific interests. However, the interviews were conducted using a semistructured interview guide and data were analysed using an implementation theory and adapted analysis coding framework with review from all authors.