Engaging stakeholders in embedded newborn health services/systems research in Kenya : continuous process involving multiple actors

The HIGH-Q Programme 2020–2025

The HIGH-Q Programme, implemented in Kenya, is a partnership between the Kenya Medical Research Institute (KEMRI)-Wellcome Trust Research Programme, the University of Oxford, the Kenya Paediatric Research Consortium, the London School of Hygiene & Tropical Medicine and the County Departments of Health. Running from October 2020 to September 2025, the HIGH-Q Programme of work comprises a co-designed, multi-disciplinary workforce intervention and evaluation aimed at improving the quality of care in select neonatal units in Kenya. The workforce intervention was built around and concurrent with a multi-country technology intervention programme known as Newborn Essential Solutions and Technologies (NEST360).¹⁸ The NEST360 technology intervention is implemented in a large subset of a network of county hospitals known as the Clinical Information Network (CIN) (See Box 1 for a summary of NEST360 and CIN); the HIGH-Q programme involved a smaller subset of these NEST360 sites.

The HIGH-Q Programme has a specific focus on neonatal care in Kenyan hospitals. The Programme is assessing how introducing additional workforce alongside the NEST360 technologies in Kenyan hospitals affects technology adoption and quality of care in light of known and substantial workforce deficits.¹⁴ The Programme’s rationale is based on the assumption that adding technologies by themselves will improve quality and safety and that the available staff can successfully implement new technologies. However, this might not be the case; hence, the research focuses on an improvement in staffing levels to assess more successful use of technologies to improve care. Completed in July 2023, the specific workforce intervention aimed to increase the number of nursing and ward assistants in newborn units (NBUs).¹⁹ Specifically, the intervention involved the employment of three additional nurses and three ward assistants in each of the four-county hospital NBUs for 15 and 7 months, respectively. Sick newborns in NBUs require continuous monitoring and treatment, but low nurse-to-patient ratios compromise care and have been linked to an increase in mortality. While adding more skilled nurses could improve care quality by reducing missed tasks and strengthening teamwork, it remains a challenge due to underfunded health systems. Nurses also spend time on lower-skilled duties, such as cleaning equipment, changing linens and diapers, and assisting mothers with breastfeeding, tasks that could be delegated to ward assistants.

Salaries for the additional intervention staff were paid by the HIGH-Q Programme for the research period, but the staff were employed and line-managed by the hospitals. The other HIGH-Q objectives examine health workers and mothers’ experiences of newborn care, how to better support the delivery and integration of post-discharge care for families, the governance process of introducing technologies and service delivery innovations and infection prevention and control in these NBUs. Primary data collection includes structured and unstructured observations of care and technology use at NBUs and interviews with healthcare professionals (at all levels of the health system) and with the carers of sick babies. Table 1 summarises HIGH-Q Programme work packages, research questions and methods. The Workforce intervention was assessed under work packages 1a and 1b.

In Kenya, health care is devolved from the national government to 47 semi-autonomous sub-national governments (commonly known as counties).²⁰ Under this governance arrangement, county governments have an executive, administrative, and legislative mandate over their areas of jurisdiction. From the early study planning stages for HIGH-Q, it was recognised that stakeholder engagement was needed at national, county, hospital and newborn unit levels, and should build upon and consider wider CIN and NEST360 activities, as well as any other research that might be ongoing at these sites during this period.

Synthesis and Interpretation of Learnings

To support our reflections on the impact of the Programme’s stakeholder engagement activities, we have applied ideas from Kujala, Sachs, Leinonen, Heikkinen & Laude (2022)’s conceptual framework, which proposes engagement activities to consist of moral, strategic, and pragmatic components.²² Their framework is a response to what is observed by the paper as a fragmented approach to analysing stakeholder engagement, which hampers research progress. The moral component involves considering the ethical dimensions of how the operations of an organisation affect individuals and groups rather than focusing on organisational interests. In the context of our research, we consider, for example, how to engage particularly marginalised stakeholders, how to ensure respectful interactions with all stakeholders, and how to ensure the research is made meaningful to diverse stakeholders. The strategic component is concerned with how managers distribute scarce resources among stakeholders in ways that guarantee value creation for the organisation and the achievement of its objectives. In the context of our research, for example, we recognise that we have deliverables to ourselves and our funders with regard to the quality of research and our contribution to knowledge. The pragmatic component could be seen as working across the moral and strategic components. Drawing on Chinyio and Akintoye (2008),²³ and Forsythe, Ellis, Edmundson et al. (2015),⁹ we considered these as pragmatic activities to build and sustain connections, including employing negotiations, making trade-offs and adjusting to the practical requirements of stakeholders.

Under this approach to reflection, stakeholder engagement activities in the Programme were assessed based on meeting the three goals of engagement: the moral, the strategic and the pragmatic components by analysing the elements of the of Programme specific activities against the three components.

Developing an initial strategic plan to guide stakeholders’ engagement

During the proposal development and the programme’s setup phases, the investigators formulated an initial framework for engaging stakeholders. This framework encompasses engagement objectives, detailed stakeholder mapping, and the nature and content of engagement activities ( Figure 1). The overall interrelated goals of stakeholder engagement were to contribute to 1) good quality science through inputs into study design, implementation and outputs; 2) ensuring ethical practice throughout the research process; and 3) maximising the uptake of research findings in policy and practice.

The engagement framework covered a full spectrum of activities. At one end of the spectrum were deep engagements involving fewer people in more interactive ways. Activities include engaging stakeholders in consultative meetings to co-design elements of the research and get advice on the stakeholders’ engagement plan itself, as well as on how to conduct the research. At the other end of the spectrum, there are wider engagement activities aimed at reaching many people through inevitably less interactive activities. Examples include radio programmes, video outputs, newspaper articles, and policy briefs.

At the heart of the engagement activities, and to advise on all our undertakings, we set up an international and a national advisory expert group with diverse expertise. The international expert group includes seven stakeholders with global health research experience and content expertise in medical sociology, health economics, organisational analysis, health workforce and systems, infectious diseases epidemiology, regional nursing policy and advocacy. The national expert group consists of eight individuals with representation from the national Ministry of Health, medical schools, medical and nursing professional communities, professional association regulators, and patient representative groups. These groups were consulted together at least annually, supplemented by periodic interactions with members as needed. The expert group contributed towards the study design. During the early stages of the study, the research team received valuable input from the members of the expert group that helped with research approaches and execution. An example of key advice on the sequence of additional nurses and ward assistants’ intervention is highlighted below:

Implementing stakeholder engagement activities across a complex and expanding range of stakeholders

The process of identifying stakeholders using the engagement matrix yielded multiple stakeholders who included the national and county level policymakers, medical and nursing professional communities, professional association regulators, health managers at various levels of the health system, frontline healthcare workers in hospitals, patients, family members and patients’ representative groups. These stakeholders, especially those close to the research, were engaged throughout the HIGH-Q research process based on our understanding of their needs, interests, and concerns ( Figure 2). Drawing on the International Association for Public Participation distinctions described above, we informed all stakeholders by providing study details and our plans to help them comprehend the research issues and support further engagement activities. We also involved representatives of all stakeholders (including research participants) to ensure their concerns were understood and considered in research activities and study learning. We consulted with representatives of many stakeholder groups to gather input regarding our plans, including through our advisory groups, and had formal collaborations with stakeholder groups drawn from the counties and hospitals. We sought to ensure that the HIGH-Q research was as responsive as possible to stakeholders’ needs and priorities, and we would not have been able to proceed if gatekeepers such as regulators and county and hospital managers had not approved the study. Nevertheless, the final decision-making on the study details remained with the study team. In these ways, we were continuously balancing ethical, strategic, and pragmatic elements of stakeholder engagement. At the workforce intervention level, the recruitment approach followed county procedures as described in a subsequent section. At the NBU level, nurse managers were fully empowered to manage duties and performance of additional nursing staff.

Figure 2. Presentation of multiple and layered stakeholders engaged, showing differing interests.

As presented in Figure 3, there were three main observations from this work. First, the bulk and the backbone of stakeholder engagement focused on four intervention hospitals. We were aware that engaging the leadership and health workers is critically important. Our initial engagement plan targeted the county directors of health (DoHs) as the first point of entry, followed by engagements at the county hospitals and down the hierarchy to the NBUs level. However, the workforce component significantly expanded the complexity of our engagement processes, as outlined in Figure 4.

Figure 3. A summary of the HIGH-Q Stakeholder engagement against Programme’s milestones.

Figure 4. A presentation of the County and Hospital structures and the evolving engagements experienced by the HIGH-Q Programme.

Our second observation is that we had to respond to field questions and realities with many informal or unplanned follow-up engagements or activities in response to issues raised by the study teams or stakeholders. Examples include:

We observed that while some of these issues could be considered primarily practical, strategic, or moral, there were often interactions and interplays between the components. For example, having too many research staff at a ward had practical implications (difficult to collect data) and moral implications (disruption to routine ward services offered by overwhelmed staff caring for vulnerable children). Our third observation was that some stakeholder groups were more challenging to engage in terms of reaching them and achieving sustained involvement than others. For instance, our engagement with parents was primarily through interviews and observations, preceded by consent and informal interactions. Linked to research governance, our study was subject to review and approval by the National Commission for Science, Technology, and Innovation (NACOSTI), and our international researchers were required to apply for and obtain research permits from the Immigration Department. Still, on research governance, we had more interactions with nursing regulators/professional bodies such as the Nursing Council of Kenya (NCK) and the National Nursing Association of Kenya (NNAK). Our engagement with these regulatory institutions involved following all requirements at international, national, and local levels for initial approval and, where necessary, annual reports and renewals are required. At the data collection level, we reached out to a range of regulators of health research and technology to conduct interviews, with varying success.

As part of connecting to an even wider audience, we plan to engage with teaching institutions and the media as part of the next phase of stakeholders’ engagement.

Different approaches and outcomes between counties with similar architecture.

In the county governance establishment, there were several important officeholders to consider in any hospital-based engagements (See Box 2 on County Governments and the Departments of Health structure relevant to the study, adapted from the County Government Act, 2012²⁴ and the Transition to Devolved Government Act, 2012²⁵). In the HIGH-Q Programme, the workforce intervention triggered a more complex set of engagements than is typically required for more observational or descriptive research. Figure 4 illustrates our original linear simple partnership plans in the centre, represented by the broken yellow line arrow. At the engagement stage, this approach morphed into many referral contacts, with many people to talk to and get permission from and eventually led to the conversion of simple agreements into complex memoranda of understanding (MoU). To get permission and approvals, we engaged with the chief officers, the county executive committee member, the county public service board, the county legal offices, the county secretary, and the governors. These engagements differed across counties, leading to differences in the way intervention staff were recruited, despite the similarity in county hierarchical and administrative architecture.

Other notable differences were in the frequency and duration of research approvals and for the other research-related milestones in the four counties. Table 2 below presents quantitative data on the frequency and the duration of engagements with different county officials in the pre-intervention (PRE-I) period, comprising 7 months, and during the intervention (DUR- I), comprising 15 months. The pre-intervention period was before the MoUs signing and the recruitment and deployment of additional staff. The intervention period consisted of deploying staff working in the units, and the data collection period. The number of engagements in this period with county and hospital-level senior office holders varied across the counties. The combination of methods employed for engagement consisted of face-to-face, phone calls, email correspondence, and letters. A majority of county and hospital executives were engaged directly and at varying durations (presented numerically in brackets and consisting of weeks of engagement from initial contact to sharing of findings) and frequencies (also presented numerically). Those closer to the intervention (the medical superintendents) were engaged multiple times, while those further from the intervention (the governors and the CECMs) engaged less often.

The differences in engagement pathways between counties were also reflected in the execution of MoUs and recruitment of staff ( Table 3). The pathways differences were because of proactive incorporation of county stakeholders’ input into the engagement process and consultations on how best to implement the study as collaborators. For example, as presented in Figure 4, the Programme agreement proposals were expanded and converted to MoUs and included additional clauses and signatories. We assessed MoUs attributes according to the number of signatories, the number of articles, and the duration in weeks from their initiation to full execution. We see that County A added an extra clause on recruitment by the County Public Service Board (CPSB), and County B was directly coordinated by the county director of health (DoH), who worked closely with the County’s legal team to develop and sign a complex MoU with multiple clauses. The process at County B took double the time of Counties A and C. The extra layer of engagement process involving the CPSB was an uncharted route before and presented tension and uncertainty with budgetary implications for the advertisement of nurses’ jobs in a newspaper with national circulation. However, these concerns were well mitigated with an efficient internal consultative process, strategy, and open engagement with the Board. Still under recruitment, County B deployed intervention staff in other departments and utilised its own staff for the intervention. Though this was a noble initiative for sustainability, the Programme had to carefully relay this feedback to the newly recruited staff whose contract indicated they would work in NBUs and would be trained before deployment.

Our experiences underscored the importance of being able to be flexible and responsive in our engagements and in the research plans. Notably, the entire process above was also complicated by routine job (re) postings and by national general elections. At the governor level, three of four governors and all four executive committee members (CECMs) at the Department of Health in the four counties vacated office after the national general elections, and in one of the counties, there were 3 different CECMs during the engagement period. All four chief officers at the departments of health (COHs) left after the elections. During the engagement period, all hospitals registered a transition in the office of the medical superintendents. The first transition of medical superintendents happened only one month after the commencement of engagement at County C, and the last a month before the end of the engagement period. These transitions of various officeholders necessitated renewed engagements, especially where the new officeholders were closely involved in the implementation, such as medical superintendents. In Counties A&B, the transition for the position of medical superintendent happened even before the MOUs had been signed.

Engagement for research feedback and study withdrawal

As illustrated in Figure 3, we undertook two cycles of research feedback at the four county hospitals during the engagement period, with further feedback and discussion of findings planned over the next 6–12 months.

The first research feedback meeting took place between November and December 2022, six months into the nursing workforce intervention and just before the introduction of ward assistants. The objective of this feedback was to give progress updates and preliminary research findings from HIGH-Q’s baseline survey research. The findings were anonymised across all four sites. The target audience included the medical superintendents, paediatricians, nursing leaders, and various departmental staff. The feedback sessions provided opportunities for hospitals to understand the HIGH-Q’S qualitative work and the consent process, and to promote discussion on areas of care the hospitals were eager to improve on. Stakeholders were keen to have hospital-specific data presentations in the future as a reaction to data anonymisation approach adopted for this initial round of feedback.

The second set of feedback occurred between December 2023 and February 2024, a year later, after completion of all the fieldwork, and following further data analysis. Before this round of research feedback, the research team spent considerable time carefully preparing for the feedback and crafting appropriate messages to hospitals. Hospital-specific data was provided together with comparative anonymised data from other hospitals. Sensitive content was provided feedback in careful ways. The NBU hospital teams were interested in discussing the clinical data, but were also concerned about the withdrawal of intervention staff once the study was over.

In response to the NBU team’s request to support with advocacy for the retention of intervention staff, and as part of the withdrawal plan from hospitals, we conducted an additional extensive set of engagements with county stakeholders at the departments of health in all four counties. We corresponded with all county and hospital executives previously involved in the study approval process - county secretary (County A), CECMS (all counties) CoH (Counties A&C) DoHs (all counties), and medical superintendents (all counties). Our engagements involved a brief highlight of existing partnership and contractual arrangements for the HIGH-Q programme, an update on the status of workforce enhancement and the short-term employment of intervention staff, an assessment of the contribution of the additional workforce to the quality of newborn care at the NBUs, and finally, exploring opportunities to retain and absorb the project nurses. These efforts culminated in a brief face-to-face meeting with three of the four CECMs at a CECM forum. Our interest in having continuity of the nursing workforce conflicted with the position of CECMs. The feedback from the CECMs cited county human resources practices which promote a competitive recruitment process, leaving no room for the direct absorption of staff from the Programme into the county public service. Options for inviting the HIGH-Q Programme nurses to be part of recruitment once public funding for staffing at the county was discussed, with promises from the CECMS that staff would be offered opportunities through an open recruitment process. These discussions, however, took place at a time of a major nationwide problem with recruiting new health staff linked to national budget caps with direct implications for the County governments. The Public Finance Management (National Government) Regulations, 2015 (Kenya)²⁶ imposes a limit of 35 per cent expenditure on county staff salaries and benefits against total county revenues. According to the latest report available, on average, the County governments have exceeded this limit.²⁷