Neonatal nursing policy and practice in Kenya: Key stakeholders and their views on task-shifting as an intervention to improve care quality. [version 1; referees: awaiting peer review]

Improving the quality of facility based neonatal care is central to Background: tackling the burden of neonatal mortality in Low and Middle Income Countries (LMIC). Quality neonatal care is highly dependent on nursing care but a major challenge facing health systems in LMICs is human resource shortage. In Kenya, task-shifting among professional care cadres is being discussed as one potential strategy of addressing the human resource shortage, but little attention is being paid to the potential for task-shifting in the provision of in-patient sick newborn care. This study identified key neonatal policy-making and implementation stakeholders in Kenya and explored their perceptions of task-shifting in newborn units. The study was exploratory and descriptive, employing qualitative Methods: methods including: document review, stakeholder analysis, observation of policy review process meetings and stakeholder feedback. A framework approach was used for analysis.


Introduction
Neonatal mortality currently accounts for over 40% of all child mortality in many countries in sub-Saharan Africa 1 . Reducing neonatal mortality is a global priority 2 and improving access to quality care is central to these efforts 3 . As access to facility based health care for maternity and neonatal services is slowly improving, particularly in urban areas (see UNICEF page on maternal and newborn health) weaknesses in facility based healthcare delivery are emerging as a major factor contributing to the neonatal mortality burden [4][5][6] .
Improving the quality of care for neonates, and specifically for sick newborns, involves particular challenges as this group often have multiple morbidity and require multiple interventions, given repetitively often over many days. In addition to carefully planned medical care, providing quality care to sick newborns is highly dependent on the availability and quality of nursing care. In countries such as the UK it is recommended that, even for babies who do not require intensive care, there should be 1 nurse for every 2 to 4 sick babies 7,8 with evidence suggesting higher mortality where such standards are not met 7 . Providing such levels of nursing care is a major challenge in low-income settings where there are considerable deficits in human resources for health 9 . Many countries in sub-Saharan Africa fail to reach the World Health Organization (WHO) recommended minimum ratio of 2.5 health workers per 1,000 population 10 . In Kenya, a recent study examined services in 22 large county hospitals and found a median ratio of inpatient children to nurses on paediatric wards of 11:1, a ratio often higher at night 11 . Comparable data were not available for newborn units but a complete absence of qualified nursing staff in some facilities has been noted in prior reports 12 .
A strategy suggested for addressing health workforce challenges in low income settings is task-shifting, defined by the WHO as: "the rational redistribution of tasks among health workforce teams", wherein "specific tasks are moved, where appropriate, from highly qualified health workers to health workers with shorter training and fewer qualifications in order to make more efficient use of the available human resources for health" 13 . That task-shifting may be effective is supported by work on HIV 14 , on non-physician clinicians 15 and a recent systematic review on community based care 16 . In many sub-Saharan African countries, there is often considerable informal sharing of tasks between existing professional cadres. For example, nurses perform diagnostic tests and prescribe medicines and non-physician clinicians perform surgery 17 . In Kenya, such forms of taskshifting are the subject of policy discussions intended to update the legal and regulatory framework guiding professional roles for health care providers 18 . However, there is as yet no discussion of shifting tasks to non-professionals who work alongside health professionals within hospitals. This study aimed to explore the potential of task-shifting as a policy option to address constraints in the provision of sick newborn care in Kenya. Specifically, the objectives were to: describe neonatal nursing care policies; identify key neonatal nursing policy development and implementation stakeholders; and explore stakeholder perceptions of the potential role of task-shifting in improving the quality of neonatal nursing care.

Methods
The study was exploratory and descriptive employing a variety of qualitative data collection methods including: document review, stakeholder analysis, observation of policy review process meetings and stakeholder feedback.

Data collection
Document review and identification of key stakeholders A review of published and unpublished Kenyan national documents relating to neonatal health was undertaken to: i) ascertain the existence and content of policies relating to the provision of inpatient nursing care for sick newborns; and ii) to identify the key stakeholders involved in inpatient neonatal care policy development. The documents were identified through an internet search (Google scholar), and in discussions with experts in the field in Kenya (nurse educators, policy makers and managers) beginning with those known to researchers in the KEMRI-Wellcome Trust Research Programme (KWTRP) as a result of recent nursing workforce studies 11 . These documents and contacts helped identify initial important stakeholders with additional stakeholders identified through snowball sampling 19 . Stakeholder sampling was further guided by a framework used previously in policy analysis 20 that helps categorize stakeholders according to their main roles in policy development and implementation. The categories included: i) statutory policy making/strategic endorsement; ii) technical advice; iii) evidence generation; and iv) consultative.

Stakeholder interviews
Stakeholders were contacted, either via phone or email, requesting an interview. The interviews followed a semi-structured, open ended format (see Supplementary File 1) designed to allow for discussion of: the factors influencing the initiation of policy change; the process of policy making; the actors involved in policy making and implementation along with their roles, responsibilities and relative influence; and views on taskshifting. DO conducted the interviews which were digitally recorded and subsequently transcribed by a transcription service and checked for accuracy. During the process of transcription all participants were anonymized by assigning interviewee codes. The audio recordings and transcripts were stored in a password protected computers.

Non-participant observation of the policy review process
The first author (DO) and two other researchers were observers in the process to develop a task-shifting policy for health care services in Kenya, primarily aimed at updating professional schemes of service among existing formally defined cadres of health worker 18 . Of five Technical Working Groups (TWG) supporting this policy process DO attended and observed the Legal and Regulatory committee meetings that was reviewing policy documents that promote or hinder task-shifting in Kenya. Two authors (JN and DG) observed the Service Delivery committee that was discussing tasks that are/could be shared across existing cadres of staff within the Kenyan health system.

Stakeholder feedback
Building on earlier nursing workforce studies 11 and informed by the stakeholder interview data, the KWTRP has convened an Expert Advisory Group (EAG) to guide current work on neonatal nursing and quality of care (Supplementary File 2). Draft results of the policy context and stakeholder analysis were presented to the group; suggested modifications were noted and incorporated into the final results presented in this paper.

Data analysis
A framework approach 20 linked to the objectives was used to analyze the stakeholder interviews. To determine which stakeholders were likely to be 'essential', 'important', and or/'necessary' 21 to involve in exploring how task-shifting might create a new cadre of staff to support inpatient neonatal care, the stakeholders were asked who they thought were the key players in nursing policy development and implementation. Stakeholder policy making and implementation power grids were developed based on an analysis of the number of times an actor or organization was mentioned in the interviews combined with the respondents' perceptions of who they thought was the most influential. Drafts were shared with the EAG to confirm the position of the actors.

Results
Neonatal & nursing care policies/guidelines Twelve documents were identified and reviewed (Table 1). Of these, seven were national strategy documents, three clinical care guideline documents and two were concerned with roles and responsibilities of nursing staff. Three national strategy documents identify averting neonatal deaths and improving No formal Kenyan policy document stating the nurse to patient ratio required in neonatal care or other higher dependency areas or the qualifications of such staff was identified.
Nursing policy development A total of 19 stakeholders were interviewed from 8 organizations ( Table 2). The main roles of each organization in policy development are illustrated in Figure 1. The Ministry of Health (MoH) and the Nursing Council of Kenya (NCK) were confirmed by the interview participants and EAG members as holding formal positions in statutory policy making with technical advice being provided by national and international academic institutions and international agencies; evidence provided by national research institutions and nationally run projects; in consultation with national professional associations, civil society and faith-based and private health providers. In addition to providing technical support, the international bodies provide financial support to health projects and interventions in Kenya.
The stakeholder power-grid ( Figure 2) demonstrates the relative influence and importance of each stakeholder in policy making. Stakeholders reported that while the MoH and NCK are  both highly involved and have high influence, international debate, supported by a technical donor partner voice, was the most frequent stimulus for instigating nursing policy change.

"first of all the UN agency are very key like UNICEF and WHO. Then we also have the USAID and then we also have the bilateral agencies, DANIDA has been very active in terms of working directly, but most of the others may not work directly with us, and then training institutions, professional bodies like KPA, nursing council and then the regulatory bodies" (Policy stakeholder, PSH005)
Once discussions around policy formulation have begun, the MoH takes the lead role, relying heavily on technical and consultative partners for advice and funding. The first step typically involves the leaders of the appropriate departments within the MoH convening one or more TWGs with participation from the stakeholders defined in Figure 1. The role of the TWGs, as described by the stakeholders and observed during TWG participation, is to examine existing evidence and solicit local knowledge to help develop a draft policy document, standard or guideline. Drafts are subsequently sent to the designated Policy Advisory Committee within the MoH for ratification.
Although represented by officials from the National Nurses Association of Kenya (NNAK), frontline nurses appear to have little influence on the process or outcome of policy making. Even though there are a number of training institutions in the country, only two were mentioned as being involved in policy formulation. The important role of faith based organizations and the private sector in provision and running of health facilities was recognized by individual stakeholders and the EAG but these groups were rarely mentioned in discussions on inpatient neonatal nursing care policy formulation.

Nursing policy implementation
Once a policy has been promulgated at the national level, it is cascaded to the County governments who incorporate it into their strategic plans and, as the focus moves from policy formulation to policy implementation, there is a shift in the relative importance of several key stakeholders (Figure 3). Two of the  groups (front line nurses and County governments) with low influence and low involvement in policy formulation join the NNAK as having high influence and involvement in policy implementation. The County governments, the NNAK and front line nurses have the potential to influence policy implementation in a variety of ways. Following devolution in Kenya the responsibility for implementing national health polices was devolved from National to County government. In theory, the County government receives policy advice from the national level (MoH) which they then adopt/adapt for implementation. Almost all resource allocation (budgeting and staffing) necessary for the running of health services is now undertaken by the County governments 33 . County health officials also provide continuous supportive supervision to front line service providers working in the health facilities. Those at county level thus have considerable power to either promote or reject national policy.
"We domesticate health policies from the national level, and oversee their implementation in the county, and basically in coordination there are a lot of things, support supervision and ehh of course rationalizing placement of personnel within the county and also ensuring that the facilities are stocked with health commodities" (County official PSH001) The NNAK is professional body that represents the welfare and practice of its members. As a registered national association it has the capacity to adopt a stance at odds with government policy if such policy appears not to be in its members' interests; playing an important role in setting professional norms that may prevent or promote policy adoption. In extreme cases, it may call for strikes especially in cases where the members' needs are not met or when the members are dissatisfied with work conditions, terms and policies.
Front line service providers, facility managers and mid-level managers, while playing little or no role in policy formulation, play the most important role when it comes to policy implementation. Clinicians and nurses provide care to the sick newborns. Their practice is guided by policy briefs, directives and codes of conduct formulated by the MOH and by the NCK. However, if these frontline providers of health care are not adequately trained or briefed on changes, if they disagree or are uncomfortable with what they are being asked to do, or if they are not provided with the means to undertake the required tasks, then implementation can be hampered.

Perspectives on task-shifting
The stakeholders were asked for their views on task-shifting as a potential strategy for addressing health workforce challenges in inpatient neonatal nursing care. The majority (16/19) supported the shifting of nursing tasks requiring low skill levels to lower cadre staff. Fourteen of the nineteen key stakeholders interviewed fell into the high influence/high involvement quadrants for either policy formulation or implementation. Of these high influence stakeholders, all except one supported the idea of a task-shifting strategy. The support for task-shifting as a solution to the human resources shortage appeared to be born out of necessity linked to the strains on staff caused by inadequate existing human resources and reluctant acceptance that substantially increasing the recruitment of professional nursing cadres is a major financial challenge for the government.
"the reality is that nurses are probably either over worked because of the numbers that they have in the ward and all that they have to do, there are probably either two nurses who are caring for up to twenty or thirty newborns." (Policy stakeholder, PSH002) While supporting task-shifting as a pragmatic strategy, all participants emphasized the vulnerability and sensitivity of sick newborns, and were clear that shifting of tasks should not encroach on the provision of skilled clinical care. Tasks mentioned as having the potential for shifting included bathing ('top tailing'), feeding, milk preparation, changing and sorting of linen. Shifting these tasks would allow the nurses time to concentrate on providing more knowledge intensive clinical nursing care, especially to the sickest newborns.
Respondents acknowledged the existence of some level of "informal task-shifting" already happening in public health facilities due to the shortage of qualified staff.
"Task-shifting is already taking place but in haphazard manner. There is nothing guiding it" (NCK official PSH008).
Even among those who supported task-shifting, concerns were expressed about the current informal way in which it was happening, with lapses in supervision and regulation providing leeway for some non-clinical staff to take on roles beyond their scope and mandate, with rising incidence of harm to patients. An example was given of a recent highly publicized case of malpractice in western Kenya to support such concerns. In this case errors in a procedure (injection) given by unqualified staff resulted in injuries to children affecting their mobility. Such examples were used by participants to emphasize that, if task-shifting were to happen, then at least it should be formalized in terms of training, a scheme of service, supervision and regulation.
A small minority of stakeholders (3/19) were strongly opposed to any task-shifting in newborn care, citing concerns about safety and levels of competency of "unskilled/untrained" low cadre staff. One was in the high influence/high involvement quadrant for policy implementation while two were from training institutions. These three stakeholders explained that nurses are recognized and regulated by the NCK while "lower cadre staff", such as nurse aids or patient attendants are not; linking the lack of regulation to poor quality care. Furthermore, their views appeared to be influenced by previous experiences of a lower cadre of nursing staff, abolished more than 10 years ago due to concerns about their providing health care services they were not qualified to deliver: "No I don't think we have a place for nurse aids, it is like taking house helps from the houses and bringing them to the hospitals, please no, they are not even allowed. They are not even recognized by the nursing body. Anybody using nurse aids you are using it at your own peril. Who are they? We used to have them, but they were causing more damage, harm to patients than good" (Training stakeholder, TSH004)

Discussion
In many low-income settings a number of health system bottlenecks exist, preventing the scale-up of essential interventions that are key to reducing neonatal morbidity and mortality 34 . Key among these is tackling the deficit in human resources for health 35 . Kenya has high-level policy goals for reducing neonatal mortality but few nurses in the country have special training in neonatal care. Even if more skilled nurses were to be trained, there is no guarantee that they would be employed in neonatal units as, paradoxically, the supply of generally skilled nurses is not the primary health workforce problem in Kenya, rather it is the ability (and perhaps the willingness) to finance a major expansion in the professional healthcare workforce overall that is limiting 36 .
Task-shifting has been successful in expanding access to specific forms of care, but concerns have also been raised. A recent review of health workers' experiences of task-shifting in sub-Saharan Africa found that the strategy had the potential to negatively impact health workers' sense of agency and ability to perform their work 37 . Furthermore, introducing task-shifting to complex, multi-professional, facility environments such as hospitals raises different challenges to those experience in deploying community health workers 37 . Success in the latter is felt to be more likely if task-shifting approaches are based on the values, preferences, knowledge and skills of all stakeholders, and on the feasibility and applicability of the intervention for particular settings and healthcare systems 16 .
In this study, among those stakeholders identified as playing a key role in neonatal nursing policy formulation there was broad acceptance of the concept of task-shifting. This acceptance was largely born of resignation that a major expansion of an increasingly professionalized nursing workforce was unlikely to be realized, and that there is an imperative for action to improve newborn survival. However, several respondents urged considerable caution, highlighting past and recent episodes of malpractice in which lower cadre staff had overstepped their roles resulting in patient harm. Negative past experiences of an intervention can slow the process of policy change 38 and it was clear that, particularly among stakeholders who would be involved in the implementation of a neonatal nursing taskshifting policy, previous negative experiences would be likely to hamper the acceptance of the introduction of a lower cadre of staff. These concerns echo the challenges identified in a recent review of health-worker task-shifting where the relinquishing of tasks within more traditional, facility based health care settings was found to include a creeping expansion of roles taken on by less well qualified personnel, unclear accountability mechanisms and tension between expanding the quantity of service provision and maintaining quality 39 .
Changing health policy is recognized as a complex and context specific process 39-41 . In Kenya, moving beyond the acceptance, in principal, of task-shifting from professional nurses to lowerlevel cadres of worker as a solution to the shortage of nurses available to provide care to sick newborns will require careful navigation in a complex policy and implementation environment with different stakeholders important in different phases of this process. It is also important to learn lessons from prior efforts to develop task-shifting solutions; using a participatory process to ensure that key stakeholders are involved in characterizing the problems and designing potential task-shifting solutions to help address these challenges. In this study, important insights were gained that could inform the design of such a workforce solution in Kenya so that it addresses legitimate concerns (such as scope of practice) and the particular context of neonatal care where patients are both highly dependent and highly vulnerable (including establishing clear lines of accountability and related supervisory arrangements).

Study Limitations
While every effort was made to include prominent policymaking and implementation stakeholders some stakeholders may have been missed. Assigning institutional actors specific capacities (eg. level of policy influence) is potentially an oversimplification of the varied ways in which institutions, and individuals within them, may affect both policy making and policy implementation. However, the form of stakeholder analysis we employed is widely used to help provide a general framework to guide understanding of, and engagement with, what can be complex networks of actors and their roles, to ensure that their legitimate interests and concerns are addressed 42-44 .

Conclusion
There was broad acceptance among key nursing policy makers in Kenya of the idea that addressing the deficit in neonatal nursing care may require some form of task-shifting. However, concerns about task-shifting were raised, particularly among key stakeholders involved in policy implementation. Any taskshifting strategy will need to be undertaken with considerable caution working in collaboration with key policy making and implementing stakeholders to navigate the complex policy and implementation environment with different stakeholders important in different phases of this process.

Ethics approval and consent to participate
Ethical approval was granted by the Kenya Medical Research Institute, Scientific and Ethics Review Unit (SSC No. 2897).
All study participants signed a written informed consent form before participating in the study.

Data availability
The data that support the findings are not publicly available due to restrictions. Public availability of data could potentially compromise participant privacy. Participants did not consent to have their full transcripts or excerpts of transcripts made publically available.