An exploration of the political economy dynamics shaping health worker incentives in three districts in Sierra Leone

The need for evidence-based practice calls for research focussing not only on the effectiveness of in- terventions and their translation into policies, but also on implementation processes and the factors in ﬂ uencing them, in particular for complex health system policies. In this paper, we use the lens of one of the health system's ‘ building blocks ’ , human resources for health (HRH), to examine the implementation of of ﬁ cial policies on HRH incentives and the emergence of informal practices in three districts of Sierra Leone. Our mixed-methods research draws mostly from 18 key informant interviews at district level. Data are organised using a political economy framework which focuses on the dynamic interactions between structure (context, historical legacies, institutions) and agency (actors, agendas, power relations) to show how these elements affect the HRH incentive practices in each district. It appears that the of ﬁ cial policies are re-shaped both by implementation challenges and by informal practices emerging at local level as the result of the district-level dynamics and negotiations between District Health Management Teams (DHMTs) and nongovernmental organisations (NGOs). Emerging informal practices take the form of selective supervision, salary supplementations and per diems paid to health workers, and aim to ensure a better ﬁ t between the actors' agendas and the incentive package. Importantly, the negotiations which shape such practices are characterised by a substantial asymmetry of power between DHMTs and NGOs. In conclusion, our ﬁ ndings reveal the in ﬂ uence of NGOs on the HRH incentive package and highlight the need to empower DHMTs to limit the discrepancy between policies de ﬁ ned at central level and practices in the districts, and to reduce inequalities in health worker remuneration across districts. For Sierra Leone, these ﬁ ndings are now more relevant than ever as new players enter the stage at district level, as part of the Ebola response and post-Ebola reconstruction. © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
In recent years, there has been increasing attention paid to the need for evidence-based practice to improve health outcomes worldwide (Pang et al., 2003). Research has focused on identifying which policies work, but has also explored the processes by which knowledge is translated to highlight potential bottlenecks for evidence-based policy (Oliver et al., 2014). While a growing literature exists to explore the use of evidence in policy-making, there is limited knowledge on how policies can be successfully translated into effective practices. However, several studies (Chaudoir et al., 2013;Durlak and DuPre, 2008) confirm that implementation does influence the outcomes of an intervention and highlight the importance of understanding which factors affect implementation by looking at elements both in the context outside the organisation of focus and within the cultural and management features of the organisation. The importance of filling the knowledge gap seems even more relevant for complex health system interventions, where the wider context can play a major role in influencing the outcome of the policies. It is therefore essential to look beyond policy-making to reflect on actual practices, and on how, by whom, and why policies are potentially reshaped in the translation process.
In this paper, we aim to analyse how features relating to the context (structure) and the actors (agency) in three districts in Sierra Leone influence the implementation of health workers' incentive policies and define HRH practices at local level. Our research question focuses on if and how the local political economy features and dynamics, and in particular the interactions between District Health Management Teams (DHMTs) and nongovernmental organisations (NGOs), may have effects which contribute to shape incentives for health workers in public facilities and, thus, the functioning of the local health systems. Our focus is not on a specific intervention, but broadly on HRH incentives, including the official policies in place to regulate the incentive package for public health workers, as well as the actual practices that influence the financial and non-financial incentives effectively available for those health workers. We believe that HRH incentive issues make a useful case study to reflect on how deeply structural and agency features can influence local-level practices in a key area, such as HRH. In order to analyse the translation process at local level, we adopt a political economy framework. The framework allows us to explore the policy implementation, going beyond a static view of one organisation (usually the DHMT), to look at the dynamics between the layers of the structural context and the multiple actors and organisations that shape practices and define the incentive package differently in each district.
This research was conducted just before the Ebola virus epidemic started in Sierra Leone, in May 2014. Our findings highlight some of the factors that may have played a role in the collapse of the health system, as we point to in the concluding section. Moreover, our research contributes to the reflection on the consequences of the changing local dynamics as new players enter the stage at district level as part of the Ebola response and post-Ebola reconstruction, of which HRH incentive practices are an essential component.

Context
Sierra Leone is a West African country of 6 million people, with a GDP per capita of 613 USD in 2012 (IMF, 2013). Between 1991 and 2002, the country was ravaged by a civil war which left the public health system in ruins (Gberie, 2005). Over the last decade, Sierra Leone's health system underwent a process of reconstruction and reform. However, the Demographic and Health Survey (DHS) for the 2008e2013 period finds that maternal mortality remains high at 1165 deaths per 100,000 live births, while under-five mortality is estimated at 156 per 1000 live births (SSL & ICF International, 2014). In terms of health workforce, in 2011 there were an estimated 0.0071 doctors and 0.0631 nurses per 10,000 people in the public sector (Wurie et al., 2014). The distribution of health workers remains inequitable with major imbalances between rural and urban areas, and health workers attraction, retention and motivation are a challenge (Witter et al., 2015).
We analysed elsewhere the trajectory and drivers of HRH policymaking in the post-conflict period (Bertone et al., 2014). It emerged that the launch of the Free Health Care Initiative (FHCI) in 2010 provided the momentum for the approval of a series of HRH reforms, which included a substantial salary uplift for all technical staff of the Ministry of Health and Sanitation (MoHS), and the cleaning of the MoHS payroll to eliminate 'ghost workers' and add those working as 'volunteers'. The HRH reform process continued with the introduction of a Performance-based Financing (PBF) scheme in 2011 (which includes a staff bonus) and a Remote Allowance for health workers based in rural areas, in 2012. In parallel to their support to the design, and in some cases the funding, of these reforms, certain donors and NGOs adopted measures to ensure the alignment and rationalisation of the health workers' incentive package. In particular, the World Bank and Global Fund abolished supplementary payments to health workers in charge of HIV/AIDS services. However, despite the relative success of the decision-making process and the design of reforms, their implementation remained filled with challenges (Witter et al., 2015).

Methods
The present research was undertaken in the districts of Kenema, Bo and Moyamba (Fig. 1), which were purposefully selected to maximize differences in poverty, urbanisation, type and remoteness of facilities, as well as number of NGOs.
This paper draws on a series of key informant interviews at district level (n ¼ 18), carried out in SeptembereNovember 2013. The interviews aimed to be as comprehensive as possible of actors at local level, including DHMTs, as well as donors and local and international NGOs' staff ( Fig. 2). One NGO in Bo was not included as not available for interviewing at district level, although some information was collected from its representative at central level and from secondary sources, and triangulated during interviews with other actors in Bo. Moreover, key actors such as Members of Parliament and politicians, civil society members and representatives of Local Councils, who have some authority over health issues under the on-going decentralisation process, were not interviewed, nor have we included in the analysis private and informal providers of healthcare which also influence the incentives in the public system (Ensor and Witter, 2001). This is due to the fact that initially key informant interviews aimed solely at providing a background to the broader research, focused on health workers at individual level. However, the interactions between DHMTs and NGOs became such a relevant and recurring theme that it was later developed into a specific research question. The omission of actors external to the health system and non-public providers is a major limitation of our work.
The key informant interviews at district level are embedded in a larger mixed-methods research, which aims to investigate the health workers' 'complex remuneration' by quantifying their overall income and exploring the consequences of income levels and fragmentation. The broader research makes use of other data. A longitudinal survey was carried out to collect information on revenues (salary, remote allowance and PBF, as well as per diems and salary top-ups, and informal incomes) for 266 primary healthcare workers (90 in Kenema,96 in Bo,80 in Moyamba). The research also involved prolonged fieldwork (September 2013eMay 2014), a series of in-depth interviews with health workers (n ¼ 39e13 in Kenema, 12 in Bo, 14 in Moyamba), as well as an earlier documentary review and 23 key informant interviews at central level. Although this study relies mostly on key informant interviews at district level, the other sources of information were important to inform the analysis. For example, preliminary results from the health workers survey are included to support the findings from key informant interviews. Ethical clearance for all research components was obtained from the London School of Hygiene and Tropical Medicine and the Sierra Leone Ethics and Scientific Review Committee.
In order to map the emerging elements and themes, the analysis makes use of a political economy framework. This framework is based on that proposed by Harris (2013), but slightly adapted to take into consideration the fact that this research is not driven by a pre-identified problem, but is rather exploratory in scope (Fig. 3).
Two main areas are identified as the subject of analysis e on the one hand, the structural features which include the historical, cultural, geographical context and the relevant 'rules of the game' (institutions), such as policies, regulations and social norms; on the other, the agency features relating to the main actors, their interests, incentives and relations of power, and the analytical concepts that may explain actors' decision logics and behaviours. In particular, as analytical concept, we apply 'agency theory', which describes a