Health Policy and Planning, 33, 2018, 41–58 doi: 10.1093/heapol/czx146 Advance Access Publication Date: 25 October 2017 Original Article Pathways to high and low performance: factors differentiating primary care facilities under Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 performance-based financing in Nigeria Shunsuke Mabuchi1,*, Temilade Sesan2 and Sara C Bennett3 1 The World Bank Group Health, Nutrition and Population Global Practice, Washington, DC, USA, 2Centre for Petroleum, Energy Economics and Law, University of Ibadan, Ibadan, Nigeria and 3International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA *Corresponding author. The World Bank Group Health, Nutrition and Population Global Practice, 1818 H Street, NW, Washington, DC20433, USA. E-mail: smabuchi@worldbank.org Accepted on 21 September 2017 Abstract The determinants of primary health facility performance in developing countries have not been well studied. One of the most under-researched areas is health facility management. This study investigated health facilities under the pilot performance-based financing (PBF) scheme in Nigeria, and aimed to understand which factors differentiated primary health care centres (PHCCs) which had performed well, vs those which had not, with a focus on health facility management practices. We used a multiple case study where we compared two high-performing PHCCs and two low- performing PHCCs for each of the two PBF target states. Two teams of two trained local research- ers spent 1 week at each PHCC and collected semi-structured interview, observation and documen- tary data. Data from interviews were transcribed, translated and coded using a framework approach. The data for each PHCC were synthesized to understand dynamic interactions of differ- ent elements in each case. We then compared the characteristics of high and low performers. The areas in which critical differences between high and low-performers emerged were: community engagement and support; and performance and staff management. We also found that (i) contex- tual and health system factors particularly staffing, access and competition with other providers; (ii) health centre management including community engagement, performance management and staff management; and (iii) community leader support interacted and drove performance improve- ment among the PHCCs. Among them, we found that good health centre management can over- come some contextual and health system barriers and enhance community leader support. This study findings suggest a strong need to select capable and motivated health centre managers, pro- vide long-term coaching in managerial skills, and motivate them to improve their practices. The study also highlights the need to position engagement with community leaders as a key manage- ment practice and a central element of interventions to improve PHCC performance. Keywords: Health facilities, health sector reform, health services, health systems, health workers, international health, manage- ment, maternal and child health, primary health care, community C The Author 2017. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. V This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com 41 42 Health Policy and Planning, 2018, Vol. 33, No. 1 Key Messages • PHCCs which manage to improve performance under PBF in Nigeria do so through dynamic interactions between con- textual factors (e.g. staffing, access, competition), strong health centre management (performance management, staff management and community engagement), and community leader support. • PHCC performance improvement under PBF requires the development and/or identification of officers in charge (OICs) with strong management capacity. • It is important to look at community leaders as a critical driver to improve PHCC performance: selection of OICs and Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 health workers, and PHCC-level interventions need to be centred on gaining support from community leaders and more broadly the community, as a central element of an integrated approach to improve performance. Introduction solving with local stakeholders (e.g. involvement of local authorities and communities; adaptation of approaches to the local situation; In low-income countries, primary-level health services are often the active involvement of local staff to identify and implement solutions only type of formal health care accessible to a majority of the popu- to problems) (Dieleman et al. 2009); (ii) building a system of lation (Schneider et al. 2006; Komatsu et al. 2008; Gormley et al. accountability (visibility of performance; rewards and sanctions 2011). Given their importance for reducing preventable deaths of based on performance) (Topp et al. 2015); (iii) motivating health children and mothers, numerous initiatives exist to improve the per- workers for change (e.g. enhancing health workers’ awareness of formance of primary health services. However, the determinants of local problems; showing visible improvements in quality of care; primary health care facility performance in poor resource setting peer pressure; staff empowerment; and salary supplements) (Rowe have not been well studied (Marchal et al. 2010), and there is a et al. 2005; Dieleman et al. 2009); (iv) building team work and cre- dearth of evidence on what works to improve health facility per- ating a sense of belonging, trust and respect, and support by the formance (Dieleman et al. 2009). While multiple factors, from the management team (Dieleman et al. 2009; Marchal et al. 2010); and strength of broader health systems, to accountability relationships (iv) leadership to help build such supportive environments (Rowe are likely to influence facility performance (Topp et al. 2015), the et al. 2005; Topp et al. 2015). These findings provide a useful list of nature of management practices within the health care facility has important elements of health centre management. However, such been argued to be a critical determinant (World Health studies provide little insight on key drivers that improve perform- Organization 2007). Further, the quality of health facility manage- ance and differentiate high and low performers, and pathways ment may be particularly important in the context of performance- through which such drivers influence performance of primary health based financing (PBF) where, typically, facility managers receive facilities. both greater incentives for good performance, and greater autonomy to innovate and manage. This article explores the connections PBF in developing countries between facility performance and health centre management within PBF has been implemented or is under discussion in > 30 countries PBF schemes. in sub-Saharan Africa (World Bank 2013). Major design features of PBF include: (i) providing finance to health facilities based on quan- Determinants of health facility performance and health tity and quality of services provided; and (ii) providing autonomy centre management in developing countries for health facilities to plan and implement activities to improve their Despite the widespread acknowledgement of poor primary health health services. PBF allows primary health care facilities to use the care facility performance in low resource settings, little is known received funds at their discretion to improve health services. Health about management practices in these contexts. Empirical studies facilities, for example, can use a part of the received funds to pur- that have explored the relationship between health facility manage- chase drugs from certified local pharmacies, refurbish facilities, buy ment practices and performance are skewed towards hospital-based equipment, carry out more outreach activities and provide monetary studies in developed countries (Shortell and LoGerfo 1981; Shortell or non-monetary incentives to patients. The rest of the funds can be 1985; Davies and Ware 1988; Shortell et al. 1994a, b, 1998; allocated to health workers based on their performance and respon- Mitchell and Shortell 1997; Davies and Nutley 2000; Donaldson sibilities (Fritsche et al. 2014). et al. 2000; Ferlie and Shortell 2001; Meyer and Collier 2001; Providing autonomy and PBF to health facilities can create large Bloom et al. 2009; Bloom and Van Reenen 2010; Dorgan et al. variations in performance among them. Well-performing facilities 2010; McConnell et al. 2013). There are few empirical studies that will receive more performance incentives, can invest to further look at management-related issues at the primary health care level in improve their performance and thus receive more funds, creating a developing countries (Topp et al. 2015), and many approaches to virtuous cycle. In contrast, health facilities that cannot use the improve health facility management in developing countries are not received finance to attract more patients, or have problems that can- based on evidence. not be addressed by having cash on hand, will receive limited finan- Rowe et al. (2005) and Dieleman et al. (2009) reviewed pub- cial incentives, which in turn will limit their ability to improve lished studies on factors that affect health worker performance and health services (vicious cycle). With increased attention to the PBF interventions that improved health worker performance in low- and approach, many impact evaluations of the overall effect of PBF on middle-income countries. Also, Marchal et al. (2010) examined health service coverage, structural and process quality, human practices in a well-performing hospital in Ghana, and Topp et al. resources and cost effectiveness have been initiated, with at least (2015) explored the factors that drive health centre performance in nine evaluations disseminated by the end of 2016 (Kandpal 2016). Zambia. Important elements related to health facility management The results from Argentina, Cameroon, Rwanda, Zimbabwe, identified through these studies included: (i) engaging and problem- Zambia and to a limited extent the Democratic Republic of Congo Health Policy and Planning, 2018, Vol. 33, No. 1 43 suggest that PBF can be highly effective in improving coverage and 1. What differentiates good and poor performers among the the quality of services across many aspects of maternal and neonatal PHCCs under the PBF scheme in Nigeria? health (Kandpal 2016). They also provided evidence of general 2. Particularly, which management factors differentiate the per- health system strengthening such as more active supervision, more formance of the PHCCs? involvement of communities, and increased health worker satisfac- 3. Through what mechanisms do these factors affect the perform- tion (Kandpal 2016). However, how PBF works in different contexts ance of the PHCCs? has been regarded as a ‘Black Box’ (Renmans et al. 2016), and little This research seeks to address these questions in order to help research has tried to understand what determines the performance policy makers and program managers decide what to prioritize and of health facilities under a PBF scheme, and in particular the role of where to start to improve health centre performance in Nigeria and Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 health facility management. similar countries. PBF in primary health care centres (PHCCs) in Nigeria General approach Nigeria is an economic giant in Africa with a GDP of US$414.5 bil- This research used a multiple case study approach. A case study is lion, and an economy that has been consistently growing at a rate of ‘an empirical inquiry that investigates a contemporary phenomenon 7% per annum. However, Nigeria has made limited progress in within its real-life context; when the boundaries between phenom- delivering critical health services. Institutional delivery and DPT3 enon and context are not clearly evident; and in which multiple coverage remain very low at 35.8 and 38.2%, and contraceptive sources of evidence are used’ (Yin 2009). In contrast to experimental prevalence rate is only 9.8% (NDHS 2013). While the average designs that seek to test a specific hypothesis through the compari- catchment population for a PHCC providing first-level essential son of treatment and control groups, the case study approach lends health services is around 7600 people, PHCCs on average see only itself well to capturing information on more explanatory ‘how’, 1.5 patients per day, even when they have >10 staff (World Bank ’what’ and ‘why’ questions (Creswell and Plan Clark 2010). This 2014). Poor performance despite relatively abundant human resour- study was informed by the extreme or deviant case sampling ces for health relates to weak accountability and motivation due to approach that looks into unusual cases (in this research, high per- the ‘fundamental problem of the lack of clarity in responsibilities for formers and low performers) that provide rich information (Patton PHCC between state and local government’ (Oxford Policy 1990). This approach allows us to ‘understand under what condi- Management 2011), coupled with health financing arrangements tions programs get into trouble and under what conditions programs whereby no cash flows to health centres. exemplify excellence’ (Patton 1990), and integrate organizational To address these problems, the World Bank-funded Nigeria contextual factors in thinking about what works (Bradley et al. State Health Investment Project (NSHIP) introduced PBF that finan- 2009). ces a comprehensive essential package of services (Supplementary Appendix S1) and also incentivizes improvements in quality of care. Sampling and selection of cases During the pre-pilot phase in 33 PHCCs in Adamawa, Nasarawa The research was carried out in two Local Government Areas and Ondo states which started in December 2011, the PBF created (LGAs) in Nasarawa and Ondo states that started the pre-pilot PBF large variations in performance among the participating PHCCs. activities in December 2011. Due to security reasons, another target For example, coverage of institutional delivery was around 10% of state for the NSHIP, Adamawa state, was dropped from the selec- catchment population before the PBF in all target PHCCs, and tion of PHCCs. The number of cases (eight cases—two high-per- began to diverge significantly between good and poor performers formers and two low-performers for each of the two LGAs) was shortly after the scheme was initiated and these differences increased determined considering the transferability of findings to other over time, with high-performers achieved 80–90% coverage while PHCCs. Although PHCCs in the target LGAs are mostly rural and low-performers struggled with 20–30% coverage (National Primary have a small number of staff, the contextual factors and manage- ment practices of the PHCCs can be diverse. By having more than Health Care Development Agency 2016). This suggests that good one high and low performer in each LGA, the research aimed to performers achieve high uptake by using PBF wisely, while poor per- develop a good understanding of what are common and distinctive formers struggle with translating the opportunities that PBF pro- factors that influence performance. vides into results. We selected high and low performers in each LGA based on the In sum, there are clear knowledge gaps in drivers that improve quantity and quality of essential health services provided by the performance of primary health facilities, and in particular how man- PHCC. The package of services that PBF incentivizes in Nigeria cov- agement practices in health facilities influence the performance of ers essential services including outpatient visit, vaccination, referral primary health facilities in poor resource setting. This article aims to to hospitals, ANC, delivery, PNC, family planning, PMTCT, STD, address these knowledge gaps, particularly in the context of PBF TB, Malaria net, etc. (see Supplementary Appendix S1 for details), schemes, through an in-depth case study in Nigeria. and total monthly PBF earnings was used as a proxy measure for the level of utilization and quality of essential health services.1 The Methods PHCCs were ranked using three methods: (i) total PBF earnings adjusted for catchment population of each PHCC; (ii) unadjusted Study objectives total PBF earnings (as official catchment population data is not This study aims to provide an in-depth understanding of the deter- always accurate); and (iii) percentage improvement in total PBF minants of PHCC performance in the context of PBF, with a focus earnings from baseline. For the ranking, 2-month data from before on management practices at the PHCCs. The study is a part of a and after the payment delay that occurred between October 2012 broader effort to understand the relationship between management and February 2013 were used. Among the 12 consistently high or and performance at PHCCs under the NSHIP. The following spe- low performing PHCCs identified through these rankings, eight cific research questions were explored: PHCCs were selected through discussion with state and LGA staff in 44 Health Policy and Planning, 2018, Vol. 33, No. 1 each state who had been supervising the PHCCs for more than engagement and support, and staff management and motivation 2 years. The PHCCs that recently experienced changes (e.g. change emerged as potential key differentiating factors of PHCC perform- of Officers in Charge (OICs2)) and a particularly large PHCC that ance, and additional factors and sub-factors related to community was not comparable with other PHCCs were excluded. engagement and staff management were added to the code book for Supplementary Appendix S2 describes details of the rankings and coding and subsequent analyses. selection of the PHCCs. In Phase 2, data were organized to produce a case description for each PHCC. The interview and observational data were compared Conceptual framework and synthesized in order to develop a comprehensive picture of each Table 1 presents a conceptual framework outlining factors that case and understand dynamic interactions of different elements in Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 would lead to differentiation in PBF performance. This was built on each case. To minimize social desirability bias, interview results, the conceptual framework developed by Health Results Innovation particularly of OICs, were compared with interviews with other Trust Fund (2015) on PBF performance, by adding constructs stakeholders (non-OIC health workers, Ward Development around health facility management from an extensive literature Committee (WDC)4 chairpersons, and LGA PHCC coordinators) review in PubMed and Google scholar.3 In addition, the lead author and observation notes. To avoid a biased interpretation, texts that also drew upon his own knowledge and experiences in supporting contradicted emerging key features were carefully reviewed, and PHCC performance as a core team member of the NSHIP at the described in the case synthesis (rival explanation). World Bank. In summary, four main explanatory factors were iden- Phase 3 focused on cross case comparisons. Individual case tified: (i) community; (ii) health systems; (iii) PBF design and imple- descriptions were summarized and compared in tables for high per- mentation; and (iv) health centre management, together with more formers (four PHCCs) and low performers (four PHCCs). For each specific items (sub-factors) within each of these categories. The case syntheses of the three key areas of differentiating factors of per- shaded rows in Table 1 reflect sub-factors related to PBF design that formance were compared: Contextual and health system factors; are unlikely to vary by PHCC in the same LGA. These four factors Community engagement and support; and Performance and Staff and sub-factors were used to design interview guides, code tran- Management. Again, variations within the good performers and scribed data, and analyse data. poor performers, especially examples that contradicted the emerging Based on the themes that emerged through data collection and interpretations or patterns were closely examined. analyses, we adjusted these four main factors developing an alterna- tive three factors that seemed to better frame the influences on per- Results formance: contextual factors; community engagement and support; and performance and staff management. Table 1 explains the Findings from this study are presented in two sections. First we pro- changes and reasons for the changes. Case study findings are pre- vide a summary profile of the eight PHCCs, outlining overall pat- sented based on these factors. terns observed in high and low performing PHCCs, and key areas of differentiating factors of performance. The second section compares Data collection approach the eight PHCCs for each area of potential differentiating factors. In order to develop a thorough understanding of the cases and increase credibility of the study, the research involved the multiple Key features and differences of good and poor sources of evidence, using a range of quantitative and more com- performers monly qualitative techniques (Creswell and Plan Clark 2010). Table 3 describes basic information of the high and low-performing Information was compiled from operational data from the PBF PHCCs studied, and Tables 4 and 5 summarize their key features. reporting system, interview data, documentary data and observa- They suggest diverse and dynamic characteristics of high and low tions (Table 2). performers. Among the high-performers, PHCC-1 has many contex- Two teams of two local researchers who have experience in qual- tual and health system-related advantages such as abundant staff itative research and speak the local languages spent about 1 week and good road access from/to communities. These favourable fea- per PHCC and carried out data collection. They were trained by the tures were fully leveraged through good management practices by lead author through discussion of interview questionnaires, observa- the OIC. She not only served as a role model for other workers in tion protocols and mock interviews, and the second author leads the hard work, and patient care and support, but also carried out data collection team. During the data collection period, the team numerous strategic actions to motivate staff and gain support from had initially daily and subsequently weekly calls with the lead the WDC and traditional leaders. She also tracked key services indi- author to debrief findings, review emerging themes and data collec- cators (e.g. institutional delivery, fully vaccinated child, ANC) on a tion plans. wall, regularly updated targets to stretch staff, compared results with targets and agreed on specific activities for further improve- ment by fully involving WDC and traditional leaders. Abundant Data analysis staff were managed with clear roles and responsibilities under three- Analysis was carried out in three phases. Phase 1 started concur- shift 24/7 operations. rently with data collection, as transcribed data, interview notes and In contrast, PHCC-2, 3, and 4 had many contextual and health observation notes were generated for each PHCC. Coding and sub- system-related disadvantages such as poor staffing (five staff for sequent case analyses used both deductive and inductive approaches. each PHCC only), rural location, and very bad road access. Transcribed interviews were imported into NVivo 10 for electronic However, they overcame or minimized such disadvantages through coding based on a code book developed by the lead and second dedicated work (e.g. making themselves available 24 h a day by authors according to the conceptual framework (Table 1). The code sleeping in or next to the PHCCs) and good management practices book is presented in Supplementary Appendix S3. Throughout the mainly driven by the OICs. Notably, in PHCC-2 and 3, rural loca- data collection process, many different elements of community tion is linked to the existence of very influential chiefs who enforced Health Policy and Planning, 2018, Vol. 33, No. 1 45 Table 1. Potential differentiating factors of PBF performance—conceptual framework (non-differentiating factors shaded in blue) Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 Health Policy and Planning, 2018, Vol. 33, No. 1 46 Health Policy and Planning, 2018, Vol. 33, No. 1 47 Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 informal laws to ban the use of unlicensed providers and eliminated However, the OIC in PHCC-1, in a semi-urban area, did not have any competition. Also, being in very remote PHCCs made it necessary other revenue generating activities as she was committed to managing for staff to sleep in the community through weekly shifts, which her PHCC well. enabled staff to build family-like relationship with communities Differences in management practices became evident only after (PHCC-3 and 4). PBF was implemented. Before the PBF scheme, the performance of Contrary to the high-performing PHCC-2, 3 and 4 in rural loca- high-performing PHCCs was equally very low and the difference tions, three of four low-performing PHCCs (PHCC-6, 7 and 8) are in with low-performers was negligible. Although there were no visible semi-urban location, with good road access from/to the catchment differences in autonomy or the availability of performance based communities. Further, all of the four low-performing PHCCs had funds that the PBF platform provided to the PHCCs, the ways the more staff than PHCC-2, 3 and 4 (6–20 staff compared with 5 staff). It PHCCs leveraged the opportunities were different between high and is important to note that semi-urban PHCCs can face additional chal- low performers. lenges such as direct competition with general hospitals and security • Prior to PBF we had nothing and our structure here was very threats (PHCC-7 and 8). Also, the level of commitment by OICs and staff can be eroded leading to absenteeism given that they have other poor but PBF has been able to change that; OIC; PHCC-1. • In fact, the situation was very poor. [. . .] Patronage was quite low revenue-making opportunities (PHCC-5, 6 and 8). Comparison of PHCC-1 and low-performing PHCCs with similar semi-urban settings because the people would see the dilapidated building and wouldn’t highlights the importance of the OIC’s commitment and management want to be treated here. [. . .]. But the difference is clear now that capabilities. While PHCC-7 and 8 suffered from competition with gen- the government has provided the PBF; Non-OIC; PHCC-2. eral hospitals and other PHCCs, PHCC-1 with its close connection with the community and cheaper services leveraged PBF bonuses to Differentiating factors of performance by key areas take patients away from these competitors. Also, in general, staff at Area 1: contextual and health system factors semi-urban PHCCs tend to have other revenue generating activities, Among the potential contextual and health system factors that can whereas workers in very remote PHCCs do not have these activities. differentiate performance in the original framework, key factors 48 Health Policy and Planning, 2018, Vol. 33, No. 1 Table 2. Data collected for each PHCC Type Data Use Operational data from Monthly quantity of the selected 21 services pro- To analyze the performance (utilization) of the health PBF reporting system vided (e.g., Outpatient visit, vaccination, institu- centres tional delivery) for all the PBF PHCCs verified by an independent agency. Quarterly quality assessment score verified by local To analyze the performance (quality) of the authority and counter-verified by the NPHCDA PHCCsManagement indicators in the quality checklist for all the PBF health centres (Since the launch of were used to verify reported management practices Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 the PBF pre-pilot). Interview data OIC for the selected PHCCs (transcribed) • To understand specific organizational contexts, man- 2 interviews, 1 group agement practices at the PHCCs, and the support interview, and 1 group received from supervisors and communities discussion per facility A group interview with 2–3 health workers (e.g., • To triangulate what OICs explained (e.g., check their nurses, midwives, or community health extension understanding of PBF targets to assess the effectiveness workers) per case (transcribed) of OIC’s communications); • To understand health workers’ perception of the health centre’s practices and changes observed (e.g., their per- ception on communication with the OIC) A chairperson of Ward Development Committee • To understand community leader and community (WDC)a per case (transcribed) engagement practices by the PHCCs, and the activities of the community and their effects Group discussions with LGA PHCC Department • To understand how supervisors viewed the PHCCs and supervisor and PBF consultant who visit the possible reasons for high and low performance. health centres regularly (not transcribed) • To understand the differences in their supervision activ- ities across the PHCCs. Documentary data Review of reports and tools used at the health To triangulate the responses of the OICs and other stake- centres, including: (i) business plan, (ii) financial holders, and assess the management practices at the statement (indices tool), (iii) PBF invoices, quality PHCCs (e.g., review meeting minutes to see if the checklist, and HMIS report, (iv) drugs records; PHCCs analyze issues, conclude with clear actions with (v) notices and graphs on the wall, (vi) staff eval- deadlines, and review the progress of what are agreed in uation sheet, and (vii) minutes of the health previous meeting)The data review results were docu- facility committees and other meetings, based on mented in the case summary note for each PHCC, and the observation protocols extracted through the individual case analysis Observations Observation of (i) facility, equipment, drugs and • Triangulate the performance data by looking at the con- waste management and (ii) monthly meetings at ditions of and services provided by the PHCCs the PHCCs based on the observation protocols • Observe the meetings to see how performance and issues are discussed, actions are agreed and assigned, such actions are reviewed in the meeting, and health workers and communities are actively involved in open discussions. Triangulate this with interview data a A committee comprised of community, youth, women leaders, etc. that is responsible for reviewing performance of a PHCC, authorizing the use of PBF funds, and assisting the PHCC to improve utilization and quality of services. Table 3. Basic information of selected PHCCs State Nasarawa Ondo High-performers PHCC-1 PHCC-2 PHCC-3 PHCC-4 Semi-urban PHCC with Rural PHCC with catchment Rural PHCC with catchment Rural PHCC with catchment catchment population population 18 000. Has population 6000. Has population 10 500. Has 7000. Has 17 staff, of only 5 staff, of which 2 are only 5 staff, or which 2 are only 5 staff, of which 3 are which 6 are skilled. Opens skilled. Opens 24 h/7days, skilled. Opens 24 h/7days, skilled. Opens 24 h/7days, 24 h/7days, and patient per and patient per day after and patient per day after and patient per day after day after PBF is 11.3 PBF is about 12.6 PBF is 8.3 PBF is 8.7 Low-performers PHCC-5 PHCC-6 PHCC-7 PHCC-8 Rural PHCC with catchment Semi-urban PHCC with Semi-urban PHCC with Semi-urban PHCC with population 6500. Has 8 catchment population catchment population catchment population staff, of which 4 are 8500. Has 20 staff, of 8000. Has 6 staff, of 10 000. Has 8 staff, of skilled. Opens from morn- which 10 are skilled. which 3 are skilled. Opens which 4 are skilled. Opens ing to evening, and patient Opens from morning to from morning to evening, 24 h/7 days, and patient per day after PBF is 3.8 evening, and patient per and patient per day after per day after PBF is 1.3 day after PBF is 3.2 PBF is 3.2 Health Policy and Planning, 2018, Vol. 33, No. 1 49 Table 4. Overview of high-performing PHCCs State Nasarawa Ondo Names PHCC-1 PHCC-2 PHCC-3 PHCC-4 Summary Semi-urban PHCC with good Remote PHCC with serious Remote PHCC with serious Remote PHCC with serious access and abundant staff. shortage of staff and bad shortage of staff and bad shortage of staff and bad The OIC carried out road access. The PHCC road access. The PHCC road access. The OIC car- numerous strategic actions benefitted from full sup- benefitted from full sup- ried out numerous strategic to motivate staff, gain sup- port by a very influential port by a very influential actions to motivate staff, Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 port from WDC and tradi- chief and a dedicated OIC. traditional leader who gain support from WDC tional leaders, and solve Although performance enforces a law against unli- and traditional leaders, problems identified in reg- management did not seem censed providers. Highly build trust and attract ular performance reviews. to be as rigorous as other 3 dedicated OIC and staff patients. Staff were from The PHCC was taking high-performers, the OIC resided in the PHCC, the community and reside patients from bigger hospi- managed to build trust reviewed performance on in the PHCC; OIC tals and PHCCs from traditional leaders weekly basis and carried reviewed performance and community, and moti- out numerous strategies to every week, and involved vate staff to achieve results gain trust and attract even LGA or state staff to patients solve problems A. Contextual and Favourable conditions: Serious shortage of staff, yet Serious shortage of staff, yet Serious shortage of staff, yet health system Relatively good access to was making health services was making health services was making health services factors catchment area, and abun- available for 24/7 by OIC available for 24/7 by using available for 24/7 by staff dant staff and staff living in and near a weekly duty roster with 2 sleeping at the PHCC Proactive influence by PHCC the PHCC people on-duty sleeping at Very bad access with 40 min on contextual and health Very bad access during rainy the PHCC each week drive from main express- system factors: e.g., hired season, yet mitigating it Very bad access with 3H way on unpaved rough extra staff from community with planned outreach vis- drive from LGA secretariat roads which get heavily to enable 3 shifts for 24/7 its to affected communities on unpaved roads with flooded during the rainy services; hired a doctor to No major competition with flooding in rainy season season. OIC mitigated it by attract patients; prioritized large-scale providers given No competition: Traditional paying staff and clients in pregnancy test to enrol its remoteness leader and WDC regulated cash or in kind often out of reluctant women; worked and removed unskilled her pocket with TBA; convinced com- providers No major competition with munity to construct a road large-scale providers given to improve access its remoteness B. Community Strong support from WDC Strong support by traditional Strong support from tradi- Strong support from tradi- engagement and traditional leaders for leader/WDC chair: The vil- tional leader: He set an tional leader: Traditional and support attracting patients and lage chief and has very unwritten law that banned leaders mandated commun- building infrastructure strong authority over sub- unskilled providers and ity members to utilize Engagement with WDC, tra- village chiefs and residents. fined those who used them. PHCC and settle any ditional leaders: Strong sup- His communications to Health committee moni- problems port was attributed to the communities enhanced uti- tored and eliminated Engagement with WDC, tra- OIC’s/PHCC’s devoted lization of PHCC competition ditional leaders: Very open work and active and open Engagement with WDC, tra- Engagement with WDC, tra- and proactive updates and engagement with them. ditional leaders: OIC built ditional leaders: The PHCC consultation, in addition to Staff were all remembered trust with the chief with requested the WDC specific gifts to appreciate them by name by community hard work and by follow- supports, e.g., addressing Community engagement: Community engagement: ing his advice. poor utilization by a spe- All staff except OIC was PHCC built trust and Community engagement: cific tribe, countering a from the community, and recruits patients through PHCC built trust and false rumour, regulating all staff were sleeping in the various measures, e.g., fre- recruited patients through household waste, and PHCC, which helped gain quent health education, free numerous measures, e.g., shared large part of PBF trust by the community. mobile clinic to remote health education, outreach bonus with them There was a free flow of communities, free and dis- even in rainy season, fee Community engagement: people to PHCC not just count services, services on reduction and transparent PHCC created bonds for health services. credit. operation, 24/7 services, through residing in the Numerous strategies to provision of incentives/ PHCC, providing gifts, end recruit patients, e.g., out- gifts, and laboratory of year parties, etc., and reach 2–3 times a week; investigation addressed barriers, e.g., fol- individual tracking of preg- low-up with pregnant nant and postnatal women; women and children; motor gifts; free services bike transport for pregnant women; services on credit (continued) 50 Health Policy and Planning, 2018, Vol. 33, No. 1 Table 4. (continued) State Nasarawa Ondo Names PHCC-1 PHCC-2 PHCC-3 PHCC-4 C. Performance Strong performance manage- Strong performance manage- Strong performance manage- Strong performance manage- and Staff ment: OIC tracked key ment by OIC where he pre- ment: OIC reviewed results ment: OIC updated and Management indicators, displayed them sented targets and results in on weekly basis with the gave stretched targets to on a wall, compared targets regular meetings with other skilled staff by encourage improvement; and results at the monthly health workers, WDC, and reviewing weekly handover tracked results every week; Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 WDC meetings and dis- at the community town notes, and addressed prob- informed them to workers, cussed approaches to halls, addressed specific lems quickly by involving and discussed ideas to achieve targets, agreed on problems (e.g., fee levels, WDC and traditional lead- improve. The PHCC specific solutions with clear growth monitoring ers. All staff interviewed involved WDC, traditional responsibilities, and fol- improvement), and fol- had clear attention to tar- leaders, and even LGA and lowed up for lowed up on results. There gets and performance of the state staff to address prob- implementation was some level of attention PHCC and committed to lems as needed. Staff were Staff management/ to targets and actuals achieving the targets fully aware of targets and motivation among workers (not as Staff management/ performance, and highly Role model: OIC was a role strong as PHCC-1, 3, 4) motivation committed to achieve the model of client service (e.g., Staff management/ Role model: OIC was a role targets paying costs for patients motivation model for dedicated work, Staff management/ out of her pocket) Role model: OIC was a role communication, and track- motivation Open environment to sugges- model with his hard work ing of pregnant women Staff support and team build- tion and correction from and proactive covering of Intrinsic motivation: Highly ing: OIC fostered a family- her subordinates staff’s absence motivated workers run two like relations with her team Motivating by targets: regu- Staff support: OIC provided weekly shifts for 24/7 by bringing food paid out lar communication and personal gifts and feedback, Staff support and team build- of her own pocket and tracking of targets training opportunities, step- ing: OIC campaigned vigo- cooking and eating same Staff support: OIC assisted down training, etc. The rously for a prize from state food with staff, and giving staff from her own pocket OIC changed the PBF for a staff; OIC changed the small money to them when needed; monthly clin- bonus allocation formula to PBF bonus allocation for- Motivating by targets: OIC ical training and step down increase allocation to staff mula in favor of staff rather regularly updated stretched training, made bonus trans- rather than OIC to moti- than OIC; staff in the same targets based on past results parent by evaluating vate them to achieve targets shift cooked food together; and community situation, through a committee, etc. coaching and flat relation- and communicated results ship with mutual feedback every week derived from the eight cases were: (i) staffing; (ii) distance and acces- Contextual and health system factors need to be viewed in com- sibility; (iii) security; and (iv) competition with other providers. bination with other factors or drivers of performance such as com- Notable differences were not found in other factors such as com- munity engagement and support, and performance and staff munity income, cultural and social norms and support from other management. There seems to be no decisive contextual and health partners and programs. Also, contrary to the hypothesis in the origi- systems factor on the performance of PHCCs. The case studies sug- nal conceptual framework (Table 1), except for staffing of the gest that the PHCCs can change or mitigate some contextual and PHCCs, few notable differences were found in support by state or health system-related disadvantages, and leverage contextual and LGA, financing, supply chain, supervision and training, and PBF health system-related advantages. For example, in rural areas, design and implementation among the PHCCs. At least for the PHCC-2, 3 and 4 addressed potential staffing and accessibility issues PHCCs studied, differences in health system and PBF design and by ensuring that staff reside in or next to the PHCCs, scheduling implementation were not the primary factors explaining differences outreach activities to avoid rains, and providing additional funds to in performance. the team for outreach during the rainy season from PBF funds or the As described in Table 4, among the high-performers, PHCC-1, OIC’s own pocket. In a semi-urban area, PHCC-1 was taking located in a semi-urban community, had favourable conditions for patients from competitors including a hospital, and did not suffer all the above factors except for competition. In contrast, PHCC-2, 3 from absenteeism of staff due to other revenue-making opportunities and 4 in rural communities had challenges in staffing, and distance that was otherwise found in low-performing PHCCs in semi-urban and accessibility, while security and competition issues were mini- areas. It was evident that various activities to attract patients such as mal. In contrast, among poor-performers, PHCC-6, 7, and 8 were active community engagement, free and discount services and serv- semi-urban with more staff and good access, but PHCC-7 and 8 suf- ices on credit, and hiring of a doctor and extra staff from the com- fered from competition with other providers and security issues. munity to build credibility helped attract patients from competitors, PHCC-5 was in a rural community with the same challenges in staff- while the OIC’s dedication and support to staff kept staff motivation ing and distance and accessibility as high-performers in rural high. In contrast, none of the high-performing PHCCs had security communities. issues, and we did not have evidence to conclude whether security Health Policy and Planning, 2018, Vol. 33, No. 1 51 Table 5. Overview of low-performing PHCCs State Nasarawa Ondo Names PHCC-5 PHCC-6 PHCC-7 PHCC-8 Summary A relatively well-placed rural Semi-urban PHCC with good A semi-urban PHCC that A semi-urban PHCC that PHCC in access, staffing access, abundant staff, less faced competition with a faced competition with and competition. competition and good sup- hospital and security issue. hospitals and public, pri- However, OIC was regu- port from other programs. Poor performance manage- vate and unlicensed pro- larly unavailable and there However, limited com- ment, staff shifts, and viders. WDC and Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 were no visible efforts in munity engagement, poor engagement with WDC traditional leaders did not community outreach, per- performance management, and traditional leaders regulate unlicensed pro- formance management, problems in bonus alloca- seem to have resulted in viders. OIC’s weak com- and staff motivation. tion to staff seem to have limited support by tradi- mitment and poor Support from WDC and resulted in staff absentee- tional leaders, difference in management also seem to traditional leaders were ad ism, poor support from performance between have contributed to low hoc WDC and traditional lead- shifts, and overall poor staff motivation and lim- ers, and drug stock-outs results. WDC Chair did ited community not live in the community, engagement which also limited his support A. Contextual and health Shortage of staff, but better Abundant staff: In addition Competition with a hospital Competition with various system factors than PHCC-2, 3, and 4; to 15 original staff, 5 staff posed a major challenge. health providers. There there were no visible (nurse/midwife and They also had competition were also private pharma- actions to address it. The CHEWs) from a national with traditional providers cies and 7 unlicensed pro- OIC was often not avail- scheme Security challenge made the viders in the community. able due to other revenue Good regular access: It had residents leave their com- One of them was well inte- generating activity access issue in rainy season, munity in the afternoon, grated in the community, Bad access to some commun- but regular access was good making it difficult to attract and known by traditional ities during rainy season, Some competition with a gen- patients leaders and WDC but it was less serious than eral hospital and 3 patient Shortage of staff, but better Security challenge: Two years PHCC-2; there were no medical vendors, but this than PHCC-2, 3, and 4 and ago 24 thieves trooped in visible efforts to address it did not seem to pose partic- they had less patients. OIC and killed people. Concern No major disadvantages in ular challenge compared qualification (registered still existed other areas: good support with other PHCCs nurse) were the highest Sufficient staff: 3 staff on from other programs; some Relatively abundant support among those researched duty all time for few but not severe competition from national and state patients with other providers programs for equipment Good access: Just off express- and medicines way to two towns and most could get there within 30minutes of walk B. Community engagement Limited engagement with and Relationship issues with tra- Limited engagement with and Limited support from WDC and support support from WDC and ditional leaders and WDC: support from WDC and and traditional leaders: A traditional leaders: Good No visible support from traditional leaders: traditional leader and relationship with WDC WDC and traditional lead- Traditional leaders’ support WDC did not regulate unli- through consultation and ers to address the competi- was only on-request basis. censed providers, given sharing of PBF bonuses. tion with patent medical WDC chair did not live in their personal relationships However, the WDC sup- vendors. WDC Chair the community and his with them. Their support ported only ‘when the need demanded share of PBF engagement was limited. was limited to ad hoc trou- arises’ bonus, which led to rela- PHCC did not report spe- ble shootings Limited community engage- tionship issues. No relation- cific performance to WDC Limited community engage- ment: Outreach was only 2- ship established with Limited community engage- ment: Outreach only once a 3 times a month; OIC was traditional leaders ment: No visible strategies month; lack of follow-up unavailable in the PHCC Limited community engage- were presented. There seem with pregnant women for regularly, which affected ment: No 24H services to be good practices in one delivery; no other visible the level of patronage by despite abundant staff; free of the two shifts, but the engagement by the PHCC community services and incentives only other shift were not per- when free drugs were avail- forming well able; PBF bonuses not used to address financial barriers for community C. Performance and Staff Poor performance manage- Poor performance manage- Poor performance manage- Poor performance manage- Management ment: No specific improve- ment: No update of plan ment: Little systematic ment: No changes in targets ments were explained as a and targets, and regular dis- process for planning for since the beginning of PBF, result of the performance cussion on the target and services, target setting, and and staff were not aware of reviews. Staff did not know results. None of the health performance tracking and the targets. There was no (continued) 52 Health Policy and Planning, 2018, Vol. 33, No. 1 Table 5. (continued) State Nasarawa Ondo Names PHCC-5 PHCC-6 PHCC-7 PHCC-8 targets and plans of the workers interviewed knew review. Meetings with staff evidence of systematic per- PHCC, and neither targets or actual and WDC were irregular, formance review. Meetings attended the meeting nor performance performance were not dis- were irregular – only one got informed about the Stock-outs: The only PHCC cussed in numbers, and staff meeting and two com- meeting results that experienced stock-outs staff were not aware of mittee meetings in 2013 Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 Staff management/ of essential medicines twice plans and targets Staff management/ motivation: in recent months Staff management/motiva- motivation: Collaboration among staff, Staff management/ tion: Large disparity in the Staff members covered for fairness in training oppor- motivation: performance of the two each other when someone tunities and technical ses- There was collaborative rela- shifts. Members of the bet- was absent sions among staff for tionship among staff and ter performing shift were Lack of efforts motivate staff: learning was working well staff received fair training more motivated and OIC showed up late for Poor motivation among staff: opportunities worked well as a team work and was idling Workers were not aware of Poor motivation among staff: Lack of efforts to motivate around the PHCC rather any efforts by the OIC in Absenteeism reported from staff: OIC was not seen to than going after patients; encouraging good perform- multiple sources; non-trans- do much to motivate staff, did not sleep in the PHCC ers; regular absence of the parent distribution of per- except for using the per- like other staff do and did OIC affected the staff moti- formance bonus; no formance bonuses to not come on weekends. She vation; no specific team proactive engagement by reward good performers was not around much of building efforts observed the OIC in motivating staff the time to supervise work Table 6. Comparison of contextual and health system factors between high- and low-performers in semi-urban and rural communities High-performers Low-performers Semi-urban PHCCs PHCC-1 PHCC-6, PHCC-7, PHCC-8 • Abundant staff (17 staff) • Abundant staff (20 staff, 6 staff, 8 staff) • Good access from/to community • Good access from/to community • No security issue • Security issue (PHCC-7 and 8) • Some competition with a hospital and a PHCC (but • Tough competition with hospitals, PHCCs, medi- taking patients from them) cine stores, unlicensed providers Rural PHCCs PHCC-2, PHCC-3, PHCC-4 PHCC-5 • Serious shortage of staff (5 staff each) • Shortage of staff but sufficient for the low patient • Very poor access from/to community load (8 staff) • No security issue • Poor access from/to community (better than • Few or no competition with other providers PHCC-2) • No security issue • Some but not severe competition with other providers issue can be overcome. However, the fact that two of the four community, while such proactive actions were not observed in low performers did not have any security issue suggest that there PHCC-7 and 8. are many other factors that contribute to low performance of PHCCs. Further, as described below, competition particularly with unli- Area 2: community engagement and support censed providers will differ according to the role that traditional Table 6 summarizes the three main patterns of community engage- leaders play in influencing community choice of provider. For the ment and support we found in the eight cases. First, in high- health system factors, in Ondo state, respondents from both high performing PHCCs, we identified a pattern where strong and multi- and low performers suggested that ad hoc provision of free drugs ple types of engagement by the OICs/PHCC staff to WDC, tradi- helped increase patient numbers when they were available, but had tional leaders and community members enhanced the level of a negative impact in terms of failing to meet raised community support from the PHCC. Second, we observed another pattern expectations when they were not available. Differences were found where traditional leaders/chiefs in rural communities spontaneously in how PHCCs responded to the availability of free drugs—PHCC-3 exercised very strong authority, enforced the use of PHCCs, and and 4 made services free, or discounted the price for patients using removed unlicensed providers. PBF funds when free drugs were not available, and received support Third, in low-performing PHCCs, in all four cases, although a from community leaders in explaining the need for user fees to the few PHCCs shared their PBF bonus with WDC as an incentive, the Health Policy and Planning, 2018, Vol. 33, No. 1 53 engagements of the OICs/PHCCs with WDC and traditional leaders and build team spirit. Of these approaches, training and coaching were limited to trouble-shooting on ad-hoc basis. Direct community were also implemented in all of the poor performers to some extent. engagement by PHCCs were also limited in these four PHCCs (e.g. However, PHCC-6 suffered from non-transparent allocation of per- outreach once a month by PHCC-8, and two to three times a month formance bonus among staff, and PHCC-7 did not involve health by PHCC-5, compared with two to three times a week by PHCC-1) workers in deciding the use of PBF bonus for improving services. with few specific strategies to recruit patients. In PHCC-5, regular Other approaches, such as the OIC acting as a role model for other absence of the OIC eroded community trust. In some PHCCs, there staff, providing rewards/gifts/assistance to staff, building of family- were clear challenges in WDC or traditional leader support, due to like relationship, and bonus re-allocation to benefit staff were not the unavailability of WDC (PHCC-7) and lack of collaboration to observed in any of the low-performing PHCCs. In PHCC-8, Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 address competition with unlicensed providers, given their personal the OIC’s regular absence while asking other staff to maintain relationships with the providers (PHCC-8). 24 h/7 days a week operation by sleeping in the PHCC seem to drive The case studies suggest diversity in the pathways to gain com- staff’s mistrust of the OIC and lack of collaboration with her. munity support. However, it was common across cases that com- munity support could not be gained without support from WDC Discussion and traditional leaders, and that poor engagement of WDCs and tra- Drivers and pathways of performance improvement ditional leaders by the OICs/PHCCs led to weak support. On the factors that differentiate good and poor performers (Research question 1), we found critical differences in the areas of Area 3: performance and staff management community engagement and support, and performance and staff Performance management. There were clear common performance management. For the community engagement and support, multiple management features among high-performers that were missing in types of engagement by OICs and PHCC staff to build trust and low-performers including planning, target setting, performance gain support from WDC, traditional leaders and community mem- tracking and review, and problem solving. All of the four high- bers, and spontaneous support from influential traditional leaders in performing PHCCs had clear commitments to achieve targets, car- some of rural PHCCs, enhanced the use of PHCCs by communities ried out rigorous performance reviews, and addressed problems by among high-performers. For performance and staff management, involving WDC, traditional leaders and other stakeholders. All four performance management activities, such as frequent updating of high-performing PHCCs updated the target quantity of essential targets, visualized tracking of results, weekly or monthly reviews of health services (e.g. outpatient visits, ANC, institutional delivery, performance and target achievement, and problem solving to fully vaccinated children) based on actual performance and the com- improve performance with clear actions and follow-ups with WDC, munity situation to keep them as ‘stretch’ targets, and reviewed traditional leaders, and other stakeholders appeared to be key differ- achievement of targets on a wall (PHCC-1), or on a monthly ences between good and poor performers. Also, multiple approaches (PHCC-1 and 2) or weekly (PHCC-3 and 4) basis among staff and by OICs to motivate staff and build team spirit through providing with WDC and traditional leaders. As a result, all health workers at an appropriate role model, setting stretch targets, providing the four high performing PHCCs (except one health worker at rewards/gifts/assistance to staff, building a family-like relationship, PHCC-2) could explain the targets and actual results in the last and re-allocating bonuses in favour of staff emerged as critical dif- month. The OICs and staff in the high-performing PHCCs were able ferences between high and low performers. Among these factors, to explain many specific actions that came out of the performance community engagement—engagement of WDC, traditional leaders review meetings, such as the use of a mobile clinic to access remote and community members, performance management and staff man- communities, negotiation with the LGA to purchase drugs from a agement approaches are health centre management factors that certified vendor close to town rather than in the capital to reduce seem to have differentiated high and low performing PHCCs the cost of drugs, adjustment of the price of services using subsidies (Research question 2). from the PBF bonus to ensure affordability to the community, and In terms of the mechanisms through which such differentiating construction of a road to an underserved community by community factors affect the performance of the PHCCs (Research question 3), groups. It should also be noted that, as shown in PHCC-1 (in although we do not have sufficient information to fully explain all Table 4(C) “Open environment”) and PHCC-4 (in Table 7), OICs the dynamic mechanisms, Figure 1 synthesizes the key pathways in the high-performing PHCCs created a flat and open environment (presented as arrows) identified through the case studies. Contextual for staff and other stakeholders to discuss issues and develop these and health system factors, particularly staffing, access, and competi- creative solutions. In contrast, none of the health workers inter- tion with other providers; health centre management including com- viewed at low-performing PHCCs were aware of or remembered munity engagement, performance management, and staff targets of the PHCCs, and they could not explain any of the actions management; and community leader support, are highlighted as the that came out of the performance review meetings. three main drivers of performance improvement (shaded boxes in Figure 1) that have direct influence on the use of essential health Staff management and motivation services by the community.5 There were also clear differences between high-performers and low- Importantly, these three main drivers of performance influence performers in staff motivation. In three of the four low-performing each other. The case studies particularly highlighted the influences PHCCs, low staff motivation and absenteeism (PHCC-5, 6), and that health centre management can have on the other drivers. High lack of trust in the OIC among staff (PHCC-8) were highlighted as performing PHCCs employed management approaches to leverage an issue. In contrast, in high performers, highly motivated health positive contextual and health system factors and mitigate negative workers clearly contributed to gaining support from WDC and tra- factors such as poor staffing and limited road access, whereas poor ditional leaders and trust-building with the community. management cancelled the positive factors and worsened the nega- Table 7 summarizes approaches observed in high-performing tive factors. Also, as mentioned above, multiple types of engagement PHCCs. All four PHCCs used multiple approaches to motivate staff by PHCCs helped build trust and gain support from community Table 7. Observed patterns of community engagement and support 54 Pattern Description PHCC Quotes (selected examples) Strong and multiple Strong and multiple types of engagement by the PHCC-1, PHCC-4 She (OIC) is very active and devoted to the facility. She also looks up to the WDC for discussion types of engage- OICs/PHCC staff to WDC, traditional leaders (PHCC-2 and 3 also on problems associated with the PHCC, relates very well with the community and the LGA ment by the OICs/ and community members enhanced the level carried out similar PHC coordinator. If there is any problem she always inform me and if there is anything that PHCCs of support from them. The OICs/PHCCs fully activities) she needs that would improve the PHCC she also tell me; WDC Chairperson; PHCC-1 involved WDCs and traditional leaders in Due to the help they render to people of the community, even if they (community members) planning, performance monitoring, problem don’t see any of the health workers they keep asking about the person. Through being nice and solving, and PBF bonus allocation, which rendering help to clients, for example, treatment on credit. The community members know all appears to have encouraged the WDC mem- the workers individually; WDC Chairperson; PHCC-1 bers and traditional leaders to support the I tell them that we just got this bonus. I don’t hide it from them, because they are the ones who PHCCs. PHCC staff also had built strong mark the scheme. I tell them, this is how much we were paid, and I give them N2, 000 each. trust with community members They tell me they have never experienced this kind of openness before with the PHC; OIC; PHCC-4 When I get here every Monday, I go round the community to greet people. [. . .] For pregnant women especially, when I haven’t seen them in two or three days, I go to visit them [. . .] Some other OIC may sit in the PHCC and not reach out to the community this way; OIC; PHCC-4 Strong support by tra- Traditional leaders/chief in rural community (a PHCC-2, PHCC-3 So I can say the chief is very much involved in the welfare of the ward as he represents the whole ditional leaders traditional leader of the community for ward as the WDC chairman; OIC; PHCC-2 PHCC-2 is also the WDC chairperson) spon- I am the Chief. If I say anything, everybody cooperates with me. My late father had only one taneously exercised very strong authority, Mai unguwa (village sub-head) but I have four (4) Mai unguwas. If I want anything done I will enforced the use of PHCCs, and removed tell them and they will go and inform people in their sub-wards [. . .] Through me they (PHCC unlicensed providers. Traditional leaders also staff) have been able to build relationship with the community. If I talk to them, other leaders influenced WDC members for them to engage of the community will do as I say; WDC Chairperson; PHCC-2 with community and monitor unlicensed Prior to PBF we didn’t have this level of communication with the community [. . .] Really it was providers because the announcement (by village head) is made that everybody is listening; OIC; PHCC-2 It is the traditional leader that will set down rules for the community members that if they do not come to the PHC for delivery, their lands which they are employed to will be collected from them; OIC; PHCC-3 WDC is just like the police. If we see you going to a quack (unskilled provider), you had better leave this community for us, so you don’t implicate us; WDC Chairperson; PHCC-3 It helps a lot. You see, that is what has made it easy for us to gain ground in this place. Because we didn’t meet quacks here; Non-OIC health worker; PHCC-3 Weak pathways to The engagements of the OICs/PHCCs with PHCC-5, PHCC-6, Community members are not able to access the OIC regularly in the PHC and that eroded his gain community WDC and traditional leaders were limited to PHCC-7, PHCC-8 cordial relationship with and the level of patronage by the community; LGA supervisor; support (Poor per- trouble-shootings on ad-hoc basis. In some PHCC-5 forming PHCs) PHCCs, there are clear challenges in WDC or We aren’t close to any of the traditional leaders; OIC; I am not aware of any support by the tradi- traditional leader support, due to the unavail- tional leaders; Non-OIC health workers; PHCC-6 ability of WDC (PHCC-7) and lack of collab- I don’t see any activity from them (WDC). The WDC chairman doesn’t live here. I don’t think I oration to address competition with saw him once throughout the month of October [. . .] The chairman didn’t attend the last meet- unlicensed providers (PHCC-8) ing; Non-OIC Health Worker; PHCC-7 The Oloja (community leader) made it clear that there is nothing that can be done about this par- ticular quack (unlicensed provider). He said they can’t chase him away because he is one of them in the community; OIC; PHCC-8 He said he would ask them if their community members had any particular reasons for not patronizing the centre, note those down, and give us feedback. We have not heard any further word on the matter since then. He has not called us; Non-OIC health worker; PHCC-8 Health Policy and Planning, 2018, Vol. 33, No. 1 Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 Health Policy and Planning, 2018, Vol. 33, No. 1 55 Table 8. Observed approaches in high-performing PHCCs in motivating staff (ͱ: observed) Approach Description PHCC-1 PHCC-2 PHCC-3 PHCC-4 Quotes (selected examples) Role model OIC motivates staff ͱ ͱ ͱ ͱ The workers in this clinic are good and have been really moti- through own behav- vated especially through the OIC who supports the commun- iours such as hard ity members that come for treatment out of her pocket; WDC work and good patient Chairperson, PHCC-1 care. He makes sure he comes on time to observe those who also comes in time and late, by so doing he encourages us too since we are all close by. So by 7:30 to 8:00 we are already in Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 the office. He shows good example; Non-OIC health worker; PHCC-2 Stretched target OIC sets ‘stretch’ targets ͱ ͱ ͱ ͱ It encourages the staff to achieve the targets because the and review and monitor achieve- bonuses boost their moral to work more. The volunteers are ment with staff even more encouraged than the fulltime staff. For example, rigorously the lab technician comes to work on Sundays if there are tests to be carried out and analyzed; OIC; PHCC-1 Yes, we have targets for the numbers of patient we expect for different services. For immunization, our target this month is 18 children and for pregnant women it is 20. Based on what the LGA gives us as our target numbers, the OIC sets the tar- get for each month. For example, if we decide that OPD tar- get is 300 for the month, then we know that we need to go out more into the community and do more outreaches to achieve the number. OPD target for October is 240; Non- OIC health worker; PHCC-4 Involvement/ OIC involves and con- ͱ ͱ ͱ ͱ There is no secrecy at the centre. All activities are open and transparency sults with staff in key clear to all staff so that in my absence any other staff can decisions, and ensure give a detailed report concerning our operations accurately; transparency OIC; PHCC-2 Flat and open OIC open to listen to ͱ ͱ ͱ Our ability to sit/meet and discuss matters when they arise environment feedbacks from work- made us able to collaborate with each other. Through meet- ers and reflect them ing and discussing issues team work is successful. I am on a statewide immunization exercise now, but it is even the OIC that is covering up for me; Non-OIC health worker; PHCC-2 We do it (setting targets) together. I tell them, this month, we should have so-and-so numbers. They sit down and say, yes, it’s true. How do we achieve the numbers? One person may say they’ll go to this village, another will say they’ll go to that village. And then, they know the tricks they use to bring peo- ple in; OIC; PHCC-4 Training and OIC creates fair and fre- ͱ ͱ ͱ ͱ PHC staff who were opportuned to attend other external train- coaching quent training and ings are mandated to carry out a step down training for other coaching opportunities staff and to make any relevant educational materials to staff acquired during the course of such trainings available in the PHCC; WDC Chairperson; PHCC-1 Reward/gift/assis- OIC provides personal ͱ ͱ ͱ ͱ It’s the way I endear myself to them. When I’m coming here, I tance to staff gifts, appreciation, buy bread and other things - from my own pocket; I don’t let cash to appreciate and/ them feel anything. I tell them our money has not yet been or support staff paid, but encourage them to keep working because it will become our burden if we don’t do it. That lifts their spirits; OIC; PHCC-4 Family-like OIC builds family-like ͱ ͱ I bring raw food materials with me when I’m coming to the relationship relations, e.g., by cook- PHCC, and all the staff here cook and eat from the same ing and eating together pot. No one knows who owns what. But in (other PHCC), every staff member brings their own pot and cooks sepa- rately; OIC; PHCC-4 Bonus re- OIC revise bonus alloca- ͱ ͱ So we rather amended the formula and raised the indices allocation tion formula set by the thereby encouraging the staff. This solution was arrived at project to benefit staff during the HF PBF meetings were we discussed and reviewed more than OIC the indices to encourage the staff to work more; OIC; PHCC- 2 Last year, for instance, when I did a lot of extra time, they gave me extra money. My boss got only a little more money than me. I was very happy; Non-OIC health worker; PHCC-3 56 Health Policy and Planning, 2018, Vol. 33, No. 1 Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 Figure 1. Pathways to improve the use of essential health services at the PHCCs under PBF leaders in high performing PHCCs. Health centre management prac- pathways through which these drivers of performance influence tices also mediated the impact of PBF bonuses on further perform- each other to improve performance. Understanding how each driver ance improvement, as how the PBF bonuses were used for health of performance can influence other drivers to improve performance centre management activities differentiated performance. The case will help policy makers and programs decide what to prioritize and studies also showed that community leaders can not only encourage where to start to improve health centre performance in developing and sometimes enforce the use of PHCCs, but also greatly influence countries. contextual factors by regulating unlicensed providers and reducing The findings of this research need to be viewed in the particular competition, and advising PHCCs on necessary actions to attract context of rural and peri-urban Nigeria and the PBF scheme. For patients. The case of PHCC-8 also shows that strong traditional example, the influence of traditional leaders differs by area, and leadership, when influenced by local politics such as personal rela- other contextual factors such as cultural barriers may play a larger tionships of community leaders with unlicensed providers, can nega- or smaller role in other places in Nigeria or in other countries. More tively affect competition and PHCC performance. importantly, as shown in Figure 1, the presence of the PBF scheme is important in driving outcomes. For example, under PBF, greater use Value and transferability of findings of essential health services by the community leads to a larger PBF This study intended to provide insights on what has been viewed as bonus for the PHCC to use for further improving PHCC perform- a ‘black box’ in past literature on the determinants of primary health ance creating a virtuous cycle, whereas in contexts without PBF, an facility performance, health facility management, and the PBF increase in service utilization may demotivate staff and deplete sup- approach in developing countries. What we found is consistent with plies thus undermining service quality (Gilson and Mclntyre 2005). findings from previous studies, while adding new insights. First, our Although PBF is implemented in many developing countries, if it is findings support the importance of involving local authorities and not typically the standard arrangement. While some health systems communities and adapting approaches to the local situation that may mimic certain aspects of PBF, for example by providing greater Dieleman et al. (2009) identified. In addition, we found that proac- autonomy to facility managers, the result may well be different in tive engagement by PHCCs to recruit patients, and community lead- contexts where there are no direct rewards based on performance. ers’ support to encourage PHCC use and regulate unauthorized The importance of health centre management and the OIC’s man- providers thus reducing competition, to be critical. agement capacity will be larger under PBF than in settings where Second, we found a system of accountability that Topp et al. PHCCs do not receive any operational funds and have limited (2015) identified, and various measures to improve staff motivation autonomy. and team work highlighted by Dieleman et al. (2009), Rowe et al. (2005) and Marchal et al. (2010). These performance and staff man- agement activities were interlinked and mutually reinforcing as Limitation of the study strong staff awareness of plans and targets motivated staff, and This study has a few limitations which imply a need for further motivated collaborative teams appear to improve performance man- research. First, although longitudinal data on performance is avail- agement and community engagement activities. Third, in addition to able, this study took a snapshot of PHCC activities. Further the above drivers of performance, we described the various research with a longitudinal study design is needed to acquire a Health Policy and Planning, 2018, Vol. 33, No. 1 57 more in-depth understanding of the dynamics of PHCC improve- the target PHCCs, including the ones in very remote areas with very poor ment. Second, the analysis of demand-side factors (e.g. how com- road access. munity members see PHCCs and make decisions on their use) was limited in this study, though interviews of health workers, WDC Funding chairpersons and observations at the PHCCs allowed us to incor- porate some dynamic interaction between PHCC’s community The health results innovation trust fund (HRITF) managed by the World Bank provided generous funding to this research. engagement approaches and the community’s reactions. Third, health systems factors were not important differentiators in this Conflict of interest statement. None declared. study as it was carried out in PHCCs with similar health systems Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 conditions. However, the potential importance of health systems Notes factors should not be underestimated. Further case studies across 1. Total PBF earnings is calculated by adding quantity- more diverse LGAs and states would cast light on how health sys- based earnings for all of the selected 21 essential health tem factors at these levels support high and low performers. Finally, as explained above, the findings need to be viewed in the services (see details in Supplementary Appendix S1) and context of PBF as an enabling environment for good management quality-based earnings based on quality assessment practices to produce results. scores. The quantity-based earnings for each service is calculated by multiplying the number of the service delivered per month by unit PBF fees for the service Conclusion that is defined by the government of Nigeria based on This study has several important policy implications. First, it sug- the importance of each service. Quantity of services is gests the importance of a platform that provides autonomy to pri- verified by teams of State Primary Health Care mary health centres. The examples of high-performing PHCCs in Development Agency (SPHCDA) staff and technical assis- Nigeria provide a clear picture of how primary health centres can tance agency staff. improve their performance with sufficient levels of autonomy and 2. OIC is a health centre manager position that all PHCCs support. It should be noted that the performance of high-performing PHCCs was equally very low and the difference with low- have in Nigeria. They are also clinical staff, and typically performers was negligible before the PBF scheme. In this regard, the community health extension workers, particularly in rural key drivers and pathways for performance improvement of PHCCs areas. identified in this study would be valid only in the context of PBF or 3. The reviewed literature include: Baldridge performance similar schemes. Second, this research highlighted a need to develop excellence program, 2011; Karsten (2010), Kramer et al. and/or identify OICs with strong management capacity. The case (2007), Management Sciences for Health (1998), McCarthy studies described how OICs can influence contextual and health sys- and Fitzpatrick (2009), NHS Institute for Innovation and tem factors, gain community leaders’ support, identify issues and Improvement and Academy of Medical Royal Colleges solve them by involving key stakeholders, and motivate staff—or (2010), Office for Health Management (2004), Omoike alternatively fail to do so. They also highlighted the particular et al. (2011), Pillay (2010), Schmalenberg and Kramer importance of performance management, staff management and (2009), Sherman et al. (2007), Squires (2001), Zori et al. motivation, and community and community leader engagement. (2010); and literature on management practices scorecard Clear differences in practices of OICs between high and low- performing PHCCs suggest that these differences cannot be including: Dorgan et al. (2010), Bloom and Van Reenen addressed through short-term training interventions alone—these (2010); and McConnell et al. (2013), Meyer and Collier differences are related to the OIC’s mindset, managerial skills, and (2001), Ohman-Stickland et al. (2007). fit with a managerial position. This underlines the need for a set of 4. WDC consists of various types of community leaders that interventions to select capable and motivated OICs, provide long- reviews PHCC performance through regular meeting, sup- term coaching of managerial skills, and motivate OICs to improve port the PHCC and sign-off the use of performance based their practices. Third, this research highlights the importance of funds by the PHCC. looking at community leaders as a main driver to improve PHCC 5. In Figure 1, community engagement is included in ‘Health performance. Interventions targeted at PHCCs need to have a core Center Management’ rather than in ‘Community engage- focus on gaining support from community leaders as part of an inte- ment and support’ as categorized in Tables 3 and 4. This gral approach to improve performance. is because community engagement refers to the manage- ment activities by PHCCs, and pathways can be described Supplementary Data more clearly in this way than in the way that separates Supplementary data are available at HEAPOL online. community engagement from health centre management. Acknowledgements The authors wish to thank the following colleagues who took time to advise References authors and discuss the topic with authors: Dinesh Nair, Asha George, David Bloom N, Propper C, Seiler S, Van Reenen J. (2009). Management Practice Holtgrave, Leiyu Shi, Kunle Alonge, Benjamin Loevinsohn, Ayodeji Oluwole and Productivity in Hospitals (Presentation). EDS Innovation Center. Odutolu, Gyorgy Fritsche. Temitope Olukowi, Obiageli Chiezey Onwusaka, Bloom N, Van Reenen J. 2010. Why do management practices differ across Abdullahi Balarabe Sallau carried out interviews and other data collection in firms and countries? Journal of Economic Perspectives 24: 203–24. 58 Health Policy and Planning, 2018, Vol. 33, No. 1 Bradley EH, Curry LA, Ramanadhan S, Rowe L, Nembhard IM, Krumholz NHS Institute for Innovation and Improvement and Academy of Medical HM. 2009. Research in action: using positive deviance to improve quality of Royal Colleges. 2010. Medical Leadership Competency Framework. health care. Implementation Science 4: 25. London: NHS Institute for Innovation and Improvement and Academy of Creswell JW, Plan Clark VL (Eds.). 2010. Designing and Conducting Mixed Medical Royal Colleges. Methods Research, 2nd ed. SAGE. Office for Health Management. 2004. Management Competency User Pack: Davies AR, Ware JE. 1988. Involving consumers in quality of care assessment. For Nurse and Midwife Managers. Office for Health Management. Health Affairs 7: 33–48. Ohman-Strickland PA, Orzano AJ, Nutting PA et al. 2007. Measuring organiza- Davies HT, Nutley SM. 2000. Developing learning organisations in the new tional attributes of primary care practices. Health Services Research, 42: 1257. NHS. British Medical Journal 320: 998–1001. Omoike O, Stratton KM, Brooks BA, Ohlson S, Storfjell JL. 2011. Advancing Dieleman M, Gerretsen B, Jan van der Wilt G. 2009. Human resource manage- nursing leadership—a model for program implementation and measure- Downloaded from https://academic.oup.com/heapol/article-abstract/33/1/41/4565567 by World Bank Publications user on 12 March 2019 ment interventions to improve health workers’ performance in low and middle ment. Nurse Administration 35: 323–32. income countries: a realist review. Health Research Policy and Systems 7: 7. Oxford Policy Management. 2011. Political Economy and Institutional Assessment Donaldson M, O’Connor, D. Bishop D. (2000). Exploring Innovation and for Results Based Financing for Health. Oxford Policy Management. Quality Improvement in Health Care Micro-Systems: A Cross-Case Patton MQ. 1990. Qualitative Evaluation and Research Methods. Beverly Analysis. Washington, DC: Institute of Medicine, National Academy Press. Hills, CA: Sage. Dorgan S, Layton D, Bloom N et al. 2010. Management in Healthcare: Why Pillay R. 2010. The skills gap in nursing management in South Africa: a sec- Good Practice Really Matters. McKinsey & Company. toral analysis: a research paper. Journal of Nursing Management 18: Ferlie E, Shortell S. 2001. Improving the quality of health care in the United 134–44. Kingdom and the united states: a framework for change. The Milbank Renmans D, Holvoet N, Garimoi Orach C, Criel B. 2016. Opening the 0 black Quarterly 79: 281–315. box0 of performance based financing in low- and lower middle-income Fritsche GB, Soeters R, Meessen B. 2014. Performance-Based Financing countries: a review of literature. Health Policy and Planning Toolkit. World Bank Training. Washington, DC: World Bank. https://open Rowe AK, Savigny DD, Lanata CF, Victora CG. 2005. How can we achieve knowledge.worldbank.org/handle/10986/17194 License: CC BY 3.0 IGO. and maintain high-quality performance of health workers in low-resource Accessed 5 October 2017. settings? Lancet 366: 1026–35. Gilson L, McIntyre D. 2005. Removing user fees for primary care in Africa: Schmalenberg C, Kramer M. 2009. Nurse manager support: how do staff the need for careful action. BMJ 331: 762–5. nurses define it? Critical Care Nurse 29: 61–9. Gormley W, Mccaffery J, Quain E. 2011. Moving forward on human resour- Schneider H, Blaauw D, Gilson L, Chabikuli N, Goudge J. 2006. Health sys- ces for health: next steps for scaling up toward universal access to tems and access to antiretroviral drugs for HIV in southern Africa: service HIV/AIDS prevention, treatment, and care. Journal of Acquire Immune delivery and human resources challenges. Reproductive Health Matters 14: Deficiency Syndrome 57: 113–5. 12–23. Health Results Innovation Trust Fund. 2015. Performance-Based Financing Sherman RO, Eggenberger T, Bishop M, Karden R. 2007. Development of a Conceptual Framework. https://www.rbfhealth.org/resource/performance- leadership competency model. The Journal of Nursing Administration 37: based-financing-conceptual-framework. Accessed 5 October 2017 85–94. Kandpal E. 2016. Completed Impact Evaluations and Emerging Lessons from the Shortell SM. 1985. High-performing healthcare organizations: guidelines for the Health Results Innovation Trust Fund Learning Portfolio. The World Bank. pursuit of excellence. Hospital & Health Services Administration 30: 7–35. Karsten MA. 2010. Coaching: an effect live leadership intervention. Nursing Shortell SM, LoGerfo JP. 1981. Hospital medical staff organization and qual- Clinics of North America 45: 39–48. ity of care: results for myocardial infarction and appendectomy. Medical Komatsu R, Low-beer D, Schwartlander B. 2008. Towards universal access: Care 19: 1041–55. scaling up the priority HIV/AIDS interventions in the health sector. Progress Shortell SM, O’Brien JL, Hughes EF et al. 1994. Assessing the progress of report global fund-supported programmes contribution to international tar- TQM in US Hospitals: findings from Two Studies. Quality Letter for gets and the millennium development goals: an initial analysis. Bulletin of Healthcare Leaders 6: 14–7. World Health Organization 85: 805–11. Shortell SM, Zimmerman JE, Rousseau DM, et al. 1994. The performance of Kramer M, Maguire P, Schmalenberg C et al. 2007. Nurse manager sup- intensive care units: does good management make a difference? Medical port—what is it? Structures and practices that promote it. Nurse Care 32: 508–25. Administration Quarterly 31: 325–40. Shortell SM, Bennett CL, Byck GR. 1998. Assessing the impact of continuous Management Sciences for Health. 2004. Management and Organizational quality improvement on clinical practice: what it will take to accelerate Sustainability Tool (MOST). Boston progress. Milbank Quarterly 76: 593–624. Marchal B, Dedzo M, Kegels G. 2010. A realist evaluation of the management Squires A. 2001. Leadership development for the new manager in the small, of a well-performing regional hospital in Ghana. BMC Health Services acute care facility. Journal of Nursing Administration 31: 561–4. Research 10: 24. Topp SM, Chipukuma JM, Hanefeld J. (2015). Understanding the dynamic McCarthy G, Fitzpatrick JJ. 2009. Development of a competency framework interactions driving Zambian health centre performance: a case-based for nurse managers in Ireland. The Journal of Continuing Education in health systems analysis. Health Policy and Planning 30: 485–99. Nursing 40: 346. World Health Organization. 2007. Towards better leadership and management McConnell KJ, Lindrooth RC, Wholey DR, Maddox TM, Bloom N. 2013. in health: report on an international consultation on strengthening leadership Management practices and the quality of care in cardiac units. JAMA and management in low-income countries. WHO/HSS/healthsystems/2007.3. Internal Medicine 173: 684–92. Working Paper No. 10. Accra, Ghana. Meyer SM, Collier DA. 2001. An empirical test of the causal relationships in World Bank. 2013. Results-based financing for health. http://www.rbfhealth. the Baldrige health care pilot criteria. Journal of Operations Management org/project/our-projects. Accessed 15 December 2016. 19: 403–25. World Bank. 2014. Service delivery indicator survey in Nigeria. Mitchell PH, Shortell SM. 1997. Adverse outcomes and variations in organiza- Yin RK. 2009. Case Study Research: Design and Methods. SAGE. tion of care delivery. Medical Care 35: NS19–32. Zori S, Nosek LJ, Musil CM. 2010. Critical thinking of nurse managers related National Primary Health Care Development Agency. 2016. PBF portal. http:// to staff RNs’ perceptions of the practice environment. Journal of Nursing nPHCCda.thenewtechs.com/. Accessed 15 December 2016. Scholarship 42: 305–13.