Organisation of primary health care systems in low- and middle-income countries: review of evidence on what works and why in the Asia-Pacific region

Patient and public involvement

Patients were not involved in the development of this literature review.

Review question

The evidence review sought to address the following questions:

1. What is the existing evidence on approaches to the organisation of PHC in the Asia-Pacific region?

2. What evidence is there of the impact or effectiveness of each of the identified approaches? What are the pathways underpinning these outcomes?

3. What are the gaps in knowledge of ‘what works’ in PHC organisation and financing in LMICs of the Asia-Pacific region?

This paper aims to answer the first two questions. A companion paper presents the results in relation to the third question, that is, gaps in evidence on the organisation of PHC service delivery, while a third paper used the same set of questions to guide a review of PHC financing.11 12

Review methods

Online supplementary file 1 presents the conceptual framework we developed to guide the literature search. We searched peer-reviewed publications and grey literature from January 2008 to April 2018; this period was chosen to ensure evidence of intervention effectiveness was recent and therefore relevant to present-day health systems. We then used a narrative approach to summarise findings, generating new insights by systematically comparing, contrasting and extrapolating emerging themes.13 14 Characteristics of successful interventions and their enabling factors were tabulated.

To complement and help contextualise the review findings, we collated a summary of PHC system characteristics in a subset of LMICs in the Asia Pacific. We selected countries for which a ‘Health in Transition’ (HiT) country profile was available (n=14) from the Asia Pacific Observatory on Health Systems and Policies, a WHO-hosted partnership; HiTs provide recent, comparable data on PHC systems.15 Supplementary information for these countries was collected via a Google search that yielded primary and secondary sources including national/local government websites, journal articles and newspaper articles. Additional sources of indicator data were identified from backwards citation chaining from government websites.

Setting and inclusion/exclusion criteria

We defined LMICs of the Asia-Pacific region as those countries within the World Bank regions of South Asia and East Asia and Pacific that had an economic classification of low, lower-middle or upper-middle income (as designated in June 2017).16 We did not adopt any specific definition of ‘primary health care’ services but, consistent with the Alma-Ata declaration, our study encompassed facility-based care at lower levels of the health system as well as preventive and promotive services delivered by community health workers (CHWs) and/or formal sector employees. Inclusion and exclusion criteria are found in box 1. Notably, we searched for intervention studies that, by their nature, often target a specific aspect of the primary care system or a specific set of clinical services. However, our analysis of these studies took a broad, systems perspective consistent with the recognised core principles of PHC.

Search strategy

Standard database searches were conducted using MEDLINE, Embase, Global Health and Cochrane Database. The search strategy is provided in online supplementary file 2. Database searches were supplemented with a hand search of grey literature from the following websites: WHO-Geneva; WHO-SEARO; WHO-WPRO; World Bank (filtered for East Asia and Pacific, and South Asia); Asian Development Bank; Alliance on Health Policy and Systems Research; Asia Pacific Observatory on Health Systems and Policies; Centre for Global Development; Results for Development; Institute for Development Studies (UK); Secretariat for the Pacific Community; Save the Children Fund, International; World Vision, International.

Analysis

The evidence review and synthesis was conducted in two stages between February and April 2018. In stage 1, the title and abstract of initial search results were screened and assessed for relevance by three members of the research team (RD, SJ, AP), removing duplicates and publications that did not meet the inclusion criteria. The full text of potentially relevant publications was then retrieved and reviewed for inclusion by two researchers (RD, AP). Reference lists of the final included publications were hand-searched for additional articles relevant to the review, and further relevant articles were identified through consultation with other authors engaged in the global PHC research consortium which produced the studies published in this edition of BMJ Global Health. Figure 1 presents the flow chart of literature selection.

In stage 2, each full-text publication was reviewed to identify key study characteristics (context, geographical scope, research method, study population, disease focus), as well as the PHC system outcomes and inputs specifically addressed by the research. Our first attempt to map literature against categories set out in the conceptual framework found it to be too detailed, with notable overlap between categories and omission of key elements, including outcome measures. We therefore developed a new framework against which to map the results of our literature search with both ‘inputs’ and ‘outcomes’ (see online supplementary file 3). Inputs were derived from the conceptual framework, with some re-organisation and re-grouping of categories, while ‘outcomes’ were identified as those health system attributes underpinning the principles of Universal Health Coverage (UHC)—quality, coverage, efficiency, responsiveness and equity. Four of these five outcome areas are common to both World Bank and WHO frameworks on health systems performance, while the fifth, equity, is a focus of health systems analysis in the era of UHC.17–19 Other domains commonly referenced as markers of health systems performance are safety and access; we assumed these to be incorporated in definitions of quality and coverage, respectively. We also used the revised framework to guide the identification of major gaps in the literature (see companion paper).11

PHC system characteristics

Online supplementary file 4 presents PHC system characteristics in 14 LMICs of the Asia Pacific. It highlights high variation in the level of economic development between countries, resulting in wide variations in health expenditure: from US$53 per capita in Laos to US$425 in China. Despite this, there are many shared organisational features across PHC systems. Notably, every country has a network of village or community health workers, though with various degrees of training and differing remuneration arrangements. Further, in at least five countries, we found no information on whether doctors were available outside the hospital context, for example, practising privately in remote locations. In a further three countries, we only found reports of doctors practising in urban settings or second-level facilities. This suggests non-physician health workers (NPHWs) of varying skill levels including nurses and midwives, and CHWs will continue to have a key role in delivering essential services, even in the better-resourced health systems of middle-income economies.

Just over half the countries reported some kind of gatekeeping policy (n=8) which requires patients to access non-emergency hospital care or specialist services via PHC.20 However, we know from other studies that enforcement of gatekeeping arrangements is often difficult in practice, leading to inefficient use of hospital resources.21

Eight countries used non-state providers (private or not-for-profit) to deliver PHC in some regions, often in rural areas. This suggests increasing practical experience in government of working in partnership with non-state providers, for example, on how to structure contracts; however, it also points to a level of fragmentation in PHC systems, with a mix of NGOs, state services and private providers all providing services. Nine countries reported some level of fee-for-service at primary care level including seven of the eight countries using non-state providers. However, five of these nine countries also have fee exemptions for poor people in either state-run and or state-contracted services, suggesting that most countries see value in removing financial barriers to access.

Literature synthesis

A total of 111 articles were included in our synthesis (figure 1) (21 reviews (9 Cochrane, 5 other systematic and 7 non-systematic reviews), 12 grey literature publications and 78 original research). A complete list of included articles is provided in online supplementary file 5. Characteristics of successful interventions and their enabling factors are summarised in tables 1–5; each table relates to one of the five outcome domains identified in our analytical framework (online supplementary file 3). Interventions that targeted multiple outcomes are referenced in more than one table. Although we intended to look at unsuccessful or ineffective interventions, there were too few published articles to reliably provide commentary on this.

Institutionalise and sustain community engagement

Strong community engagement and participation were integral to early conceptualisations of PHC, including ‘health for all’ as set out in the Alma-Ata declaration and reiterated in the recent Astana declaration.3 81 82 Although the studies we reviewed indicate that community engagement is important, processes to enhance engagement were not well explored. Engagement needs to be conceptualised broadly, to encompass the involvement of communities in the planning, design, governance and delivery of services.83 Processes that stress the importance of partnership over consultation are required.84 This means going beyond one-off exercises to gather information on community needs and priorities—though this is important—to sustained engagement resulting in the transfer of power from health system to communities, allowing communities a role in agenda-setting and actively engaging community members in decision-making.83 84 A stronger evidence base would help inform the tailoring of different approaches to community engagement in different contexts, as well as how to monitor and measure the effectiveness of these efforts.

Develop more robust NPHW workforce strategies

‘Non-physicians’ are a broad group, encompassing CHWs with just a few weeks’ training through to experienced nurses, midwives and other ancillary professions. Strengthening the skills and capacities of this highly varied workforce is nonetheless a common focus of the PHC literature given they are on the frontline of service delivery, often working without supervision. Studies we reviewed suggest training and support provided to NPHWs (including lay and CHWs) is typically one-off, ad hoc and focused on building skills in a particular area. Recent studies in high-income countries suggest efforts to strengthen primary care, including through task-shifting, are not well aligned with workforce development plans,85 86 which typically prioritise medical and specialist workforce development. It is also likely that many LMICs do not pay sufficient attention to medium-term and long-term needs of their NPHW, or strategically link the development of this workforce to evolving models of PHC and in particular the increasing focus on chronic care.6 Comprehensive and differentiated workforce development strategies for NPWHs are needed, which recognise that CHWs will require very different approaches to formal sector workers such as nurses and midwives, but also that all groups require a clear career pathway and remuneration framework that considers both financial and non-financial rewards. In addition to training, these strategies should consider how to build ‘team’ approaches and strengthen supportive supervision. Researchers can play a key role in providing the evidence to support the implementation of these strategies.

Operationalise an integrated PHC service package

Increasingly, countries are defining a ‘package’ of services linked to the achievement of UHC.87 Brokering agreements on an entire package of services covered under UHC reforms is a complex process that ideally involves the generation of locally relevant evidence of cost-effectiveness. Against this background, interventions delivered through PHC as opposed to hospital setting can be highly cost-effective and acceptable to the community.1 79 88 89 Until recently, the focus has been on acute care and maternal and child health services, but the importance of integrating NCD prevention and control into the PHC model of care is increasingly recognised.7 79 90 Disease-agnostic, integrated service delivery models are therefore a priority. However, we found limited evidence of comprehensive systems-based approaches to operationalising an integrated package. This means consideration of, for example, staff roles, skills acquisition, supervision and workload, medicines and equipment supply, establishing appropriate care pathways for referral and follow-up, data collection and monitoring, and cost considerations including stimulus funding to transform models of care. Rather, studies focused on a single aspect of the package, such as introducing a new NCD service. A more systems-based approach may be underway in countries, but our review suggests any learnings are not being shared through the academic literature. Researchers have an important role in capturing and disseminating such knowledge.

Build managerial capacity within PHC systems

Good governance—encompassing accountability to local communities, availability of information on provider performance, robust financial management and procedures to report misuse of resources—is critical to effective service delivery but remains a chronically under-studied element of health systems development and financing.91–93 Good health sector management practice is an important aspect of governance. It ensures facilities and the equipment within them are well maintained, medicines are ordered on time and properly stored, outreach activities are integrated and executed effectively, and so on.91 Strong managerial capacity is also key to effective contracting of non-state providers to deliver services. Despite this, we found little evidence on how to build managerial systems and capacities to strengthen PHC organisation—or indeed on how ‘good management’ is defined—suggesting that these functions are neglected at PHC level. Arguably, managerial capacity is most important at PHC level given scarcity of resources and, in remote areas, distance from administrative support in capitals.94 More evidence is needed on how to improve capabilities such as planning, budgeting, target setting and monitoring, and these capabilities need to be prioritised in PHC strengthening efforts. Better management will in turn strengthen accountability—as clear target setting and measurement underpins performance monitoring.92

Modernise and integrate information systems for PHC

Information systems can play a key role supporting the organisation and delivery of PHC services, through assessing the health needs of populations, guiding the planning and implementation of health interventions, and facilitating coordination of care.8 95 Yet, in many LMIC contexts, systems for data collection, analysis and dissemination are weak and fragmented, and cannot support even basic monitoring of vital statistics and health trends.96 The risk of relying on technological ‘fixes’ to systemic problems is long recognised. Over 25 years ago, Sandiford et al noted the potential of new means of generating health data to guide PHC reform, but warned this would not be realised without greater attention to how health information is used and by whom.97 This observation remains valid today: while we saw promising evidence of the potential of digital health strategies to support PHC systems—for example, through data collection, processing and real-time data analysis—the ‘human elements’ required to support adoption were often neglected. Most interventions did not recognise the context of relatively weak information systems nor the broader challenges of establishing appropriate data governance standards. Long-standing challenges of ensuring data informs decision-making remain, while new challenges of maintaining confidentiality while promoting transparency and linking data from multiple sources have emerged.98 A PHC ‘ecosystem’ approach is therefore needed.

Limitations

The majority of studies reviewed were small in scale and limited in scope, making it difficult to generalise findings. Nevertheless, cross-country analysis of country experience is a common and accepted approach to generating cross-cutting lessons, including for PHC.99 Although we reviewed many literature sources, it is possible that the search strategy missed some relevant literature. We focused on intervention studies with the objective of gathering evidence of ‘what works’, and as a result we did not consider narrative reviews or policy analyses which may provide insight on how the broader structure, financing and governance of the health system affects the quality of PHC services. Given the highly dynamic and variable levels of financing of health systems in the region, this will be a particularly important consideration when assessing the suitability of certain interventions. We conducted an extensive search of the grey literature but could not access internal government evaluations of PHC service delivery models given these are not publically available. Due to the high variation in studies and literature sources, we did not conduct a formal appraisal of evidence quality. Finally, the literature remains dominated by evidence from larger countries such as India and China; complex health systems analyses remain relatively rare and there are few systematic reviews or cross-country comparisons to draw out best practices.