Characteristics of included studies
After title, abstract and full text screening, 50 studies met the inclusion criteria (PRISMA flowchart provided in Annex 3). Included studies were published between 2005 and 2020. There was a steady increase in the number of publications over the years, with a peak in 2015 (n = 8) (Fig. 1).
The majority of studies (n = 36) focused on health financing reforms (e.g., social or national health insurance), and the remaining (n = 14) on UHC in general. As per our inclusion criteria, all included studies were in LMICs (based on the World Bank classification [40]). The largest number of studies focused on Ghana (n = 6), Mexico (n = 5), India (n = 5) and Thailand (n = 4) (Fig. 2), potentially reflecting major successful UHC reforms that have taken place in those countries. Indeed, research focus often seemed to be related to a policy or reform that had been successfully introduced or implemented, and very few papers looked at unsuccessful or failed reforms (e.g., [41]).
Approaches to PEA and study methods
Papers included in the review adopted a variety of PEA and policy analysis approaches. While specific approaches and frameworks used differed, all aimed to understand the contextual factors and the dynamics among actors, ideas and institutions and the contestation of power and resources, with prominence given to some or all these elements. All studies included an empirical component (which was one of the inclusion criteria), so that many adopted a case study or a comparative case study design, using qualitative or mixed-methods approaches.
Most studies (n = 48, 94%) adopted a retrospective approach, looking back at policy processes that had happened in the past and had reached a conclusion. Only three papers carried out a prospective analysis. They used this approach to understand bottlenecks and facilitators to achieve UHC and to propose ways to improve the efficiency of health financing. One of the papers drew on a panel discussion with relevant stakeholders (e.g., Ministry of Health and Finance, WHO representatives) [42]; the other two studies aimed to revise a resource allocation formula and its implementation [43] and to predict the role played by BRICS [Brazil, Russia, India, China, South Africa] countries in the UHC movement [44].
Stakeholder engagement in the research process
Researchers as stakeholders
In an attempt to reflect on the role of researchers as stakeholders, and recognizing that the design and intention of a study might depend on researchers' values, ideologies, positionality, affiliation, role, and so on, we looked at the affiliation of the authors (first author only) in relation to the study setting. Our findings show that 54% (n = 27) of first authors conducted the studies in a country different from their affiliation (of which 85% were affiliated with HIC institutions and studied a LMIC; the remaining were based in a LMIC institution and studied another LMIC), while for the rest of the first authors (46%, n = 23) the country under study corresponded to their country’s affiliation. Regarding the survey, out of 17 responses received, 11 authors were based in HIC institutions and led studies in LMICs. Five LMIC-affiliated authors conducted studies in their country of affiliation and one author in a different LMIC country. From this analysis it appears that more than a third (40%) of first authors and survey respondents conducted work in the same country as their affiliation. However, we do not have enough information on the authors’ backgrounds to fully unpack their position as stakeholders in the UHC arena. As a consequence, in the analysis below, we refer only to non-researcher stakeholders in our considerations. We further reflect on this point in the “Discussion” section.
Research funding
Funding was clearly reported for 30 papers (58%), but it was not detailed whether the funders were part of the political economy landscape being studied. Funding for the research seems to commonly originate from external funders, such as bilateral and multilateral development agencies (World Bank, United Kingdom’s Department for International Development [DFID], WHO), foundations (Rockefeller Foundation) and research funding organizations (International Development Research Centre [IDRC], Overseas Development Institute [ODI], European Union’s Seventh Framework Programme). Only a few were funded by local research centres or universities. There is no clear evidence of whether the research funders might have influenced the research processes and findings.
Research design
Engagement of stakeholders at the stage of research design is rarely reported in the literature we reviewed. Indeed, only one study, looking at the process of establishing the national health insurance scheme in Uganda, reported stakeholder engagement at the research design stage [45]. It was mentioned that the study was part of the work programme of the Ugandan health sector, defined and approved by the government, donors and all stakeholders, as indicated in the second National Health Policy and Health Sector Strategic and Investment plans.
However, the published literature may provide an underestimation of the stakeholders’ involvement at the design stage, as we obtained further information from the authors’ survey. Five respondents mentioned that either stakeholders were engaged in the design process by providing the conceptual framework for the study, or they initiated the research process by demanding specific evidence. In other cases, policy-makers were part of the research team and contributed to the research design, or the researchers were also policy-makers. Several authors in the studies reviewed acknowledged their role and participation in the policy reform of the country or being closely connected with the policy-making process.
Data collection
Several papers (n = 22, 43%) did not include any stakeholders in the data collection process (i.e., only document review was conducted). The remaining (n = 29, 57%) involved stakeholders, mainly as key informants being interviewed and/or sharing relevant documentation. We could not find evidence of stakeholders who were driving or leading the process of data collection.
The average number of stakeholders interviewed was 20. The actors involved in data collection were mostly government representatives or ministries (e.g., ministry of health, finance) and health workers/health system officials (Fig. 3). The actors least involved were civil society, researchers and the private sector, such as insurance companies.
Data analysis
In terms of data analysis, reporting in the published literature might be somewhat underestimating the role of stakeholders for the interpretation and analysis of the data. Three papers reported the involvement of stakeholders for the validation of findings. For instance, with reference to social health insurance in Nigeria, Onoka et al. [46] mentioned that they conducted a post-study workshop where they presented the preliminary findings to study participants in order to discuss and validate the analysis and reach consensus on the results of the analysis. Similarly, Tangcharoensathien et al. [47] involved policy actors, civil society representatives and academia in the data analysis for their study in Thailand, while Musango et al. [42] based their study on a joint panel discussion with ministries, WHO Regional Office for Africa and others who participated in the meeting. In addition to this (unreported in the respective published documents), a few (n = 3) authors confirmed in the survey that they had shared preliminary findings with stakeholders, who were able to provide feedback.
Knowledge translation: communication and dissemination of findings, and monitoring and evaluation of research uptake
Two studies, by Musango et al. and Twea et al., reported how findings were shared and disseminated [42, 43]. In these studies, the dissemination was part of the discussions that took part as part of the research process. However, as for the elements above, these findings are likely to reflect an underreporting, potentially due to limited space in published papers or the information not being considered a priority for academic publications. The authors’ survey brought more information to our attention. For instance, several authors (n = 11) reported that they disseminated results via workshops with policy-makers or health workers, via meetings (e.g., with parliament committee, municipal officers, panels) and via conferences on health financing. They also acknowledged that essential factors for successful policy uptake are early engagement with stakeholders, coproduction of research questions, and local partners establishing key linkages with stakeholders.
However, follow-up and monitoring of the usage of research findings and actual research uptake appear to be somewhat limited, even allowing for a time lag between publication and research uptake. Information based on the survey shows that some of the authors (n = 7) were aware that findings were being used, for example by development partners (World Bank) or local parliaments, or they mentioned that the authors themselves held a position that allowed for uptake of research findings and recommendations.
Recommendations included in articles reviewed
Following the point above, it seems important to analyse the types of recommendations that were made in the studies reviewed—including looking into whether they were methodological or policy recommendations and whether they pointed to general lessons learned versus context-specific ones.
Almost all papers (n = 43, 84%) included some type of recommendations grounded in the analysis conducted. Recommendations were made on research, such as operational or methodological aspects, as well as on policy- or governance-related aspects, or both types. Policy recommendations represented the majority of those included (n = 35, 68%). Overall, the approach was general (rather than context-specific), referring for example to the need to strengthen political commitment and leadership, to adopt multisectoral approaches, or to improve resource allocation and integration of financing mechanisms to achieve UHC. Other recommendations, at the intersection of PEA methods and policy, underscored the importance of taking into account interests, ideologies and power of politically and economically influential groups, the contestation of ideas and consensus among social groups, and the need to consider contextual features, such as fragmentation of health systems and the implications for UHC, among others. Only a few papers (n = 11, 21%) included context-specific recommendations.
Recommendations were mostly aimed at policy-makers and researchers adopting PEA approaches, with only a few papers explicitly targeting donors and international organizations.