Implementation science in resource-poor countries and communities

Background Implementation science in resource-poor countries and communities is arguably more important than implementation science in resource-rich settings, because resource poverty requires novel solutions to ensure that research results are translated into routine practice and benefit the largest possible number of people. Methods We reviewed the role of resources in the extant implementation science frameworks and literature. We analyzed opportunities for implementation science in resource-poor countries and communities, as well as threats to the realization of these opportunities. Results Many of the frameworks that provide theoretical guidance for implementation science view resources as contextual factors that are important to (i) predict the feasibility of implementation of research results in routine practice, (ii) explain implementation success and failure, (iii) adapt novel evidence-based practices to local constraints, and (iv) design the implementation process to account for local constraints. Implementation science for resource-poor settings shifts this view from “resources as context” to “resources as primary research object.” We find a growing body of implementation research aiming to discover and test novel approaches to generate resources for the delivery of evidence-based practice in routine care, including approaches to create higher-skilled health workers—through tele-education and telemedicine, freeing up higher-skilled health workers—through task-shifting and new technologies and models of care, and increasing laboratory capacity through new technologies and the availability of medicines through supply chain innovations. In contrast, only few studies have investigated approaches to change the behavior and utilization of healthcare resources in resource-poor settings. We identify three specific opportunities for implementation science in resource-poor settings. First, intervention and methods innovations thrive under constraints. Second, reverse innovation transferring novel approaches from resource-poor to research-rich settings will gain in importance. Third, policy makers in resource-poor countries tend to be open for close collaboration with scientists in implementation research projects aimed at informing national and local policy. Conclusions Implementation science in resource-poor countries and communities offers important opportunities for future discoveries and reverse innovation. To harness this potential, funders need to strongly support research projects in resource-poor settings, as well as the training of the next generation of implementation scientists working on new ways to create healthcare resources where they lack most and to ensure that those resources are utilized to deliver care that is based on the latest research results.

Many of the physical constraints that impede the routine delivery of effective health interventions to those who can benefit are (by definition) far more severe in resource-poor than in resource-rich countries. For instance, for each citizen, the resource-poor countries of sub-Saharan Africa spend only a fraction of the amount on health that the resource-rich countries of Western Europe spend, and the numbers of doctors and nurses per population are orders of magnitudes lower in Africa than in Europe (Fig. 1). At the same time, amenable mortality-i.e., the mortality that existing effective healthcare technologies could eliminate if they were delivered successfully to all those who can benefit-is far higher in resource-poor countries than in resource-rich ones ( Fig. 1) [1,2]. This "inverse care law" in cross-country comparison-the "availability of good medical care tends to vary inversely with the need for it in the population served" [3]-is of course merely a global version of the classic inverse care law, which operates across communities within both resource-rich and resource-poor countries. In this editorial, we are addressing specific features of implementation science for both resource-poor countries and resource-poor communities, recognizing that scarcity and deprivation affecting the delivery of evidence-based healthcare exist worldwide and across all geographic areas and that there is a continuum from resource poverty to resource wealth in all countries.
An obvious approach to reduce the high levels of amenable mortality in resource-poor countries and communities is to increase the financial resources available for healthcare. This approach, however, requires either substantial economic growth-which may fail to emerge in both resource-poor countries [4] and communities [5]-a redistribution of existing resources across sectors-which is difficult to achieve for obvious political reasons [6]-or external assistance-which cannot be relied on over the long term as donor priorities shift frequently [7,8]. Another approach is to create new resources to deliver effective health interventions given the existing financial constraints. Implementation science can contribute to this approach as the science of the discovery, design, and evaluation of novel approaches to deliver evidence-based healthcare practice.

Creating resources
The goal of implementation science is to discover and test approaches "to promote the systematic uptake of research findings and other evidence-based practices into routine practice, and, hence, to improve the quality and effectiveness of health services" [9]. Many of the frameworks that provide theoretical guidance for implementation science feature resources and physical capacity to deliver evidence-based practice-such as health workers, drugs, supply chains, and healthcare facilities-as part of the context of implementation [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27]. In these frameworks, assessments of the resources context are used to guide analysis or action, e.g., to (i) predict the feasibility of implementation of a novel evidence-based practice [16,25,28], (ii) explain implementation success and failure [11-13, 24, 26, 29], (iii) adapt a novel evidence-based practice to local constraints [15,19,20,23,30], and (iv) design the implementation process to account for local constraints [17,22,30]. As such, in these theoretical frameworks-and in the implementation science for resource-rich settings they have been derived from and guide-resources are viewed as important contextual factors. Implementation science for resource-poor settings shifts this view from "resources as context" to "resources as primary research object" [31]. Table 1 shows examples of implementation science in resource-poor countries and communities testing approaches to expand human resources for health-through tele-education, telemedicine, task-shifting to lower-skilled health workers, task-shifting to clients, new models of care, and technological innovation-and to increase laboratory capacity and supplies. A large body of implementation science in resource-poor countries and communities has focused on creating resources for evidence-based healthcare. This research is likely to continue with vigor because "there need to be minimal human resources, financing, drugs, and supply systems before effective interventions can be delivered" [31]. In particular, research developing and testing community health worker programs [32]-which are widely viewed as one of the few viable solutions to the persistent health worker shortages in many resource-poor countries and communities [33][34][35]-and information and communication technologies-which can provide affordable training and decision support for health workers anywhere-will continue to attract increasing implementation research funding [36][37][38].

Changing behavior
In contrast to research aimed at increasing resources, to date, comparatively few studies in resource-poor settings have investigated approaches to change the behavior and utilization of those resources to ensure that research findings are translated into routine practice. A 2017 "overview of systematic reviews" on "implementation strategies for health systems in low-income countries" published in the Cochrane Database of Systematic Reviews is a case in point [39]. The 18 systematic reviews on different strategies to change health worker behavior in this overview article-education materials [40], internet-based learning [41], educational meetings and workshops [42][43][44][45], educational outreach [46][47][48], local opinion leaders [49], audit and feedback [50], reminders [51], tailored interventions [52], and multi-faceted    interventions [42,47,50,53]-synthesized 820 primary studies. Among these primary studies, which can be viewed as the global knowledge base on strategies to change health worker behavior, only 13 (or 1.6%) took place in a low-income country and only 82 (10.0%) took place in a middle-income country. There is thus strong potential for resource-poor countries to learn from the experiences in resource-rich countries. Clearly, some evidence generated in resource-rich settings is highly relevant for resource-poor settings-if "the implementation strategies considered … address a problem that is important in low-income countries, would be feasible, and would be of interest to decision-makers in low-income countries" [39]. Equally clearly, however, studies systematically investigating the transferability of the large body of evidence on strategies to change health worker behavior generated in resource-rich countries are urgently needed. In addition to the obvious resources gradient, reasons why evidence on effective practice cannot be transferred from resource-rich to resource-poor settings may include important differences in political and institutional factors [54][55][56]. While transfer of evidence from any one to any other context will always need to take account of these factors, there will often be particularly large differences in the answers to questions such as those posed by the "Tailored Implementation for Chronic Diseases Checklist" (TICD Checklist) when considering evidence transfer from resource-rich to resource-poor settings: Do "influential people", "political stability", and "corruption" "facilitate or hinder implementation of necessary changes?" [30]. In many cases, successful implementation of evidence-based practice in resource-poor settings will thus require research to learn how to best adopt strategies that have proven effective in resource-rich settings, as well as the discovery and evaluation of wholly new approaches.

Creativity and reverse innovation
Resource constraints, however, are not only an important object of implementation research in resource-poor countries and communities, but they are also a powerful stimulus for creativity [57]. The psychological and marketing literature shows that creativity thrives when choices are restricted [58][59][60]. It is likely that the severe human and physical resources constraints in the health systems of resource-poor countries and communities have boosted discovery in implementation science for health. Routine healthcare in resource-poor countries and communities is often provided by nurses and community health workers, without access to basic medical equipment, in primary care clinics or in homes without reliable referral chains to higher-level care. As a result of these constraints and the large differences between "ideal" and "real-world" delivery in resource-poor countries and communities, innovation is likely to thrive, because greater creativity is required to ensure that scientific innovations can be delivered in routine healthcare practice. The implementation research leading to novel approaches to deliver HIV care in resource-poor countries and communities illustrates this creativity. Implementation researchers have worked with implementers to discover, design, and test such highly innovative approaches as social clubs [61][62][63][64][65][66], street dispensing machines [67,68], and drones [69,70] to deliver HIV antiretroviral drugs, as well as mobile phone technology to provide HIV prevention education [71][72][73]. In many other areas, major and minor innovations are continuously increasing capacity and quality of care in resourcepoor countries and communities, such as the multitude of novel eHealth [74,75], mHealth [76][77][78][79], and telemedicine [80] applications. This creativity under constraints leads to potential for "reverse innovation" [81,82], i.e., innovation arising first in resource-poor RCT randomized controlled trial, UBA uncontrolled before-after study, CBA controlled before-after study, NRC non-randomized controlled study, WSuV withinsubject validation study, PMTCT prevention of mother-to-child transmission of HIV program settings and only later spreading to resource-rich settings. According to a recent review, important areas for future "reverse innovation" in healthcare include "rural health service delivery; skills substitution; decentralisation of management; creative problem-solving; education in communicable disease control; innovation in mobile phone use; low technology simulation training; local product manufacture; health financing; and social entrepreneurship" [83]. In several research areas-e.g., skills substitution and innovation in mobile phone use ( Table 1)-evidence is likely to continue to increase substantially in resource-poor-but not in resource-richsettings, opening up opportunities for "reverse" flows of innovation and experience.

Methods innovations
The definitional characteristic of resource-poor settings, resource poverty, also has implications for the methods of implementation science, stimulating the development of new approaches. For instance, the stepped-wedge cluster randomized controlled trial-in which clusters, such as communities or clinics, are randomized to an exposure sequence over time rather than to one time-invariant exposure as in the traditional parallel-arm trial-was first envisioned, developed, and used for a study in The Gambia in 1987 [84]. The stepped-wedge trial remains a methods mainstay of implementation science in resource-poor countries today [85][86][87][88][89]. One of the motivations for choosing a stepped-wedge over a parallel-arm design is that in the latter all communities "within the study eventually receive the intervention, thereby improving equity and acceptability" [90]. In contrast, traditional parallel-arm cluster randomized trials withhold the intervention that is tested from the communities in the control arm over the entire course of the study. This assignment can lead to political opposition to a study, because community members perceive value in the intervention to be tested. Such political opposition, in turn, is typically stronger in resource-poor than in resource-rich communities, because the former often lack many of the basic amenities and services that the latter have good access to. Other methods innovations in implementation science in resource-poor countries have been driven by a lack of resources for science. On average, low-income countries spend far less money on science and have far fewer scientists per population than high-income countries [91] (Fig. 1). To overcome resource constraints in research, implementation scientists have developed novel approaches to collect and analyze data using information and communication technologies. These innovations include field workers and community health workers using mobile phones to collect survey data [92], screen for diseases [93], and record healthcare utilization events [94].
Resource poverty can also cause or exacerbate variation in the scale-up of novel interventions across communities and-because of rationing-across individuals [95]. Such exposure variations, in turn, offer opportunities for innovative quasi-experiments to evaluate implementations of health interventions. Examples of such quasi-experimental designs include regression discontinuity-which can be used when threshold rules are used to determine eligibility for an intervention [96,97]-and difference-in-differenceswhich exploits changes in intervention exposure in one set of communities while the exposure in another set remains unchanged [98,99]. Quasi-experiments have the added advantage that they are typically far cheaper to carry out than experiments which require a prospective research infrastructure and substantial investment in trial processes. Finally, quasi-experiments take place in "reallife" without the distorting influences of experimental intervention which can introduce artificiality into the implementation context [100]. As such, quasi-experiments have been popular to establish causal impacts of interventions in resource-poor countries and communities [101], but they are of course equally valuable in resource-rich settings [102].

Creating research capacity
Implementation science is unlikely to be an exception to the general rule that resource-poor countries have far fewer researchers per population than resource-rich countries (Fig. 1). It may be possible to overcome the resulting "inverse care law" of implementation sciencecapacity is lowest where need is highest-with innovative solutions for training the next generation of implementation researchers in resource-poor countries. Major international funders, such as the Fogarty International Center of the US National Institutes of Health, are currently making large investments in South-South and South-North partnerships for implementation science training [103]. Several universities in the Global South have recently started to offer master and doctoral degrees in implementation science, such as the University of Nairobi (Kenya), University of Ghana, University of Zambia, University of the Witwatersrand (South Africa), BRAC University (Bangladesh), Universidad de Antioquia (Colombia), Universitas Gajdah Mada (Indonesia), and the University of Beirut (Lebanon) [104]. Another important opportunity to increase capacity for implementation science are massive open online courses (MOOCs), which provide (free or inexpensive) training in implementation science through online learning platforms (see Table 2 for two examples). Reflecting the reality of implementation science projects in resource-poor countries, these research programs include training in theory and formative research for intervention design; process, impact, and economic evaluation methods; and approaches for knowledge dissemination and policy translation. Despite these promising initiatives, the availability of researchers in resource-poor countries who have been rigorously trained in quantitative, qualitative, and mixed methods for implementation research remains low [105].

Science for policy
An important counterpoint to the triad of high need, high potential, and low capacity for implementation science in resource-poor countries and communities is the powerful opportunities for policy impact that engagement with policy makers offer. In many resource-poor countries, policy makers and stakeholders are closely involved in implementation research, ranging from the conception of research ideas to the interpretation of findings and from leading research agenda setting exercises with scientists [106,107] to principal investigator roles in scientific studies [87]. Close collaboration between implementation scientists and policy makers is not constrained to resource-poor settings [108], but it is likely particularly strong in those settings because of the higher need for implementation evidence when the capacity to deliver interventions is extremely scarce as well as a culture of testing the delivery of scientific innovations in "demonstration projects" to guide policy decisions and the design for long-term routine practice. For instance, many African countries are currently considering adopting HIV pre-exposure prophylaxis (PrEP) as routine health policy but are unsure which delivery models work best in their specific contexts. To fill this knowledge gap, more than 50 PrEP demonstration projects in Africa are currently experimenting with alternative delivery models [109,110].

Conclusion
In any setting, the results of implementation science can lead to improved routine healthcare practice. In resource-poor countries and communities, however, the need for such results is arguably higher than in resource-rich countries, while the capacity to carry out implementation research is lower. Despite this "inverse care law of implementation science," several specific opportunities for implementation science in resource-poor settings exist. First, intervention and methods innovations thrive under constraints. Second, reverse innovation transferring novel approaches from resource-poor to research-rich settings will gain in importance. Third, policy makers in resource-poor countries tend to be interested in collaborating closely with scientists on implementation research projects aimed at informing national and local policy. To realize these opportunities, several actions are needed. Funders need to increase their commitments to implementation science in resource-poor settings [111]. Funders and universities need to increase their investment in training the next-generation of implementation scientists who devote their careers to discovering and testing novel approaches to create and influence healthcare resources where they lack most. Finally, journal editors need to signal strongly that they are interested in featuring results from rigorous implementation science originating in resource-poor settings, to ensure that some of the brightest graduate students can be recruited into this field. The results of such actions will likely lead to a double benefitgenerating major scientific advances and contributing to improved health among the world's poor.  Ethics approval and consent to participate Not applicable

Consent for publication
Not applicable

Competing interests
The authors declare that they have no competing interests.