Environmental Health Science and Engineering for Policy, Programming, and Practice

Actions at the interfaces among communities of experts and decision makers are sometimes referred to as boundary science, and good practices in boundary spanning are emerging (Bednarek et al. 2018). Because both scientists and policy makers have insufficient training, skills, and resources to work at this interface (Cash et al. 2003; Quevauviller 2010), knowledge brokers and public entrepreneurs have emerged. These formal or informal roles vary widely in scope and scale.

Knowledge brokers are people or groups familiar with more than one professional world (e.g., science, practice, policy) who find, assess, and interpret evidence; formulate recommendations; transfer information; facilitate interaction; and identify emerging research questions (Ward et al. 2009). These individuals or institutions are accountable to the operational norms of both scientists and decision makers (Cash et al. 2003). Some knowledge brokers have been practicing in US EHSE for decades, such as the National Academies of Sciences, Engineering, and Medicine (1863), Resources for the Future (1959), and the Southern California Coastal Water Research Project (SCCWRP) (1969).

An international example of knowledge brokers was explored by Brocklehurst (2013), who brought together senior advisers to finance ministers from six African countries in a Chatham House Rule focus group. The report describes ministers of finance keen to serve as custodians of the public purse and interested in evidence on value for money and objective comparison of policy options. Politicians and their advisers were exasperated by (rather than welcoming of) the lack of detail, specificity, and precision in flat briefings. Participants felt staff in sector ministries (e.g., water resources, rural development) failed to relate their concerns to the prevailing political context. Ministers were further frustrated by weak monitoring and evaluation of both service delivery and impact, and disliked ad hoc requests for last-minute briefings, rather than planned and timely dialogue.

The UN Special Rapporteur on the human rights to water and sanitation is an institutionalized, individual knowledge broker. Incumbent experts have produced many documents to bridge science and application among human rights and practice constituencies (Heller et al. 2020). These include careful reflection on contentious ideas (such as the right to free water, effects of megaprojects, affordability, and privatization of services) that facilitate the practical application of human rights principles. However, analysis of the results of country missions and associated reports concluded these communications alone were insufficient to trigger substantive (structural) change (Heller et al. 2020).

Some institutes and NGOs established to influence policy have declared missions as knowledge brokers. Three US-based examples are Water Advocates, WaSH Advocates, and Global Water 2020. Their shared characteristics include commitment to a finite time span and exit strategy (enabling them to focus on addressing issues rather than self-promotion); sufficient core funding (liberating them from competing demands for fundraising and the risk of pandering to donor interests); and engagement with diverse stakeholders across science application interfaces.

In contrast to knowledge brokers, public entrepreneurs are individuals or institutions (e.g., policy analysts, educators, authors, nonprofit or for-profit organizations, researchers, lobby groups, or think tanks) who generate, translate, and introduce new ideas into the public sector (Roberts and King 1991). These include professional associations, which often represent the interests of their members and develop and assert policy stances. For example, the International Water Association (IWA), responding to pending publication of a novel policy development [water safety plans (WSPs)] in the 2004 edition of the WHO Guidelines for Drinking Water Quality (WHO 2004), adopted an aligned policy position in its Bonn Charter for Safe Drinking Water (IWA 2004). Both organizations acted as public entrepreneurs in introducing a new idea. WHO managed the 10-year policy development initiative, involving more than 500 people (frequently knowledge brokers themselves) from approximately 100 countries, as well as other potential public entrepreneurs (e.g., World Plumbing Council, International Organization for Standardization). IWA, with membership classes for both individual professionals and water suppliers, translated it into a form that was more accessible and relevant to water supply organizations.

A further type of public entrepreneurship is publication of recommendations through articles or letters, typically by experts or authority figures. These appear in news periodicals and lay media formats (e.g., as an op-ed), as well as scientific journals. One example concerns whistleblowing by scientists and physicians about the highly publicized issue of lead in Flint, Michigan’s drinking water (e.g., Hanna-Natisha 2019). A multiauthor academic example, Howard et al. (2020), explored COVID-19 and water, sanitation, and hygiene, suggesting policy adaptations and preparedness measures to respond to this and future pandemics.

Evidence-based policy, evidence-based practices, and EBDM (and their evidence-informed analogs) indicate policy and practice decisions informed by rigorous, empirical science. They benefited from the developing social sciences, such as public policy, economics, and sociology, in the 20th century (Amjad et al. 2015). In science, the evidence-based movement has roots in evidence-based medicine, with its emphasis on randomized controlled trials (RCTs) (a methodology aimed at controlling sources of confounding) in the 20th century, and growth of evidence-based ideals in the 1990s (CHSRF 2000). In policy, the movement influenced wider debate from the late 1990s, for instance, about how early childhood influences and interventions could alter later quality of life (Nutley et al. 2007). The evidence-based movement led to policy initiatives such as Sure Start in the UK in 1998, which provides government support for early education, childcare, health care, and families; and No Child Left Behind in the US in 2001, which applied standardized measurable goals to improve individual educational outcomes and provided for disadvantaged students. Mixed results from such attempts at policy construction demonstrate the potential harm of lip service to a largely unachievable reducibility of real-world issues to a singular evidence-based solution, rather than applying and updating policy strategies tailored to both evidence and context.

One outcome of the evidence-based movement was establishment of the Cochrane Collaboration in 1993, which sets guidelines for systematic review and synthesis of published RCTs. This Global Network of researchers, funders, practitioners, patients, and caregivers works to produce credible, accessible health information to support informed decision-making (Cochrane, n.d.). The Campbell Collaboration, formed in 1999, similarly synthesizes social and educational evidence to support improved decisions and programming (Campbell Collaboration, n.d.). These structured tools have been popularized for a specific set of problems (e.g., where high-quality research has been conducted in comparable settings but not synthesized) but are not applicable to every problem (Setty et al. 2019).

Organizations that adopted explicit missions around knowledge brokering include:

EBDM efforts seek to avoid policy or programming failures by combining professional expertise and the needs of clients with the best advice from high-quality research studies, rather than basing decisions on tradition, anecdote, assumption, or hypothetical reasoning (Greenhalgh et al. 2014). Logically, EBDM should minimize bias by applying germane scientific insights, making policies and their applications safer, more consistent, and more cost effective. Because evidence typically contributes to decision-making only partially and among other factors, some prefer the term evidence-informed (De Goede et al. 2012; Nutley et al. 2007).

Adopting EBDM aligns with the declared approaches of many EHSE decision makers globally, exemplified by many governments’ totemic assertions to follow the science in response to COVID-19. At the WHO, the role of evidence was brought to the fore around 1998, triggering progressive clarification of evidence requirements for developing public health guidance, supported by distinct organizational positions, divisions, and standards. Among international NGOs, World Vision has created an evidence and learning unit to strengthen monitoring and evaluation, research and learning, and knowledge management. Similarly, Plan International’s global strategy calls for high-quality, rigorous research and accountable, transparent, and evidence-based decision-making (Plan International, n.d.). In the US, the Foundations for Evidence-Based Policymaking Act was signed into law in 2019 to “improve the use of evidence and data to generate policies and inform programs in the federal government,” requiring all agencies “to develop evidence-based policy and evaluation plans” (USEPA, n.d.-a). These intentions respond to major shifts in public and political priorities (Vernon 2017).

While a stated commitment to EBDM is low cost, low effort, and appeals to diverse stakeholders, its implementation may be complex and has unpredictable consequences.

Overall, understanding progress on and impacts of EBDM requires agreement on appropriate evaluation methods and terminology (Setty et al. 2019), and the process is yet poorly characterized. Upon reflection, the effects of the evidence-based movement have been principally twofold: a critical reflection on quality of evidence and its synthesis in the science sphere; and a formalization of the already-implicit role of evidence in the decision-making sphere. In both cases, effects may arise more through reflection and peer pressure to adhere to modified professional norms than through formal processes and commitments.

With increased attention to EBDM, various models of the science application interface have emerged, often building on one other. Examples include the knowledge-driven model, the problem-solving model, the interactive model, the political/tactical model, and the enlightenment model (Young et al. 2002). In general, one-way, rational-linear models of research production and uptake have been criticized for overreliance on a positivist and objectivist epistemology, where science discovers a singular truth, which is used as the sole basis for decisions (Huberman 1994; Nutley et al. 2007). Post-positivist models instead incorporate more frequent, multidirectional, and complex interactions and feedback loops; recognize the importance of values alongside evidence; and highlight the necessity of an explicit paradigm to understand evidence. Postmodern frameworks acknowledge suspicion of reason and the existence of privileged and marginalized groups (Nutley et al. 2007).

Realism recognizes multiple interpretations of truth, seeking instead “what works, for whom, in what circumstances, and to what extent” (Groot et al. 2017). Although research institutions might be presumed to be the main purveyors of science, realist approaches recognize that all parties both produce and use evidence, and that direct interaction among actors is the basic vehicle for evidence cycling. Thus, Gupta (2014), for example, characterizes science application interactions as a ladder, based on their structure and continuity, with greater consensus and legitimacy toward the top of the ladder idealized as continuous participatory assessments, and interactions lacking centralized assessment or governance toward the bottom.

A constructive interpretation of these diverse models is that some are more useful in certain science application scenarios than others.

Examples reflecting linear models include the systems for deriving guideline values or standards for chemical contaminants of drinking water, wherein the procedures of WHO (2009) and USEPA (USEPA, n.d.-b) are codified to describe overall approach, grading of evidence, and technical derivation. Here, the benefit of agreeing to common rules exceeds the risk of lost control because stakeholders’ interests are advanced by incorporating evidence into regulation, deriving credibility from visible transparency and consistency, accelerating decision-making, and minimizing conflict by adopting explicit procedures.

Complex evidence uptake examples typically involve many stakeholders with potentially-competing interests and values. Examples include the negotiation of international development policy, planning processes around major infrastructure (such as dams and transport), and intersectoral policies (such as decarbonizing energy production in response to climate change). The country missions carried out by UN special rapporteurs also reflect this model, involving diverse stakeholders with competing interests and imbalanced power. In the early years of the UN, development policy adoption appears to have been relatively straightforward, perhaps reflecting the few stakeholder groups (primarily national governments) and low risk associated with largely inconsequential development decade outcomes (Bartram et al. 2015), as well as the little evidence available to compete with other inputs to decision-making. Changes following adoption of the Millennium Declaration (2000) and the Millennium Development Goals (MDGs) (2001) reflected trends in many countries toward greater engagement in governance by civil society and private sector actors, more available evidence, and increasing impact of multilateral goals. Whereas the MDGs were developed by a small group of individuals loosely inspired by the Millennium Declaration in meetings at the UN headquarters, the UN organized a complex multistakeholder process to negotiate their successors—the Sustainable Development Goals (SDGs). A generalizable lesson from these UN examples concerns the value of deliberate, early efforts to identify and engage stakeholders.

Some examples that appear linear evolve to become complex. For example, in work with the international NGO World Vision, research into metal contamination of drinking water wells (Fisher et al. 2021) was initially conceived as a simple linear process (if problem detected, then evaluate and apply corrective and preventive measures). It evolved to become more complex and politically relevant because lead was widely detected. Other stakeholders became involved, including the governments of known-affected and potentially-affected countries, other implementing agencies, and national and international regulators. Evidence was critical in the trigger and became the fulcrum for discussion, but the detail of the evidence was secondary to the negotiation of communications and roles. The evidence did not implicate any specific country or implementing organization, minimizing finger-pointing and fostering negotiation. In contrast, much commentary about the detection of arsenic in community wells in Bangladesh [arising from a well-intentioned national government well-drilling program supported by the United Nations Children’s Fund (UNICEF)] was somewhat distracted by efforts to identify fault.

No single model is applicable to all cases; diverse models contribute to exploring and understanding different scenarios of science application, and all are simplifications of real-world processes. Linear models are readily applied in single, delegated authorities. Complex models reflect situations in which interactions among science, application, and politics are extensive and call for attention to diverging interests. These distinctions reflect in part the philosophical distinction of law (simplistically: rule following) from politics (simplistically: management of trade-offs and competing interests) (Gray 2009).

Diffusion of innovation theory [the science of how new ideas, technologies, and practices spread (Rogers 2003)] parallels science application. Its five process steps are knowledge, persuasion, decision, implementation, and confirmation (Nutley et al. 2007). Depending on when adopters of an innovation accomplish this process, they are categorized as innovators, early adopters, early majority, late majority, and laggards (Rogers 2003). Characteristics that influence uptake of an innovation include its relative advantage, compatibility with existing values and practices, simplicity and ease of use, trialability, and observable results (Robinson 2012). Borrowing from the Bass diffusion model, some adoption is initially driven by media reports and novel information access (among innovators), while most adopters (imitators) are driven by conversations with peers (Rogers 2003).

Modern implementation science distinguishes passive diffusion from active dissemination, which allocates specific effort and resources to encourage uptake of evidence-based practices (Setty et al. 2019).

The already-cited example of WHO and IWA’s speedy dissemination of a policy initiative offers a prime ESHE example of active dissemination. WHO grounded the WSP innovation in familiar, established concepts [e.g., sanitary inspection, hazard analysis and critical control point (HACCP), failure mode analysis, continuous quality improvement]; involved many stakeholders in policy formulation; collaborated with innovators (e.g., Australia, Iceland) to document and build on their experiences; and fostered early adopters (e.g., United Kingdom, New Zealand). A decade after the innovation was cointroduced in 2004, “WSPs [were] being implemented to varying degrees in 93 countries representing every region of the world, with 30% of countries at an early adoption stage and others implementing on a national scale” (WHO 2017). Further, “46 countries report[ed] having policy or regulatory instruments in place that promote or require WSPs, and another 23 countries report[ed] that such instruments are under development.” Factors associated with diffusion included political readiness and compatibility with cultural values.

Another example is household water treatment and safe storage (HWTS) (ensuring a household’s ability to transport, handle, and treat its own drinking water). It has several beneficial characteristics that appeal to widespread values: enables individual behaviors, targetable on needy populations, rapidly deployable in emergencies, not reliant on expensive infrastructure, and marketable. While HWTS aggregates several potentially-competing technical approaches to household water purification (e.g., boiling, filtration, sunlight and ultraviolet irradiation, chlorine dosing), stakeholders cooperate through the International Network to promote household water treatment and safe storage, which serves as both a knowledge broker and public entrepreneur (WHO, n.d.). Early dissemination efforts emphasized potential for disease reduction; however, over time, evidence from longer-term and more rigorous studies questioned the benefits and sustainability of HWTS under normal circumstances (Hunter 2009). From these examples, implementation and dissemination situations are far from straightforward and benefit from active understanding of the complexity of evidence, contextual factors, actors, roles, and measures of success.

Several well-studied perspectives and approaches exist for academic engagement with communities (e.g., community-based participatory research) and peer groups (e.g., participatory action research) (Setty et al. 2019; Perkmann et al. 2021). This concept extends to research conducted by stakeholders outside of academia (e.g., continuous quality improvement in the private sector; governmental research institutes). In common, these perspectives and approaches emphasize the value of frequent, meaningful, sustained engagement between scientists and concerned stakeholders (through all stages of research) and the beneficial impact of co-construction on research relevance, conduct, and uptake (Cash et al. 2003; Gupta 2014; Setty et al. 2020). Engaged science incorporates the views of multiple parties in research agendas from the outset, preferably via joint, explicitly structured problem identification (Bryant et al. 2014; Viergever et al. 2010; Weichselgartner and Kasperson 2010). Some evidence suggests engagement benefits scientific productivity (Perkmann et al. 2021).

While engagement is intuitively attractive, it poses challenges in practice. Scientific neutrality may be perceived as incompatible with close stakeholder engagement, and alignment with one stakeholder or stakeholder group may alter credibility among others (Shields et al., forthcoming; Pielke 2007). Regrettably, research projects often take place in isolation of stakeholders, especially during phases such as proposal and data analysis (Setty et al. 2020).

One example study of water, sanitation, and hygiene (WaSH) in selected Pacific Island countries was predicated on engaged research with communities seeking to improve their WaSH conditions (Barrington et al. 2016). Researchers made extensive and effective efforts to engage with communities identified by a local NGO from the earliest stages. Early interactions—potentially subject to courtesy bias—confirmed interest in WaSH improvements. As confidence grew, however, community members revealed additional and higher priorities elsewhere (e.g., building pathways, reducing unemployment, building a new church). While the research team endeavored to broker contact and encourage action by other agencies, ultimately the tension between an honestly preconceived project focus and the revealed preferences of community stakeholders remained unresolved. This experience confirms the value of engagement in identifying valid priorities and suggests true engagement fits poorly with short-term project-funding cycles.

Explicit grading criteria for assessing the quality of evidence from studies aid rules-based approaches to decision-making.

Study types have often been organized into a hierarchy, misleadingly represented by a pyramid. Such depictions undermine the value of complementary insights from different method types, and fail to convey the relationships and progressions among them as evidence accumulates (Setty et al. 2019). Recognizing that not all studies of the same nature should hold the same weight, the Grading of Recommendations Assessment, Development and Evaluation (GRADE) Working Group was formed in 2000 to address simplistic grading systems used in health care research. Its methods have since been adopted by more than 100 organizations globally, including WHO (Alonso-Coello et al. 2016) and the Cochrane Handbook for Systematic Reviews of Interventions (Higgins and Green 2011). This approach still reflects a dominant focus on study type. Beginning with four levels of evidence quality determined by study type, reviewers upgrade evidence (based on a large effect size, the influence of plausible confounding, and/or demonstration of a dose-response gradient) or downgrade evidence [responding to limitations in the design and implementation, ability to make direct inferences, unexplained heterogeneity or inconsistency in results, imprecision of results (e.g., wide confidence intervals), and the probability of publication bias] (Higgins and Green 2011). Another tool, adopted by the US National Toxicology Program for human and animal study quality evaluation (NTP 2015), digs further into individual studies’ risk of bias (internal validity) based on 11 factors that might compromise the credibility of the link between exposure and outcome.

Some international approaches to qualifying evidence for use in standard setting (analogous to GRADE) were formalized in the 2000s. For example, for chemicals in drinking water, WHO prefers human studies over animal studies (in principle, although high-quality human studies are few); and prefers peer-reviewed literature, including commercially sensitive literature after review by an international body such as the Joint Food and Agriculture Organization of the United Nations (FAO)/WHO Meeting on Pesticide Residues (WHO 2009). Uncertainty factors are applied where appropriate to account for inter- and intraspecies variation, study or database adequacy, and the nature and severity of health effects.

Several constraints obstruct application of these types of study quality criteria to EHSE, and may impede use of the best-available evidence for decision-making. Even the basic assumption that RCTs are always superior to observational epidemiological studies for managing bias in environmental exposures studies is questioned (Steenland et al. 2020). Some evidence in EHSE falls outside the GRADE structure, and in some instances evidence from preferred study types is nonexistent and potentially ethically unattainable. For example, evidence for the derivation of standards for chemical contaminants of drinking water cannot be ethically attained through RCTs that expose human populations to known risk. Thus, natural experiments (unintentionally exposed human populations) and animal studies are relied on. Natural experiments are necessarily retrospective, so exposure is typically poorly characterized, identifying controls is challenging, and exposure may be through multiple routes, short term, or restricted to certain life stages (e.g., adults in the workplace). Laboratory animal experiments enable better exposure control but require extrapolation from nonhuman species.

Further, much important EHSE insight derives from qualitative and mixed methods research, which is often undervalued or explicitly excluded from evidence hierarchies. While the research method is commonly misconceived as the dominant element of rigor, different study types are appropriate for different research questions, and applied, qualitative, and translational research types can all be carried out rigorously (Setty et al. 2019). Cooper (2016) emphasizes the need for greater social science inputs, especially for more complex science applications, and quantitative and qualitative research perspectives are necessary and complementary in many EHSE domains. Research quality also depends on the actors’ breadth and depth of knowledge, the actors’ communication and coordination skills, and external (potentially uncontrollable) circumstances and events.

Using medical approaches to categorize EHSE evidence may discourage valuable scientific study designs, such as phenomenology and case studies, which elaborate on the meaning behind statistical relationships and identify counterexamples to generic knowledge, respectively. Post-positivism acknowledges that scientific methods can reduce, but not eliminate, the risk of bias. Uncritical application of evidence based on statistical significance or automated decision algorithms may miss nuance in real-world problems relevant to decision makers (Greenhalgh et al. 2014). Conversely, embracing mixed methods research and triangulating evidence types can illuminate the consequences of implicit method assumptions and increase robustness of findings.

The newer GRADE Evidence to Decision Framework (EtD) (Alonso-Coello et al. 2016), often carried out by an expert panel, systematically assesses evidence and the role it plays among other factors in decisions. Criteria, each backed by justification, account for factors external to the evidence, including priority of the problem, benefits and harms, certainty of the evidence, outcome importance, balance, resource use, equity, acceptability, and feasibility. Poor external applicability of the EtD called for a comparable exercise or adaptation to the diversity of evidence applicable to public health and EHSE problems, wherein the WHO-INTEGRATE framework proposes a generic approach suitable for complex individual-, population- and system-level health interventions (Stratil et al. 2020).

On an individual study level, the important factors in evidence quality have been standardized for a minority of study types (Setty et al. 2020). Recommendations for method adherence and reporting rigor could be better developed (especially with audience needs in mind) and made accessible to both authors and reviewers involved in the publication process.

In some EHSE applications, evaluation of individual studies leads directly to decision-making. Continuing the example of norms for chemicals in drinking water, guideline development typically relies on determination of the most applicable (single) study (WHO 2009). However, decisions would preferably be supported by synthesized bodies of evidence. Congruency among multiple studies (where findings provide similar implications, or differences relate logically to contextual differences) is one of the nine principles laid out by Hill (1965) for determining causality, where environmental exposures lead to human health effects. Synthesizing bodies of evidence and understanding their coherence (or otherwise) has increasing importance in the context of “fake news” and the long-established tendency of politicians to use outlier scientific voices to undermine scientific consensus (e.g., for anthropogenic climate change).

The systematic review and meta-analysis approaches adopted initially in medicine (Cochrane, n.d.) have also been embraced in EHSE. A systematic literature review uses explicit and comprehensive methods to formulate a question; identify, select, and critically appraise relevant research; and synthesize data (Khan et al. 2003). However, because of method diversity, meta-analysis is rarely feasible in EHSE.

Describing scientific uncertainty is similarly problematic in EHSE evidence synthesis, in part because it is often conflated with context dependency. The outcomes of many EHSE interventions depend on context. Water treatment effectiveness, for example, varies with water quality characteristics, pathogen die-off dependency on temperature, and the effect of preexisting health status on disease reduction. When studies from diverse contexts are aggregated, clear differences may be incorrectly labeled as uncertainty, casting doubt on the merit of the intervention.

Critiques of evidence synthesis further observe that the volume of evidence has become massive, even unmanageable. Young et al. (2002) state, “The rush of enthusiasm for evidence-based policy making overlooks the fact that a great deal of research has already been carried out on a wide range of social problems, providing policy makers with pointers that they rarely follow.” Potential implications are twofold. First, evidence alone is often insufficient to elicit decision-making, suggesting that most influence arises when evidence is available at an influential moment in time (Rose et al. 2020). Second, better use should be made of evidence, where available, rather than on generation of new evidence (Setty et al. 2019). Certainly, there are areas replete with duplicative studies (each adding little to the accumulated evidence base) due to conscious or unconscious repetition across geographies, research groups, literature outlets, languages, and disciplinary silos.

In contrast, sparse evidence exists in many critical areas. In some cases, only insufficient, poor-quality, imbalanced or potentially misleading evidence is available to inform decision-making. For example, recent reviews explored environmental health conditions during the emergency (Shackelford et al. 2020), transition (Cooper et al. 2021) and protracted (Behnke et al. 2020) phases of forced population displacement. Even though more than 70 million people are forcibly displaced at any given time, the reviews revealed a dearth of meaningful or high-quality evidence. Reviews of conditions in orphanages (Moffa et al. 2019a), prisons (Guo et al. 2019), and homeless shelters (Moffa et al. 2019b) pointed to similar conclusions.

Recognition (and, if possible, mapping) of evidence gaps is crucial to avoid placing priority on better-characterized issues through the streetlight effect (Setty et al. 2020; Whaley et al. 2020). For example, the Collaboration for Evidence-Based Healthcare and Public Health in Africa (CEBHA+) sets out a pragmatic approach to identify and fill evidence gaps that address policy and practice needs in resource-limited research settings (Rehfuess et al. 2016). Scoping reviews (or evidence maps) explore the breadth and depth of evidence where it is sparse (Pham et al. 2014; Whaley et al. 2020).

Saliency, or relevance, refers to timeliness (Cash et al. 2003; Sarkki et al. 2014). Because most applied research is reactive, evidence is likely to become available after decisions have been made. Achieving saliency requires a paradigm shift from rigid adherence to individual project funding cycles to broader program continuity and ongoing readiness.

One example of intentional research delivery timing comes from an initiative to deliver germane WaSH-related evidence at the time of SDG negotiation. Seven themes were laid out in an opinion article (Bartram 2008) and ultimately 18 related research papers were published in the lead-up to SDG adoption. At the time, each provided a substantive addition to a sparse evidence base during complex multistakeholder negotiations. It would be difficult to demonstrate these studies’ influence among other diverse efforts; however, all themes covered were reflected in the SDGs in some form. This experience illustrates that saliency does not demand scientific sophistication—several of the publications were scoping reviews or compilation and interpretation of existing publicly accessible data. Rather, matching the policy window observations of Rose et al. (2020), packaging the right information for the right audience at the right time tapped into previously inaccessible insights.

Further examples of planned salience come from deliberate attempts to garner foresight. Foresight is difficult and neither scientists nor decision makers are normally trained in it. However, some events may be individually unpredictable and entirely foreseeable. Whether and when a waterborne disease outbreak will occur in a certain location is unpredictable; that such outbreaks will occur in general is foreseeable. What disease agent will cause the next global pandemic is unpredictable; that further global pandemics will occur and that the causes will be disproportionately zoonotic viruses is foreseeable. Perhaps the weak response to such general foresight arises in part because of its weak specific salience.

Recognition of salient issues, such as new policies yet to be implemented, can prompt reassessment of research needs and priorities. One example surveyed a network of senior and operational government, civil society, external support, private, and research and learning stakeholders participating in Sanitation and Water for All (SWA). The exercise sought to identify priority knowledge needs just after the SDGs were adopted, and to describe evidence-use challenges (Setty et al. 2020). Low confidence was observed concerning the target on managing untreated wastewater and fecal sludge, demonstrating an evidence demand. Respondents preferred multinational information sources and reported little direct demand for local university research. They also valued combined brief and lengthy information formats (e.g., summaries with attached technical explanation). Persons outside research and learning institutions more often perceived information sources as contradictory or unreliable, illustrating the need for a bridge between scientists and others.

Saliency benefits from timely identification of influencing opportunities. Furthermore, saliency may benefit from academic freedom, i.e., the ability to pursue exploratory research, regardless of specific topic funding. Such freedom may be eroded, for example, through highly constrained funding conditions. Interactions among researchers and funders or decision makers can help raise and gauge salient evidence needs in a way that balances applied needs with the basic scientific tenet of expanding knowledge. For example, the research development cycle at SCCWRP is revisited quarterly by a 14-member board comprising regulatory and management agencies to ensure all parties’ interests are considered (SCCWRP 2020). While health sciences funding agencies, for example, increasingly support science application, a focus on communication at the end-of-grant stage may overlook engagement during critical initial stages (Smits and Denis 2014).

Legitimacy refers to adherence to the quality norms of the research community (Cash et al. 2003). Deficiencies in individual research outputs occur, for example, from inappropriate (even if popular) methods, ticking boxes on rigor checklists while losing sight of context and needed outcomes, or implementing methods poorly.

The HWTS example illustrates some issues concerning legitimacy. Most evidence purveyors acted as issue advocates and many were closely invested in a specific technical approach.

Journal publication is highly associated with legitimacy, and requirements for academic career progression set expectations for evidence supply, creating a high incentive to publish. Some poor-quality or unnecessarily repetitive research is therefore conducted and published, in part because voluntary peer review capacity (fundamental to the premise that peer review sorts for and ensures quality) is overwhelmed. The consequences can be substantive if research is improperly vetted, as with the Lancet’s retraction of an influential paper on chloroquine treatment for COVID-19 (Editors of the Lancet Group 2020) and consequently changed editorial policies. This example arose on an issue under public scrutiny (e.g., Davey 2020). In general, though, the scientific community is poor at collective self-regulation, in part because the economics of publication create no industry incentive to limit output and research supporters often expect publication as a tangible and countable deliverable.

As outlined previously, some study types have carried greater prestige and legitimacy. Often the “purer” study methods applied earlier in the science application pipeline persist in popularity even when other methods would provide more valuable information (Brown et al. 2017; Fig. 1 of Setty et al. 2019). In some instances, this is beneficial—for example, in continuing to explore the occurrence of a phenomenon under different conditions. In other instances, inappropriate study methods may be repeated despite known weaknesses and the opportunity cost. This is problematic if the number of papers creates the impression of a weight of evidence. Following popular research trends may also foster weak practices and poor self-reflection among researchers, and absorb financial, intellectual, and community resources in low-return efforts.

The disproportionate value placed on experimental research close to causal relationships and fundamental processes stands in bizarre contrast to the frequent demand for greater societal relevance. Even the most high-quality RCTs tell us little about how to implement the intervention. Generating evidence around close-to-application interventions tends to be viewed as a component of practice rather than a proper subject for research, despite its direct relevance to widespread costs and outcomes.

Credibility relates to the quality of being trusted, convincing, or believable (Cash et al. 2003). In contrast to legitimacy, it primarily concerns the perspective of the decision maker; that is, why decision makers should have confidence in evidence to which they are asked to respond. Credibility may arise from biased or true perceptions around individual studies, bodies of evidence, researchers, research groups, institutions, publishers, or settings.

The media influence credibility; however, occasional overinterpretation, exaggeration of confidence in findings, and unreasonably extrapolated implications undermine credibility and leave some scientists wary of interactions (Nutley et al. 2007).

One effort to increase transparency, individual accountability, and implicitly credibility around applied WaSH research is the Nakuru Accord (Fig. 1). Some of the commitments it describes will prove challenging to career researchers. For example, Barrington et al. (2012) suggests “engineers are not neutral bystanders in these processes, though this is often how they see themselves. Rather, they are political actors whose decisions and actions can well determine the extent to which outcomes are likely to be socially and environmentally just.” Such pledges to openly learn from failures could aid deimplementation of unsuccessful initiatives and thereby accelerate progress in EHSE. Similarly, conflict of interest training and reporting, ethics standards and procedures, and open science initiatives are growing in prominence and gradually shifting professional norms.

Scientists tend to discuss uncertainty extensively (e.g., EFSA et al. 2019), in contrast to the confidence expected of decision makers. Van den Hove (2007) argues that transparency about limitations does not weaken science, but rather reinforces its quality. Communicating scientific knowledge should “systematically include reflective information on boundaries, uncertainties, indeterminacy, and ambiguity, as well as acknowledgment of ignorance and of the irreducible plurality of valid standpoints” (van den Hove 2007).

Much credibility arises from early and meaningful engagement. Efforts to further build credibility would benefit from researcher training in communication and knowledge brokering. This contrasts with the lauding of the pure scientist and undervaluing or criticizing the stance of issue advocate (Pielke 2007), which are widespread stereotypes, especially in EHSE.

General goodwill and interpersonal connections between scientists and decision makers can be naturally strong (e.g., facilitated by neighboring offices and frequent interaction), nonexistent, or contentious.

Convening encourages direct interaction between researchers and practitioners, which has been shown to enhance research use even in an antagonistic environment (Nutley et al. 2007), especially given strong leadership and coherence, a supportive political and social context, and responsiveness to the needs of both parties. It promotes meaningful, shared decisions, which are at the root of the evidence-based philosophy (Greenhalgh et al. 2014) and form the foundations of engaged science.

Convening implies action by individuals and institutions, such as a knowledge broker, alliance, or partnership network, who have the credibility and means to bring together others who would not spontaneously or voluntarily meet otherwise. It typically seeks to proactively enhance mutual understanding and recognition of counterparts’ valid viewpoints to prevent more fractious interactions (Quevauviller 2010; Cairney 2016).

Some EHSE conferences (e.g., Stockholm International Water Institute’s Stockholm Water Week and the Water and Health Conference of the University of North Carolina at Chapel Hill) have evolved or were designed, respectively, to include platforms (e.g., side events or workshops) for convening by participants who might not otherwise be able to meet in the same time and place. Face-to-face interactions most encourage research use (Nutley et al. 2007). However, they are constrained by time and travel costs and other restrictions, such as those arising from the COVID-19 pandemic. These practical considerations exacerbate and perpetuate power imbalances (e.g., between the global north and global south).

Some UN entities (such as WHO and UNICEF) use their formal convening powers frequently. This includes both technical-normative convening of scientists (e.g., WHO initiative on emerging issues in water and infectious disease; WHO-UNICEF advisory group on SDG monitoring of water, sanitation, and hygiene) and political intergovernmental convening (e.g., the intergovernmental European Environment and Health conferences leading to the 1999 European Protocol on Water and Health, which brought together policy makers and scientists from multiple sectors and promotes evidence-based national target setting to meet the legal requirements of the protocol).

Much convening involves scientists in the stance Pielke (2007) characterizes as science arbiter (i.e., a responsive resource ready to answer factual questions conceived by the decision maker, but without influence over those questions nor interacting with stakeholders). Less frequent but beneficial in some instances is Pielke’s honest broker of policy alternatives (increasing and characterizing the range of options but leaving the selection among them to the decision maker). Both roles are frequently played by groups, for example, in the form of expert panels or science advisory boards. Because convening typically involves bringing together dissimilar stakeholders, power balance is important (SPLASH, n.d.-a). In practice, the voice of presumed eventual beneficiary populations is often weak or absent.

Translation concerns rendering science into the vocabulary and modes of communication of other stakeholders, including recognition that some characteristics, findings, and implications differ in importance among groups. Cash et al. (2003) suggest “mutual understanding between experts and decision makers is often hindered by jargon, language, experiences, and presumptions about what constitutes persuasive argument.” Quevauviller (2010) noted differences between the norms of science and policy, although generalizations do not apply to every individual and norms of practice evolve. These differences include pertinent aims, time scales, people, communication, success measures, evidence sources, quality control mechanisms, information synthesis approaches, and views of interaction.

Translation is needed throughout the research process. The goal is not one-sided, but mutual comprehension (Cash et al. 2003). Both active listening and information delivery are crucial—for example, scientists may craft a research presentation or report without gathering information about the audience and what they would like to know, as described by Brocklehurst (2013). Translation is impossible without knowledge of the language of the intended stakeholder recipient, and no single translation is appropriate for all audiences. Structured standards for involvement of stakeholders throughout the research process can help ensure communication needs are discussed to steer effective translation (Weichselgartner and Kasperson 2010).

Translation is challenging in part because most individuals and organizations work within discipline or subject boundaries, communicating with others who share similar vocabulary and sometimes values. Alone, they are thereby unsuited for the translation function, which is often associated with conveners and knowledge brokers. Translation is a major function for some knowledge brokers; for example, ASCE’s report card for America’s infrastructure adopts the familiar format of a school report card with assigned letter grades (ASCE, n.d.), while the Scholars Strategy Network requires members to develop op-eds in their area of expertise. In international development, foreign consultants often translate and adapt international policy and guidelines for national and local governments. These examples typify translation as recipient-focused dissemination through multiple channels and formats with commitment to stakeholder engagement.

Regrettably, much applied science communication begins with already-secured findings. One checklist to aid research translation suggests: adopting a strategic approach to dissemination and reviewing existing organizational practice; defining target users, working in collaboration with in-country partners; undertaking a user information needs analysis; ensuring a viable dissemination strategy through planned activities at all stages of the project and beyond; and planning to monitor and evaluate dissemination activities and outcomes (SPLASH, n.d.-b).

The perspective of a scientist preparing a paper or report differs substantially from that of those reading and considering applying the evidence. Common standards, criteria, checklists, or methods for reporting, such as the message box developed by COMPASS, could help to develop research communication skills (COMPASS 2020). Turnhout (2018) recommends repackaging knowledge using modern terms to fit categories considered politically salient. Information can usefully be packaged in multiple formats, such as policy briefs, visual presentations, and technical addendums, that directly reflect the requests of end users (Gagliardi et al. 2015; Setty et al. 2020).

Mediation involves active rather than passive resolution of conflicts, stalemates, and other issues that impede information flow between experts and decision makers (Cash et al. 2003). Such conflicts arise from trade-offs among saliency, legitimacy, and credibility, and competing priorities beyond the nature of evidence available. Constituency silos may lead to misunderstandings and friction (Roux et al. 2006). The skills required for mediation (e.g., moderation of group forums and knowledge-building exercises, active listening, cognitive interviewing, summarization of peer review feedback, diplomacy) differ from those needed for scientific discovery, and may require targeted professional development.

Mediation is especially critical in values-dominated, politicized decision-making, such as decisions about large dam construction—a touchstone for the early environmental movement. In this case an attempt at evidence-based resolution—the World Commission on Dams—failed, likely related to its low credibility and complicated recommendations, slowing improvement in the massive undersupply of water storage in low-income countries. In the negotiation of the Nukus Declaration on Sustainable Development Issues in the Aral Sea Basin, signed by the Central Asian heads of state in 1995, shuttle diplomacy involved individuals going between and reporting back to different stakeholder groups to achieve consensus. Such mediators translate feedback and interpret must-have versus nice-to-have requests to help reach consensus. In some instances, as here, mediation takes place rapidly and intensively (over one night); in others, it occurs sporadically over an extended time.

The views expressed here do not necessarily reflect those of our present or former employers.