Aligning evidence generation and use across health, development, and environment
Heather Tallis
1, Katharine Kreis
2, Lydia Olander
3, Claudia Ringler
4, David Ameyaw
5, Mark E Borsuk
6, Diana Fletschner
7, Edward Game
8,9, Daniel O Gilligan
4, Marc Jeuland
10, Gina Kennedy
11, Yuta J Masuda
12, Sumi Mehta
13, Nicholas Miller
14, Megan Parker
2,
Carmel Pollino
15, Julie Rajaratnam
2, David Wilkie
16, Wei Zhang
4, Selena Ahmed
17, Oluyede C Ajayi
18, Harold Alderman
4,
George Arhonditsis
19, Ines Azevedo
20, Ruchi Badola
21, Rob Bailis
22, Patricia Balvanera
23, Emily Barbour
24, Mark Bardini
25, David N Barton
26, Jill Baumgartner
27, Tim G Benton
28, Emily Bobrow
29, Deborah Bossio
30,
Ann Bostrom
31, Ademola Braimoh
32, Eduardo Brondizio
33, Joe Brown
34, Benjamin P Bryant
35, Ryan SD Calder
6,
Becky Chaplin-Kramer
35, Alison Cullen
31, Nicole DeMello
36,
Katherine L Dickinson
37, Kristie L Ebi
38, Heather E Eves
39, Jessica Fanzo
40, Paul J Ferraro
41, Brendan Fisher
42,
Edward A Frongillo
43, Gillian Galford
42, Dennis Garrity
44, Lydiah Gatere
45, Andrew P Grieshop
46, Nicola J Grigg
15, Craig Groves
47, Mary Kay Gugerty
31, Michael Hamm
48,
Xiaoyue Hou
32, Cindy Huang
49, Marc Imhoff
50, Darby Jack
51, Andrew D Jones
52, Rodd Kelsey
53, Monica Kothari
2,
Ritesh Kumar
54, Carl Lachat
55, Ashley Larsen
56,
Mark Lawrence
57, Fabrice DeClerck
58, Phillip S Levin
13, Edward Mabaya
59, Jacqueline MacDonald Gibson
60,
Robert I McDonald
36, Georgina Mace
61, Ricardo Maertens
62, Dorothy I Mangale
63, Robin Martino
64, Sara Mason
3,
Lyla Mehta
65, Ruth Meinzen-Dick
4, Barbara Merz
36, Siwa Msangi
4, Grant Murray
66, Kris A Murray
67,
Celeste E Naude
68, Nathaniel K Newlands
69, Ephraim Nkonya
4, Amber Peterman
70, Tricia Petruney
71, Hugh Possingham
8,72, Jyotsna Puri
73, Roseline Remans
74, Lisa Remlinger
75,
Taylor H Ricketts
42, Bedilu Reta
76, Brian E Robinson
77, Dilys Roe
78, Joshua Rosenthal
79, Guofeng Shen
80,
Drew Shindell
81, Ben Stewart-Koster
82, Terry Sunderland
83, William J
Sutherland
84, Josh Tewksbury
85, Heather Wasser
60, Stephanie Wear
86, Chris Webb
87, Dale Whittington
60,
Marit Wilkerson
88, Heidi Wittmer
89, Benjamin DK Wood
90, Stephen Wood
36,91, Joyce Wu
92, Gautam Yadama
93and Stephanie Zobrist
2Althoughhealth,development,andenvironmentchallengesare interconnected,evidenceremainsfracturedacrosssectors duetomethodologicalandconceptualdifferencesinresearch andpractice.Alignedmethodsareneededtosupport SustainableDevelopmentGoaladvancesandsimilaragendas.
TheBridgeCollaborative,anemergentresearch-practice collaboration,presentsprinciplesandrecommendationsthat helpharmonizemethodsforevidencegenerationanduse.
Recommendationsweregeneratedinthecontextofdesigning andevaluatingevidenceofimpactforinterventionsrelatedto fiveglobalchallenges(stabilizingtheglobalclimate,making foodproductionsustainable,decreasingairpollutionand respiratorydisease,improvingsanitationandwatersecurity, andsolvinghungerandmalnutrition)andserveasastarting pointforfurtheriterationandtestinginabroadersetofcontexts anddisciplines.Weadoptedsixprinciplesandemphasize threemethodologicalrecommendations:(1)creationof compatibleresultschains,(2)considerationofallrelevanttypes ofevidence,and(3)evaluationofstrengthofevidenceusinga unifiedrubric.Weprovidedetailedsuggestionsforhowthese recommendationscanbeappliedinpractice,streamlining effortstoapplymulti-objectiveapproachesand/orsynthesize evidenceinmultidisciplinaryortransdisciplinaryteams.These recommendationsadvancethenecessaryprocessof reconcilingexistingevidencestandardsinhealth,
development,andenvironment,andinitiateacommonbasis forintegratedevidencegenerationanduseinresearch, practice,andpolicydesign.
Addresses
1TheNatureConservancy,GlobalScience,100ShafferRd.,SantaCruz, CA95060,USA
2PATH,2201WestlakeAveSuite200,Seattle,WA98121,USA
3TheNicholasInstituteforEnvironmentalPolicySolutions,Duke University,Durham,NC27708,USA
4InternationalFoodPolicyResearchInstitute,1201EyeSt,NW Washington,DC20005-3915,USA
5InternationalCenterforEvaluationandDevelopment,AppleCross Road,Lavington,Nairobi,Kenya
6DepartmentofCivilandEnvironmentalEngineering,DukeUniversity, Durham,NC27708,USA
7LandesaCenterforWomen’sLandRights,1424FourthAvenue,Suite 300,Seattle,WA98101,USA
8TheNatureConservancy,GlobalScience,SouthBrisbane,QLD4101, Australia
9UniversityofQueensland,CenterforBiodiversityandConservation Science,St.Lucia,QLD4067,Australia
10SchoolofPublicPolicyandDukeGlobalHealthInstitute,Duke University,Durham,NC27708,USA
11BioversityInternational,ViadeiTreDenari472/a,00054Maccarese, Rome,Italy
12TheNatureConservancy,74WallSt,Seattle,WA98121,USA
13VitalStrategies,61Broadway,Ste1010,NewYork,NY10006,USA
14TheNatureConservancyinWisconsin,633WestMainStreet, Madison,WI53703,USA
15CSIROLandandWater,GPOBox1700,Canberra,ACT,Australia
16WildlifeConservationSociety,FountainCircleTrail,Bronx,NY10460, USA
17MontanaStateUniversity,345ReidHall,MontanaStateUniversity, Bozeman,MT59717-3370,USA
18EU-ACPTechnicalCentreforAgriculturalandRuralCooperation, 6708PWWageningen,TheNetherlands
19DepartmentofPhysicalandEnvironmentalSciences,Universityof Toronto,Toronto,OntarioM1C1A4,Canada
20CarnegieMellonUniversity,CollegeofEngineering,5000Forbes Avenue,BakerHall129,Pittsburgh,PA15213,USA
21WildlifeInstituteofIndia,Chandrabani,248001DehraDun, Uttarakhand,India
22StockholmEnvironmentInstitute,11CurtisAvenue,Somerville,MA 02144-1224,USA
23InstitutodeInvestigacionesenEcosistemasySustentabilidad, UniversidadNacionalAuto´nomadeMe´xico,Apdo.Postal27-3,Sta.Ma deGuido,Morelia,Michoaca´n58350,Mexico
24SchoolofGeographyandtheEnvironment,UniversityofOxford, OxfordOX13QY,UK
25KhulisaManagementServices,4630MontgomeryAvenue,Suite510, Bethesda,MD20814,USA
26NINA,NorwegianInstituteforNatureResearch,Gaustadalle´en21, NO-0349Oslo,Norway
27InstituteforHealthandSocialPolicyandDeptofEpidemiology, BiostatisticsandOccupationalHealth,McGillUniversity,Montreal, QuebecH3A1A3,Canada
28UniversityofLeeds,LeedsLS29JT,UK
29MEASUREEvaluation,CarolinaPopulationCenter,UniversityofNorth CarolinaatChapelHill,123WFranklinSt,BuildingC,Suite330,Chapel Hill,NC27516,USA
30TheNatureConservancy,GlobalLands,22956ECliffDrive,Santa Cruz,CA95602,USA
31UniversityofWashington,EvansSchoolofPublicPolicy&
Governance,ParringtonHall,410015thAveNE,Seattle,WA98195- 3055,USA
32WorldBank,1818HStNW,Washington,DC20433,USA
33IndianaUniversityBloomington,DepartmentofAnthropology,Student Building130,701E.KirkwoodAvenue,Bloomington,IN47405-7100,USA
34SchoolofCivilandEnvironmentalEngineering,GeorgiaInstituteof Technology,790AtlanticDrive,Atlanta,GA30332-0355,USA
35TheNaturalCapitalProjectandWaterintheWest,Stanford University,371SerraMall,Stanford,CA94305,USA
36TheNatureConservancy,4245NorthFairfaxDrive,Arlington,VA 22203,USA
37ColoradoSchoolofPublicHealth,UniversityofColoradoAnschutz MedicalCampus,Aurora,CO80045,USA
38CenterforHealthandtheGlobalEnvironment,Universityof Washington,Seattle,WA98105,USA
39VirginiaTech,CenterforLeadershipinGlobalSustainability,900North GlebeRd.,Arlington,VA22203,USA
40SchoolofAdvancedInternationalStudiesandtheBermanInstituteof Bioethics,JohnsHopkinsUniversity,Washington,DC20036,USA
41JohnsHopkinsUniversity,BloombergSchoolofPublicHealth,Carey BusinessSchool,andWhitingSchoolofEngineering,3400North CharlesStreet,Baltimore,MD21218-2608,USA
42GundInstituteforEnvironmentandRubensteinSchoolofEnvironment andNaturalResources,UniversityofVermont,Burlington,VT05405,USA
43DepartmentofHealthPromotion,Education,andBehavior,University ofSouthCarolina,Columbia,SC29208,USA
44WorldAgroforestryCentre(ICRAF),ICRAFHouse,POBox30677, Nairobi,Kenya
45AgricultureandFoodSecurityCenter,ColumbiaUniversity,61Route 9W,Palisades,NY10964,USA
46DepartmentofCivil,ConstructionandEnvironmentalEngineering, NorthCarolinaStateUniversity,Raleigh,NC27695,USA
47ScienceforNatureandPeoplePartnership,735StateStreet,Suite 300,SantaBarbara,CA93101,USA
48MichiganStateUniversity,NaturalResourcesBuilding,480WilsonRd, Rm312B,MSU,EastLansing,MI48824,USA
49CenterforGlobalDevelopment,2055LStreetNW,FifthFloor, Washington,DC20036,USA
50UniversityofMaryland,5825UniversityResearchCourt,Suite4001, CollegePark,MD20740-3823,USA
51Columbia,EnvironmentalHealthSciences,722West168Street,11th Floor,NewYork,NY10032,USA
52DepartmentofNutritionalSciences,SchoolofPublicHealth, UniversityofMichigan,AnnArbor,MI48109,USA
53TheNatureConservancy,CaliforniaScience,555CapitolAvenue, Suite1290,Sacramento,CA95814,USA
54WetlandsInternationalSouthAsia,A-25SecondFloor,Defence Colony,NewDelhi110024,India
55DepartmentofFoodTechnology,SafetyandHealth,GhentUniversity, CoupureLinks653,9000Gent,Belgium
56BrenSchoolofEnvironmentalScience&Management,Universityof California,SantaBarbara,CA93106-5131,USA
57InstituteforPhysicalActivityandNutrition(IPAN),SchoolofExercise andNutritionSciences,DeakinUniversity,Geelong3220,Australia
58AgrobiodiversityandEcosystemServicesProgram,Bioversity International,ParcScientifiqueAgropolisII,34397MontpellierCedex5, France
59CornellUniversity,CornellInternationalInstituteforFood,Agriculture andDevelopment,Ithaca,NY14850,USA
60GillingsSchoolofPublicHealth,UniversityofNorthCarolinaatChapel Hill,148ARosenauHall,CB#7431,ChapelHill,NC27599,USA
61CentreforBiodiversityandEnvironmentResearch,UniversityCollege London,GowerStreet,LondonWC1E6BT,UK
62DepartmentofEconomics,HarvardUniversity,1805Cambridge Street,Cambridge,MA02138,USA
63TheChildhoodAcuteIllnessandNutritionNetwork,Universityof Washington,908JeffersonSt.,Seattle,WA98104,USA
64BiodiversityResultsandIntegratedDevelopmentGainsEnhanced Project(BRIDGE),1300IStreetNW,Suite400E,Washington,DC20005, USA
65InstituteofDevelopmentStudies,LibraryRoad,BrightonBN19RE,UK
66DukeUniversityMarineLab,135DukeMarineLabRoad,Beaufort,NC 28516,USA
67GranthamInstitute–ClimateChangeandtheEnvironment, DepartmentofInfectiousDiseaseEpidemiology,SchoolofPublic Health,ImperialCollegeLondon,UK
68CentreforEvidence-BasedHealthCare,DivisionofEpidemiologyand Biostatistics,FacultyofMedicineandHealthSciences,Stellenbosch University,FrancievanZijlDrive,Tygerberg7505,SouthAfrica
69AgricultureandAgri-FoodCanada,SummerlandResearchand DevelopmentCentre,4200Highway97,POBox5000,Summerland,BC V0H1Z0,Canada
70UNICEFOfficeofResearch–Innocenti,PiazzaSS.Annunziata12, 50122Florence,Italy
71PathfinderInternational,9GalenStSuite217,Watertown,MA02472, USA
72ARCCentreofExcellenceforEnvironmentalDecisions,TheUniversity ofQueensland,Brisbane,QLD4072,Australia
73IndependentEvaluationUnit,GreenClimateFund,SouthKorea
74BioversityInternational,W.DeCroylaan42,3001Heverlee,Belgium
75WashingtonEnvironmentalCouncil,14023rdAve#1400,Seattle,WA 98101,USA
76AddisAbabaUniversity,AAU,CenterforEnvironmentalScience Studies,AratKilo,1176AddisAbaba,Ethiopia
77McGillUniversity,DepartmentofGeography,BurnsideHallBuilding, Room705,805SherbrookeStreetWest,Montreal,QuebecH3A0B9, Canada
78InternationalInstituteforEnvironmentandDevelopment,80–86Gray’s InnRoad,LondonWC1X8NH,UK
79FogartyInternationalCenter,NationalInstitutesofHealth,Bethesda, MD20892,USA
80CollegeofUrbanandEnvironmentalSciences,PekingUniversity, Beijing100871,China
81TheNicholasSchooloftheEnvironment,DukeUniversity,Durham, NC27708,USA
82AustralianRiversInstitute,GriffithUniversity,170KesselsRoad, Nathan4111,Australia
83CentreforInternationalForestryResearch(CIFOR),POBox 0113BOCBD,Bogor16000,Indonesia
84UniversityofCambridge,DepartmentofZoology,Room3.05David AttenboroughBuilding,DowningSt,CambridgeCB23EJ,UK
85FutureEarth,ColoradoGlobalHub,Boulder,CO80309,USA
86TheNatureConservancy,DukeMarineLab135DukeMarine, Beaufort,NC28516,USA
87TheNatureConservancy,26–28ElyPlace,LondonLNDEC1N6TB,UK
88TheNatureConservancy,201MissionSt,SanFrancisco,CA94105, USA
89HelmholtzCentreforEnvironmentalResearch,UFZ,Departmentof EnvironmentalPolitics,Leipzig,Germany
90HeiferInternational,1899LSt.NW,Suite325,Washington,DC20036, USA
91YaleSchoolofForestryandEnvironmentalStudies,NewHaven,CT 06511,USA
92TheAustralianNationalUniversity,Resource,Development&
EnvironmentGroup,Canberra,Australia
93BostonCollege,SchoolofSocialWork,McGuinnHall,Room132, McGuinnHall,ChestnutHill,MA02467,USA
Correspondingauthor:Tallis,Heather([email protected])
CurrentOpinioninEnvironmentalSustainability2019,39:81–93 ThisreviewcomesfromathemedissueonOpenissue
EditedbyEduardoBrondizio,ProfessorOphaPaulineDube,and WilliamSolecki
Received:12December2018;Accepted:15September2019
https://doi.org/10.1016/j.cosust.2019.09.004
1877-3435/ã2019TheAuthors.PublishedbyElsevierB.V.Thisisan openaccessarticleundertheCCBY-NC-NDlicense(http://creative- commons.org/licenses/by-nc-nd/4.0/).
Introduction
Numerous studies have shown the strong links among health, development, and environmental sustainability [e.g. 1,2]. Overlooking these links in research and management can lead to negative unintended conse- quences[3–7];aswellasmissedsynergiesandalimited
viewofviableinterventionstoaddressachallenge[8,9].
Inresponsetoincreasedawarenessoftheselinkagesand theperilsof ignoring them,intergovernmental commit- ments (e.g. Sustainable Development Goals (SDGs), Paris climate agreement) [10] increasingly recognize thefundamentalimportanceofaccountingforfeedbacks and linkages among these sectors. Many efforts have called for integration [e.g. 1,2,8,9,10], yet agendas are dominated by narrowly defined goals [11], funding remains highly sector-specific [12], technical expertise and networks are largely isolated [9,13], professional incentivesfocusonin-sectoradvancement,andthetrain- ingandevidencebasesunderpinningresearchadvances, policies,andactionsremainfragmented[14].
Here,wefocusondescribingandremovingsomebarriers that reinforce a fragmented evidence base, stymieing joint research and action planning across the health, development,andenvironment sectors[2].Each sector already approaches problems by conducting evidence- basedresearch,design,andplanning. Asthecomplexity ofglobalchallenges(suchasclimatechange,large-scale humanmigration,foodandwaterinsecurity,airandwater pollution, urbanization, desertification, and emerging infectiousdiseases)increases,multidisciplinaryandtrans- disciplinaryapproachesexpandandmanyrelevantframe- worksandmethodshaveemerged(e.g.networkanalysis [15]; system integration [16]; ecosystem services [17];
planetaryhealth[2];onehealth[18];nexus approaches [19]; multi-objective planning [20], analysis [21] and decision-making [22]; and socio-ecological action situa- tions[23]).However,theirpracticalusebyindividualsor teams continues to be hampered by the fractured evi- denceavailableandthevaryingandsometimesconflict- ingmethodsusedbydifferentdisciplines.
Thekindsofmultidisciplinaryandtransdisciplinarycol- laborationsneededtosolvetoday’sglobalchallenges[24]
require time to align on terms, methods and standards beforework canproceed. This need for alignment can slowprogress and limitadoptionof existing approaches [24].Inanefforttostreamlinealignmentofmethodsand provideapracticalstartingpointforfurtheriteration,we present a set of principles and methodological recom- mendations for evidence generation and use across health,development,andenvironmentsectors.Wedraw fromreview of the recentliterature and consensusof a diverse set of experts from relevant disciplinary and practice backgrounds (see Supplementary material, TableS1).Ourrecommendationsaddressthreecommon methodologicalbarrierstoevidenceuse;(1)inconsistent design of logic models when developing or assessing interventions; (2) disagreement about admissible evi- dence for evaluatingconfidence;and (3) differentstan- dardsforwhatconstituteshighconfidenceinagivenset of evidence for assessing intervention impacts. Each is describedfurtherbelow.
The first set of methodological challenges we address relatestounderstandinghowaninterventionislikelyto contributeto change(s)inasystem [25].Withintypical research and planning processes, the health, develop- ment,andenvironment sectorseachemploy someform of logical framework to explore the impacts of system changesorinterventions.Frameworkscantaketheform oflogicmodels,logframes,theoriesofchange,orresults chainsindevelopment[e.g.26]andhealthevaluations[e.
g.27], asubsetof social,physical or biological network modelsaddressingcausalinteractions[e.g.15],andmen- tal models, results chains or means-ends diagrams in environmentalplanningandresearch [e.g.28,29].Here, weusetheterm‘resultschain’foralllogicalframeworks thatvisuallyrepresentthecausallogic ofhow interven- tions lead to consequences (positive and negative) throughaseriesofexpectedchanges[20,28].
There is an increasing emphasis on including and representingfeedbacksandinteractionswithinasystem in results chains [30] and depicted causalrelationships canbefurtherexpandedortranslatedintomathematical models (e.g. Bayesian network models, earth system models, or many other types). Relationships within models can be quantified with data drawn from an increasingly wide range of sources (e.g. survey data, direct observations, smart sensors, remote-sensing drones, satellites, big data processed by computer algorithms,etc. [31–35]).
Whileresults chains ofsome form are used by health, development,and environmentsectors,methodological challenges and variations limit their effective use for cross-sector problems. The creation of results chains fromsinglesectorentrypointscanfailtoidentifynega- tiveunintendedconsequencesthatposeriskstoproject success or to other aspects of the system. Cases of unintended impacts from one sector on another are abundant.Forexample,expansionofbiofuelstoreduce fossilfueluseandstabilizetheglobalclimatecancause local food insecurity [3]. In other examples, nature conservationintendedtosavebiodiversitycanuninten- tionally worsen inequalities in local communities by reducing access to land or resources [4] or by driving inconsistent access to markets or resources [5]. Eco- nomic development programs aimed at improving irri- gationcanincreasewaterdepletion,environmentaldam- age, and agricultural risk in some cases [6] and can increasemalaria riskinothers [7].
In addition, single sector results chains can overlook positive unintended consequences and synergies (also calledco-benefits), leadingto conservativeexpectations abouttotalsystemimpacts,miscalculationoftotalreturn in investment,and missed opportunitiesfor implemen- tationwithothersectors[8,9].Forexample,reproductive health and conservation programs can have greater
impactsonbothhealthandtheenvironmentwhenimple- mentedtogethercomparedtothesameprogramsimple- mented in parallel [8]. When research or practitioner groupsdoexpandonsinglesectorresultschains,lackof knowledgecanleadtogenericrepresentationsofcausal pathways and impacts (e.g. a conservation intervention leadingdirectly to‘communityresilience’ oradevelop- mentinterventionleadingto a‘healthierenvironment’).
Planning for and evaluatinginterventions fromasingle sectorperspectivealsoleadstoamyopicviewofsolutions, resultinginoverlookedinterventionsandmisinterpreta- tions of what the most effective solution may be. For example, a hypothetical case of environment, develop- ment, and health results chains constructed for single- sector outcomes (Figure 1a) shows how this view can overlookthepotentialfortheenvironmentanddevelop- ment interventions to deliver on health benefits
(Figure1b).Ifsectorsusedconsistentmethodstocreate results chains, a systems view could more readily be taken, revealing bothpositive andnegative unintended consequencesinothersectorsandidentifyingthefullset of viablecandidate interventions.
Asecondsetofmethodologicalchallengerelatestodiffer- encesinthetypesofevidenceconsideredadmissiblefor determining confidence in potential impacts. Results chainsarecommonlyusedasabasisforstructuredsynthesis of evidence to evaluate the confidence in intervention effectiveness[20,26].Toimproveconsistency,sectorssup- port effortstostandardizetheinterpretationof evidence within their own communityso thatresearchers, practi- tioners, and policy makers can work from a consistent understanding (e.g. Cochrane, Campbell Collaboration, 3ie, Conservation Evidence, Environmental Evidence).
Nascent efforts (e.g. Evidence Synthesis International,
Figure1
Mechanical thinning
Fire intensity
Outdoor air pollution
Population exposure
Respiratory disease symptoms
Micro solar subsidy
Micro solar adoption
Fuelwood use rate
Local household electrification
Indoor air pollution
Respiratory disease symptoms Tree density
for population of concern
Tree density for population of
concern
Respiratory inhalers provided
Respiratory disease symptoms Inhaler
adoption (a)
(b)
Current Opinion in Environmental Sustainability
Simplifiedsingle-sector(a)andcross-sector(b)viewsofthreeinterventions.
Typicalresultschains,suchasthehighlysimplified,hypotheticalchainsin(a),relateinterventions(greynodes)toexpectedsector-specific impactsontheenvironment(greennode),development(bluenode)orhealth(orangenode).Byexpandingtheviewacrosssectors(b),results chainscanhelpidentifyabroadersetofsolutionsandamorecompleteunderstandingofconsequences.Solidarrowsrepresentpositive relationships,dottedarrowsrepresentnegativerelationships.
Global EvidenceSynthesisInitiative) areemergingto more fullyalignexistingevidencestandardsacrosssectors,but majorchallengesremaininharmonizingmethods.
First, there are different views among (and sometimes within)disciplines on thetypes of information that are admissible as evidence for this use. For example, the healthsectorreliesonaspecificsetofmethodstoinform theevidence base on interventions or treatments, with large, randomized controlled trials serving as the gold standard[36,37].Viewsinthemedicalfieldareexpand- ing.Forexample,CochraneReviewsnowallowinclusion of non-randomizedstudiesandother formsof quantita- tive studies, economic data, qualitative studies, and equityconsiderations[36],whilemethodsfor additional evidencetypesare underdevelopment.Large,random- ized trials are often not feasible, nor sensible in the environmentsector;hencealternativeformsofevidence arecommonlyused[38].Economicandsocialdevelop- ment researchers hold diverse views, some aligning closelywithhealthcommunitiesinpursuingexperimen- tal or quasi-experimental methods, while others adopt case studies, mixed and comparative methods, mathe- maticalmodels, triangulationand causalmechanisms as viableevidenceforms[39].
As each sector or discipline follows its own standards, different subsets of evidence are admittedfor analyses, possiblyleading to different levels of confidence in the sameintervention.Forexample,considerforestfuelman- agement (such as thinning and debris removal) as an interventionforreducingfires,smokeexposureandrespi- ratorydiseaserisk.Availableevidenceoneffectivenessof this intervention consists of several large-scale pseudo- experiments and models [e.g. 40,41]. Some ecologists would readily admit this evidence, while some health expertswouldnot,leadingtoevaluationsofdifferentsub- setsofevidence,andlikelyinconsistentconclusions.
Within these same standards, we find the third major methodologicalbarrierweaddress;differencesinhowto assessthestrengthofadmittedevidence.Evaluationsof the strength of evidence are commonlydone to create confidencestatements,whichcaninformdecisionsabout whether and how to proceed with an intervention. For example,if there islow confidence in alink in achain (Figure 2) that is high risk and/or of importance to stakeholders,decisionmakersmaychoosenottogoahead withanaction,identifyadditional interventions,modify theinvestmenttomitigaterisks,orinvestinmonitoring andevaluationtoincreaseunderstanding.Manymethods for establishing confidence statements have been advanced, some through standard setting bodies (e.g.
GRADE[42],IPCS/WHO[43]).Effortsintheenviron- mentsectorhavebeenmorediffuse(e.g.[44],IPCC[45], IPBES[46],USNationalClimateAssessment[47]),and thereisnoacceptedevidence standard-settingbody.
Differences in standardsand lack of consensusmake it challengingtouseanyoneexistingmethodforconfidence statementswhenevidence isused frommultiplesectors.
Somemethodsaresetupformulti-disciplinaryapplication (likeIPCC,IPBES,USNCA),buteachisbuiltforpurpose ratherthanworkingfroma consistentsetof methods or assumptions.Thiscanmaketheiruseincompatibleacross disciplines.Forexample,theIPCCandtheInternational Agencyfor ResearchonCancerrubricshave madesome cross-sectorconsiderations,buttreattheorydifferentlyasa type of evidence [48]. Bespoke standards also limit the comparisonoftrendsovertimeorthecomparisonofinter- ventions across sectors (e.g. each topical IPBES report createsitsownconfidencestatementmethod).
An emergent research-practice collaboration, called the Bridge Collaborative, was created and joined by the authors of this paper to address some of the noted challengesin evidence use acrosssectors.Aswe sought to find consensus across disciplines and streamline the alignmentprocessforfutureefforts,threeaspectsofthe Bridge Collaborative process made the findings here novel:(1) thebreadth of global challenges, sectors and disciplinaryperspectivesincluded;(2)thefocusoncon- sensus across this broad range of disciplines and chal- lengesratherthansynthesisordiscussionofdifferences;
and(3) theuse of iteration betweenspecificchallenges andgeneralizableagreements.
Througharapid,iterativeprocess,over100expertsfrom 80research,practice,privatesectorandmultilateralorga- nizations engaged in six multi-sector working groups.
Collaborativemembersleador engageinmanyexisting networksandcross-sectorefforts(e.g. Locus;Scalingup Nutrition;Agriculture-Nutrition Communityof Practice (Ag2Nut);OneHealth;EAT;FutureEarth;GlobalEvi- dence Synthesis Initiative; Planetary Health Alliance;
Cochrane;ConservationEvidence;Food,Energy,Envi- ronment, and Water Network; CGIAR Agriculture for NutritionandHealth;CGIARWater,Land,andEcosys- tems; USAID’s BRIDGE Project;others), providingan opportunity for groups to learn from, find generalities among,andamplifytheseinitiatives.
Theprocessfocusedonreachingconsensusaroundmeth- odsthatarerelevanttoawiderangeofglobalchallenges andacceptableacrossdisciplinesandsectors.Thegroup did not focus on synthesis and summary but rather on agreement, elevating principles and methods that all participants endorsed from their various perspectives.
Pasteffortsto findsuch consensustypically focusedon a single challenge (e.g. climate change, food security), ratherthanlookingbroadlyacrossadiversesetofglobal challenges. Working group foci included: stabilize the global climate; make food production sustainable;
decrease air pollution and respiratory disease; improve
sanitation and water security; and solve hunger and malnutrition(two groups).
The nine-monthconsensusprocess startedwithawork- shopattendedbytheco-leadsofallsixworkinggroupsand theBridgeCollaborativeSecretariat.Eachworkinggroup thenprogressedindependentlytoreviewrecentrelevant disciplinaryliteratureanddrawfromtheirownexperiences togeneraterecommendationsforprinciplesandmethodo- logicalsolutions.Thesixinitialsetsofrecommendations werecompiledandsynthesizedbytheBridgeCollabora- tiveSecretariatandusedasthebasisfordiscussionsinanin- person meetingof allworking group co-leads.Liveline editing continued until consensus was reached on all recommendations.Additionalfeedbackwasincorporated fromaroundofreviewbyallcontributingauthors,anda secondroundofreviewfromworkinggroupco-leads.The process allowed for effective iteration between topical working group foci that grounded thinking in practical challengesandthecreationofgeneralizedrecommenda- tions that tested the applicability of suggestions across contextsanddisciplines.
Althoughourframingandparticipantswerediverse(see Supplementary material,Table S1), they were notrep- resentative of all disciplines, sectors or relevant
challenges. We present the following principles and recommendationsasastartingpointfor furtheriteration andtesting inabroadersetof contextsanddisciplines.
Principlesfor effectivecross-sector collaboration
Methodologicalsolutionstothechallengesreviewedabove arelikelytoemerge fromand beappliedthroughsomeform ofcross-sectorcollaboration.TheBridgeCollaborative,as onesuchcollaboration,adoptedandreinforcedsixprinci- plesthatweredeemedvaluableforadvancingcross-sector interactionsaroundevidenceuse[9].Theseprinciplesmay aidtransdisciplinaryandcross-sectorgroupsapplyingthe methodologicalrecommendationsthatfollow.
Useevidencetoinformdecisions
Thehealth,development,andenvironmentsectorshave long recognizedthebenefitsofevidence-baseddecision making [49,50].
Actnowandlearnbydoing
Weacknowledge thatintentionallearning bydoingcan improve actions and impact evenwhilethere is incom- plete understanding, evidence, or political or social alignment. This principle forms the basis of adaptive management, evidence-based management, and action
Figure2
Intervention Other Factors
Other Factors
F
L F
L M
M H
H
Intended Outcome or Impact
Unintended Outcome or
Impact Intended
Change in System
Unintended Change in
System
Current Opinion in Environmental Sustainability
Generalizedresultschainconstructedusingrecommendationsforcompatibleresultschainsandevidenceevaluation.
Arrowsreflectanincrease(solidarrow)ordecrease(dottedarrow)intheendpointnode,arrowweightindicateseffectsize(thickerarrowsshow largereffectsizes,thinnerarrowsshowweakereffectsizes),andarrowcolorindicatestimescaleofchange(blackarrowschangequickly,grey arrowschangeslowly).Additionalgraphicalsymbolscanbeaddedtoreflecttheconfidenceintheassumptionunderlyinganarrowgivenavailable evidenceevaluatedusingtheunifiedrubric.Confidencecanbehigh(H),moderate(M),fair(F)orlow(L).
researchapproacheschampionedextensivelybytheenvi- ronment[51],developmentandhealthfields[52].These approachesallemphasizetheneedtoplanforlearning,as itisnotguaranteedto happenonitsown.
Seekandrespectotherperspectives
Many barriers to multi-sectoral action will be reduced overtimebyadoptionoftheprinciplethatgoalsin one sector may be met more effectively, efficiently or sus- tainablybyembracingideas, interventions,methods, or conceptsfromothersectors [12,14].Preliminaryexperi- encesoftheBridgeCollaborativesuggestthatevenbrief (<1 day) opportunities for people with expertise and experiences from different sectors to problem solve together can lead to rapid transformation in problem framing,strategicplanning,and evidenceuse.
Beintentionalaboutinclusion
The value of inclusion of people from diverse back- grounds (disciplinary, geographic, race,culture,gender, age, etc.) and information from diverse sectors and sources has been shown in many fields. Guidance and tools for increasing inclusion are well established for usewithinhealth,development,andenvironmentsectors [e.g.53,54].Existing guidancemaybeequallyusefulin cross-sectorengagements.
Strivetodonoharm
Cross-sectoral efforts that fail to prevent or mitigate negative outcomes for other sectors, groups, or future generationsarelikelytobeshort-livedandineffectiveat balancing multiple objectives. Tools and methods for identifying tradeoffs and synergies are available [55]
and could be applied more widely. When negative impacts or inequitable outcomes are expected, they should beavoided or reduced and assistanceshould be providedto thosewhoareharmed [55–57].
Shareinformationopenlyandtransparently
Lack of openness and transparency across sectors may leadtomistrust,misunderstandings,increasedtransaction costs, inefficiency, overlooked options, and short-lived partnerships[58].Weencouragealltosharedata,frame- works,conceptsandsoftwarequickly,openly,andtrans- parently (respecting anonymity, privacy, and security concerns), and to recognize, articulate, and challenge barriersto doingso.
Methodologicalrecommendations for cross- sectorevidence use
The Bridge Collaborative made methodological recom- mendations to advance three key challenges in the detailedpracticeof usingevidencefrom multipledisci- plinesininterventiondesign:(1)createmorecompatible resultschains;(2)agreeonadmissibleevidence;and(3) use a consistent standard for confidence statements.
These recommendations focus on removing remaining barrierstotheuseofevidenceacrossmultipledisciplines andchallenges.
Creationofcompatibleresultschains
Whilegeneralguidanceforuseofresultschainsisabun- dant, it varies across and withinsectors, often creating confusingorconflictingstartingpointsforteamsapplying multi-objectivemethodsortakingamultidisciplinary or transdisciplinary approach [20,26–28,59]. To streamline the use of evidence across sectors, we generated eight recommendationsforharmonizingmethodsandimprov- ing the cross-sectoral compatibility of results chains (Box1).Intheir simplestform,theserecommendations suggestthatresults chainsshould bemadeup of nodes that represent drivers (including interventions), media- torsoroutcomes(intermediate orfinal),andarrowsthat represent hypothesized causal relationships (Figure 2).
Thisalignswithsomerecommendations[e.g.20,26]but differsfromothersthataremorespecializedforparticular disciplinaryuses(forexample,directedacyclicgraphsin epidemiology[60]).
Whiletherecommendationsmayseembasic,theauthors consideredeachoneimportant tocreate enoughconsis- tencyforcomparisonandintegrationacrosssectors,orto surfaceandaddresschallengesthatcommonlyarisewhen extendingresultschainsfromsingle-sectortocross-sector applications. For example, time scales of impacts may varydramaticallyacrosssectors and commonlyresultin some unintendedconsequences (e.g. longer term envi- ronmentalorequityimpactsarecommonlyoverlookedfor nearertermdevelopmentorhealthgains).Assuch,time scales should be represented when possible (Box 1, Recommendation 3). Thesetemporal trade-offs canbe demonstrated through the example of promoting women’shusbandryofanimalswithlowerenvironmental footprints (e.g. chickens instead of goats or cattle) that mayhave short-term effects onchildren’s growth rates andothernutritionaloutcomesandlonger-termimpacts
Box1 Guidanceforcompatibleresultschains
1Arrowspointfromcausetoeffectforeachlink.
2Arrowscangraphicallyrepresenteffectsizeand/orwhethereffect ispositiveornegative.
3Arrowscangraphicallyreflectexpectedtimescaleofchange.
4Eacharrowreflectsonlyonehypothesizedandtestablecausal relationship.
5Nodescapturedriversand/orconsequences.
6Nodesdonotcapturethedirectionofchange,butarrowscan(see
#2).
7Nodesdonotrepresentactors,stakeholders,orcontextwithout beingassociatedwithadriverorconsequence.
8Impactsincludedinthechainaremeasurableorobservable.
onincomeresiliency,women’sempowerment,education attainment,andenvironmental conditions.
Severalresultschainrecommendationssupportaconsistent andusefullevelofsensitivityandspecificityacrosssectors, helpingtoavoidtheuseofvagueconceptssuchas‘human well-being’, ‘community resilience’, or ‘wildlife’. While usefultounderstandgeneralconnections,thesetermsare notsufficientlyprecisetoguidehypothesisdevelopment, interventionselection,ormetricdevelopment.Werecom- mendavoidingthesegeneralitiesbycreatinglinksin achain that reflect only one hypothesized and testable causal relationship (Recommendation 4). In some instances, it maybeusefultoconstructchainswithlinksthatdoreflect morethanoneexpectedcausalrelationshipwhencomplex- ity underlying the link is expressedelsewhere(e.g. in a complex,dynamicmodel),evidenceforspecificlinkshas been explored and found to be lacking, or when it is necessary to simplify for larger scale considerations or communicationwithstakeholders.Wefurtherrecommend thatnodesonlyreflectspecificgroupsofpeopleorelements ofcontextiftheyarespecifiedasadriverorconsequence (Recommendation7),andthatpositedimpactsbemeasur- ableorobservable(Recommendation8).Forexample,an initialvagueideathatconservationmayimpact‘localcom- munities’onfurtherprobingmayrevealthattheexpected impactisongenderequityinassetsinlocalcommunitiesor diversity of food sourcesin localcommunities. The latter are much morespecificandmeasurableelements.Graphical inclusionofallsuggestedtypesofinformation(Figure2) may bemoreconfusingthanclarifyinginsomecontexts.
Theintentoftheserecommendationsistospurthinking aboutcriticalelementsforconsiderationandtoencourage researchersandpractitionerstoexploreanddocumenteach oftheseelementsasuseful.
Applying theserecommendationswould leadto thepro- duction of results chains able to consistently represent interventionsandpotentiallyquantifyimpactsformultiple sectors (Figure 2). Beyond the simplified, hypothetical examplesprovidedhere,therecommendationshavebeen usedtocreateresultschainsformorecomplexcontextswith feedbacksandinteractionsthatinclude;pesticidetaxesand habitatsubsidiesasalternativeinterventionsinsustainable agriculture [25],solarenergyinstallationonpublic lands [25,61],oysterreefrestorationinvestmentsintheGulfof Mexico[62],andsaltmarshhabitatrestoration[63].These applicationsprovidesomesuggestionthattherecommen- dations are relevantto abroader set of challenges. The generalizabilityoftheserecommendationswillbefurther improvedthroughcontinuedtestinganditeration.
Admissibleevidence:whatcanbeincluded?
Once results chainsare created, one candeterminethe strength of confidencein causalpathways andpotential impacts.Thefirststepincreatingconfidencestatements is to determine what qualifies as admissible evidence.
Recognizingtheneedforinclusive,cross-sectorproblem solving,werecommenddrawingonallrelevanttypesof evidencefrominvolvedsectors.Weconsideradmissible evidence to include quantitative studies, qualitative studies, theory, model results, expert, and tacit knowledge(includinglocalknowledge,traditionalknowl- edge,subjectmatterexpertise),andmeasurementresults.
Though some advocate for a more narrow definition of evidence, other groups support a similarly broad definition [44,64–66].
Ensuringcoverageofallrelevantandavailableevidence will requireinclusion ofperspectivesfrommultipledis- ciplines, sectors, and sources. Relevant guidance exists for includinglocal andtraditionalknowledgein climate change initiatives [67], health and economic or social development approaches [e.g. 68], and conservation assessments [e.g. 69]. Searches for evidence may be broadened by looking across multiple languagesources as wellasexpanding keywordlistsandexpertandlocal networks.
Strengthofevidence:whatcreateshighconfidence?
The secondstep increatingconfidencestatementsisto assess the strength of admitted evidence. To address inconsistenciesinthisstepacrosssectors,werecommend assessing confidence(Figure 2) by applying a common andconsistentrubric(Table1).Hereweprovidearubric withconfidencecriteriathatdrawfrommultipleexisting frameworks (e.g. [45], IPCC [49], IPBES [46], US National Climate Assessment [47], GRADE [49], IPCS/WHO[43]),andwereagreeabletoBridgeCollabo- rative members spanning the health,development and environmentsectors(Table1).Inthisrubric,confidence isbasedonthediversityoftypesofevidence,consistency of results across evidence, status of methods used to generateevidence,andapplicabilityofavailableevidence to thestudycontext.
Thisrubricimprovesonsomecritiquesofexistingframes [43,70]butleavesothersunaddressed[70].Oneadvance is to more clearlyspecify elements of high-quality evi- dence,heredetailedascertaintyofmethodsandapplica- bilityofevidence.Inaddition,ourspecificationofconfi- dence criteria may improve consistency of evidence interpretation by trans-disciplinary project teams and majorassessmentprocessesthatdonothaveastandard- izedconfidencerubricoralignmentbody(e.g. theenvi- ronmental community, and environmental assessments suchas thoseconductedbyIPBES).
The proposed rubric includes four confidence levels (Table1).Highconfidencecanbestatedwhenmultiple typesofevidence(e.g.randomizedcontroltrials,system- atic reviews,model results, and qualitative focus group results)supportahypothesis,resultsareconsistentacross sources, types of evidence and contexts, methods used
