Intervention strategies to reduce the burden of non ... - The BMJ

Intervention strategies to reduce the burden of non-communicable diseases in Mexico: cost effectiveness analysis. BMJ 2012; 344 doi: ... Skiptomaincontent Research Intervention... Interventionstrategiestoreducetheburdenofnon-communicablediseasesinMexico:costeffectivenessanalysis CCBYNCOpenaccess Research Interventionstrategiestoreducetheburdenofnon-communicablediseasesinMexico:costeffectivenessanalysis BMJ 2012; 344 doi:https://doi.org/10.1136/bmj.e355 (Published02March2012) Citethisas:BMJ2012;344:e355 Article Relatedcontent Metrics Responses Peerreview JoshuaASalomon,associateprofessorofinternationalhealth1,NatalieCarvalho,doctoralstudent2,CristinaGutiérrez-Delgado,deputydirectorgeneralofeconomicsandhealth3,RicardoOrozco,analyst4,AnnaMancuso,clinicalsocialworker5,DanielRHogan,postdoctoralfellow6,DianaLee,doctoralstudent7,YukiMurakami,healtheconomist/policyanalyst8,LakshmiSridharan,residentininternalmedicine9,MaríaElenaMedina-Mora,directorgeneral4,EduardoGonzález-Pier,directoroffinance101DepartmentofGlobalHealthandPopulation,HarvardSchoolofPublicHealth,665HuntingtonAvenue,Boston,MA02115,USA2HarvardUniversity,Cambridge,MA3UnidaddeAnálisisEconómico,SecretaríadeSalud,MéxicoDF,Mexico4InstitutoNacionaldePsiquiatríaRamóndelaFuenteMuñiz,MéxicoDF5BostonCenterforRefugeeHealthandHumanRights,BostonMedicalCenter,Boston,MA6HarvardSchoolofPublicHealth,Boston,MA7UniversityofCalifornia,Berkeley,CA8OrganisationforEconomicCo-operationandDevelopment,Paris,France9UniversityofCalifornia,SanFrancisco,CA10InstitutoMexicanodelSeguroSocial,MéxicoDFCorrespondenceto:JASalomonjsalomon{at}hsph.harvard.eduAccepted6November2011AbstractObjectiveToinformdecisionmakingregardinginterventionstrategiesagainstnon-communicablediseasesinMexico,inthecontextofhealthreform.DesignCosteffectivenessanalysisbasedonepidemiologicalmodelling.Interventions101interventionstrategiesrelatingtoninemajorclustersofnon-communicabledisease:depression,heavyalcoholuse,tobaccouse,cataracts,breastcancer,cervicalcancer,chronicobstructivepulmonarydisease,cardiovasculardisease,anddiabetes.DatasourcesMexicandatasourceswereusedformostkeyinputparameters,includingadministrativeregistries;diseaseburdenandpopulationestimates;householdsurveys;anddrugpricedatabases.ThesesourcesweresupplementedasneededwithestimatesforMexicofromtheWHO-CHOICEunitcostdatabaseorwithestimatesextrapolatedfromthepublishedliterature.MainoutcomemeasuresPopulationhealthoutcomes,measuredindisabilityadjustedlifeyears(DALYs);costsin2005internationaldollars($Int);andcostsperDALY.ResultsAcross101interventionstrategiesexaminedinthisstudy,averageyearlycostsatthepopulationlevelwouldrangefromaround≤$Int1m(suchasforcataractsurgeries)to>$Int1bnforcertainstrategiesforprimarypreventionincardiovasculardisease.Widevariationalsoappearedintotalpopulationhealthbenefits,from<1000DALYsavertedayear(forsomecomponentsofcancertreatmentsoraspirinforacuteischaemicstroke)to>300 000avertedDALYs(foraggressivecombinationsofinterventionstodealwithalcoholuseorcardiovascularrisks).Interventionsinthisstudyspannedawiderangeofaveragecosteffectivenessratios,differingbymorethanthreeordersofmagnitudebetweenthelowestandhighestratios.Overall,communityandpublichealthinterventionssuchasnon-personalinterventionsforalcoholuse,tobaccouse,andcardiovascularriskstendedtohavelowercosteffectivenessratiosthanmanyclinicalinterventions(ofvaryingcomplexity).Evenwithinthecommunityandpublichealthinterventions,however,therewasa200-folddifferencebetweenthemostandleastcosteffectivestrategiesexamined.Likewise,severalclinicalinterventionsappearedamongthestrategieswiththelowestaveragecosteffectivenessratios—forexample,cataractsurgeries.ConclusionsWidevariationsincostsandeffectsexistwithinandacrossinterventioncategories.Foreverymajordiseaseareaexamined,atleastsomestrategiesprovidedexcellentvalueformoney,includingbothpopulationbasedandpersonalinterventions.IntroductionIn2003,MexicointroducedamajorhealthreformthatcreatedtheSystemofSocialProtectioninHealth(SSPH).1SSPHgeneratednewfinancialrulestofundpopulationbasedinterventionsforallMexicansregardlessoftheirinsurancestatusandpersonalhealthcareinterventionsforthosewithoutaccesstosocialsecurity(abouthalfofthetotalpopulation).ThelatterwerefinancedthroughaninsurancebasedcomponentcalledSeguroPopular.Throughthereforms,publicexpenditurefortheuninsuredincreasedsubstantially,2withtheoverallbudgetfortheMinistryofHealthrisingbyalmostfourtimesinrealtermsbetween2001and2010.3Bytheendof2010,thenumberofSeguroPopularbeneficiarieshadreachedover43million,88%ofthepreviouslyuninsuredtargetpopulation.4Duringtheplanninganddesignofthereform,rigorousevidencewasneededonthemagnitudeofdifferenthealthproblemsandonthebenefitsandcostsofdifferenthealthinterventions.Decisionstoincludenewinterventionsinexplicitlydefinedpackagesofserviceswereinformedbyadeliberativeprocessthatincludedananalyticalprioritysettingexercisebasedonmeasurementoftheburdenofdiseaseandcosteffectivenessofdifferentcandidateinterventions.5StatesinMexicoarerequiredtoprovideallinterventionsincludedinthreeminimumpackagesofservicescomprisingcommunityandpublichealthinterventions;lowandmediumcomplexityclinicalservices;andhighcomplexityclinicalservices,eachfinancedthroughseparatefundingmechanisms.Aspartoftheprocessofdefiningthecontentofthesepackagesduring2004–6,weundertookcosteffectivenessanalysesforawidearrayofhealthinterventionsspanningmajorcausesofdiseaseburdeninMexico.Theoverallprocessofdefiningthecontentoftheservicepackageswasintendedtobeevidencebased,equitable,transparent,andcontestable.Afterthereform,evidenceoncosteffectivenesscontinuestoinformdecisionmakingregardingamendmentstothepackagesofservicescoveredbySSPH;statelevelpoliciesregardingcoverageofinterventionsadditionaltothedefinedminimumpackages;andbroaderdebatesovertheadvantagesanddisadvantagesofexplicitpackagingbasedinpartoneconomicevidence—forexample,amongsocialsecurityinstitutionsinMexicothatdonotcurrentlybasecoveragedecisionsonexplicitpackagesofinterventions.Thispaper—aspartofaseriesonthecosteffectivenessofinterventionsfornon-communicablediseaseandinjuryineconomicallydevelopingregionsoftheworld—reportsoncosteffectivenessanalysesfor101interventionstrategiesdirectedatninemajorclustersofnon-communicablediseasesinMexico.MethodsOuranalysesfocusedoninterventionsrelatedtothefollowingninediseaseareasthataremajorcontributorstotheoverallburdenofdiseaseinMexico56:depression,heavyalcoholuse,tobaccouse,cataracts,breastcancer,cervicalcancer,chronicobstructivepulmonarydisease,cardiovasculardisease,anddiabetes.InterventionsforanalysiswereselectedinconsultationwiththeMinistryofHealthinMexico,basedonpolicyprioritiesandongoingdebatesregardingthecontentofpackagesofservicesinSSPH.Interventiondefinitionsweredevelopedaccordingtostandardsofqualitycareandavailableevidenceoneffectiveness.Thetable⇓summarisesthemaintypesofinterventionsanalysedinthisstudy,notingthespecificpackageofservicesforwhicheachtypeofinterventionwasconsidered.Thetechnicalappendixonbmj.comprovidesafulllistinganddefinitionsofthe101specificinterventionsthatweevaluated.Summaryofinterventionstrategiestoreduceburdenofnon-communicablediseasesinMexico,bymajordiseaseclusterViewthistable:ViewpopupViewinlineTheoverallanalyticalapproachusedinthisstudyadheredtorecommendationsforundertakinggeneralisedcosteffectivenessanalysisintheWHO-CHOICEframework.7WherepriorregionalanalyseswereavailablefromtheWHO-CHOICEproject,8910111213weusedsimilarinterventiondefinitions,modellingapproaches,andcostingmethods.Herewesummarisethemethodsandassumptionsusedintheanalyses.Furtherdetailsareprovidedinthetechnicalappendixonbmj.com.AnalyticoverviewTocomputetheeffectivenessofaninterventiontargetingaparticularhealthproblemintermsofnetchangesinpopulationhealth,wefirstdefinedthecurrentepidemiologyofthehealthproblem,buildingonalinkedstudytomeasuretheburdenofdiseaseinMexico.6Interventioneffectivenesswasexpressedintermsofchangesindiseasemodelparameters—thatis,aschangesinratesofincidence,prevalence,casefatality,orremissionorchangesinhealthstatevaluationsthatreflecttheseverityofaparticularhealthoutcome.Weusedapopulationmodeltotranslateinformationondiseasedynamicsintogeneric,comparablemeasuresofpopulationhealth,expressedasdisabilityadjustedlifeyears(DALYs).InlinewithWHOglobalburdenofdiseaseestimation,14theMexicannationalburdenofdiseaseassessmentrelatedtothisproject,6andstandardWHO-CHOICEmethods,7DALYsavertedwerediscounted(at3%ayear)andageweighted(seealsothegeneralappendixonbmj.com).Costswereevaluatedfromasocietalperspective,withinthreebroadcategories:patientcosts,programmecosts,andtrainingcosts.Costswereexpressedasinternationaldollars($Int)at2005prices,discountedat3%ayearaccordingtoCHOICEstandards.7Internationaldollarsrepresentahypotheticalcurrencythatallowsforthesamequantitiesofgoodsorservicestobepurchasedregardlessofcountry,standardisedonpurchasingpowerintheUnitedStates.DatasourcesWeusedMexicandatasourceswhenpossible,drawingextensivelyonfourmainlocalsources:administrativeregistries,populationestimates,householdsurveys,anddrugcostdatabases.AdministrativeregistrydataInformationonagespecificmortalityandcausesofdeatharecompiledfromdeathcertificatesbytheMinistryofHealthandtheInstitutoNacionaldeEstadísticayGeografía.Additionally,publicinstitutionsprovidinghospitalservicesmaintaindischargeregistries,whichincludepatientlevelinformationonsociodemographiccharacteristics,causesofadmissiontohospitalclassifiedbyICD-10codes(internationalclassificationofdiseases,10threvision),diagnosticandtreatmentprocedurescodedaccordingtoICD-9-CM,resultofprocedures,mortalityinhospital,reasonsfordischarge,numberofhospitalbeddays,andinsurancestatus.Forthepresentanalyses,weusedmortalitydatafromtheyear2004.WeuseddischargeregistriesfromMinistryofHealthhospitalsfortheyears2000–5andfromhospitalsaffiliatedwiththeInstitutoMexicanodelSeguroSocial(IMSS)fortheyears2004and2005.PopulationestimatesPopulationestimatesfor2000–5wereprovidedbytheConsejoNacionaldePoblación(CONAPO),whichdevelopsyearlyprojectionsofpopulationnumbersbyage,sex,state,andinsurancestatus.ProjectionsaremadebasedonofficialsurveysincludingtheNationalSurveyonFertilityandHealth,censusesandspecialsurveysrelatingtomigration.PopulationhouseholdsurveysTheEncuestaNacionaldeSaludyNutrición2005–6(ENSANut)wasanationallyandstaterepresentativesurveythatsampled47 695householdsand206 700individuals.ModulesinENSANutincludedinformationonhouseholdcharacteristics,healthinsurance,riskfactors(smokingandalcoholuse),healthstates,anduseofservices.Biomarkerswerecollectedonconcentrationsofcholesterol,plasmaglucose,andhaemoglobinA1candbloodpressure.ENSANut2005–6wasusedinthisanalysisprimarilytomeasurecurrentcoverageofselectedinterventions.DrugcostdatabasePublicinstitutionsprovidinghealthcareinMexicoarerequiredtopurchaseonlydrugsincludedintheCuadroBásicodeMedicamentos(MexicanPositiveList,MPL).Publicpurchasingregulationsrequirepublicbiddingformultiplesourcedrugs,althoughthereisnogeneralconsolidatedpurchasesystemforthewholepublichealthsector,asallinstitutionsareessentiallyindependent.Weusedthe2005purchasepricedatabaseavailablefromIMSS,whichisthelargestpublicpurchaserinMexico.EstimationofinterventioneffectsEpidemiologicalestimatesFormostinterventions,currentepidemiologyofrelevantdiseasesorinjurieswasdefinedbasedonresultsfromtheMexicanburdenofdiseaseanalysis,6expressedintermsofageandsexspecificincidence,prevalence,casefatality,remission,andmortality.Vitalregistrationdatafor2004wereusedtoestimatemortalitybyage,sex,andcause.Totalmortalityfiguresbyageandsexwereadjustedwithstandarddemographictechniquestoaccountforunder-recordingofdeathsatcertainages,misreportingofageonthedeathcertificate,andmigration.1516Estimatesforcausesofdeathwereadjustedfollowingstandardalgorithmsforredistributingdeathscodedto“illdefined”diseaseorinjurycategories,cancersofunknownsites,andcardiovasculardisease,aswellasmiscodingofdiabetestocardiovascularorotherchronicdiseases.6171819Estimatesoftheincidenceofdifferentdiseasesandtheirrelevantsequelaebyageandsexwerederivedfromacombinationofsourcesandimputationapproachesappliedtodifferentgroupsofcauses.6InterventiondefinitionsTheselectionandspecificationofinterventions,incaseswhereaWHO-CHOICEregionalanalysiswasavailable,correspondedcloselytotheinterventionsdefinedintheprioranalysis.IncaseswherenoWHO-CHOICEregionalanalysiswasavailable,thechoiceofinterventionswasguidedbyconsultationwithexpertsinMexicoorbyexistingnormsforclinicalpracticeinMexico.Definitionofinterventions,ingeneral,wasasexplicitaspossibleandspecifiedbothimmediatecomponentsofinterventiondeliveryaswellasthetypesofactionsthatwouldbeundertakeninresponsetodownstreamconsequencesofthediseaseprocessoroftheinterventionitself.InterventioneffectivenessThedefinitionofthecurrentepidemiology,combinedwithinformationoncurrentcoverageandeffectivenessofinterventions,servedastheanalyticstartingpointfordefiningallinterventionscenarios,includingthenullscenario.FollowingthestandardWHO-CHOICEapproach,wederivedinterventioneffectivenessfrommeta-analysesandsystematicreviewswherethesewereavailable.ForinterventionsthatwereexaminedpreviouslyinWHO-CHOICEregionalanalyses,wemaintainedconsistentassumptionsaboutinterventionefficacyunlesstherewassufficientevidencetosuggestdifferentoutcomesinMexico.Forincorporationintothepopulationmodeldescribedbelow,interventioneffectivenesswastranslatedintochangesinrelevantdiseasemodelparameters.Forexample,theeffectsofprimarypreventionwereexpressedaspercentreductionsinageandsexspecificincidenceratesfromaparticularcondition,whiletreatmentinterventionswereallowedtoaffecttransitionstootherdiseasestates,remissionrates,casefatalityrates,orhealthstatevaluations.Detailsontheparametersofeffectivenessforspecificinterventionanalysesareprovidedinthetechnicalappendixonbmj.com,includingafulllistingofdatasourcesoneffectiveness.PopulationhealthoutcomesPopulationhealthoutcomesunderdifferentinterventionscenariosweremodelledwiththemultistatepopulationmodelPopMod20(seegeneralappendixonbmj.com)oranalogoustoolsdevelopedforthisprojectwheremoredetaileddiseasemodelswererequired.Interventionswerecomparedwithanullscenario,whichwassimulatedbyalteringthebaselineepidemiologicalparameterstoremovetheestimatedeffectsofcurrentinterventioncoverage.Forinterventionscenarios,effectivenessestimateswereadjustedtoaccountfortargetpopulationcoveragelevelsandproviderandpatientadherencetointerventions.Eachinterventionscenarioassumedimplementationoftheinterventionfora10yearperiod,butthepopulationmodelcapturedalleffectsovera100yeartimehorizon.Forboththenullandinterventionscenarios,theinputstothepopulationmodelincludedratesofincidence,remission,andcasefatality;estimatesofprevalence;andhealthstatevaluationsforrelevantoutcomes.Outputsfromthemodelincludedestimatesoftheresidencetimeineachdiseasestateandtotalnumberofhealthylifeyearslivedbyage,sex,andcalendaryear.Theseoutputswereusedtocalculateinterventioneffectivenessintermsofchangesinaggregatepopulationlevelhealthoutcomes.WherealternativestoPopModwereusedforspecificanalyses,thesealternativemodellingapproachesaredescribedinthetechnicalappendixonbmj.com.EstimationofinterventioncostsWeusedaningredientsapproachtocosting,bywhichthequantitiesofinputsthatareusedindeliveringaparticularserviceorinterventionaremultipliedbytheirunitpricestoobtaintotalcosts.FollowingthestandardWHO-CHOICEapproach,weconsideredthreebroadcategoriesofcosts:patientcosts,programmecosts,andtrainingcosts.PatientcostsPatientlevelcostsincludedhospitalbeddays,hospitalvisits,healthcentrevisits,ancillarycare,laboratoryanddiagnostictests,drugs,andothercostsrelatedtospecificinterventions.Resourcequantitieswerederivedthroughreviewofthepublishedliterature,andfrompracticeguidelines.DrugpricesweretakenfromtheIMSSpurchasepricedatabase.Unitpricesfornon-tradedgoods,includingpatientservicessuchashospitalbeddays,werestandardisedacrossinterventionanalysesandderivedfromtheWHO-CHOICEpricedatabase,whichincludescountryspecificestimatesbasedonaneconometricanalysisofmultinationaldatasetsoncosts.21ProgrammecostsWeconsideredseveralkeycategoriesofprogrammeactivities.Basicadministrationincludesplanningandoverheadcosts,inadditiontostaffrequiredtoeffectivelymonitor,evaluate,andsupervisetheprogramme.Thesecostsdependonwhethertheinterventionrequireslegislation,thelevelatwhichadministrationisrequired,andthecomplexityofmonitoringandevaluationneeded.Othercategoriesofprogrammecostsrelevanttocertaininterventionsincludedmediacampaigns,otherinformation,educationorcommunicationactivities,andlawenforcement.Formostinterventions,weadoptedestimatesofresourceuseandpricesforprogrammecostcomponentsfrompreviousWHO-CHOICEregionalanalyses.Detailsonassumptionsofprogrammecostsforspecificanalysesareinthetechnicalappendix(seebmj.com).TrainingcostsTrainingcosts,relevantforsomeinterventions,dependonthelengthoftrainingrequired,thenumberofsupervisoryvisitsneededayear,andthecapacityforasingletrainingsession.AllinterventionsinthisstudythatrequiredtrainingcostswerelinkedtopreviousWHO-CHOICEregionalanalyses,andwemaintainedtheassumptionsfromthesepreviousanalyses.EstimationofcosteffectivenessWecomputedtotalcostsforagiveninterventionasthesumofallpatient,programme,andtrainingcosts.Thenullscenariobydefinitionincludesnocosts,socostsofallotherinterventionscanbeinterpretedasbeingincrementalonthenull.Thetotalhealthbenefitsofaninterventionwerecomputedbycomparingthenumberofhealthylifeyearslivedinthepopulationinaparticularinterventionscenariowiththetotalnumberofhealthylifeyearslivedinthepopulationunderthenullscenario.Inthebasecase,weuseddiscountedageweightedDALYsastheunitofaccountforhealthylifeyears,butwealsoconductedasensitivityanalysiscomparingtheseresultswiththosewithoutageweighting.Forallinterventions,wereportannualisedquantitiesforbothtotalcostsandtotalbenefits.Fortotalcosts,thesewerederivedsimplybydividingthecostsoverthe10yearinterventionperiodby10.Fortotalbenefits,thesewerederivedbytakingthefulldifferenceinhealtheffectsoverthe100yearmodelledperiodanddividingthisby10.Inthisway,theannualisedcostsandbenefitscanbeinterpretedasthecostsandbenefitsassociatedwithasingleyearofintervention.Wereportaveragecosteffectivenessratiosforinterventions,orderedbyincreasingoveralleffects.Averagecosteffectivenessratiosareinterpretableasthenetcostsperunitofnetbenefitassociatedwithdeliveringtheintervention,comparedwithdoingnothing.Wealsoreportincrementalcosteffectivenessratioswherethesemeasuresarerelevant—thatis,inevaluatingmutuallyexclusiveinterventionsthatrepresentcompetingchoices.Incrementalcosteffectivenessratioswerecomputedforaninterventionwithrespecttothenextmosteffectivealternativeaftereliminatingstrategiesthatweredominated(thatis,thosethatweremorecostlyandlesseffectivethanotheroptions)orthosethatwereweaklydominated(thatis,hadhighercosteffectivenessratiosthanmoreeffectiveoptions).Followingthestandardbenchmarksforvalueformoneyproposedininternationalworkoncosteffectiveness,wecomparedcosteffectivenessratiosagainstthresholdsdefinedinreferencetothegrossdomesticproduct(GDP)percapitainMexico,whichwas$Int10 770in2005.InterventionswereconsideredtobehighlycosteffectivewhentheyhadratiosthatfellbelowthepercapitaGDPandwereregardedasbeingpotentiallycosteffectiveiftheyhadratiosbetweenoneandthreetimespercapitaGDP.ResultsOverviewThefigure⇓summarisesinformationoncosts,populationhealtheffects,andcosteffectiveness(comparedwiththenull)forallinterventions.Acrossthe101interventionsexaminedinthisstudy,theaverageyearlycostsatthepopulationlevelrangedfrom≤$Int1m(forcataractsurgeriesandsomeelementsofcervicalcancertreatment)to>$Int1bn(forhighcoverageoftreatmentforhypercholesterolaemiaoraggressivemanagementofabsolutecardiovascularrisks).Comparingthetotalannualisedpopulationhealthbenefitsacrossinterventions,weagainobservedawiderangeofoutcomes,from<1000DALYsavertedayear(forsomecomponentsofcancertreatmentsoraspirinforacuteischaemicstroke)to>300 000DALYsavertedayear(foraggressivecombinationofinterventionstodealwithalcoholuse,whichwastheleadingriskfactorfortheburdenofdiseaseinMexicoin2004,6andcardiovascularrisks).Costs,populationhealtheffects,andcosteffectivenessof101interventionstrategiesinMexico,byinterventionpackageDownloadfigure Openinnewtab Downloadpowerpoint Inthefigure⇑,withbothaxesdisplayedonalogscale,diagonallinesmovingfromthelowerlefttotheupperrightdirectionarecosteffectivenessisoquants,whichmeansthatanypointonthesamelineasanotherpointhasthesameaveragecosteffectivenessratiocomparedwiththenull.Thus,thesefiguresofferaneasygraphicaldisplayofbroadbandsofcosteffectivenessacrossinterventions.Theinterventionsinthisstudyspannedawiderangeofaveragecosteffectivenessratios,differingbymorethanthreeordersofmagnitudebetweenthelowestandhighestratios.Atthelowend,someinterventionscost$Int100 000perDALY,suchashighintensitytreatmentforchronicobstructivepulmonarydiseaseoraspirinforacuteischaemicstroke.ComparisonacrossinterventionpackagesTheinterventionsinthisanalysispertainedtothreedifferentpackagesofinterventionscoveredbythehealthreformscheme:communityandpublichealthinterventions;lowandmediumcomplexityclinicalinterventions;andhighcomplexityclinicalinterventions.Thereformspecifiedthateachofthesecategoriesofinterventionsshouldbefinancedfromaseparatefund.Thefigure⇑distinguishesinterventionsfallingintothesethreebroadcategories.Thecommunityandpublichealthinterventionsincludednon-personalinterventionsforalcoholuse,tobaccouse,andcardiovascularrisks;andscreeningforbreastcancerandcervicalcancer.Overall,thecommunityandpublichealthinterventionstendedtohavehighbenefitsandmediumleveloverallcostsandtendedtohavelowercosteffectivenessratiosthanmanyoftheinterventionsintheothertwocategories.Evenwithinthecommunityandpublichealthinterventions,however,therewasmorethana200-folddifferencebetweenthehighestandlowestcosteffectivenessratios.Lowandmediumcomplexityhealthcareinterventionsincludedpersonalservicesforalcoholuse,tobaccouse,andcardiovasculardiseaseprevention;treatmentforchronicobstructivepulmonarydisease;interventionsfordepression;cataractsurgery;andsecondarypreventionfordiabetes.Highcomplexityhealthcareinterventionsincludedtreatmentforischaemicheartdisease,stroke,andcongestiveheartfailureandtreatmentforbreastcancerandcervicalcancer.Intermsoftotalcostsandtotalbenefits,thehighcomplexityinterventionstendedtohavelowertotalsthanthelowandmediumcomplexityinterventions,mainlybecausethetargetpopulationsfortheseservicesweresmalleronaverage.Intermsofcosteffectivenessratios,ontheotherhand,therewaslittlediscernabledifferencebetweenthetwogroupsofinterventions.Costeffectivenessratiosbothforlowandmediumcomplexityinterventionsandforhighcomplexityinterventionsspannedrangesofmorethanthreeordersofmagnitude.ResultsbydiseaseclusterFullresultsontheyearlycosts,populationeffects,andcosteffectivenessratiosforall101interventionsareprovidedintheappendixtableonbmj.com.Herewesummarisethecosts,effects,andaveragecosteffectivenessresultsbyinterventioncluster.Theaveragecosteffectivenessratiocanbeunderstoodasthecosteffectivenessofaninterventioncomparedwithanullscenario(nointervention).Whenwecomparemutuallyexclusiveinterventionswithinacluster,wealsodescribeincrementalcosteffectivenessratios,whichwerebasedonthenetcostsandneteffectsofaninterventioncomparedwiththenextmosteffective,non-dominatedintervention.Depression—Weevaluatedfourmaininterventionsforthetreatmentofdepression—olderantidepressantdrugs(tricyclicantidepressants),newerantidepressants(selectiveserotoninreuptakeinhibitors),psychotherapy,andproactivecasemanagement—aswellasvariouscombinationsoftheseinterventions.Wefoundthatproactivecarecombinedwitholderornewerantidepressantshadthebiggestimpactonpopulationhealth,avertingalmostdoublethenumberofDALYsavertedbypsychotherapyorantidepressantsalone.AllexaminedstrategieshadaveragecosteffectivenessratiosbelowthepercapitaGDPinMexicoforeachDALYaverted,makingthemcosteffectivebyinternationalstandards.Regardingallinterventionsandcombinationsasmutuallyexclusivecompetingchoices,wefoundthattwointerventionsdominatedallothers:newerantidepressantscomparedwiththestatusquo,whichhadanincrementalcosteffectivenessratio