P.A. Hall
a,* , R.C.R. Taylor
b, L. Barnes
caHarvardUniversity,CenterforEuropeanStudies,27,KirklandStreet,CambridgeMA,USA
bTuftsUniversity,DepartmentofSociology/CommunityHealthProgram,112,PackardAvenue,MedfordMA,USA
cOxfordUniversity,NuffieldCollege,NewRoad,OxfordOX1NF,UK Received19April2013;accepted31May2013
Abstract
Background.– Theobjectiveofthisstudyistooutlineacapabilitiesapproachtothesocialdeterminantsofpopulationhealthandtocompareits explanatorypowerandimplicationsforpublicpolicy-makingwithpsychosocialapproaches.
Methods.– Amodellinkingthestructuresofeconomicandsocialrelationstohealthoutcomesisdevelopedandlogisticmethodsusedto confirmitsbasevalidityforarepresentativesampleof16,488citizensin19developeddemocraciesdrawnfromtheWorldValuesSurveysof1990 and2005.Self-reportedhealthisthedependentvariable.Age,gender,education,employmentstatus,self-mastery,income,autonomyatwork,ties tofamilyandfriends,subjectivesocialstatus,associationalmembershipsandsenseofnationalbelongingareconsidered.
Results.– Atbaseline,riskratiosreflectingmovementfromthe25thto75thpercentileinthedistributionofthevariableindicatethatincreases inincomereducethelikelihoodofpoorhealth(0.78;0.73–0.82)asdoeshigherautonomyatwork(0.90;0.85–0.94)butsodoesaccesstosocial resourcesreflectedintiestofamilyandfriends(0.89;0.86–0.92),associationalmemberships(0.93;0.89–0.98),subjectivesocialstatus(0.77;
0.54–0.90)whiletheabsenceoffeelingsofnationalbelongingincreasesthelikelihoodofpoorhealth(1.14;1.06–1.23).
Conclusion.– Theresultssuggestthatpopulationhealthisdependentonthedistributionofsocialaswellaseconomicresourcesalongthe dimensionspredictedbyacapabilitiesmodel.Governmentsshouldbeattentivetotheimpactofpolicyonthedistributionofsocial,aswellas economic,resources.
#2013ElsevierMassonSAS.Allrightsreserved.
Keywords:Capabilities;Populationhealth;Gradient;Psychosocial;Collectiveimaginary;Stress;Healthpolicy Re´sume´
Positionduproble`me.– L’objectifdel’e´tudeestdepre´senteruneapprocheparles«capabilite´s»desde´terminantssociauxdelasante´ des populationsetdecomparersavaleurexplicativeetsesimplicationsdansl’e´laborationdespolitiquespubliquesaveclesapprochespsychosociales.
Me´thodes.– Cettee´tudede´veloppeunmode`lequie´tudielarelationentrelesstructuresdesrelationse´conomiquesetsocialesetl’e´tatdesante´.
Lavalidite´ del’approchesurune´chantillonrepre´sentatifde16488citoyensissusde19paysoccidentaux(e´chantillontire´ desWorldValues Surveysde1990et2005)este´tudie´eparre´gressionlogistique.Lavariablea` expliquerestl’e´tatdesante´ auto-de´clare´.Lesvariablesexplicatives sont:l’aˆge,lesexe,leniveaud’e´tudes,lestatutprofessionnel,lamaıˆtrisedesoi,lesrevenus,l’autonomieautravail,lesliensaveclafamilleetles amis,lestatutsocialsubjectif,l’appartenance a` uneassociation,lesentimentd’appartenancea` unenation.
Re´sultats.– Lesrisquesrelatifsassocie´sa` l’augmentationderevenusentrele25eet75epercentilere´duisentlaprobabilite´ d’avoirunmauvais e´tatdesante´ (0,78;0,73–0,82),cequieste´galementlecasd’uneplusgrandeautonomieautravail(0,90;0,85–0,94),del’acce`sa` desressources socialessetraduisantparlesliensa` lafamilleetauxamis(0,89;0,86–0,92),del’appartenancea` uneassociation(0,93;0,89–0,98)etdustatut
Availableonlineat
www.sciencedirect.com
Revued’E´ pide´miologieetdeSante´ Publique61S(2013)S177–S183
*Correspondingauthor.
E-mailaddress:phall@fas.harvard.edu(P.A.Hall),rtaylor@tufts.edu(R.C.R.Taylor),lucy.barnes@nuffield.ox.ac.uk(L.Barnes).
0398-7620/$–seefrontmatter#2013ElsevierMassonSAS.Allrightsreserved.
http://dx.doi.org/10.1016/j.respe.2013.05.016
socialsubjectif(0,77;0,54–0,90).Enrevanche,l’absencedesentimentd’appartenancea` lanationaugmentelerisquedemauvaisesante´ (1,14; 1,06–1,23).
Conclusion.– Lesre´sultatssugge`rentquelasante´ despopulationsde´penddeladistributiondesressourcessocialesete´conomiquesselonles parame`trespre´ditsparunmode`ledes«capabilite´s».Lespouvoirspublicsdevraienteˆtreattentifsa` l’impactdespolitiquessurladistributiondes ressourcessocialesete´conomiques.
#2013ElsevierMassonSAS.Tousdroitsre´serve´s.
Motscle´s: Capabilite´s;Sante´ despopulations;Gradient;Psychosocial;Imaginairecollectif;Stress;Politiquesdesante´
1. Introduction
Inrecentyears,well-developedliteratureshaveestablished that,alongsidethematerialfactorslongcitedascontributorsto health and illness, there are also social determinants of populationhealth[1,2].Theseeffectsmanifestthemselvesina varietyof waysbut are prominentcontributors topatternsof inequalities in health. One of the recurrent findings in population healthis the presence at the aggregate level of a gradientlinkingincomeorotherdimensionsofsocioeconomic statustopeople’shealth[3,4].Thisrelationshipholdsforawide range of measures of health, from life expectancy to self-reportedhealth, andis repeated across space and time, even whentheprincipal causesofmortality shift[5].
The observationthathealthhassocialdeterminantscarries important implications for the typesof policies governments should pursue if they seek to reduce inequalities in health.
However, the issue of what policies to adopt turns on the questionofwhattheprincipalsocialdeterminantsofhealthare, andthereisasyetnoconsensusaboutthis[6,7].Theobjectof thisarticle is toidentify some formulationsabout the social determinantsofhealth,withemphasisonthosethatcontribute to the health gradient, and to explore their implications for publicpolicy-making. Weconsider the dominant ‘psychoso-cial’ approach to the health gradient and then outline an alternativetoitbasedona‘capabilities’perspective,whichhas potentialformodelingsocialdeterminantswithmoreprecision.
Thisanalysis carriesa wide rangeof implications for public policy-making.
2. Thepsychosocialperspective
Following pioneering work on the social determinants of populationhealth[1],researchbyMarmotandhiscollaborators [8–10], often based on the Whitehall study of British civil servants,aswellascross-nationalstudiesbyWilkinsonandhis collaborators[11,12]havedonemuchtoestablishwhatwewill termapsychosocial approachto the health gradient [13,14].
Althoughthe specificformulations about causal mechanisms linkingsocialarrangementstothehealthgradientvaryacross studies,thecorecontentionofthisapproachisthatallsocieties, ofhumans as well as some other primates, contain a social hierarchythatgivesriseinitslowerrankstostatusanxietywith adverse effects on health via a set of well-established physiologicalpathwayswherebysocialexperience‘getsunder theskin’[15–17].Inshort,aubiquitoushealthgradientresults from status anxieties, typically associated with feelings of
relativedeprivation,generatedbyasocialhierarchyreflecting differencesof socialstatuspresentinallsocieties.
A considerable body of evidence suggests that, to some extent at least, the core contentions of this psychosocial approach have validity. Even when a wide range of factors correlated withsocial rank are controlled,lower rankis still associatedwithpoorerhealthoutcomes[10,13].However,the psychosocialapproachsuffers fromtwolimitations.First,the causalmechanismslinking socialranktopoor healtharenot alwayswellspecified.Therearetwopartstothiscausalchain.
Theultimatesegmentspecifyinghowexperiencesofstressor anxiety associated with lower rank induce physiological reactions with adverse effects on health has been relatively wellidentified andtested [15–17]. However, thepenultimate segment specifying how social arrangements give rise to experiences of stress or anxiety is left vague in most of the relevantstudies.Thelinkbetweenstatusconcernsandanxiety isplausiblebutrelatively-untestedinempiricalterms;anditis notclearwhysocialhierarchyshouldalwaysgiverisetostatus concerns. In societies marked by traditional customs of deference, for instance, possession of a lower social rank maynotalwaysinducefeelingsof relativedeprivation[18].
Second, the universalism of the psychosocial approach, whichpositsa ubiquitoussocial hierarchyfromwhichstatus anxiety invariably follows, militates against precision in comparative inquiry. Where the relationshipin question is a physiological effect following, for instance, from chemical reactions inthe brainor body,it makes sense toexpect it to apply, ceteris paribus, toall men or women. But, wherethe relationship isonebetween aset of socialarrangements and emotionalstates,propositionscouchedinuniversaltermstend toobscuretherangeofvariationinsocialarrangementsacross settingsthatmaybepertinenttotheoutcomes.Thereisaneed formorepreciseformulationsaboutthemultipledimensionsof socialhierarchythatmightgeneratetherelevanthealtheffects and about how they vary across societies, as well as more attentiontothewaysinwhichotherfeaturesofsocialcontext mightmediatetheextenttowhichlowersocialrankgivesriseto statusanxietyandexperiencesofstress.
Wilkinson[11,12]addresses thisproblemtosomedegree, but in terms that attach overwhelming importance to the distribution of income. His formulations suggest that status anxieties willbemoreprevalentinsocietieswhereincome is distributed more unequally, presumably on the premise that higher levels of income inequality generate a steeper status hierarchyand/ormoreintensefeelingsofrelativedeprivation.
However,it isdifficulttotestthesepropositionsaboutcausal
mechanisms and assuming their validity obscures some important issues. The extent to which status depends on incomeisitselfsomethingthatmayvary acrosssocietiesand over time. In societiesthatattach high value toconspicuous consumption,forinstance,statusmaybecloselytiedtoincome;
but thereare societies inwhich income or the possession of consumer goods is not the principal marker of status. Even whereincomematters,socialstatusistypicallya multidimen-sionalphenomenon:people whodo notsecure it via income may do so on other grounds, for instance, through their craftsmanshipatwork, theirprowessonthesportsground or their roles as good parents [19]. Thus, instead of assuming income is aproxy for status, research should investigatethe relationship between these two variables; and, instead of reducing the relevant dimensions of social structure to the distribution of income, researchers should consider a wider rangeofdimensionsofsocialrelationsandfurtherformulations abouthow theymightbearonhealth.
3. Thecapabilities perspective
To advance these objectives, we elaborate an alternative perspectiveonthesocialdeterminantsofhealththatisbroadly compatible with psychosocial perspectives but allows for a more expansive conception of the dimensions of social relationsthatconditionhealthandoffersmoreprecisionabout the causal mechanisms linking those dimensions to the experiences of stress and anxiety that have adverse health effects.Thisisa‘capabilitiesperspective’onpopulationhealth initiallyintroducedinHallandLamont[20].
Thestartingpointforthisperspective,likeitspsychosocial counterpart, is the observation that the principal causes of mortalityinadvancedpost-industrialsocietiesarethechronic cardiovascular diseases and cancers that display strong associations with developments in the hypothalamic-pitui-tary-adrenocortical, sympathetic-adrenal-medullary, and immune systems, reliably related to cumulative experiences ofstressandassociatedemotionalstatesofanxiety,angerand frustration.Inotherwords,‘thewearandtearof dailylife’is closelylinkedtoimportanttypesofmorbidityandmortalityin developed societies [15,17]. The observation that cross-nationalvariations inpopulation health are more substantial intheworking-agepopulationthanamongtheelderlyoryoung inthesesocietiesfurthersuggeststhatwearandtearofthissort maybeanimportantdeterminantof healthinsuch societies.
Therefore, the problem is to explain variations in experiences of stress and related emotional states across individuals,socioeconomicgroupsandsocieties.Theemphasis hereisonthevariationacrosssocioeconomicgroupsassociated withinequalitiesinhealthandthesocialdeterminantsofsuch variation.Atthe coreofthe capabilitiesapproachisamodel that highlights the life challenges people face and the capabilitiestheybringtothosechallenges.
Thecentralpremiseofthismodelisthecontentionthatall people face a similar set of life challenges, such as those associated with securing decent housing and a livelihood, findingaspouseandcaringforchildrenoragingdependents.To
thesechallenges,peoplebringasetofcapabilitiesthatenables themtocopewiththesechallengesmoreorlesseasily.Where theyhaveamplecapabilitiesfor copingwithsuch challenges, peoplewillhavefewerexperiencesofstress.Conversely,where their capabilities are more limited, people will experience higher levels of stress, anxiety, anger and frustration with correspondingadverseconsequencesfortheirhealth.Thus,the life expectancy andhealthstatus of apersondepends onthe relativebalancebetweenthelifechallengesfacingthatperson andthecapabilitiesbroughttosuch challenges.
Capabilities are tosomeextent afunction of attributesof personality that are rooted in early childhood if not genetic makeup;andinthatrespectthesocialconditionsaffectingearly childhood development are relevanttothese outcomes anda potential source of systematic cross-national variation [21].
Evidencethatthedevelopmentofexecutivefunctionassociated with the pre-frontal cortex can be conditioned by social circumstancespointstoonepathwayfor sucheffects[22].
However,thesocialdeterminantsemphasizedbythismodel aretheinstitutionalandculturalframeworksconstitutiveofthe structureofsocialrelationsineachcountry.Theseframeworks condition health because they provide and distribute social resources on which individuals draw to cope with life challenges. Whereaperson’sposition withinthese structures supplieshimorherwithmoresuchresources,thatpersonwill copemoreeffectivelywithlifechallenges,thereby experienc-inglowerlevelsofstressovertimeandbetterhealth.Although we emphasize the national features of such structures, there mayalsoberegional variation.
It iswellunderstoodinthe literature on healthstatusthat people are situatedwithin a structureof economicrelations, constituted by the institutional practices and frameworks distinctive to a country’s political economy. That structure distributes economic resources, in the form of income, job security, workplaceautonomyandunemploymentbenefits,to mention onlythoseresourcesmostpertinent tohealth[2,23].
Moreover,therearesystematicdifferencesacrossnations,not only in the distribution of income, but in other economic resources, rootedininstitutionalvariationsacross varietiesof capitalism[24].Thesevariationsmatterforhealthinequalities because each of these resources enhances a person’s capabilities for coping with life challenges. Income is a multipurposeresourceemphasizedinmanyaccounts;andjob securityreducessomeofthepotentialstressorsapersonfaces.
Forinstance,itmayimproveanindividual’sabilitytotaketime off work to cope with illness or care for dependents, while several dimensions of workplace autonomy are closely correlated with healthoutcomes [23]. Thus,at each position within anationalstructureof economicrelations,apersonis suppliedwithcertainlevelsofeconomicresourceswith health-relatedvalue;andtherecanbesystematicvariationsinthelevel of resources suppliedat analogouspositions inthisstructure across countries.
However,ineachcountrythereisalsoastructureofsocial relations, constituted by parallel institutional and cultural frameworks thatconveycorrespondingsocialresources. This structurehasverticalandhorizontalplanes.Itsverticalplaneis
P.A.Halletal./Revued’E´ pide´miologieetdeSante´ Publique61S(2013)S177–S183 S179
characterizedbyasocialhierarchythatdistributessocialstatus orprestige,muchaspsychosocialapproachesposit.Butinthe capabilitiesmodelthesocialhierarchyisadistinctdimension ofsocialrelations,whichmayormaynotbetightlycoupledto thedistributionofincome;anditfigures,notonlyasasourceof statusanxiety,but asasourceof capabilities, onthe premise that coping with life challenges requires the cooperation of othersandpeople withhigherstatus securesuch cooperation morereadily[25,26].
On thehorizontalplane,the structureofsocialrelationsis composed,first,of setsof socialtiesthatconnectindividuals directlywithfamily,friendsandothersinsociety.Thesesocial networksmaybemoreorlessextensiveordense,andeachtype ofnetworkismoreusefulforsomepurposesthanothers[27].
Butthesedirectformsofsocialconnectednessareimportantto health becausethey offer individuals social resources in the formoflogisticalandemotionalsupportthatisconstitutiveof theircapabilities[28,29].In addition,as Durkheimobserved [30], people are connected by an overarching cultural framework or conscience collective, which we term a
‘collectiveimaginary’,composedofthenarrativesthatconnect acommunity’spasttoitsfuture,specifythesocialboundaries of the community, and indicate what its members owe one another[31,32].Thiscollectiveimaginary cansupplypeople withasenseofbelongingandpurposethatisalsoconstitutiveof theircapabilitiesor,bydefiningsomegroupsasmarginaltothe community, it can limit their capacity to secure cooperation fromothersandtheircapabilitiesmoregenerally[33,34].
Onceagain,thecorecontentionisthateachpositionwithin thisstructureof socialrelationsconfersspecificsetsofsocial resourcesonwhichpeopledrawtocopewithlifechallenges, therebyconditioningtheirhealth,andtherecanbesystematic cross-national variation in this structure of social relations.
There is some evidence, for instance, that the ratio for the densityofsocialtiesbetweentheupperandlowersocialclasses islargerinFrancethanit isinmostotherEuropeansocieties [35].Fig.1 displaysthismodelof thesocial determinantsof health.
4. Preliminaryempiricalassessment
If one implication of the most prominent psychosocial models is that a person’s health is strongly conditioned by income,the implicationofthe capabilitiesperspectiveisthat healthstatusshouldbeaffected,notonlybyincome,butbya wider range of dimensions of the structure of social and economicrelationsindependentlyoftheeffectsofincome.In research fully reported elsewhere [35], we test these implications with multivariate logistic regressions and fixed country effects on representative samples of 16,488 citizens from 19 post-industrial democracies drawn from the World ValuesSurveysof1990and2005.
The dependentvariableis aconventionalmeasureof self-reported health status, dichotomized to reduce measurement error,andtheexplanatoryvariablesofsubstantiveinterestare social connectedness (measured by the importance the respondentattachestotiestofamilyandfriends),associational membership(measuredbythenumberofassociationstowhich therespondentbelongs),autonomyatwork(measuredbyhow much freedom the respondent has to make decisions or to perform a job), social status (measured by self-rated social class), sense of national belonging (measured by pride expressed in being a citizen of the nation) and income (measured by average income in the income decile of the respondentexpressedinUSdollarsatpurchasingpowerparity andloggedtoreflecttheusualshapeofthisrelationship).Age, gender,levelofeducationandanindicatorforself-masteryare controls.
Table1reportstheriskratiosforthelikelihoodofreporting poor healthassociatedwithmovesalongeachofthe relevant dimensions whenthe othervariablesare heldat theirmeans.
The magnitude of each move is reported in Table 1 and whereverpossible reflects a move from the 25thto the 75th percentileinthedistributionofthevariable.Theresultsprovide broadsupportforacapabilitiesapproachtopopulationhealth.
Fig.1. Acapabilitiesmodelforpopulationhealth.
Fig.1. Acapabilitiesmodelforpopulationhealth.