User Guide - t · PDF fileUser Guide. Centre for Research ... The four dimensions are: • residential instability • material deprivation ... example, if the goal is to identify

RRASPRéseau de recherche sur l’amélioration de la santé des populations

PHIRNPopulation Health Improvement Research Network

version 1.0User Guide

Centre for Research on Inner City HealthSt. Michael’s Hospital

209 Victoria Street Toronto, ON M5B 1W8

Developers Flora I. Matheson, PhD

James R. Dunn, PhD Katherine L. W. Smith, MHSc

Rahim Moineddin, PhD Richard H. Glazier, MD, MPH

[email protected]

t a b l e o f c o n t e n t sThe Ontario Marginalization Index: Background.............................. 2

Uses for ON-Marg.................................................................... 2

Technical details....................................................................... 3

ON-Marg dimensions................................................................ 4

How to use the ON-Marg dimensions............................................. 5 Summary score for the ON-Marg dimensions............... 5 Calculating an average ON-Marg score value for higher-order geographical units.............................. 6Using ON-Marg for analysis........................................................ 7 I. Exploring the relationship between outcomes and area-level marginalization............................... 7 II. Using ON-Marg as an individual-level proxy........... 8 III. Mapping the index................................................ 8 IV. Comparing the marginalization of 2 or more groups 8 V. Comparing rates of events..................................... 8

Limitations............................................................................... 9

Ethics and confidentiality............................................................. 9Acknowledgements................................................................... 9Appendix I.............................................................................. 10

References.............................................................................. 11

LIST OF TABLES & FIGURES

Table 1. Dimensions of marginalization and their respective indicators.... 4

Figure 1. Population in each quintile in the south-east sub-LHIN............. 7

Ontario Marginalization Index 1 User Guide Version 1.0

u s e s f o r o n - m a r g


The Ontario Marginalization Index (ON-Marg) is a census- and geographically-based index that seeks to:

• showdifferencesinmarginalization betweenareas;and

• understandinequalitiesinvarious measuresof healthandsocialwell-being, eitherbetweenpopulationgroupsor betweengeographicalareas.

ON-MargisanOntario-specificversionof theCanadianMarginalizationIndex(CAN-Marg),whichhasbeeninusesince2006andisavailablefromwww.canmarg.ca.

ON-Margismultifaceted,allowingresearchersandpolicyandprogramanalyststoexploremultipledimensionsof marginalizationinurbanandruralOntario.

The four dimensions are:

• residential instability• material deprivation• ethnic concentration • dependency

Theindexwasdevelopedusingatheoreticalframeworkbasedonpreviousworkondeprivationandmarginalization.Itwasthenempiricallyderivedusingprincipalcomponentsfactoranalysis.Ithasbeendemonstratedtobestableacrosstimeperiodsandacrossdifferentgeographicareas(e.g.citiesandruralareas).Ithasalsobeendemonstratedtobeassociatedwithhealthoutcomesincludinghypertension,depression,youthsmoking,alcoholconsumption,injuries,bodymassindexandinfantbirthweight.1-7

t h e o n t a r i o m a r g i n a l i z a t i o n i n d e x : b a c k g r o u n d

ON-Marg can be used for:

1) Planning and needs assessment.Forexample,if thegoalistoidentifyservicegaps,ON-Margcanbeusedtoidentifywhereratesof hospitalizationsforaparticulardisease,suchasdiabetes,arehighandadditionalservicesmightbeneeded.

2)Resource allocation.Forexample,marginaliza-tionindexescouldbeusedinfundingformulaeforprimaryhealthcareservices.

3)Monitoring of inequities.Forexample,marginalizationindexescanprovideawaytomonitorchangesinareasovertimetolookforimprovementortoidentifyareasthatmaybeindecline.

4) Research.Forexample,inthehealthsectorthereisalonghistoryof usingsmallareaindexestodescribetherelationshipbetweenmarginalizationandhealthoutcomes;greatermarginalizationisassociatedwithhighermortalityrates,andhigherratesof manydiseases.8-12

Following a literature review,42variableswereselectedfromthe2001Canadiancensusforpotentialinclusionintheindex(seeAppendixI).PrincipalcomponentfactoranalysisyieldedfourfactorswithEigenvaluesgreaterthan1.Of theoriginalvariables,18wereincludedinthefourfactors(seeTable1).The2001indexwascreatedfromtwocorefileswith49,153disseminationareas(DAs)and4,757censustracts(CTs).Theindexwasreplicatedusing2006datawith52,973DAsand5,017CTs.

Factorloadingswereusedtocomputeaseparateindexforeachof thefourdimensions.Eachdimensionisanasymmetricallystandardizedscale.

ON-Margappliestoareas,notindividualpeople.ScoresforeachdimensionareavailableforeverycensustractanddisseminationareainOntario,exceptwheredataissuppressed.

ON-Marg is available for download in Excel 2002 format for the 2001 and 2006 census years. Files are named:

ON-Marg_2001.xlsON-Marg_2006.xlsON-Marg_geographic_crosswalk_2001.xlsON-Marg_geographic_crosswalk_2006.xls

The2001and2006ON-Margfileshavetabs for the following geographies:

• disseminationareas

• censustracts

• censusdivisions

• censussubdivisions

• LHINs

• sub-LHINs

• publichealthunits

• consolidatedmunicipalservicemanager areas

ON-MargvaluesforgeographieslargerthancensustractswerederivedfromtheoriginalDAfactorscores.

QuintilevaluesarenotprovidedforLHINs,publichealthunits,censusdivisionsandconsolidatedmunicipalservicemanagerareasbecausetherearetoofewgeographicunitswithineacharea,whichdisguisesheterogeneity.Forexample,aLHINwith14CTsorDAshastoofewgeographicunitstocreatemeaningfulquintiles.

t e c h n i c a l d e t a i l s


A census tract (CT)is a small, relatively stable geographic unit with a population of 2,500 to 8,000 people constructed similarly with respect to eco-nomic status and social conditions. Census tracts are located in census metropolitan areas and in census agglomerations having an urban core population of 50,000 or more as of the most recent census.13

A dissemination area (DA) is a small, relatively stable geographic unit composed of one or more adjacent dissemination blocks. It is the smallest standard geographic area for which all census data are disseminated. DAs cover all the territory of Canada.13

o n - m a r g d i m e n s i o n s

D I M E N S I O N S

I N

D

I

C A

T

O

R S

Dependency Ethnic concentration*Material deprivationResidential instability

Proportionof thepopulationlivingalone

Proportionof thepopulationwhoareaged

65andolder

Proportionof thepopulationwhoarerecentimmigrants(arrivedinthe5yearspriortocensus)

Proportionof thepopulationwhoself-identifyasavisible

minority

Dependencyratio(totalpopulation0-14and65+/totalpopulation15

to64)

Proportionof thepopulationwhoarenotyouth(aged16+)***

Averagenumberof personsperdwelling***

Proportionof dwellingsthatareapartment

buildings

Proportionof thepopu-lationwhoaresingle/divorced/widowed***

Proportionof dwellingsthatarenotowned***

Proportionof thepopulationwhomovedduringthepast5years

Proportionof thepopulationnot

participatinginlabourforce(aged15+)***

Proportionof thepopulationaged20+withoutahigh-school

diploma**

Proportionof familieswhoareloneparent

families

Proportionof thepopulationreceivinggovernmenttransfer

payments

Proportionof thepopulationaged15+who

areunemployed

Proportionof thepopula-tionconsideredlow-

income****

Proportionof householdslivingindwellingsthatareinneedof majorrepair

Table 1. Dimensions of marginalization and their respective indicators

* Aboriginal indicators did not load on any of the factors.

** For the 2006 index, the indicator is the proportion of the population aged 25+ without a certificate, diploma or degree. This is due to a change in the Statistics Canada definition.

*** Indicators were reverse coded, meaning they were coded opposite of the measure (e.g. % married/common law becomes %single/divorced/separated/widowed).

**** “Low income” is defined as below the low income cutoff (LICO), a Statistics Canada measure that is adjusted for community size, family size and inflation.


TheON-Margdimensionscanbeusedseparatelyorcombinedintoacompositeindex(seethenextsection).Whetheryouuseindividualdimensionsorthecombinedindexwillbedeterminedbytheresearchquestion.

For each dimension, ON-Marg is provided in two forms:

I. Factor scores (interval scale):Factorscoresareconstructedfromtheprincipalcomponentfactoranalysis.TheywerederivedfromtheCanadianMarginalizationIndex(CAN-Marg,seewww.canmarg.ca)and,whenthefullCanadianindexisused,representastandardizedscalewithameanof 0andastandarddeviationof 1.Lowerscoresoneachdimensioncorrespondtoareasthataretheleastmarginalized;higherscoresoneachdimensioncorrespondtoareasthatarethemostmarginalized.

II. Quintiles (ordinal scale):Quintileshavebeencreatedbysortingthemarginalizationdataintofivegroups,rankedfrom1(leastmarginalized)to5(mostmarginalized).Eachgroupcontainsafifthof thegeographicunits.Forexample,if anareahasavalueof 5onthematerialdeprivationscale,itmeansitisinthemostdeprived20percentof areasinOntario.

Thequintileswerecreatedprovince-widetoenablecomparabilityacrosstheprovince.However,if youareinterestedinaparticularcityorurbanarea,itmaybepossibletore-createthequintilesusingtheindividualfactorscoresforthatcity/urbanarea.

Theobjectivesof youranalysisandthemethodsyouareusingwilldeterminewhetheryouusefactorscoresorquintilesinyouranalysis.Forexample,amappingexercisemightbebestpresentedusingquintiles,whereasaregressionmodelmightbenefitfromthedetailof thefactorscores.

SUMMARY SCORE FOR THE ON-MARG DIMENSIONS

Usersmaywishtoexamineoverallmarginalizationusingasummatedscore.Thiscanbedoneusingthequintilescalesforeachdimension.

In order to calculate the summated score, follow these steps:

Compare the correlations between each dimension with the outcome.Thisallowsyoutodetermineif theassociationsareinthesamedirection(eitherallpositiveorallnegative).If theassociationsareeitherallpositivelyorallnegativelyassociatedwiththeoutcomethenanaveragemarginalizationscorecanbecomputed.If oneormoredimensionsareintheoppositedirectionitisnotrecommendedtocombinethedimensions.Forexample,if ethnicconcentrationisnegativelyassociatedwiththeoutcomeof interest,thismayrepresentaprotectivefactor(e.g.ahealthyimmigranteffect)anditmaynotbeappropriatetocombineethnicconcentrationwiththeotherdimensionsthatarepositivelyassociatedwiththeoutcomeandthereforerepresentriskfactors.


h o w t o u s e t h e o n - m a r g d i m e n s i o n s

1


Sum the quintile values across the four dimensions.

Divide by 4(whichisthenumberof dimensions).

Thesestepswillproduceascorerangingfrom1to5where1reflectslowlevelsof marginalizationand5reflectshighlevelsof marginalization.

Summary Score = (instability_quintile + deprivation_quintile + dependency_quintile + ethniccon_quintile) / 4

Caution:Factor scores cannot be used to obtain a summary score.

CALCULATING AN AVERAGE ON-MARG SCORE VALUE FOR HIGHER-

ORDER GEOGRAPHICAL UNITS

Commonly-usedOntariogeographieshavealreadybeenaggregatedandareavailableaspartof theON-Margpackage.

However,someresearchandpolicyquestionsrequiregeo-codingatcustomgeographicunitsYoucanusetheDAandCTdatainON-Margandthemethodsdescribedinthissectiontocreatevaluesforyourowngeographies,usingpopulation-weightedaveragescores.

Example: calculating weighted average scores for a single Ontario urban health region from 2006 CT- or DA-level marginalization scores.

Definethehealthregionintermsof thecomponentCTsand/orDAs.

Usingthepopulationcounts,taketheweightedaverageof eachfactorscorevalueacrossalltheCTsorDAsinthehealthregion.UsetheCT(orDA,dependingonyouranalysis)worksheetof theON-MargfileON-Marg_2006.xlstoobtaintheweightedaverageforthehealthregion,followingthesesteps:

MultiplyeachCTorDAmarginalizationscorevaluebythepopulationwithintheCTorDAforthehealthregion.

Sumthemultipliedvaluesfroma).Thisbecomesthenumerator.

SumthepopulationvaluesfromeachCTorDAtoobtainatotalpopulationcountforthehealthregion.Thisbecomesthedenominator.

Dividethetotalfrom(b)bythetotalfrom(c).Thisisyourweightedaverage.

Weighted average deprivation score:

∑(ON-Marg_CT_2006* ONPop_CT_2006)

∑(ONPop_CT_2006)

Youcannowusetheseweightedaveragestocreatequintiles.

Caution: Weighted averages can disguise heterogeneity within large geographic areas.Forexample,whentheweightedaveragemethodisusedtodeterminethedeprivationquintilefortheSouth-eastsub-LHINinToronto,theresultis5(mostdeprived).Figure1,however,showsthetruevariationinthissub-LHINbyusingsummedDApopulationcountsbyquintile,notweightedaverages,toshowthenumberof peopleineachquintile.Theresultinggraphshowstherearepocketsof low,moderateandhighdeprivationintheSouth-eastsub-LHINthatwouldbemaskedbyusingthesummaryscoreof 5.

1

a

b

c

d

2

3

3

2


50,000

0

10,000

20,000

30,000

40,000

1 2 3 4 5

P O

P U

L A

T I O

N

least deprived

most deprived

Figure 1. Population in each quintile in South-east sub-LHIN of the Toronto Central LHIN, based on DA population

D E P R I V A T I O N Q U I N T I L E S

I. EXPLORING THE RELATIONSHIP BETWEEN OUTCOMES AND AREA-

LEVEL MARGINALIZATION

Outcomes can include the following:

• individualhealthstatus;

• individualriskorprotectivefactors;*

• ratesof disease,oranyhealthrelatedevent.

Research questions that could be answered include:

1)Whatistheassociationbetweenhealthoutcomes,suchasmortalityanddiabetesrates,andarea-levelmarginalization?

2)Whatistheassociationbetweenhealthbehaviours,suchassmokingandalcoholconsumption,andarea-levelmarginalization?

3)Whatistheassociationbetweenaccesstoroutinesurgicalprocedures,suchasjointreplacement,andarea-levelmarginalization?

To answer such questions, merge the outcomefilewithON-Marg,followingthesteps below:

Preparetheoutcomefile:

Ensuretheaddressesareerror-free.

Geocodeeachobservationinyour outcomedataset(e.g.mortality,crime events,hypertension)toCTorDA.OftenthisisaccomplishedusingthePCCF4+SASprogramcreatedby StatisticsCanada.14NoweveryrecordisassociatedwithaparticularCTorDA.

u s i n g o n - m a r g f o r a n a l y s i s

1a

b

*Healthbehaviourssuchassmokinganddrinkingareoftenexaminedasriskand/orprotectivefactors,butcanalsobeoutcomesof interest.

MergeyourhealthoutcomedatasetwiththeON-MargCTorDA,thuslinkingeachgeocodedoutcomewiththeappropriateareamarginalizationscores.

II. USING ON-MARG AS AN INDIVIDUAL-LEVEL PROXY

Insomeinstances,ON-Margcanbeusedasaproxyforindividual-leveldatawhenactualdataisnotavailable.If individual-levelsocio-economicstatusdataisunavailable,forexample,DA-levelfactorscoresorquintilesfordeprivationcanbeassignedtoeachindividualbasedontheDAinwhichtheindividualresidesandusedasaproxyforsocioeconomicstatus.

Caution: To minimize measurement error, use the smallest spatial area available.Inthecaseof ON-Marg,thisisDAdata.Thereasonissimilartothatprovidedunderthe“caution”forweightedaveragesonpage6.Asthesizeof thegeographicunitincreases(e.g.CTsandsub-LHINs),thepotentialforecologicalfallacyincreasesaswell,sincenoteveryoneinamarginalizedareaismarginalized.

Ineffect,usingareaslargerthantheDAwillweakenanyrelationshipbetweenindividual-andarea-levelmarginalization.Thelargerthegeographicarea,thelesslikelyitisthatanindividual’ssocio-economicstatuswillactually

correspondtothedeprivationscoreof theareainwhichs/helives.

III. MAPPING THE INDEX

TheindexcanbedisplayedgeographicallyusingmappingsoftwaresuchasArcGISorMapInfo.

IV. COMPARING THE MARGINALIZATION OF 2 OR MORE

GROUPS

If youwanttocomparelevelsof marginalizationbetweentwoormoregroups(e.g.hypertensiveversusnon-hypertensive;diabeticversusnon-diabetic)youcancomparethedistributionsof quintiles(orfactorscores)usinganon-parametrictest.Thistestisusedbecausequintilevaluesareordinal,andtheprincipalcomponentscoresareskewed.

V. COMPARING RATES OF EVENTS

If youarecomparingratesof eventswithmarginalization(e.g.mortalityratesinaregioncomparedacrossthefivemarginalizationscalevalues)youcancalculatearankcorrelationcoefficient,orsimplyplotyourresults.NotethatthedenominatorsforyourratescanbeobtainedfromtheCTorDApopulations.

Ethicalapprovalforthedevelopmentof ONMarg2001and2006wasobtainedinApril2005fromtheResearchEthicsBoardof St.Michael’sHospital.

AccesstothedatausedinthisstudywasprovidedbyStatisticsCanadaundertheDataLiberationInitiativeattheUniversityof Toronto.

e t h i c s & c o n f i d e n t i a l i t y


2

Missing data:ThereissomemissingdataintheDAandCTfilesduetodatasuppression(e.g.income).Additionally,insomeareas,inputvariableshaveavalueof 0.Forexample,aDAmaynothaveanyrecentimmigrants.

Time period of data:Datafortheindexisfromthe2001and2006censusyearsandusersshouldbeawareof thiswhenselectingthemostappropriateyearfortheirownanalyses.Forexample,if youroutcomedatawascollectedin

2005or2007youwouldusethe2006indextoensuredatacomparability.If youroutcomedatawascollectedin2000or2002,youwouldusethe2001index.

Coverage of census:Somepopulations,forexampleAboriginalpeoplelivingonreserves,maybeunder-countedinthecensus.ON-Margmaynotbeassensitiveforthesepopulations.RefertoStatisticsCanadatoseeif censuscoveragewillimpactyouranalyses.

l i m i t a t i o n s


Wegratefullyacknowledgethecontributionsof ouradvisorypanelmemberswhohelpedusdevelopthisuserguide.WewouldalsoliketoacknowledgeandthankPublicHealthOntario(formerlytheOntarioAgencyforHealth

ProtectionandPromotion)andtheCanadianPopulationHealthInitiative(CanadianInstituteforHealthInformation).

a c k n o w l e d g e m e n t s

Advisory Panel

Chris BorstCity of Hamilton, Social Development & Early Childhood ServicesLisa CorscaddenCanadian Population Health InitiativeCynthia DambaToronto Central Local Health Integration NetworkEtan DiamondOntario Municipal Social Services AssociationEric HolowatyPopulation Studies and Surveillance, Cancer Care OntarioMarc LefebvreCentre de Santé Communautaire du Grand SudburySara MayoHamilton Social Planning and Research Council

Caitlin MillsOntario Health StudyTodd NorwoodCancer Prevention and Control, Cancer Care OntarioJane PolskyCentre for Research on Inner City HealthPaul RitvoCancer Prevention and Control, Cancer Care OntarioRuth SandersonPublic Health OntarioMichael SpinksSouth East Local Health Integration NetworkJulie StrattonRegion of Peel, Public HealthSusitha WanigaratneCancer Prevention and Control, Cancer Care Ontario


a p p e n d i x I : c e n s u s v a r i a b l e s

ShadingrepresentsvariableschosenforuseinON-Marg

1. Proportion of the population who moved during the past 5 years

2. Proportion living in same house as 1 year ago

3. Proportion of population lone parent families

4. Proportion of population living alone

5. Dependency ratio (total population 15 to 64/total population 0-14 and 65+)

6. Proportion of population youth (aged 5-15)

7. Proportion foreign born

8. Proportion Aboriginal

9. Proportion of the population who are recent immigrants (arrived in the 5 years prior to census)

10. Proportion with no official language

11. Proportion unemployed (aged 15+)

12. Labour force participation rate (aged 15+)

13. Proportion who self-identify as a visible minority

14. Proportion aged 15-24 not attending school

15. Proportion aged 20+ without high school diploma

16. Proportion of the population considered low income using the low income cutoff (LICO)

17. Average household income

18. Proportion of the population receiving government transfer payments

19. Proportion with no religious affiliation

20. Average dollar value of dwelling

21. Proportion of dwellings that are apartment buildings

22. Proportion of owner households spending 30% or more of household income on major payments

23. Proportion of tenant households spending 30% or more of household income on rent

24. Proportion of dwellings that are owned

25. Proportion of occupied units that are rentals

26. Proportion of population self-employed

27. Proportion of population female

28. Proportion of population married/common law

29. Proportion of households living in dwellings that are in need of major repair

30. Proportion of population aged 15+ doing unpaid housework

31. Proportion of population aged 15+ looking after children without pay

32. Proportion of population aged 15+ providing unpaid care/assistance to seniors

33. Raw population count

34. Average number of persons per dwelling

35. Average number of persons per room

36. Ratio of employment to population

37. Average income

38. Proportion of persons separated, divorced or widowed

39. Proportion of children younger than 6 years

40. Persons per square kilometer

41. Unemployment rate in private households with children under 6 years

42. Proportion of the population who are aged 65 and older


r e f e r e n c e s

1. WhiteHL,MathesonFI,MoineddinR,DunnJR,GlazierRH.Neighbourhooddeprivation andregionalinequalitiesinself-reportedhealthamongCanadians:Areweequallyatrisk? HealthPlace;2010:8.

2. UrquiaML,FrankJW,GlazierRH,MoineddinR,MathesonFI,GagnonAJ.Neighborhood

contextandinfantbirthweightamongrecentimmigrantmothers:amultilevelanalysis.AmJ PublicHealth.2009;99:285-93.Epub2008Dec4.

3. MathesonFI,WhiteHL,MoineddinR,DunnJR,GlazierRH.Drinkingincontext:theinflu

enceof genderandneighbourhooddeprivationonalcoholconsumption.JEpidemiolCom munityHealth;2011:17.

4. MathesonFI,MoineddinR,DunnJR,CreatoreMI,GozdyraP,GlazierRH.Urbanneighbor hoods,chronicstress,genderanddepression.SocSciMed.2006;63:2604-16.Epub2006Aug 22.

5. MathesonFI,MoineddinR,GlazierRH.Theweightof place:amultilevelanalysisof gender, neighborhoodmaterialdeprivation,andbodymassindexamongCanadianadults.SocSci Med.2008;66:675-90.Epub2007Nov26.

6. MathesonFI,WhiteHL,MoineddinR,DunnJR,GlazierRH.Neighbourhoodchronicstress andgenderinequalitiesinhypertensionamongCanadianadults:amultilevelanalysis.JEpide miolCommunityHealth2009;64:705-13.

7. LeePPS.InjuryandNeighbourhoodMarginalization:DoesitMatterWhereYouLive.Un publishedmaster’sthesis.Universityof Toronto,Toronto,Canada,2009.

8. EamesM,Ben-ShlomoY,MarmotMG.Socialdeprivationandprematuremortality:regional comparisonacrossEngland.Bmj.1993;307:1097-102.

9. McLooneP,BoddyFA.DeprivationandmortalityinScotland,1981and1991.Bmj.1994; 309:1465-70.

10. TownsendP,Davidson,N.InequalitiesinHealth:TheBlackReport.Harmondsworth:Pen guin,1982.

11. TownsendP,Philimore,P.,Beattie,A.HealthandDeprivation:InequalityandtheNorth. NewYork:CroomHelm,1988.

12. KitagawaEM,Hauser,P.M.DifferentialMortalityintheUnitedStates:AStudyinSocioeco nomicEpidemiology.Cambridge,Mass.:HarvardUniversityPress,1973.

13. StatisticsCanada.2006CensusDictionary.http://www12.statcan.ca/census-recense ment/2006/ref/dict/index-eng.cfm.

14. StatisticsCanada.PostalCodeConversionFilePlus(PCCF+).http://www.statcan.gc.ca/ bsolc/olc-cel/olc-cel?catno=82F0086X&lang=eng.

Documents

User Guide - t · PDF fileUser Guide. Centre for Research ... The four dimensions are: • residential instability • material deprivation ... example, if the goal is to identify