UBC Faculty Research and Publications

Agreement between aggregate and individual-level measures of income and education: a comparison across… Marra, Carlo A; Lynd, Larry D; Harvard, Stephanie S; Grubisic, Maja Mar 31, 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
52383-12913_2010_Article_1589.pdf [ 210.09kB ]
Metadata
JSON: 52383-1.0224079.json
JSON-LD: 52383-1.0224079-ld.json
RDF/XML (Pretty): 52383-1.0224079-rdf.xml
RDF/JSON: 52383-1.0224079-rdf.json
Turtle: 52383-1.0224079-turtle.txt
N-Triples: 52383-1.0224079-rdf-ntriples.txt
Original Record: 52383-1.0224079-source.json
Full Text
52383-1.0224079-fulltext.txt
Citation
52383-1.0224079.ris

Full Text

RESEARCH ARTICLE Open AccessAgreement between aggregate and individual-level measures of income and education: acomparison across three patient groupsCarlo A Marra1,2*, Larry D Lynd1,2, Stephanie S Harvard3 and Maja Grubisic2AbstractBackground: The association between lower socioeconomic status and poorer health outcomes has beenobserved using both individual-level and aggregate-level measures of income and education. While both arepredictive of health outcomes, previous research indicates poor agreement between individual-level andaggregate-level measures. The purpose of this study was to determine the level of agreement between aggregate-level and individual-level measures of income and education among three distinct patient groups, specificallyasthma, diabetes, and rheumatoid patients.Methods: Individual-level measures of annual household income and education were derived from three separatesurveys conducted among patients with asthma (n = 359), diabetes (n = 281) and rheumatoid arthritis (n = 275).Aggregate-level measures of income and education were derived from the 2001 Canadian census, including bothcensus tract-and dissemination area-level measures. Cross-tabulations of individual-level income by aggregate-levelincome were used to determine the percentage of income classifications in agreement. The kappa statistic (simpleand weighted), Spearman’s rank correlations, and intra-class correlation coefficient (ICC) were also calculated.Individual-level and aggregate-level education was compared using Chi-Square tests within patient groups. Pointbiserial correlation coefficients between individual-level and aggregate-level education were computed.Results: Individual-level income was poorly correlated with aggregate-level measures, which provided the worstestimations of income among patients in the lowest income category at the individual-level. Both aggregate-levelmeasures were best at approximating individual-level income in patients with diabetes, in whom aggregate-levelestimates were only significantly different from individual-level measures for patients in the lowest incomecategory. Among asthma patients, the proportion of patients classified by aggregate-level measures as having auniversity degree was significantly lower than that classified by individual-level measures. Among diabetes andrheumatoid arthritis patients, differences between aggregate and individual-level measures of education were notsignificant.Conclusions: Agreement between individual-level and aggregate-level measures of socioeconomic status maydepend on the patient group as well as patient income. Research is needed to characterize differences betweenpatient groups and help guide the choice of measures of socioeconomic status.Keywords: Socio-economic status income, education, aggregate-level, individual-level asthma, diabetes, rheuma-toid arthritis* Correspondence: cmarra@exchange.ubc.ca1Faculty of Pharmaceutical Sciences, University of British Columbia,Vancouver, BC, CanadaFull list of author information is available at the end of the articleMarra et al. BMC Health Services Research 2011, 11:69http://www.biomedcentral.com/1472-6963/11/69© 2011 Marra et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.BackgroundSocioeconomic status (SES) has been shown to beassociated with health outcomes among the generalpopulation [1-3] as well among various patient groups,including asthma [4-10], rheumatoid arthritis (RA)[11,12], and diabetes mellitus (DM) patients [13].Health outcomes associated with SES include level ofasthma control [4], hospital admissions [5,8], emer-gency department and physician visits [6,7,9], andasthma-related mortality among asthmatics, diseaseactivity, physical and mental health, and quality of lifeamong individuals with RA [11,14], and hospitaliza-tions among diabetics [13].The association between lower SES and poorer healthoutcomes has been observed using individual-level mea-sures of SES as well aggregate-level measures, such asthose available from census data. While SES measures atboth of these levels may be predictive of health out-comes, the validity of using aggregate-level measures asa proxy for individual-level measures is debatable. InCanada, studies aimed at quantifying the relationshipbetween individual-level SES measurements and aggre-gates derived from Canadian census data have generallyindicated poor agreement. This finding is consistentacross studies using individual-level measures derivedfrom self-report [15], from structured interviews [16],and from public health insurance data [17,18]. Thesestudies indicate that aggregate measures from theCanadian census function to mask variation in individual-level measures, the latter being more sensitive to povertyand poor health outcomes. Studies using US census datafurther suggest that aggregate-level SES measures reflect aconstruct distinct from individual-level ones [19,20].Despite their limitations, aggregate-level measures areconsidered appropriate for use when individual-leveldata are lacking [21]. Particularly in research usingadministrative data, aggregate measures are often theonly available means to adjust for SES. In this context,the question remains whether aggregate-level measuresperform equally well as individual-level proxies acrossdifferent patient groups or whether the discrepancybetween aggregate and individual-level measures is exag-gerated in some populations. The question also remainswhether there are differences across patient groupsaccording to the aggregate-level measure used.In Canada, the smallest geographical area for which allcensus variables are available is the dissemination area(DA), which typically contains 400 to 700 residents [22].Studies may also utilize data from larger census units,such as the census tract (CT), containing 2500 to 8000residents [23]. While all of Canada is divided into DAs,only regions with a population of 50,000 or more aredivided into CTs, which may lead to differences betweenDA-and CT-level data.The purpose of this study was to determine the levelof agreement between aggregate-level and individual-level measures of income and education among threedistinct patient groups, specifically asthma, DM, and RApatients.MethodsDataIndividual-level measures of annual household incomeand education were derived from three separate self-report surveys conducted among patients with asthma(n = 359), DM (n = 281) and RA (n = 275), respectively.The methods for these surveys have been published pre-viously [14,24,25]. All patients were recruited from Brit-ish Columbia, Canada, and the samples are consideredto be representative of the English-speaking, adult mem-bers of these patient groups in this region. Ethicalapproval for all three studies was obtained from theUniversity of British Columbia.Patients with asthma completed the surveys in years2000 and 2005, patients with RA in 2002, and patientswith DM in 2008. All patients completed the surveyindependently. All surveys included the same items onincome and education, which pertained to annualhousehold income prior to any deductions and to certi-ficates, degrees, or diplomas obtained. All surveys col-lected patients’ age, sex, and residential postal codes.Patients whose residential postal codes were missingwere excluded from the study.Aggregate-level measures of income and educationwere derived from the 2001 Canadian census data,available online from the University of British Columbia.Aggregate-level income was based on the median house-hold income for the census level, which is theself-reported annual household income prior to anydeductions [26]. Education was evaluated by highestlevel of schooling, which in the Canadian census refersto the self-reported “highest grade or year of elementaryor secondary (high) school attended, or to the highestyear of university or college education completed”.Patients with the census classification ‘university, withuniversity degree’ were categorized as having a univer-sity degree and all other patients were categorized asnot having a university degree [26].Statistical MethodsDemographic variables between the three patient groupswere compared using ANOVA for continuous variablesand Pearson’s Chi-Square and Fisher’s exact tests forcategorical variables, where appropriate. StatisticsCanada’s Postal Code Conversion File was used to linkpatients’ postal codes to their corresponding CT andDA [27]. The inflation factor from the Canadian Consu-mer Price Index was used to adjust individual-levelMarra et al. BMC Health Services Research 2011, 11:69http://www.biomedcentral.com/1472-6963/11/69Page 2 of 7incomes from across survey years to their 2001 incomeequivalents [28]. Both individual-level and aggregate-level income were categorized as less than $20,000,between $20,000 and $50,000, and greater than $50,000[24]. To determine the agreement between aggregate-level and individual-level income, 3 × 3 cross-tabulationsof individual-level income category by CT-level incomecategory and of individual-level income category byDA-level income category were produced. This outputwhat used to determine the percentage of cases forwhich individual-level and DA/CT-level income cate-gories were in agreement. The kappa statistic (simpleand weighted) and Spearman’s rank correlations werecalculated to determine the degree of nonrandom agree-ment between individual-level and DA/CT-level income.The intra-class correlation coefficient (ICC) was alsocalculated using the 2-way mixed model for absoluteagreement [29]. For the ICCs, levels of agreement wereadopted as proposed by Fleiss, i.e., <0.40 poor, 0.40-0.75fair to good, and ≥0.75 excellent [30].Both individual-level and aggregate-level educationwere categorized as at least a university degree or lessthan university degree. In the analysis of aggregate-leveleducation, only the census population in age groups20-24 and higher was included. Individual-level and aggre-gate-level education was compared using Chi-Square testswithin patient groups. Point biserial correlation coeffi-cients between individual-level and DA/CT level educationwere also computed.ResultsPatient groups and Number of Corresponding CTs andDAsPatients in the asthma sample (n = 359) resided in 198discrete CTs and 321 discrete DAs, representing a totalof 1,552,655 and 188,235 census respondents, respec-tively. Patients in the DM sample (n = 281) resided in169 discrete CTs and 261 discrete DAs, representing atotal of 1,166,395 and 163,095 census respondents,respectively. Patients in the RA sample (n = 276)belonged to 144 discrete CTs and 226 discrete DAs,representing a total of 833,735 and 157,500 censusrespondents, respectively.Individual and Aggregate-level PatientSociodemographicsTable 1 shows the sociodemographic characteristics ofeach of the three patient groups. Patients in the asthmasample were significantly younger than those in the RAand DM samples (p < 0.0001) and there was a signifi-cantly greater proportion of females in the RA sample(p < 0.0001). Individual-level household income washighest among patients with DM, followed by patientswith asthma and RA, respectively (p < 0.0001, with allpairwise comparisons p < 0.0001). There were no signif-icant differences in the proportions of patients classedby CT or DA as having a university degree between thepatient groups. The proportion of RA patients reportinga university degree was significantly lower than both theTable 1 Characteristics of the study participantsAsthman = 359Rheumatoid Arthritisn = 276Diabetesn = 281p-value*Characteristic Mean (sd†) or n (%) Mean (sd) or n (%) Mean (sd) or n (%)Age (years) 36.8 (8.4) 61.2 (13.8)a 56.9 (13.1) <0.0001Sex (males) 128 (35.7) 58 (21.2)b 147 (52.3) <0.0001Self-reported Income<20,000 90 (29.8) 44 (19.5) 34 (14.6)20,000-50,000 84 (27.8) 99 (43.8) 76 (32.6)>50,000 128 (42.4) 83 (36.7) 123 (52.8) <0.0001DA‡-Household Income<20,000 24 (6.7) 8 (3.0) 7 (2.5)20,000-50,000 180 (50.4) 128 (47.8) 112 (40.1)>50,000 153 (42.9) 132 (49.3) 160 (57.4) 0.001CT§- Household Income<20,000 10 (2.8) 0 4 (1.5)20,000-50,000 192 (53.8) 76 (38.0) 86 (32.2)>50,000 155 (43.4) 124 (62.0) 177 (66.3) <0.0001Individual-level University degree 124 (34.5) 45 (17.3)c 82 (29.8)d <0.0001Aggregate-level University Degree (expected)-DA 91 (25.3) 50 (18.1) 71 (25.2) 0.06Aggregate-level University Degree (expected)- CT 93 (25.9) 59 (21.3) 74 (26.4) 0.39a2 missing; b16 missing; c6 missing; d6 missing;*all p-values were obtained using Chi-square test except for age where ANOVA was used. †Standard deviation;‡ Dissemination Area; § Census Tract.Marra et al. BMC Health Services Research 2011, 11:69http://www.biomedcentral.com/1472-6963/11/69Page 3 of 7proportions of asthma patients (p < 0.0001) and DMpatients (p < 0.0001). The difference between theasthma and diabetes samples in individual-level univer-sity education was not significant (p = 0.21).Agreement: Individual-level and DA/CT Income MeasuresIn all patient groups, the proportion of patients whoreported incomes under $20,000 per year was signifi-cantly higher than the proportion of patients classed inthis income category by DA or CT (all p-values<0.0001). In the asthma group, the proportion ofpatients who reported an income between $20,000 and$50,000 per year (27.8%) was also significantly lowerthan the proportions of patients classed in this incomecategory by DA (50.4%; p < 0.0001) or by CT (53.8%; p< 0.0001); the proportions in the highest income cate-gory (>$50,000 per year) were similar. In the RA group,the proportion of patients reporting an income over$50,000 per year was significantly lower than the pro-portions of patients classed in this income category byDA (p = 0.005) or by CT (p < 0.0001). Among DMpatients, there were no significant differences betweenthe proportions of patients reporting incomes between$20,000 and $50,000 or >$50,000 per year and the pro-portions classed in these categories by DA and CT,respectively.The Spearman’s rank correlations, weighted kappacoefficients, and intra-class correlations indicating theassociation between individual-level and CT-level incomemeasures, and individual-level and DA-level income mea-sures, are shown in Table 2. Following the designationsproposed by Fleiss (<0.40 poor, 0.40-0.75 fair to good,and ≥0.75 excellent), the ICCs generally indicated pooragreement between individual-level and aggregate-levelincome measures among all patient groups.The extent of perfect agreement between individual-level and DA-level and CT-level groupings of income isillustrated in Tables 4 and 5, respectively. Among allpatient groups, both for CT-level and DA-level incomemeasures, agreement with individual-level measures wasless frequent within the lowest income grouping. Ofnote, of the nearly 20% of RA patients who reported anincome under $20,000 in the survey, none were cor-rectly classed in this income grouping by DA or CT(Tables 4 and 5). Overall, across all patient groups therewere no significant differences between DA-level andCT-level income groupings in the proportion of casesfor which there was perfect agreement with individual-level measures (Table 3).Agreement: Individual and Aggregate-level EducationMeasuresTable 6 shows the comparison of individual-leveluniversity education to DA-and CT-level measures ofuniversity education. In patients with asthma, CT-andDA-level census data indicated nearly equal proportionsof patients with a university degree. However, this pro-portion was significantly lower than the proportion ofasthma patients who reported having a university degreein the survey (35%) (p = 0.01). For the RA and DMpatient groups, differences between individual-level andaggregate-level measures of education, respectively, werenot significant.Point biserial correlations between individual-leveluniversity degree and CT-level measures of education(i.e., proportion of the population with a universitydegree) were weak within all patient groups (asthma =0.31; DM = 0.18; RA = 0.28). Compared to CT-levelmeasures, DA-level measures of education were notmore highly correlated with individual-level measures(asthma = 0.28; DM = 0.12; RA = 0.25).DiscussionThis study is the first to compare the agreementbetween individual-level and aggregate-level measures ofincome and education among three distinct patientgroups. The results suggest that the ability of aggregate-level measures to approximate individual-level measuresof SES may vary by the patient group as well as patientincome.In this study, individual-level income was poorly cor-related with CT-and DA-level measures, which is con-sistent with several other reports using Canadian censusdata [16,17,31,32]. Our findings are also similar to thoseTable 2 Spearman’s rank correlation, intra-class correlation and weighted kappa coefficients for the association ofarea-based and self-reported household incomesCensus tract Dissemination arears ICC (95%CI) k (95%CI) rs ICC (95%CI) k (95%CI)Asthma 0.28 0.25 (0.13,0.35) 0.20 (0.13,0.27) 0.35 0.29 (0.17,0.39) 0.24 (0.16,0.32)Rheumatoid Arthritis 0.23 0.13(-0.01,0.27) 0.13(0.02,0.23) 0.29 0.15 (0.03,0.28) 0.16(0.06,0.25)Diabetes 0.35 0.26 (0.13,0.38) 0.27 (0.17,0.37) 0.33 0.27 (0.15,0.38) 0.23 (0.13,0.33)rs = Spearman’s rank correlation.ICC(3,1) = Intra-class correlation coefficient (2-way mixed model, absolute agreement).k = Weighted kappa coefficient.Marra et al. BMC Health Services Research 2011, 11:69http://www.biomedcentral.com/1472-6963/11/69Page 4 of 7of Southern and colleagues [31] who observed that cen-sus-level measures provided the worst estimations ofincome among lower-income households. In all threepatient groups, significantly more patients reportedbeing in the lowest income category than were classedas such by either aggregate-level measure. Amongasthma patients, this discrepancy reflects CT-and DA-level measures having classed lowest-income patients inthe middle income category, while among RA patients itpoints to aggregate-level measures having classedmiddle-income patients in the highest income category.Thus, aggregate measures of income tended to classifypatients in higher income categories relative to indivi-dual-level measures. Notably, among RA patients who inthe survey reported being in the lowest income category,individual-level incomes were never in agreement withthe corresponding DA-or CT-level measures.Both CT-and DA-level measures were best atapproximating individual-level income in patients withDM, in whom aggregate-level values were only signifi-cantly different from individual-level values for patientsin the lowest income category. The frequency of per-fect agreement between aggregate-level and individual-level measures of income was also highest overallamong DM patients, the only patient group in whichboth CT-and DA-level values agreed with Individual-level values in more than fifty percent of cases. As wellas performing best among DM patients, both CT andDA-level measures best approximated individual-levelmeasures of income among all patients in the highestincome categories. Accordingly, across patient groupsand income categories, the greatest proportion of casesin perfect agreement with aggregate-level measureswas observed among DM patients of the highestincome category.With respect to education, all point biserial correla-tions between individual-level and aggregate-level mea-sures were weak, with no differences between CT-andDA-level measures. Despite the weak correlations acrossall patient groups, among RA and DM patients therewere no significant differences between individual-leveland aggregate-level measures in the proportions ofpatients classified as having a university degree. Theonly difference in these proportions was observed withinthe asthma group, where the proportion of patients whoreported having a university degree in the survey wassignificantly higher than the proportions in the corre-sponding CT-and DA-level populations.These findings should be taken in context with thelimitations of the study. First, individual-level incomeand education among our three patient groups was self-reported and could not be verified, and thus reportingbias may have affected the SES measures that werederived from surveys. However, the same can be said ofcensus measures, which are also self-reported; inCanada, census measures are the most accessible popu-lation-based data and no ‘objective’ measures of incomeand education are available for the Canadian population.It should also be noted that, despite the risk of bias,self-reported measures of SES remain powerful predic-tors of health outcomes [33]. Ultimately, the absoluteaccuracy of the individual-level measures does not affectthe conclusions regarding their agreement with aggre-gate-level measures.In this study, individual-level measures are assumed tobe better than aggregate-level measures, an assumptionthat follows from evidence that individual-level mea-sures are more strongly associated with health outcomes[19]. However, this assumption could be inaccurateunder some circumstances. It is possible that amongsome patients, income reported in a cross-sectional sur-vey is not representative of prior income, e.g., incomebefore retirement among older patients or income priorto disease onset among patients with work disability.This could explain the pattern of non-agreementobserved here among RA patients with the lowest indi-vidual-level incomes, as RA patients are known to havea high burden of work disability [34,35]. In these cases,aggregate-level measures could reflect a prior incomeTable 4 Percentage of cases in perfect agreementbetween individual-level and DA income groupingsAsthman(%)Diabetesn(%)Rheumatoid Arthritisn(%)p-value*<20,000 12 (13) 2 (6) 0 (0) 0.03**20,000-50,000 52 (63) 37 (49) 56 (58) 0.23>50,000 77(60) 89 (72) 53 (64) 0.12*Chi-square test; **Fisher’s exact test.Table 5 Percentage of cases in perfect agreementbetween individual-level and CT income groupingsAsthman(%)Diabetesn(%)Rheumatoid Arthritisn(%)p-value*<20,000 5 (6) 2 (7) 0 (0) 0.44**20,000-50,000 54 (64) 36 (49) 31 (44) 0.03>50,000 76 (60) 99 (83) 49 (74) 0.0002*Chi-square test; **Fisher’s exact test.Table 3 Percentage of cases in perfect agreementbetween individual-level and DA/CT income groupingsAsthman(%)Diabetesn(%)Rheumatoid Arthritisn(%)p-value*DA 141 (47) 128(55) 109 (49) 0.17CT 135 (45) 137 (62) 80 (49) 0.0005p-value* 0.62 0.14 0.99*Chi-square test.Marra et al. BMC Health Services Research 2011, 11:69http://www.biomedcentral.com/1472-6963/11/69Page 5 of 7sustained over a longer period and therefore be morerepresentative of SES. In addition, individual-level edu-cation may not be a good measure of SES among somepatients, such as women among the oldest old, whosehusband’s educational attainment may be a bettermeasure of SES than their own [36]. In this context,aggregate-level measures of education, which reflectcontextual effects, may be more representative of anindividual’s SES. Finally, the aggregate-level data usedhere is from 2001, while individual-level data was col-lected in 2000, 2002, 2005 and 2008. Although methodswere employed to correct income for inflation, noadjustment could be made to address the time lapsebetween the collection of individual-level measures andthis could be a source of bias. However, given the meanage of the survey participants in the three patientgroups, income and education status may be expectedto have been relatively stable across the study periodsand comparable to 2001 census measures.ConclusionsThis study shows that the agreement between indivi-dual-level and aggregate-level measures of SES maydepend on the patient group as well as patient income.While research is needed to characterize patterns of dif-ferences between patient groups to help guide thechoice of SES indicators, the use of both individual-leveland aggregate-level measures is advised in studies ofhealth outcomes.AcknowledgementsData for this study was derived from projects funded by the CanadianArthritis Network and the B.C. and Yukon Lung Association. These fundingbodies played no role in the study design, collection, analysis andinterpretation of data, nor in the writing of the manuscript or decision tosubmit the manuscript for publication.Author details1Faculty of Pharmaceutical Sciences, University of British Columbia,Vancouver, BC, Canada. 2Centre for Health Evaluation and Outcome Sciences,Providence Health Research Institute, Vancouver, BC, Canada. 3School ofPopulation and Public Health, University of British Columbia, Vancouver, BC,Canada.Authors’ contributionsCAM was involved in the conception of the study, participated in the studydesign and data interpretation and was involved in revising the manuscriptcritically for important intellectual content and approving the final version.LDL was involved in the conception of the study, participated in the studydesign and data interpretation and was involved in revising the manuscriptcritically for important intellectual content and approving the final version.SSH was involved in data interpretation, drafted the manuscript and gavefinal approval for the version to be published. MG performed the statisticalanalysis, interpreted the data and revised the manuscript critically forimportant intellectual content.Authors’ informationCAM is a Government of Canada Research Chair in PharmaceuticalOutcomes and a Michael Smith Foundation for Health Research Scholar inHealth Services Research. LDL is a Canadian Institutes of Health ResearchNew Investigator and a Michael Smith Foundation for Health ResearchScholar in Population Health. Both CAM and LDL are Associate Professors inthe Faculty of Pharmaceutical Sciences at the University of British Columbia.SSH is a researcher in the Faculty of Medicine at the University of BritishColumbia. MG is a biostatistician in the Faculty of Pharmaceutical Sciences atthe University of British Columbia.Competing interestsThe authors declare that they have no competing interests.Received: 26 May 2010 Accepted: 31 March 2011Published: 31 March 2011References1. Lee DS, Chiu M, Manuel DG, Tu K, Wang X, Austin PC, Mattern MY,Mitiku TF, Svenson LW, Putnam W, Flanagan WM, Tu JV: CanadianCardiovascular Outcomes Research Team: Trends in risk factors forcardiovascular disease in Canada: temporal, socio-demographic andgeographic factors. CMAJ 2009, 181(3-4):E55-66.2. Kozyrskyj AL, Kendall GE, Jacoby P, Sly PD, Zubrick SR: Association BetweenSocioeconomic Status and the Development of Asthma: Analyses ofIncome Trajectories. Am J Public Health 2009.3. Franks P, Gold MR, Fiscella K: Sociodemographics, self-rated health, andmortality in the US. Soc Sci Med 2003, 56(12):2505-2514.4. Lynd LD, Sandford AJ, Kelly EM, Pare PD, Bai TR, Fitzgerald JM, Anis AH:Reconcilable differences: a cross-sectional study of the relationshipbetween socioeconomic status and the magnitude of short-acting beta-agonist use in asthma. Chest 2004, 126(4):1161-1168.5. Watson JP, Cowen P, Lewis RA: The relationship between asthmaadmission rates, routes of admission, and socioeconomic deprivation.Eur Respir J 1996, 9(10):2087-2093.6. Boulet LP, Belanger M, Lajoie P: Characteristics of subjects with a highfrequency of emergency visits for asthma. Am J Emerg Med 1996,14(7):623-628.7. Erzen D, Carriere KC, Dik N, Mustard C, Roos LL, Manfreda J, Anthonisen NR:Income level and asthma prevalence and care patterns. Am J Respir CritCare Med 1997, 155(3):1060-1065.8. Chen Y, Dales R, Krewski D: Asthma and the risk of hospitalization inCanada: the role of socioeconomic and demographic factors. Chest 2001,119(3):708-713.9. Friday GA Jr, Khine H, Lin MS, Caliguiri LA: Profile of children requiringemergency treatment for asthma. Ann Allergy Asthma Immunol 1997,78(2):221-224.10. Cesaroni G, Farchi S, Davoli M, Forastiere F, Perucci CA: Individual andarea-based indicators of socioeconomic status and childhood asthma.Eur Respir J 2003, 22(4):619-624.11. Jacobi CE, Mol GD, Boshuizen HC, Rupp I, Dinant HJ, Van Den Bos GA:Impact of socioeconomic status on the course of rheumatoid arthritisTable 6 Percentage of patients with a university degree by individual- and aggregate-level measuresIndividual-levelUniversity Degreen(%)CT-level UniversityDegree (expected)n(%)p-value* DA-level UniversityDegree (expected)n(%)p-value*Asthma 124 (34.5) 93 (25.9) 0.01 91 (25.3) 0.01Diabetes 82 (29.8) 74 (26.4) 0.40 71 (25.2) 0.25Rheumatoid Arthritis 45 (17.3) 59 (21.3) 0.26 50 (18.1) 0.85*Chi-square test.Marra et al. BMC Health Services Research 2011, 11:69http://www.biomedcentral.com/1472-6963/11/69Page 6 of 7and on related use of health care services. Arthritis Rheum 2003,49(4):567-573.12. Marra CA, Lynd LD, Esdaile JM, Kopec J, Anis AH: The impact of low familyincome on self-reported health outcomes in patients with rheumatoidarthritis within a publicly funded health-care environment. Rheumatology(Oxford) 2004, 43(11):1390-1397.13. Booth GL, Hux JE: Relationship between avoidable hospitalizations fordiabetes mellitus and income level. Arch Intern Med 2003, 163(1):101-106.14. Marra CA, Lynd LD, Esdaile JM, Kopec J, Anis AH: The impact of low familyincome on self-reported health outcomes in patients with rheumatoidarthritis within a publicly funded health-care environment. Rheumatology(Oxford) 2004, 43(11):1390-7.15. Southern DA, McLaren L, Hawe P, Knudtson ML, Ghali WA, InvestigatorsAPPROACH: Individual-level and neighborhood-level income measures:agreement and association with outcomes in a cardiac disease cohort.Med Care 2005, 43(11):1116-1122.16. Demissie K, Hanley JA, Menzies D, Joseph L, Ernst P: Agreement inmeasuring socio-economic status: area-based versus individualmeasures. Chronic Dis Can 2000, 21(1):1-7.17. Sin DD, Svenson LW, Man SF: Do area-based markers of povertyaccurately measure personal poverty? Can J Public Health 2001,92(3):184-187.18. Hanley GE, Morgan S: On the validity of area-based income measures toproxy household income. BMC Health Serv Res 2008, 8:79.19. Geronimus AT, Bound J: Use of census-based aggregate variables toproxy for socioeconomic group: evidence from national samples. Am JEpidemiol 1998, 148(5):475-486.20. Diez-Roux AV, Kiefe CI, Jacobs DR Jr, Haan M, Jackson SA, Nieto FJ,Paton CC, Schulz R: Area characteristics and individual-levelsocioeconomic position indicators in three population-basedepidemiologic studies. Ann Epidemiol 2001, 11(6):395-405.21. Dominguez-Berjon F, Borrell C, Rodriguez-Sanz M, Pastor V: The usefulnessof area-based socioeconomic measures to monitor social inequalities inhealth in Southern Europe. Eur J Public Health 2006, 16(1):54-61.22. Centre for Health Services and Policy Research. Census Geography.[http://www.chspr.ubc.ca/resources/census/geographies/census].23. Statistics Canada. Appendix 6: Geographical Areas. [http://www12.statcan.gc.ca/census-recensement/2011/consultation/DissDiffGuide/Appendix-Annexe-eng.cfm].24. Lynd LD, Sanford AJ, Kelly EM, Paré PD, Bai TR, Fitzgerald JM, Anis AH:Reconcilable differences: a cross-sectional study of the relationshipbetween socioeconomic status and the magnitude of short-acting beta-agonist use in asthma. Chest 2004, 126(4):1161-8.25. Guimaraes C, Marra CA, Colley L, Gill S, Simpson S, Meneilly G, Queiroz RH,Lynd LD: Socioeconomic differences in preferences and willingness-to-pay for insulin delivery systems in type 1 and type 2 diabetes. DiabetesTechnol Ther 2009, 11(9):567-573.26. [http://www12.statcan.ca/english/census01/Products/Reference/dict/index.htm].27. Statistics Canada: Census of Canada: Postal Code Conversion File, PCCF+Version 4J. 2006, Postal Codes, 2001. 2007.28. [http://www.bank-banque-canada.ca/en/cpi.html].29. Shrout PE, Fleiss JL: Intraclass correlations: Uses in assessing raterreliability. Psychol Bull 1979, 86(2):420-428.30. Fleiss JL: Statistical methods for rates and proportions Oxford England: JohnWiley & Sons; 1973.31. Southern DA, McLaren L, Hawe P, Knudtson ML, Ghali WA, InvestigatorsAPPROACH: Individual-level and neighborhood-level income measures:agreement and association with outcomes in a cardiac disease cohort.Med Care 2005, 43(11):1116-1122.32. Locker D, Ford J: Using area-based measures of socioeconomic status indental health services research. J Public Health Dent 1996, 56(2):69-75.33. Lantz PM, Golberstein E, House JS, Morenoff J: Socioeconomic andbehavioral risk factors for mortality in a national 19-year prospectivestudy of U.S. adults. Soc Sci Med 2010, 70(10):1558-1566.34. Sokka T, Kautiainen H, Pincus T, Verstappen SM, Aggarwal A, Alten R,Andersone D, Badsha H, Baecklund E, Belmonte M, Craig-Muller J, daMota LM, Dimic A, Fathi NA, Ferraccioli G, Fukuda W, Geher P, Gogus F,Hajjaj-Hassouni N, Hamoud H, Haugeberg G, Henrohn D, Horslev-Petersen K, Ionescu R, Karateew D, Kuuse R, Laurindo IM, Lazovskis J,Luukkainen R, Mofti A, Murphy E, Nakajima A, Oyoo O, Pandya SC, Pohl C,Predeteanu D, Rexhepi M, Rexhepi S, Sharma B, Shono E, Sibilia J,Sierakowski S, Skopouli FN, Stropuviene S, Toloza S, Valter I, Woolf A,Yamanaka H, the QUEST-RA study group: Work disability remains a majorproblem in rheumatoid arthritis in the 2000s: data from 32 countries inthe QUEST-RA Study. Arthritis Res Ther 2010, 12(2):R42.35. Verstappen SM, Watson KD, Lunt M, McGrother K, Symmons DP, Hyrich KL,the BSR Biologics Register: Working status in patients with rheumatoidarthritis, ankylosing spondylitis and psoriatic arthritis: results from theBritish Society for Rheumatology Biologics Register. Rheumatology(Oxford) 2010.36. Guilley E, Bopp M, Fah D, Paccaud F: Socioeconomic gradients inmortality in the oldest old: A review. Arch Gerontol Geriatr 2010.Pre-publication historyThe pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1472-6963/11/69/prepubdoi:10.1186/1472-6963-11-69Cite this article as: Marra et al.: Agreement between aggregate andindividual-level measures of income and education: a comparisonacross three patient groups. BMC Health Services Research 2011 11:69.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitMarra et al. BMC Health Services Research 2011, 11:69http://www.biomedcentral.com/1472-6963/11/69Page 7 of 7

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.52383.1-0224079/manifest

Comment

Related Items