UBC Faculty Research and Publications

Income inequities in end-of-life health care spending in British Columbia, Canada: A cross-sectional… Cunningham, Colleen M; Hanley, Gillian E; Morgan, Steven G Mar 16, 2011

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

Item Metadata

Download

Media
52383-12939_2010_Article_189.pdf [ 277.86kB ]
Metadata
JSON: 52383-1.0221623.json
JSON-LD: 52383-1.0221623-ld.json
RDF/XML (Pretty): 52383-1.0221623-rdf.xml
RDF/JSON: 52383-1.0221623-rdf.json
Turtle: 52383-1.0221623-turtle.txt
N-Triples: 52383-1.0221623-rdf-ntriples.txt
Original Record: 52383-1.0221623-source.json
Full Text
52383-1.0221623-fulltext.txt
Citation
52383-1.0221623.ris

Full Text

RESEARCH Open AccessIncome inequities in end-of-life health carespending in British Columbia, Canada:A cross-sectional analysis, 2004-2006Colleen M Cunningham1*, Gillian E Hanley1,2, Steven G Morgan1,2AbstractBackground: This study aimed to measure the income-related inequalities and inequities - the inequalities thatremain after accounting for differences in health need - in expenditure on fully publicly covered (hospital andambulatory) and partially publicly covered (prescription drugs) services for those in their last year of life in theprovince of British Columbia (B.C.), Canada. We focused on a decedent population for three reasons: to minimizeunmeasured need differences among our cohort and therefore isolate income effects; to explore inequities for ahigh-spending window of health care use; and, because previous studies have found conflicting relationshipsbetween income and decedent health care spending, to further quantify this relationship.Methods: We used linked administrative databases to describe spending on health services by income for all58,820 deaths of B.C. residents 65 and older from 2004 to 2006. Regression analyses examined the associationbetween income and health care spending, adjusting for age, sex, health status, cause of death, and other relevantfactors. We then used concentration indexes to measure both inequalities and inequities separately for three keytypes of services. Analyses were also run separately for men and women.Results: On average, per capita expenditure on acute health care in the last year of life was $20,705 (CDN2006). Inneed-adjusted regression analyses, we found decedents in the highest income quintile had 11% lower hospitalexpenditures, 15% higher specialist expenditures and 23% higher prescription drug expenditures than decedents inthe lowest income quintile. Concentration index analysis suggested that spending for all types of care wasconcentrated among those with higher income before adjusting for need. Need-adjusted equity results mirroredregression findings and suggested patterns of inequities that were more pronounced among male decedents thanfemales.Conclusions: Despite the universal health care system in B.C., we found patterns of inequity in spending byincome in the last year of life, even for fully publicly covered services. These results, parallel to relationshipsbetween income and spending from previous studies of the B.C. population, suggest persistent income-relatedinequities in the health care Canadians receive throughout their lives.BackgroundEquity in access to health care, or equal access to carefor those with equal need, has long been a concernamong health services researchers and is a primarymotivation for universal health insurance systems.Because use of and spending on health care are largelyinfluenced by need for care, and health status varies byincome [1,2], differences by income in expenditure onhealth care, or inequalities in spending, are not necessa-rily problematic. However, inequalities that remain afteraccounting for differences in need, or inequities in useof health care, signal potential inconsistencies betweenactual health care use or access to care and policy ideals.Previous research has found that income-related dif-ferences in the use of health services, both inequalitiesand inequities, appear to exist in countries with (e.g.Canada and most countries in Europe) and without (e.g.the U.S.) universal health coverage [3]. Recent studies* Correspondence: ccunningham@chspr.ubc.ca1Centre for Health Services and Policy Research, University of BritishColumbia, 201 - 2206 East Mall, Vancouver B.C., V6T1Z3 CanadaFull list of author information is available at the end of the articleCunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12© 2011 Cunningham et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.from the U.S., Europe, and Canada that have controlledfor differences in need using self-reported health status(usually measured on a five-point scale from excellent topoor) along with either self-assessed disease burden oractivity limitations have found evidence of significantinequities across many health care services [4-6]. Forexample, a study based on 1998-2001 non-elderlyrespondents to the U.S. National Health Interview Sur-vey found use of ambulatory care to be inequitable, withthose of higher income using more services than needwould predict [4]. In comparing use of physicians across12 European countries using pan-European survey data,van Doorslaer et al. (2004) found pro-poor inequities invisits to GPs and pro-rich inequities in visits to specia-lists in virtually all countries [5]. For Canada, using datafrom the 2003 wave of the Canadian Community HealthSurvey (CCHS), Allin (2008) found pro-rich inequitiesin specialist visits, and for probability of a visit to a GP,and pro-poor inequities in inpatient care for Canada,but no evidence of income inequities in total visits toGPs or for inpatient care separately for the province ofB.C. [7]. A recent study using administrative claimsdatabases for the full population of B.C., which useddiagnosis code based case-mix tools to adjust for healthstatus, corroborated the CCHS-based findings of pro-rich inequities in specialist care, but found pro-poorinequities in GP visits and inpatient care [8]. While therelationship between income and prescription drug usehas been studied extensively, little work has been doneto directly measure inequities. In sum, while thereappears to be some consensus regarding pro-richinequities for specialist visits, the trends in inequity find-ings for other acute care services, even those specific toB.C., are less clear.Reviewers and authors of past studies of equity inhealth care argue that a key methodological limitationin constructing measures of inequity lies in the difficultyin adequately controlling for variations in need, both insurvey and administrative data based studies [4,8,9]. Theinability to properly measure and control for differencesin need for health care could imply bias in previousmeasures of inequity.We aimed to measure and compare income-relatedinequities in the use of medical services, hospital care,and prescription drugs in B.C. in the last year of life.We focused on end-of-life care for a cohort of seniors(age 65 and older) as this allowed for much greaterneed standardization than the research that has exam-ined the entire population. End-of-life cohorts havebeen used by Wennberg, Fisher and colleagues to helpreduce bias in the measurement of regional variations inhealth services use and cost [10-12]. As such, they alsoappear to be good candidates for investigating socioeco-nomic disparities. As well, this is an important windowof health care use, and of increasing policy importanceas baby-boomer populations in Canada and other devel-oped countries approach old age and policies surround-ing end-of-life care gain prominence [13].Previous studies of income-related differences in useof health care at end-of-life have found mixed results.Studies of U.S. Medicare populations have found weaknegative relationships between area-level income andhealth care spending [14-16], while studies of Swiss andSwedish residents found spending on health care at endof life to increase with household-level income [17,18].Thus, we also aimed to further explore the end-of-lifespending and household income relationship by examin-ing all forms of acute health care spending together andseparately by type, among decedents 65 and older.As supplementary analyses, we performed sex-disaggregated regressions and computed sex-specificmeasures of inequality and inequity. We explored therelationship between income and health services useseparately by sex at end-of-life because women andmen, on average, tend to use different levels of healthcare throughout the life course and die at different agesand of different causes. Thus, we aimed to investigatewhether income-related inequities in health servicesspending also varied by sex.We used linked administrative databases to study theincome-related inequities in spending on four mainfacets of care: hospital care, general practitioner services,specialist services, and prescription drugs. In our analy-sis, we controlled for age, cause of death, sex, area ofresidence, and clinical complexity using standard healthequity measures that facilitated comparison to previousequity studies from Canada, the U.S. and Europe.MethodsDataWith permissions from relevant data stewards at theB.C. Ministry of Health Services and the College ofPharmacists of B.C., we obtained de-identified linkedhealth data from Population Data B.C. and B.C. Pharma-Net [19]. These databases cover all B.C. residents exceptthose whose health care (and therefore health data) isunder federal jurisdiction: status Indians, military per-sonnel, and the Royal Canadian Mounted Police (under5% of the total population). For all other individuals, thedatasets contain demographic information, records of allnon-emergency hospital admissions, records of all fee-for-service medical visits (general and specialist care),and all prescriptions filled outside of acute care hospitals(regardless of payer). All residents of B.C. receive uni-versal, first-dollar public insurance for hospital care andfor general and specialist medical services. Prescriptiondrugs for B.C. residents are covered through a universalpublic drug plan with income-based deductibles, privateCunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12Page 2 of 9employer or individually-sponsored insurance, and out-of-pocket payments (private payments representapproximately 55% of total drug spending in B.C.) [20].Ethics approval for this study was obtained from theBehavioural Research Ethics Board at the University ofBritish Columbia.CohortTo define our cohort of decedents, we retrospectivelyidentified deaths in the calendar years 2004, 2005, and2006 using vital statistics data. We excluded deaths forthose who were under 65 to further minimize the unob-served health differences within our decedent cohort.For the same reason, we excluded those who died ofexternal causes, including accidents and suicides. Wealso excluded from our analysis decedents who did notlive in B.C. for at least 275 days in the year before deathto minimize misclassification bias relating to unobserveduse and unmeasured health status. Finally, to reduce theextent to which use of medical services is undercounted,we excluded residents of local health areas whereingreater than 30% of all medical services are providednon-fee-for-service, and therefore fall outside of ourdataset (representing ~0.5% of all deaths). Our finalcohort included 58,820 decedents.Dates of death in our datasets contained only year andmonth for privacy reasons. We therefore defined the‘year’ prior to death to include the month of death plusthe 12 preceding months.VariablesOur analysis focused on the amount of health careresources used, which we measured in terms of totalcost of acute care services. For the year prior to eachsubject’s death, we calculated the cost of non-emergencyhospital care, fee-for-service care from general practi-tioners, fee-for-service care from specialist physicians,and prescriptions filled outside of acute care hospitals.Costs for prescription drugs were included regardless ofcoverage and included the pharmacist fees associatedwith each prescription filled. Costs for pharmaceuticalsand fee-for-service medical care were based on actualfees and prices paid. However, because hospitals areblock funded in Canada, the cost of hospital care wascomputed using standard costing methods for cases ofdiffering resource intensity [8,21,22]. We ran analysesfor five dependent variables: total acute costs (the sumof hospital, medical and prescription drug expenditures),and separately hospital, fee-for-service general practi-tioner, fee-for-service specialist care, and prescriptiondrugs. All costs were inflation-adjusted to 2006Canadian dollars.Our independent variable of primary interest washousehold income percentile. For approximately 95% ofour cohort, our income percentiles were based onhousehold-specific income data for 2003 that was vali-dated using Canada Revenue Agency data by the B.C.Ministry of Health Services. For the remaining sample,our income percentiles were based on average incomesreported in year-2003 tax returns by households acrossCensus Dissemination Areas (comprising 400-700 per-sons) [23]. For regression analyses, we created dummyvariables for each income quintile in order to allow fornon-linearities in relationships between income anddependent variables. Incomes were re-grouped into sex-specific percentiles and quintiles for sex disaggregatedanalyses.Based on diagnosis codes in medical and hospitalrecords for the last year of life, we used AggregatedDiagnostic Groups (ADGs) of the Johns Hopkins ACG®Case-Mix system to measure each decedent’s generalclinical complexity [24]. Each hospital record containsup to 25 diagnosis codes and each record of a fee-for-service medical visit contains one diagnosis code. TheACG system maps these diagnoses into 32 ADGs ofvarying clinical severity. Groups with the highestexpected resource use are called major ADGs. Wecounted separately the number of major ADGs and thenumber of non-major ADGs for each individual.We controlled for age at death using dummy variablesrepresenting 5-year age bands (e.g. 65 to 69), with acombined group for those 100 and older, as we havepreviously identified non-linear relationships betweenage and use of health care [25,26]. We accounted forwhether an individual resided in a long term care facilityor received palliative care based on program indicatorsfrom the drug claims database. We also controlled forthe effects of urban/rural differences in resource avail-ability and care use; to be considered urban, the major-ity of residents of the local area had to have resided inthe dominant city (comprised of at least 25,000 resi-dents). Finally, because cancer deaths are likely to resultin distinct patterns of end-of-life care [27], and cancerand cardiovascular-related deaths account for the largemajority of deaths and vary (seemingly symmetrically)with income within our sample, we controlled forcancer cause of death using vital statistics records.Because pooled decedent data (2004-2006) were used inorder to increase sample size, we also controlled foryear of death in all analyses.Statistical AnalysisIn our regression analyses, we used generalized linearmethods (GLM) using the log link and the gamma dis-tribution, selected using standard methods for healthcost data [28,29]. GLM allowed us to relax the strictnormal error assumption of linear least squares regres-sion (which right-skewed, zero clustered healthCunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12Page 3 of 9expenditure data often violate) and avoid the complica-tions of retransforming estimates to eliminate potentialbias in non-linear or log-transformed least squares[30,31]. Because nearly our entire cohort had positiveexpenditures for the services examined, we used one-step GLM models.We used concentration indexes (CIs) to measure thedifferences in spending on services by type, and horizon-tal inequity (HI) index measures to quantify inequities inspending unrelated to need [32,33]. In the context of thepresent study, a CI summarizes the extent to which thedistribution of health care resources (costs) differs fromuniformity across income groups (percentiles). A CI willbe zero when resources are distributed equally acrossincome groups, negative when resources are concen-trated among those with low incomes, and positivewhen resources are concentrated among those with highincomes.Some of the resource concentration captured by theCI may reflect income-related inequalities in healthneed. A measure of inequity (as opposed to inequality)must account for income-related differences in healthstatus and thereby quantify how observed patterns ofcare deviate from the principle of ‘equal treatment forequal need’ [5]. To construct such a measure, the HIindex [32], we first used regression analysis to predictthe amount of health care we would have expected eachindividual to receive based on her or his age, sex, healthstatus, and cause of death, holding other variablesincluding income, nursing home residence, and regionof residence at the sample mean. The predicted valuesfrom this analysis are estimates of the amount of healthcare resources that a decedent would have received iftreated the same as the average person with the sameneed profile [33]. We used these values to computeneed-predicted concentration indexes for each categoryof health spending.We then computed the HI for expenditure for eachservice as the difference between the observed inequalityin spending and the inequality in spending that wouldbe predicted based on need. The HI thus measures howmuch of the observed inequality in the distribution ofspending on health care is left unaccounted for by theunequal distribution of health-related need. The HImeasure will be zero when health care spending is dis-tributed across incomes in proportion to how healthcare need is distributed. A negative (positive) HI meansthat health care spending is more concentrated in lower(higher) income groups than health care need wouldpredict. One can quantitatively interpret the HI asroughly the share of expenditure that would need to betransferred from the half of the distribution with greaterthan predicted spending to the other in order for spend-ing to equitably reflect health need [34]. We reportedboth the CI that measures the income-related inequal-ities in actual health care expenditures and the HI.While we expected regression and equity analyses togenerate similar findings, we discuss both: regressionfindings put our results in the context of previous end-of-life literature, while equity analyses allowed us todirectly compare the degree of inequity measured, usingHI estimates, to previous equity results.ResultsTable 1 describes our study population by income quintile.Those with the lowest incomes were older, female, non-urban dwelling, more likely to live in a residential carefacility, less likely to use the palliative care program, andhad fewer diagnosed conditions than those with the high-est incomes. Some of these differences by income (e.g., ageand palliative care) are likely to be related to differences incause of death: low-income decedents were more likely todie of cardiovascular and respiratory conditions but lesslikely to die of cancer than high-income decedents.Table 1 also lists rates of use, various use intensity mea-sures, and mean per capita expenditure by service type.Use of hospitals, specialist services and prescription drugswere higher in higher income groups than for those oflower income, while nearly all decedents visited a generalpractitioner, regardless of income. Expenditures for eachtype of service examined were higher in the highestincome group compared to the lowest.Table 2 lists the adjusted associations between incomequintile and health care spending in the last year of lifefor sex-stratified and pooled models (which include sexas a covariate). These models adjust for age, health sta-tus, cause of death, use of long term care facilities, ruralresidence, and year of death (see the Appendix Table inAdditional file 1 for full regression results). For compar-ison with previous end-of-life literature, the first columnof results describes total health care spending. For allcosts combined in both the sex-pooled and sex-disag-gregated models, there was no statistically significantdifference in total health care spending in the last yearof life between the lowest, second lowest and middleincome quintiles after other factors were accounted for.For women and in the sex-pooled, model total measuredspending on care in the last year of life for those in thehighest two income quintiles was slightly below (by4-6%) those in the lowest income quintile. For men,there was no statistically significant difference.After other factors were accounted for, income gradi-ents for spending during the last year of life differed sig-nificantly by type of health care. In sex-pooled resultsthe gradient for hospital spending was negative whereasgradients for spending on general practitioner services,specialist services, and prescription drugs were generallypositive. Because hospital spending represents, onCunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12Page 4 of 9average, 80% of total measured health costs in the lastyear of life, hospital and total spending regressions arefairly similar. Spending on hospital care was approxi-mately 10% lower for decedents in the highest twoincome quintiles compared to the lowest incomequintile.In the sex-pooled regression analysis and separatelyfor men the gradient in spending on general practitionerservices among the last year of life was positive, butdecreasing. Decedents in the middle income quintileshad 5-6% higher spending than those in the lowestquintile while decedents in the highest quintile had 2%higher spending than those in the lowest quintile. Theincome gradient for spending on general practitionerservices was not significant for women.After other factors were accounted for, spending onspecialist medical care and prescription drugs during thelast year of life was more positively correlated withTable 1 Description of study population, by income quintileIncome quintile1 (lowest) 2 3 4 5 (highest) TotalN 11,764 11,764 11,764 11,764 11,764 58,820Died 2004, % 34 (0.44) 34 (0.44) 34 (0.44) 33 (0.43) 31 (0.43) 33 (0.19)Died 2005, % 33 (0.43) 34 (0.44) 33 (0.43) 32 (0.43) 33 (0.43) 33 (0.19)Died 2006, % 33 (0.43) 33 (0.43) 34 (0.44) 34 (0.44) 35 (0.44) 34 (0.19)Female, % 71 (0.42) 63 (0.45) 48 (0.46) 43 (0.46) 38 (0.45) 52 (0.21)Age at death, mean 83.5 (0.09) 84.1 (0.07) 82.1 (0.07) 81.0 (0.07) 80.4 (0.07) 82.2 (0.03)Urban resident, % 81 (0.24) 79 (0.25) 78 (0.26) 78 (0.25) 84 (0.20) 80 (0.11)Major ADGs, mean count 2.6 (0.01) 2.7 (0.01) 2.8 (0.01) 2.8 (0.01) 2.9 (0.01) 2.8 (0.01)Minor ADGs, mean count 5.8 (0.03) 6.1 (0.03) 6.2 (0.02) 6.2 (0.03) 6.3 (0.02) 6.1 (0.01)Cancer cause of death, % 20 (0.37) 22 (0.38) 26 (0.41) 29 (0.42) 32 (0.43) 26 (0.18)Cardiovascular cause of death, % 28 (0.41) 27 (0.41) 25 (0.40) 24 (0.39) 23 (0.39) 25 (0.18)Respiratory cause of death, % 14 (0.32) 13 (0.31) 13 (0.31) 11 (0.29) 11 (0.29) 12 (0.14)Residential care, % 28 (0.41) 26 (0.40) 22 (0.38) 20 (0.37) 18 (0.35) 23 (0.17)Palliative care Rx coverage, % 11 (0.29) 12 (0.30) 15 (0.33) 17 (0.35) 21 (0.37) 15 (0.15)USEHospital, % 68 (0.43) 72 (0.42) 73 (0.41) 73 (0.41) 75 (0.40) 72 (0.18)General practitioner, % 98 (0.14) 99 (0.10) 99 (0.10) 99 (0.10) 99 (0.09) 99 (0.05)Specialist physicians, % 92 (0.25) 94 (0.22) 94 (0.21) 95 (0.21) 96 (0.19) 94 (0.10)Prescription drug, % 86 (0.32) 86 (0.32) 86 (0.32) 88 (0.30) 90 (0.28) 87 (0.14)Nights in hospital, mean 19 (0.28) 21 (0.29) 21 (0.30) 20 (0.28) 20 (0.29) 20 (0.13)General practitioner contacts,mean24.2 (0.18) 26.4 (0.18) 26.7 (0.18) 26.9 (0.18) 26.8 (0.18) 26.2 (0.08)Specialist contacts, mean 17.4 (0.24) 19.1 (0.24) 20.9 (0.25) 22.7 (0.26) 25.6 (0.27) 21.1 (0.11)Count of drug types (ATCL3), mean 8.14 (0.05) 8.38 (0.05) 8.42 (0.05) 8.59 (0.05) 8.91 (0.05) 8.49 (0.02)Total days of prescriptions,mean1428 (11.5) 1460 (11.3) 1442 (11.3) 1428 (11.0) 1447 (10.7) 1441 (5.0)Number of prescriptions, mean 60 (0.82) 61 (0.83) 54 (0.71) 50 (0.63) 47 (0.59) 54 (0.33)EXPENDITURES ($2006)Total 19,063(238)20,183(239)21,213(254)20,870(243)22,201(268)20,705(111)Hospital 15,165(220)16,035(222)16,787(236)16,249(224)17,111(247)16,269(103)General practitioners 949 (7.7) 1,050 (7.9) 1,076 (7.9) 1,071 (7.8) 1,059 (7.6) 1,041 (3.5)Specialist physicians 1,380(18.6)1,460(18.1)1,674(19.9)1,803(21.2)2,072(23.1)1,678 (9.1)Prescription drugs 1,526(17.1)1,593(17.0)1,639(17.4)1,726(19.4)1,939(22.7)1,685 (8.4)ADGs = Aggregated Diagnostic Groups.Standard errors in parentheses.Cunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12Page 5 of 9income than spending on general practitioners, particu-larly among men. Spending on specialist care during thelast year of life by males in the highest quintile was 20%higher than such spending by males in the lowestincome quintile. Similarly, spending on prescriptiondrugs by high-income men during their last year of lifewas approximately 30% higher than prescription drugspending by men in the lowest income quintile. Forwomen, spending on specialist medical care and on pre-scription drugs during the last year of life was signifi-cantly higher among those in the highest incomequintile compared to those in the lowest income quin-tile; however, the gradient across income quintiles wasnot significant for specialist care and non-linear for pre-scription drugs.Table 3 lists the concentration indexes and horizontalinequity measures for hospital services, general practi-tioners, specialists and prescription drugs. For the sex-pooled model, resource use was concentrated in higherincome groups, for all services examined, beforeadjusting for need (CI). After adjusting for need, spend-ing on hospital care and general practitioner serviceswas disproportionately concentrated in those of lowerincome (HI -0.031 and -0.006, respectively). While lessdramatic than the concentration index measures, spend-ing on specialist services and prescription drugsremained concentrated in those with higher incomes(HI 0.034 and 0.033). Across types of care studied, thedegree of measured horizontal inequity in health carespending was greater for men than women, particularlyfor specialist medical care and prescription drugs, whichare disproportionately concentrated in higher incomedecedents.DiscussionUsing population-based sources of administrative healthcare data, we found evidence of significant inequities inthe need-adjusted distribution of health services at end-of-life. When health need is accounted for, decedents ofhigher income have higher expenditures than those ofTable 2 Regression results: adjusted relationship between income quintile and spending on health services in the lastyear of life for a cross-section of British Columbians in their last year of life, 2004-2006Total Population (n = 58,880)All servicescombinedHospitalcareGeneralpractitionersMedicalspecialistsPrescriptiondrugsIncome 1(reference)— — — — —Income 2 0.024 0.009 0.050*** 0.040** 0.072***Income 3 0.006 -0.020 0.061*** 0.055*** 0.073***Income 4 -0.040** -0.091*** 0.050*** 0.065*** 0.106***Income 5 -0.037* -0.115*** 0.020* 0.137*** 0.208***Females (n = 30,087)All services combined Hospital care General practitioners Medical specialists Prescription drugsIncome 1(reference)— — — — —Income 2 -0.014 -0.051 0.005 0.017 0.055**Income 3 -0.004 -0.030 0.032** 0.033 0.067**Income 4 -0.058** -0.111** 0.014 0.041 0.030Income 5 -0.059** -0.132*** -0.009 0.076*** 0.122***Males (n = 28,013)All services combined Hospital care General practitioners Medical specialists Prescription drugsIncome 1(reference)— — — — —Income 2 0.038 0.022 0.068*** 0.068*** 0.118***Income 3 0.014 -0.015 0.096*** 0.083*** 0.151***Income 4 -0.030 -0.096** 0.063*** 0.104*** 0.222***Income 5 -0.026 -0.116*** 0.022 0.200*** 0.316***Model adjusted for age, health status, cancer death, year of death, urban dwelling, residential care, and sex (sex-pooled model only). See Additional file 1 for allregression coefficients.Regression models are GLM with log link (dependent variables thus modeled as log of expenditure in last year of life).* p < 0.05, ** p < 0.01, *** p < 0.001.Cunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12Page 6 of 9lower income for prescription drugs and specialists; forhospital services, spending was concentrated in those oflower income more so than the income distribution ofneed would predict. Spending on general practitionersshowed a very small concentration in those of lowincome, although regression results highlighted a non-linear relationship between need-adjusted spending andincome. For all types of services examined, inequitiesappeared more pronounced among male decedents thanfemales.Results of our regression analyses showing a negativerelationship between income and total acute healthcarespending at end-of-life are consistent with previous ana-lyses of total Medicare expenditures for sample decedentpopulations [14,15]. Yet, the results for total acutespending contrast a recent analysis of a full populationof Stockholm decedents which found spending at end-of-life on the same suite of services to increase withincome [17]. That analysis did not include detailedinformation on individual co-morbidities; our findings ofa negative income relationship for overall end-of-lifespending after need adjustment are largely influenced byaccounting for health status, and allowing for a non-linear relationship between age at death (which varieswith income) and spending. Sensitivity analyses in whichwe omitted health status controls (major and non-majorADGs) found no significant relationship betweenincome and total measured health care spending (dataavailable upon request). By using improved health statusmeasures combined with full population data and allow-ing for flexible relationships between health care spend-ing and covariates, our analysis improves on theseprevious end-of-life studies.Using an end-of-life cohort for our equity analysis, ourmeasures of horizontal inequity for hospital and medicalservices care were smaller than but similar to measuresfrom survey-based studies of full European and NorthAmerican populations [3,5,6]. Our results concerninginequities in end-of-life spending on hospital, generalpractitioner and specialist care are of the same direction,although generally larger, than those found usingadministrative data for the full population (includingnon-decedents) of B.C. in 2002 [8]. Results for specialistservices were similar to those found using survey datafrom 2003. Hospital care results in that study were notsignificant for B.C.; however they found pro-poorinequities in hospital care for Canada as a whole [7].While small sample sizes might have limited province-specific results in that study (their B.C. sample was12,367), point estimates of HI suggested pro-rich inequi-ties in use of inpatient care as measured by total nightsin hospital.Previous research that compared horizontal inequityin universally covered services (including hospital, GPand specialist services) to partially covered services(including pharmaceuticals) in Denmark found statisti-cally significant inequities only for partially covered ser-vices, with amount spent concentrated in those withhigher incomes [35]. Point estimates of horizontalinequity for hospital and specialist services, while insig-nificant, were similar in size to ours; however, the HIfor pharmaceuticals was three times as large in the Dan-ish study. Such differences are likely related to differ-ences in public drug plans: in B.C., public coverage isbased on household income and the amount spent onpharmaceuticals, while in Denmark public reimburse-ment is related only to the absolute amount spent onpharmaceuticals.While a higher concentration of spending on specia-lists and pharmaceuticals by higher income decedents islikely not surprising, given other socio-economic differ-ences that may make accessing such services easier forthose of higher income, including education, privatesupport systems, employer-based insurance and relativeaffordability of any out-of-pocket payments, it is perhapsmore surprising to find the reverse income gradient forhospital services. Some of this may be substitution ofhospital care for other health services. As research fromthe US suggests, low income is associated with a highernumber of ambulatory care-sensitive condition admis-sions at the hospital level for Veterans Affairs patients[36]. Further research focusing on hospital care couldilluminate how use of varying types and differentTable 3 Inequality and inequity in spending on hospital,general practitioners, specialists, and pharmaceuticals atend-of-lifeTotal PopulationInequality (CI) Inequity (HI)Hospital 0.017 -0.031General practitioner 0.017 -0.006Specialist physician 0.078 0.024Prescription drugs 0.044 0.033FemalesInequality (CI) Inequity (HI)Hospital 0.006 -0.028General Practitioner 0.009 -0.006Specialist physician 0.062 0.022Prescription drugs 0.040 0.020MalesInequality (CI) Inequity (HI)Hospital 0.002 -0.031General practitioner 0.015 -0.008Specialist physician 0.057 0.030Prescription drugs 0.065 0.055Bold values statistically significant at p < 0.05.CI = Concentration Index, HI = Horizontal Inequity.Cunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12Page 7 of 9intensities of end-of-life care within hospitals contri-butes to income-related inequities in spending.Our results should be interpreted within the contextof several notable limitations. First, we lack informa-tion on continuing care, whether provided in skillednursing facilities or at home, both of which maysubstitute for some forms of hospital care at the end-of-life. Including publicly funded care would havepotentially steepened the income gradient found inanalysis of combined health services, as residence incare homes appears higher for lower income decedents(see Table 1); however, previous research suggests anunclear relationship between income and continuingcare use [37-39].A second limitation of our analysis relates to ourincome measure. Ideally, we would have had incomevalues for each decedent for the year preceding theirdeath. However, given the stability of income over timeand the short period of our analysis, misclassificationfrom using 2003 incomes was likely minimal.We analysed spending on each health service, whichacts as a proxy of resource use intensity, but does notmeasure the quality of service. A quality adjustment todollars spent might have produced a stronger incomegradient in medical services. A study of home visits byfamily practitioners to end-of-life cancer patients inNova Scotia found that while nearly all patients hadcontact with a family practitioner in their last sixmonths of life, those living in lower income areas wereless likely to receive home visits compared to high andmiddle income area residents [40]. Quality adjustmentsmay also have mitigated income findings if those receiv-ing more care in terms of dollars spent were somehowgetting care of lower quality or excessive levels of care.In other words, because we lack a measure of “optimal”or “appropriate” care, our findings of a pro-poor distri-bution in hospital spending at end-of-life may reflectsome overuse of such care by persons with low incomes.Similarly, the pro-rich findings with respect to specialistmedical care and prescription drugs might representover-provision of such care to wealthier patients. How-ever, a recent study of Canadian populations suggeststhat including a measure of unmet need in equity ana-lyses, allowing for heterogeneity in the level of optimalcare, did not affect equity findings [41]. The robustnessof inequity findings suggests that identifying potentialsystemic reasons for the observed patterns of inequity isimportant in designing policies to address thesepatterns.ConclusionsThis paper is the first to our knowledge to compara-tively measure the level of inequity in health care spend-ing for decedents across various types of care usingstandard health equity measures. It is also the first tocomparatively examine inequities in spending by typeseparately for men and women decedents. We foundhigher spending on specialist services and pharmaceuti-cals at end-of-life for those with higher incomes, whichmirrors pro-rich patterns of spending and use in non-decedent populations. This suggests continuity ofincome inequities in health care use over the life courseand the potential lock-in of care inequities over time.The concentration of hospital spending among those oflower incomes may be reflective of trade-offs in carebetween specialist services, general practitioners, andhospitals, or substitution of hospital care for home-based health and social care [42], that may signal ineffi-ciencies. While our measures of inequity, put in interna-tional context, are relatively small, findings of inequitiesin a population for which unobserved health differencesare minimized confirm the presence of systemic inequi-ties in access to care.Additional materialAdditional file 1: Table S1. Full regression results - Adjustedrelationship between spending on health services in the last year of lifeand determinants for a cross-section of British Columbians in their lastyear of life, 2004-2006. The Appendix Table provides expanded results forregressions from Table 2 (including non-income coefficient estimates).AcknowledgementsThis study was funded by an operating grant ("Return on investment frompharmaceutical care: Measuring population-based causes and consequencesof prescription drug utilization and expenditure”) from the CanadianInstitutes of Health Research and by contributions of the BC Ministry ofHealth Services to the UBC Centre for Health Services and Policy Research.Sponsors had no role in the project or in decisions to publish results.Author details1Centre for Health Services and Policy Research, University of BritishColumbia, 201 - 2206 East Mall, Vancouver B.C., V6T1Z3 Canada. 2School ofPopulation and Public Health, University of British Columbia, 201 - 2206 EastMall, Vancouver B.C., V6T1Z3 Canada.Authors’ contributionsAll authors contributed to project conception and study design. CMC wasresponsible for data analysis, interpretation of results, and was the principalwriter of the manuscript. GEH contributed to interpretation of results andmanuscript preparation and revision. SGM was responsible for theacquisition of the data and contributed to interpretation of results andmanuscript preparation and revision. All authors approved the final versionof the manuscript.Competing interestsThe authors declare that they have no competing interests.Received: 26 November 2010 Accepted: 16 March 2011Published: 16 March 2011References1. Frohlich KL, Ross N, Richmond C: Health disparities in Canada today:Some evidence and a theoretical framework. Health Policy 2006,79:132-143.Cunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12Page 8 of 92. Mackenbach JP, Kunst AE, Groenhof F, Borgan JK, Costa G, Faggiano F,Jozan P, Leinsalu M, Martikainen P, Rychtarikova J, Valkonen T:Socioeconomic inequalities in mortality among women and amongmen: An international study. Am J Public Health 1999, 89:1800-1806.3. van Doorslaer E, Masseria C: Income-related inequality in the use of medicalcare in 21 OECD countries Paris: Organisation for Cooperation inDevelopment; 2004, 44.4. Shin H, Kim J: Differences in income-related inequality and horizontalinequity in ambulatory care use between rural and non-rural areas:Using the 1998-2001 U.S. National Health Interview Survey data.International Journal for Equity in Health 2010, 9:17.5. van Doorslaer E, Koolman X, Jones AM: Explaining income-relatedinequalities in doctor utilisation in Europe. Health Econ 2004, 13:629-647.6. van Doorslaer E, Masseria C, Koolman X, for the OECD Health EquityResearch Group: Inequalities in access to medical care by income indeveloped countries. CMAJ 2006, 174:177-183.7. Allin S: Does equity in healthcare use vary across Canadian provinces?Healthcare Policy 2008, 3:83-99.8. McGrail K: Income-related inequities: Cross-sectional analyses of the useof medicare services in British Columbia in 1992 and 2002. OpenMedicine 2008, 2.9. Goddard M, Smith P: Equity of access to health care services: Theory andevidence from the UK. Social Science & Medicine 2001, 53:1149-1162.10. Wennberg JE, Fisher ES, Stukel TA, Skinner JS, Sharp SM, Bronner KK: Use ofhospitals, physician visits, and hospice care during last six months of lifeamong cohorts loyal to highly respected hospitals in the United States.BMJ (Clinical research ed) 2004, 328:607.11. Fisher ES, Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL: Theimplications of regional variations in Medicare spending. Part 2: Healthoutcomes and satisfaction with care. Annals of Internal Medicine 2003,138:288-298.12. Fisher ES, Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL: Theimplications of regional variations in Medicare spending. Part 1: Thecontent, quality, and accessibility of care. Annals of Internal Medicine 2003,138:273-287.13. Health Canada: Canadian Strategy on Palliative and End-of-life Care: FinalReport of the Coordinating Committee Ottawa: Health Canada; 2007.14. Shugarman LR, Campbell DE, Bird CE, Gabel J, Louis TA, Lynn J: Differencesin Medicare expenditures during the last 3 years of life. JGIM: Journal ofGeneral Internal Medicine 2004, 19:127.15. Hogan C, Lunney J, Gabel J, Lynn J: Medicare beneficiaries’ costs of carein the last year of life. Health Affairs 2001, 20:188-195.16. Hanchate A, Kronman AC, Young-Xu Y, Ash AS, Emanuel E: Racial andethnic differences in end-of-life costs: Why do minorities cost more thanwhites? Archives of Internal Medicine 2009, 169:493-501.17. Hanratty B, Burström B, Walander A, Whitehead M: Inequality in the face ofdeath? Public expenditure on health care for different socioeconomicgroups in the last year of life. Journal of Health Services Research & Policy2007, 12:90-94.18. Felder S, Meier M, Schmitt H: Health care expenditure in the last monthsof life. Journal of Health Economics 2000, 19:679-695.19. Population Data BC. [http://www.popdata.bc.ca/data].20. Canadian Institutes for Health Information (CIHI): Drug Expenditure inCanada, 1985 to 2009 Ottawa: CIHI; 2010, 158.21. McGrail K, Green B, Barer M, Evans R, Hertzman C, Normand C: Age, costsof acute and long-term care and proximity to death: Evidence for 1987-88 and 1994-95 in British Columbia. Age and Ageing 2000, 29:249-253.22. Payne G, Laporte A, Foot DK, Coyte PC: Temporal trends in the relativecost of dying: Evidence from Canada. Health Policy 2009, 90:270-276.23. Hanley G, Morgan S: On the validity of area-based income measures toproxy household income. BMC Health Services Research 2008, 8:79.24. Weiner JP, Starfield BH, Steinwachs DM, Mumford LM: Development andapplication of a population-oriented measure of ambulatory care case-mix. Medical Care 1991, 29:452-472.25. Morgan S, Cunningham C, Hanley G, Mooney D: The British Columbia RxAtlas. 2 edition. Vancouver: Centre for Health Services and Policy Research,University of British Columbia; 2009.26. Morgan S, Cunningham C, Hanley G, Mooney D: The British ColumbiaMedical and Hospital Atlas Vancouver: Centre for Health Services and PolicyResearch, University of British Columbia; 2009.27. Lorenz KA, Lynn J, Dy SM, Shugarman LR, Wilkinson A, Mularski RA,Morton SC, Hughes RG, Hilton LK, Maglione M, Rhodes SL, Rolon C, Sun VC,Shekelle PG: Evidence for improving palliative care at the end of life: Asystematic review. Annals of Internal Medicine 2008, 148:147-159.28. Buntin MB, Zaslavsky AM: Too much ado about two-part models andtransformation? Comparing methods of modeling Medicareexpenditures. Journal of Health Economics 2004, 23:525-542.29. Manning WG, Mullahy J: Estimating log models: To transform or not totransform? Journal of Health Economics 2001, 20:461-494.30. Manning WG: The logged dependent variable, heteroscedasticity, andthe retransformation problem. Journal of Health Economics 1998,17:283-295.31. McCullagh P: Generalized linear models London; New York: Chapman andHall; 1983.32. O’Donnell O, van Doorslaer E, Wagstaff A, Lindelow M: Analyzing healthequity using household survey data analyzing: A guide to techniques and theirimplementation World Bank; 2008.33. Wagstaff A, van Doorslaer E: Measuring and testing for inequity in thedelivery of health care. Journal of Human Resources 2000, 35:716-733.34. Koolman X, van Doorslaer E: On the interpretation of a concentrationindex of inequality. Health Econ 2004, 13:649-656.35. Gundgaard J: Income-related inequality in utilization of health services inDenmark: Evidence from Funen County. Scandinavian Journal of PublicHealth 2006, 34:462-471.36. Finegan MS, Gao J, Pasquale D, Campbell J: Trends and geographicvariation of potentially avoidable hospitalizations in the Veterans health-care system. Health Serv Manage Res 2010, 23:66-75.37. Branch LG, Ku L: Transition probabilities to dependency,institutionalization, and death among the elderly over a decade. Journalof Aging and Health 1989, 1:370-408.38. Johnson RJ, Wolinsky FD: Use of community-based long-term careservices by older adults. Journal of Aging and Health 1996, 8:512-537.39. Finlayson M: Changes predicting long-term care use among the oldest-old. The Gerontologist 2002, 42:443-453.40. Burge F, Lawson B, Johnston G: Home visits by family physicians duringthe end-of-life: Does patient income or residence play a role? BMCPalliative Care 2005, 4:1.41. Allin S, Grignon M, Le Grand J: Subjective unmet need and utilization ofhealth care services in Canada: What are the equity implications? SocialScience & Medicine 2010, 70:465-472.42. Seow H, Barbera L, Howell D, Dy SM: Using more end-of-life homecareservices is associated with using fewer acute care services. Medical Care2010, 48:118-124.doi:10.1186/1475-9276-10-12Cite this article as: Cunningham et al.: Income inequities in end-of-lifehealth care spending in British Columbia, Canada: A cross-sectionalanalysis, 2004-2006. International Journal for Equity in Health 2011 10:12.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/submitCunningham et al. International Journal for Equity in Health 2011, 10:12http://www.equityhealthj.com/content/10/1/12Page 9 of 9

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.52383.1-0221623/manifest

Comment

Related Items