Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The horizontal equity of health care in New Zealand

The horizontal equity of health care in New Zealand David Peacock Public Health Medicine Registrar, New Zealand equal need’ is a clearly articulated goal of the New Zealand public health system, this study is an attempt to determine if access to public health care services in New Zealand is, for people of equal health need, independent o income. f Method: Information on health status, income and health service utilisation for just over 6,000 New Zealanders was obtained from the national Household Health Survey 1992-93.Using standardised expenditure concentration curves and a concentration index, the distribution of health service use by individuals in different income groups, as a proxy for access, was illustrated and quantified. Results: The results suggest either appropriate or slightly excess use of services by the poor given their estimated health need. Due to analytical problems caused by data deficiencies, these results must be regarded a s tentative. Conclusion: For the period under study, no evidence was found to indicate significant access barriers to publicly funded health care for people on different incomes. This study has served to demonstrate one approach to measuring inequality and analysing the relationship between inequality and inequity. Given the reforms to the health sector since 1993, ongoing monitoring o equity of access to f health care services i essential. s Implications: Fiven the income-related disparities in health that do exist, the public health community should endeavour to develop techniques to monitor the delivery of publicly funded health care to ensure that further inequity is not borne by the poor. (AustNZJPublic Health 1999;23: 126-130) Nancy Devlin Department of Economics, University of Otago, New Zealand Rob McGee Department of Preventive and Social Medicine, University of Otago, New Zealand just health system is important. Many regard equity of access to health care as a cornerstone of a fair system and many countries, including New Zealand, recognise the need for government intervention to achieve a distributionally just health care system.’-sEnsuring equity of access is one such intervention commanding wide support.6 Against this background, there have been several studies in New Zealand to investigate aspects of equity.’-I3 In general, these studies have focused on disadvantaged groups in society. By contrast, the present study aims to examine the equity of the New Zealand public health care system at a national level: to determine whether access to the New Zealand public health system, for individuals of equal health need, is independent of income (i.e. horizontal equity, that ‘equals should be treated equally’). The results of this study may also provide a benchmark for similar work in the future. Before describing the study in detail, ‘need’, ‘utilisation’ and ‘inequality measurement’ need to be briefly addressed. self-reported health as a measure of health state and status, its validity has been investigated in a number of s t ~ d i e s . ” - ~ ~ T hstudese ies provide sufficient evidence linking ‘poor’ self-reported health with mortality to suggest that self-reported health may be a valid indicator of need. However, there is less evidence on the link between self-reported health and morbidity. Further, this study assumes that self-reported health is reported similarly by rich and poor. This assumption may be challenged. However, ifhealth evaluations do vary according to income, the differences may not be large.21 Additionally. in this survey, there was a clear inverse relationship between self-assessed health and income.1° Need Although determining health needs is difficult, this study uses self-reported health as a surrogate measure of health care need, an approach widely used in research on horizontal equity.l4-I6 Because of the growing interest in using Correspondence to: Utilisation and access Similarly, measuring access is problematic. Most often, access is defined either in terms of the opportunity cost of using a service or, alternatively, by a measure of the welfare loss consequent upon using a service.?? There are empirical difficulties in either approach. Therefore, as used in other studies, the present work uses utilisation of health services as a proxy measure of the accessibility of those service^.^^,^^ However, utilisation of health services by different individuals only validly reflects their access to services if their preferences are the same.?‘ This is a significant criticism. Nevertheless, as the work outlined in this paper is intended Submitted: April 1998 Revision requested: September 1998 Accepted: January 1999 Dr David Peacock, Department of Preventive i Social Medicine, University of Otago, PO Box 3 913, Dunedin, New Zealand. Fax: +64 3 479 7298; e-mail: DPeacock@heaIthotago.com AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 1999 VOL. 23 NO. 2 The horizontal eauitv of health care in New Zealand 0 20 40 60 80 1OC Cumul % of DoDulatlon ranked bv income Figure 1:Standardised expenditure concentration curve for hypothetical distribution of expenditure. as the first in a longitudinal series to study changes over time, this criticism may carry less weight.22 Measuringinequality Figure 1 illustrates the typical approach used in the literature in measuring inequality across varying levels of income. Respondents are placed in income quintiles and inequity is illustrated and quantified using standardised expenditure concentration curves (SECC in Figure 1) and a concentration index. Standardised expenditure is the expected expenditure on health care for members of each income band after direct standardisation for age and morbidity. Differences in need are therefore explicitly addressed. Standardised expenditure on health care allows different mixes of health service use to be measured by a common currency and reflects health care utilisation. In plotting the standardised expenditure concentration curve, income quintiles are ranked on the horizontal axis from poorest to richest, and the cumulative proportion of standardised health care expenditure on each income quintile plotted on the vertical axis. The extent of inequality is quantified by the concentration index, which is twice the area between the SECC and the diagonal. It is negative if SECC lies above the diagonal, indicating inequity favouring the poor, and positive below, indicating inequity favouring the rich. If the standardised expenditure concentration curve crosses the diagonal, the concentration index quantifies the net deviation: the implicit value judgement inherent in this characteristic is that inequity favouring one group can be offset by inequity favouring another. This assumption may be controversial. income was measured in the HHS: different income bands were used depending on whether the household contained a sole or multiple income earners. However, sensitivity analysis suggests that any inaccuracies introduced were not large.The upper bounds of the four lowest income quintiles were approximately $18,000, $28,000, $40,000 and $60,000 a year. All income figures were gross, i.e. before tax. Unfortunately, as the HHS did not include information on household composition, it was not possible to adjust for the number and relationships within the household. All results were directly age-standardised into four age bands: under 15 years; 16 to 44 years; 45 to 64 years; and greater than 64 years. Respondents were asked to categorise their current health state as ‘excellent’, ‘good’, ‘not so good’ or ‘poor’. Due to small numbers, the two categories of ‘not so good’ and ‘poor’ were combined into a category of ‘fair’. Additionally, analyses were performed based on the presence of acute or chronic illness. For each respondent, health service expenditure over the previous year was calculated for general practice consultations (GP), and public hospital out-patient visits (OP) or public hospital inpatient (IP) nights.To obtain a common currency ofpublic health service utilisation, an average cost was estimated for each unit of service: $3 1 for a GP consultation or OP and $250 for an IP night, respectively (all costs in $NZ).26.27 However, in view of the considerable uncertainty in these estimates, sensitivity analyses were performed using different estimates. For each income quintile, the total standardised expenditure on all individuals therein was calculated. However, the utilisation of IP services should be interpreted cautiously because access to these is not always through primary care; emergency admissions do occur. The concentration index, HIwyp,(after the authors who developed it) was calculated separately for overall health expenditure, GP visits, OP attendances and IP admissions.?* Results The HHS dataset contains the responses from 8,910 households. Unfortunately, 1,845 records omitted the respondent’s age, a further 703 records lacked income details and another 66 records were rejected because of internal data inconsistencies; 6,296 records remained for processing (70.7% of total responses). The rejected records included more females (50.4% vs. 53.5%), a greater proportion of respondents with ‘not-so-good’ or ‘poor’ health (12.4% vs. 8.4%) and more single or widowed people. In the absence of income and age details, it is unknown what effect these differences may have had on the results. The sample was generally representative of the New Zealand population. The percentage of females and males in the sample were 50.4% and 49.6%; comparable figures from the 1991 Census were 50.7% and 49.3%, respectively. Similarly, the ethnic composition of the sample closely reflects that of the total New Zealand population; the percentages being 79.9% Pakeha, 12.8% Maori and 4.4% Pacific Islanders. This compares to the 1991 Census distribution of 79.5%, 12.9% and 4.5%, respectively. The age distribution is given inTable 1. In the Method Study data came from the Household Health Survey (HHS), a nation-wide survey developed to give a profile of the health status, health service use, distribution of risk factors and basic sociodemographic factors of the New Zealand population.”,25 Calculation of household income was complicated by the way 1999 VOL. 23 NO. 2 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Peacock, Devlin and McGee Table 1: Age distribution of HHS Sample and New Zealand population. Age population (Yo) <15 years Sample (%) 1991 NZ .. 15 to 44 years 45 to 64 years Over 64 years sample used. a inajorrty of respondents were either never or cur9/, rently married. 32% and 1 0 0respectively, 4.5% were widowed and only approxiinatcly 29” were separated or divorced. Overall, 52%, 40% and 8.5% of respondents were in health states ‘excellent’, ‘good’ and ‘fair’, respectively. The size of the latter group varied inversely with income; 16% in the poorest group compared to 3% among the most afnucnt. The majority o f respondents, 79%, had utilised public health care services over the previous year, but the utilisation of health services was grossly positively skewed; 75% of respondents utilised less than S120 worth of cardyear and a tiny majority consuincd a very large amount. Hospital services were used by many: 13% ofrespondents a s OP and X% as 1P” Table 2 and Figures 2-5 illustrate the distribution of utilisation for all income quintiles standardised for health state and age.The results for GP utilisation are likely to be the most accurate. Due to niuch lower utilisation, the results for the use of hospital outpatient and in-patient services are less accurate and may be of indicative value only. Table 2 shows that all the concentration indices values are negative, indicating that the standardised expenditure concentration curves lie above the diagonal, the line of equality. There is about 3% greater utilisation in income quintiles 1 and 3 than might be expected by their burden of ill-health. For G P services, the maximutn amount of use over expectation by the poor was about 1%. Both the poorest and most amucnt quintiles appear to utilise OP services about 2% less than might be expected on the basis of their health need. Finally, income quintiles 1 and 3 have a greater utilisation of IF care than expected of 4 - 9 4 this is the service with the most uneven utilisation of services. ?‘he concentration curves for all services follow. As a measure 0 20 40 60 80 100 Cumul % of population ranked by income Figure 2: Concentration curve for expenditure on all services by income quintiles for health states ‘Excellent’, ‘Good’ or ‘Fair’; standardised for health state and age (Index of Horizontal Inequity HI,,, = -0.096). ofthe effect of standardisation, the unstandardised HIU,pfor titilisation of all services is -0.178. The concentration index value obtained for total service use, -0.096, is unlikely to have occurred by chance (Figure 2). Income quintiles I and 3 apparently overutilise services by approximately 3% and the highest income quintile consumes only 13% of services rather than its expected share of 20Y0. However, by far the largest contributor to this inequality is the larger use by the poor of IP services (Figure 5 ) . Utilisation of GP care does not vary by income (Figure 3). As 79% of respondents had consulted a GP at least once over the previous year, across all health states, the results for utilisation of primary care are probably the most reliable. In Figure 4, the standardised expenditure concentration curve crosses the diagonal: the poor have fewer, and the affluent more, OP visits than expected. Most benefit accrues to income quintile 3 (middle New Zealand). The position and slope of the curve at the point of intersection indicates that utilisation by this group is the most significant contributor to the apparent pro-rich inequality. Figure 5 indicates that IP care in a public hospital shows the Table 2: Distribution of the expenditure on all services by Income Quintile for health states ‘Excellent’, ‘Good’, and ‘Fair’.The results have been standardised for health state and age. Income quintile Cumulative Yo of total sample __ 20 Cumulative Yo total expenditure 23 0 __ Cumulative Yo GP visits (GP) __ 20 5 40.1 Cumulative Yo Out-patient visits (OP) Cumulative YOIn-patient nights (IP) - - Lowest ~~ __.._ ~ ~~~ ___ ~_ -~ .. 43 1 - _ 38 o __-____ 63 9 _ _ ~ _ _ ~ Highest “’W”, ao ~~ __ - _ ~ _ _ ~_ .. _. 100 ~- __ 60.0 _ _ ~ 80.4 -______ 100 - -- - _______ . . . - -0 163= ~. -~ -0 096a -0.005 -0 005 ______________~ Note (a) p<0 001 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 1999 VOL. 23 NO. 2 The horizontal equity of health care in New Zealand 0 20 40 60 80 100 Cumul % of population ranked by Income Figure 3: Concentration curve for expenditure on GP services by income quintiles for health states ‘Excellent’, ‘Good’ or ‘Fair’; standardised for health state and age = -0.005). (Index of Horizontal Inequity H,, I cumul% of populationranked by Income Figure 4: Concentration curve for expenditure on OP services by income quintiles for health states ’Excellent’, ‘Good’ or ‘Fair’; standardised for health state and age. = (Index of Horizontal Inequity HIwvp -0.005). greatest degree of inequality of any service with a concentration index of -0.162.This distribution makes the largest contribution to the total inequality observed; if this service is excluded, then the concentration index for total utilisation is very close to zero. Sensitivity analyses, using different estimates of service costs, did not materially affect the results. Discussion The analyses suggest that the poor utilise public health care services somewhat more than expected by their health need. However, before discussing this result further, a number of significant methodological difficulties and potential biases need to be addressed. The biggest problem with this study was inadequate sample .H 8o $ f3 $ ag a(: z 2 cumul% of population ranked by income Figure 5: Concentration curve for expenditure on IP services by income quintiles for health states ‘Excellent’, ‘Good’ or ‘Fair’; standardised for health state and age I = -0.162). (Index of Horizontal Inequity H,, 1999 VOL. 23 NO. 2 size. In particular, standardisation for gender was not possible and only three health states could be used. Even with these restrictions, the mean expenditures for individuals in ‘fair’health sometimes had a potential sampling error of more than 50%. The resolution of the study suffers accordingly. Income calculation also presented difficulties. For example, respondents who share households with other income earners but without pooling incomes may have been misclassified into a higher income level based on total household income. However, such a misclassification would tend to bias results towards the null. Further, no adjustment could be made for family composition when determining household income. However, given that poorer households have fewer income earners and proportionately more dependants than affluent h o ~ s e h o l d s they are likely to be even poorer ,~~ than indicated by their gross household income. Therefore, misclassification by income is unlikely to have been substantial. A further potential inaccuracy was introduced by the way the annual number of GP visits was measured in the HHS: the maximum number of visits per respondent had to be taken as 12. This restriction may under-state the number of visits of those in ‘not so good’ or ‘poor’ health; however, such a limitation would tend to under-estimate any pro-poor inequality that may exist. Interpretation of the results obtained is not simple. For example, the contribution of in-patient care to the apparent pro-poor inequality observed. It is possible that the poor spend longer in hospital for reasons that are not necessarily directly related to their health, for example, inadequate home support. Further, although public hospital use is unequal, it is not necessarily inequitable: the more affluent may be choosing to receive treatment in the private sector, hence their apparent underutilisation of the public system may not represent inequity. However, the small number who used private hospital services, less than 2% in this study, seems unlikely to explain the distribution obtained. Finally, the structure of the health service itself may determine AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Peacock, Devlin and McGee if pro-poor inequality is inequitable or not. If the function of the New Zealand public health care system is simply to provide a decent minimum of care (a safety net) then ‘overuse’by the poor would be expected. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

The horizontal equity of health care in New Zealand

Download PDF
5 pages

Loading next page...
 
/lp/wiley/the-horizontal-equity-of-health-care-in-new-zealand-kVWFtqk2I3

References (32)

Publisher
Wiley
Copyright
Copyright © 1999 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.1999.tb01221.x
Publisher site
See Article on Publisher Site

Abstract

David Peacock Public Health Medicine Registrar, New Zealand equal need’ is a clearly articulated goal of the New Zealand public health system, this study is an attempt to determine if access to public health care services in New Zealand is, for people of equal health need, independent o income. f Method: Information on health status, income and health service utilisation for just over 6,000 New Zealanders was obtained from the national Household Health Survey 1992-93.Using standardised expenditure concentration curves and a concentration index, the distribution of health service use by individuals in different income groups, as a proxy for access, was illustrated and quantified. Results: The results suggest either appropriate or slightly excess use of services by the poor given their estimated health need. Due to analytical problems caused by data deficiencies, these results must be regarded a s tentative. Conclusion: For the period under study, no evidence was found to indicate significant access barriers to publicly funded health care for people on different incomes. This study has served to demonstrate one approach to measuring inequality and analysing the relationship between inequality and inequity. Given the reforms to the health sector since 1993, ongoing monitoring o equity of access to f health care services i essential. s Implications: Fiven the income-related disparities in health that do exist, the public health community should endeavour to develop techniques to monitor the delivery of publicly funded health care to ensure that further inequity is not borne by the poor. (AustNZJPublic Health 1999;23: 126-130) Nancy Devlin Department of Economics, University of Otago, New Zealand Rob McGee Department of Preventive and Social Medicine, University of Otago, New Zealand just health system is important. Many regard equity of access to health care as a cornerstone of a fair system and many countries, including New Zealand, recognise the need for government intervention to achieve a distributionally just health care system.’-sEnsuring equity of access is one such intervention commanding wide support.6 Against this background, there have been several studies in New Zealand to investigate aspects of equity.’-I3 In general, these studies have focused on disadvantaged groups in society. By contrast, the present study aims to examine the equity of the New Zealand public health care system at a national level: to determine whether access to the New Zealand public health system, for individuals of equal health need, is independent of income (i.e. horizontal equity, that ‘equals should be treated equally’). The results of this study may also provide a benchmark for similar work in the future. Before describing the study in detail, ‘need’, ‘utilisation’ and ‘inequality measurement’ need to be briefly addressed. self-reported health as a measure of health state and status, its validity has been investigated in a number of s t ~ d i e s . ” - ~ ~ T hstudese ies provide sufficient evidence linking ‘poor’ self-reported health with mortality to suggest that self-reported health may be a valid indicator of need. However, there is less evidence on the link between self-reported health and morbidity. Further, this study assumes that self-reported health is reported similarly by rich and poor. This assumption may be challenged. However, ifhealth evaluations do vary according to income, the differences may not be large.21 Additionally. in this survey, there was a clear inverse relationship between self-assessed health and income.1° Need Although determining health needs is difficult, this study uses self-reported health as a surrogate measure of health care need, an approach widely used in research on horizontal equity.l4-I6 Because of the growing interest in using Correspondence to: Utilisation and access Similarly, measuring access is problematic. Most often, access is defined either in terms of the opportunity cost of using a service or, alternatively, by a measure of the welfare loss consequent upon using a service.?? There are empirical difficulties in either approach. Therefore, as used in other studies, the present work uses utilisation of health services as a proxy measure of the accessibility of those service^.^^,^^ However, utilisation of health services by different individuals only validly reflects their access to services if their preferences are the same.?‘ This is a significant criticism. Nevertheless, as the work outlined in this paper is intended Submitted: April 1998 Revision requested: September 1998 Accepted: January 1999 Dr David Peacock, Department of Preventive i Social Medicine, University of Otago, PO Box 3 913, Dunedin, New Zealand. Fax: +64 3 479 7298; e-mail: DPeacock@heaIthotago.com AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 1999 VOL. 23 NO. 2 The horizontal eauitv of health care in New Zealand 0 20 40 60 80 1OC Cumul % of DoDulatlon ranked bv income Figure 1:Standardised expenditure concentration curve for hypothetical distribution of expenditure. as the first in a longitudinal series to study changes over time, this criticism may carry less weight.22 Measuringinequality Figure 1 illustrates the typical approach used in the literature in measuring inequality across varying levels of income. Respondents are placed in income quintiles and inequity is illustrated and quantified using standardised expenditure concentration curves (SECC in Figure 1) and a concentration index. Standardised expenditure is the expected expenditure on health care for members of each income band after direct standardisation for age and morbidity. Differences in need are therefore explicitly addressed. Standardised expenditure on health care allows different mixes of health service use to be measured by a common currency and reflects health care utilisation. In plotting the standardised expenditure concentration curve, income quintiles are ranked on the horizontal axis from poorest to richest, and the cumulative proportion of standardised health care expenditure on each income quintile plotted on the vertical axis. The extent of inequality is quantified by the concentration index, which is twice the area between the SECC and the diagonal. It is negative if SECC lies above the diagonal, indicating inequity favouring the poor, and positive below, indicating inequity favouring the rich. If the standardised expenditure concentration curve crosses the diagonal, the concentration index quantifies the net deviation: the implicit value judgement inherent in this characteristic is that inequity favouring one group can be offset by inequity favouring another. This assumption may be controversial. income was measured in the HHS: different income bands were used depending on whether the household contained a sole or multiple income earners. However, sensitivity analysis suggests that any inaccuracies introduced were not large.The upper bounds of the four lowest income quintiles were approximately $18,000, $28,000, $40,000 and $60,000 a year. All income figures were gross, i.e. before tax. Unfortunately, as the HHS did not include information on household composition, it was not possible to adjust for the number and relationships within the household. All results were directly age-standardised into four age bands: under 15 years; 16 to 44 years; 45 to 64 years; and greater than 64 years. Respondents were asked to categorise their current health state as ‘excellent’, ‘good’, ‘not so good’ or ‘poor’. Due to small numbers, the two categories of ‘not so good’ and ‘poor’ were combined into a category of ‘fair’. Additionally, analyses were performed based on the presence of acute or chronic illness. For each respondent, health service expenditure over the previous year was calculated for general practice consultations (GP), and public hospital out-patient visits (OP) or public hospital inpatient (IP) nights.To obtain a common currency ofpublic health service utilisation, an average cost was estimated for each unit of service: $3 1 for a GP consultation or OP and $250 for an IP night, respectively (all costs in $NZ).26.27 However, in view of the considerable uncertainty in these estimates, sensitivity analyses were performed using different estimates. For each income quintile, the total standardised expenditure on all individuals therein was calculated. However, the utilisation of IP services should be interpreted cautiously because access to these is not always through primary care; emergency admissions do occur. The concentration index, HIwyp,(after the authors who developed it) was calculated separately for overall health expenditure, GP visits, OP attendances and IP admissions.?* Results The HHS dataset contains the responses from 8,910 households. Unfortunately, 1,845 records omitted the respondent’s age, a further 703 records lacked income details and another 66 records were rejected because of internal data inconsistencies; 6,296 records remained for processing (70.7% of total responses). The rejected records included more females (50.4% vs. 53.5%), a greater proportion of respondents with ‘not-so-good’ or ‘poor’ health (12.4% vs. 8.4%) and more single or widowed people. In the absence of income and age details, it is unknown what effect these differences may have had on the results. The sample was generally representative of the New Zealand population. The percentage of females and males in the sample were 50.4% and 49.6%; comparable figures from the 1991 Census were 50.7% and 49.3%, respectively. Similarly, the ethnic composition of the sample closely reflects that of the total New Zealand population; the percentages being 79.9% Pakeha, 12.8% Maori and 4.4% Pacific Islanders. This compares to the 1991 Census distribution of 79.5%, 12.9% and 4.5%, respectively. The age distribution is given inTable 1. In the Method Study data came from the Household Health Survey (HHS), a nation-wide survey developed to give a profile of the health status, health service use, distribution of risk factors and basic sociodemographic factors of the New Zealand population.”,25 Calculation of household income was complicated by the way 1999 VOL. 23 NO. 2 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Peacock, Devlin and McGee Table 1: Age distribution of HHS Sample and New Zealand population. Age population (Yo) <15 years Sample (%) 1991 NZ .. 15 to 44 years 45 to 64 years Over 64 years sample used. a inajorrty of respondents were either never or cur9/, rently married. 32% and 1 0 0respectively, 4.5% were widowed and only approxiinatcly 29” were separated or divorced. Overall, 52%, 40% and 8.5% of respondents were in health states ‘excellent’, ‘good’ and ‘fair’, respectively. The size of the latter group varied inversely with income; 16% in the poorest group compared to 3% among the most afnucnt. The majority o f respondents, 79%, had utilised public health care services over the previous year, but the utilisation of health services was grossly positively skewed; 75% of respondents utilised less than S120 worth of cardyear and a tiny majority consuincd a very large amount. Hospital services were used by many: 13% ofrespondents a s OP and X% as 1P” Table 2 and Figures 2-5 illustrate the distribution of utilisation for all income quintiles standardised for health state and age.The results for GP utilisation are likely to be the most accurate. Due to niuch lower utilisation, the results for the use of hospital outpatient and in-patient services are less accurate and may be of indicative value only. Table 2 shows that all the concentration indices values are negative, indicating that the standardised expenditure concentration curves lie above the diagonal, the line of equality. There is about 3% greater utilisation in income quintiles 1 and 3 than might be expected by their burden of ill-health. For G P services, the maximutn amount of use over expectation by the poor was about 1%. Both the poorest and most amucnt quintiles appear to utilise OP services about 2% less than might be expected on the basis of their health need. Finally, income quintiles 1 and 3 have a greater utilisation of IF care than expected of 4 - 9 4 this is the service with the most uneven utilisation of services. ?‘he concentration curves for all services follow. As a measure 0 20 40 60 80 100 Cumul % of population ranked by income Figure 2: Concentration curve for expenditure on all services by income quintiles for health states ‘Excellent’, ‘Good’ or ‘Fair’; standardised for health state and age (Index of Horizontal Inequity HI,,, = -0.096). ofthe effect of standardisation, the unstandardised HIU,pfor titilisation of all services is -0.178. The concentration index value obtained for total service use, -0.096, is unlikely to have occurred by chance (Figure 2). Income quintiles I and 3 apparently overutilise services by approximately 3% and the highest income quintile consumes only 13% of services rather than its expected share of 20Y0. However, by far the largest contributor to this inequality is the larger use by the poor of IP services (Figure 5 ) . Utilisation of GP care does not vary by income (Figure 3). As 79% of respondents had consulted a GP at least once over the previous year, across all health states, the results for utilisation of primary care are probably the most reliable. In Figure 4, the standardised expenditure concentration curve crosses the diagonal: the poor have fewer, and the affluent more, OP visits than expected. Most benefit accrues to income quintile 3 (middle New Zealand). The position and slope of the curve at the point of intersection indicates that utilisation by this group is the most significant contributor to the apparent pro-rich inequality. Figure 5 indicates that IP care in a public hospital shows the Table 2: Distribution of the expenditure on all services by Income Quintile for health states ‘Excellent’, ‘Good’, and ‘Fair’.The results have been standardised for health state and age. Income quintile Cumulative Yo of total sample __ 20 Cumulative Yo total expenditure 23 0 __ Cumulative Yo GP visits (GP) __ 20 5 40.1 Cumulative Yo Out-patient visits (OP) Cumulative YOIn-patient nights (IP) - - Lowest ~~ __.._ ~ ~~~ ___ ~_ -~ .. 43 1 - _ 38 o __-____ 63 9 _ _ ~ _ _ ~ Highest “’W”, ao ~~ __ - _ ~ _ _ ~_ .. _. 100 ~- __ 60.0 _ _ ~ 80.4 -______ 100 - -- - _______ . . . - -0 163= ~. -~ -0 096a -0.005 -0 005 ______________~ Note (a) p<0 001 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 1999 VOL. 23 NO. 2 The horizontal equity of health care in New Zealand 0 20 40 60 80 100 Cumul % of population ranked by Income Figure 3: Concentration curve for expenditure on GP services by income quintiles for health states ‘Excellent’, ‘Good’ or ‘Fair’; standardised for health state and age = -0.005). (Index of Horizontal Inequity H,, I cumul% of populationranked by Income Figure 4: Concentration curve for expenditure on OP services by income quintiles for health states ’Excellent’, ‘Good’ or ‘Fair’; standardised for health state and age. = (Index of Horizontal Inequity HIwvp -0.005). greatest degree of inequality of any service with a concentration index of -0.162.This distribution makes the largest contribution to the total inequality observed; if this service is excluded, then the concentration index for total utilisation is very close to zero. Sensitivity analyses, using different estimates of service costs, did not materially affect the results. Discussion The analyses suggest that the poor utilise public health care services somewhat more than expected by their health need. However, before discussing this result further, a number of significant methodological difficulties and potential biases need to be addressed. The biggest problem with this study was inadequate sample .H 8o $ f3 $ ag a(: z 2 cumul% of population ranked by income Figure 5: Concentration curve for expenditure on IP services by income quintiles for health states ‘Excellent’, ‘Good’ or ‘Fair’; standardised for health state and age I = -0.162). (Index of Horizontal Inequity H,, 1999 VOL. 23 NO. 2 size. In particular, standardisation for gender was not possible and only three health states could be used. Even with these restrictions, the mean expenditures for individuals in ‘fair’health sometimes had a potential sampling error of more than 50%. The resolution of the study suffers accordingly. Income calculation also presented difficulties. For example, respondents who share households with other income earners but without pooling incomes may have been misclassified into a higher income level based on total household income. However, such a misclassification would tend to bias results towards the null. Further, no adjustment could be made for family composition when determining household income. However, given that poorer households have fewer income earners and proportionately more dependants than affluent h o ~ s e h o l d s they are likely to be even poorer ,~~ than indicated by their gross household income. Therefore, misclassification by income is unlikely to have been substantial. A further potential inaccuracy was introduced by the way the annual number of GP visits was measured in the HHS: the maximum number of visits per respondent had to be taken as 12. This restriction may under-state the number of visits of those in ‘not so good’ or ‘poor’ health; however, such a limitation would tend to under-estimate any pro-poor inequality that may exist. Interpretation of the results obtained is not simple. For example, the contribution of in-patient care to the apparent pro-poor inequality observed. It is possible that the poor spend longer in hospital for reasons that are not necessarily directly related to their health, for example, inadequate home support. Further, although public hospital use is unequal, it is not necessarily inequitable: the more affluent may be choosing to receive treatment in the private sector, hence their apparent underutilisation of the public system may not represent inequity. However, the small number who used private hospital services, less than 2% in this study, seems unlikely to explain the distribution obtained. Finally, the structure of the health service itself may determine AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Peacock, Devlin and McGee if pro-poor inequality is inequitable or not. If the function of the New Zealand public health care system is simply to provide a decent minimum of care (a safety net) then ‘overuse’by the poor would be expected.

Journal

Australian and New Zealand Journal of Public Health – Wiley

Published: Apr 1, 1999

There are no references for this article.