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

Learn More →

More or less equal? Comparing Australian income–related inequality in self–reported health with other industrialised countries

More or less equal? Comparing Australian income–related inequality in self–reported health with... Abstract Objective: To measure the distribution o f self-reported health by income in order to compare the level o health inequality in f Australia wth other industrialized countries. Method: Using data from the two Len Smith National Centre for Epidemiology and Population Health, Australian National lJniversi& Australian Capital Territory National Health Surveys undertaken in 1989-90 and 1995, concentration indexes were calculated to quantify the distribution of self-reported heatth by equivalent income. The concentration index for Australia was compared with those reported for nine industrialized countris in Europe and North America. Results: The estimated incomerelated concentration indexes were 4 . 1172 in 1989-90and 4 1 0 9 4 in 1995. Conclusion: The level o heakh ihequslrity f is not significantly different from the US of the UK, but significantly greater than seven other European nations. lmplkations: Australia has significant income related health inequalities and the distribution o health appears to be m e f unequal than in many other industrialized nations. There is a need to further investigate and quantify those features of the anglophone societies that set them apart from some other industrialized nations. n recent years, several studies have compared health inequality across industrialized countries using selfreported health (SRH) as a measure of health status.lo4 One of the most comprehensive was undertaken by van Doorslaer et. a!. as part of the ECuity Project in which a concentration index was used to measure the distribution of SRH by income in East Germany, the United States and seven Western European countries, based on data from national health surveys conducted between 1987 and 1992.3 The purpose of this study is to use Australian data from the two most recent National Health Surveys conducted in 1989-90 and 1995 to calculate concentration indexes that would enable comparison of incomerelated health inequality in Australia with these nine other industrialized countries. can take values between +1 and -1. A negative concentration index signifies socioeconomic related health inequality; its minimum value is minus one when all morbidity is concentrated in those individuals in the lowest socioeconomic poUp? A typical SRH question asks respondents to rate their overall health as either: ( i ) Excellent; (ii) Very Good; (iii) Good; (iv) Fair; or (v) Poor. Because these responses provide only an ordinal ranking of health status, they are difficult to analyze. One approach is to combine response categories above and below an arbitrary cut-off point in order to transform SRH into a dichotomous variable.*t4For example, the proportion of respondents reporting their health as ‘good’or better could be compared with those reporting less than ‘good’. A limitation of this approach is that the chosen cut-off point affects the magnitude of the measure of inequality6 To overcome this problem van Doorslaer et. al. in their study3 employ a different approach of deriving a continuous latent health variable capable of generating the categorical responses to the SRH question. Based on the assumption that the distribution of health is skewed, they assume this latent variable has a standard lognormal distribution. Scores for each health category can be derived by calculating threshold values that ensure the standard normal distribution is divided up in proportion to the number of individuals in each category. The mean value of the latent variable for Method Background A concentration index is a commonly used measure of health inequality that is calculated from a concentration curve, which plots the cumulative proportion of a measure of morbidity against the cumulative proportion of the population ranked from the lowest to highest in terms of socioeconomic status. The concentration index is defined as twice the area between the concentration curve and the diagonal (which represents an equal distribution of health across the population), and the index t3ubrnttt.d: November 1Q9g Revidon roques- April 2000 Accqpted: June 2000 Correspondence to: Philip Clarke, Health Economics Research Centre, Institute of Health Science, Old Road University of Oxford, Oxford OX3 7LF, United Kingdom Fax: +44-1865-226842. Email: philip.clarke@dphpc.ox.ac.uk AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2000 VOL. 24 NO. 4 Comparing Australian income-related inequality in self-reported health with other industrialised countries each category can then be estimated as the exponent of each normal score multiplied by negative one (a worked example of this procedure is provided in a recent paper by Wagstaff and van Doorslaer6). All individuals are then assigned a latent health score that matches their response category and the mean health score is calculated for each income group. A concentration index can be calculated from individual level data or alternatively from grouped socioeconomic data such as income deciles. The latter approach facilitates the use of a convenient regression-based method to estimate the concentration index in a similar manner to the closely related regression-based relative index of ineqdq.’ However, if the concentration index is estimated using this approach,the standard error of the regression cannot be used to test whether there are significant differences in health inequality across countries, because the ranking of the data introduces serial correlation into the error term which violates one of the assumptions of least squares regression. Distribution free asymptotic estimators for the standard errors must used in hypothesis testing.’ To remove any confounding effects of demographic variables on the concentration index, van Doorslaer et aL3 use direct age-sex standardization. This method involves applying the age-sex specific illness rates of each income group to the agesex distribution of the whole country and produces a distribution of illness by income group that would apply if all income groups had the same age-sex composition. these surveys contain information on each individual’s equivalent income decile and SRH status. Equivalent income is derived by dividing the g r m income of each income unit (a person or a group of related persons who share a common income) by a standard Australian equivalence scale, to produce a measure that reflects each income unit’s relative standard of living.’ Both surveys contain a categorical SRH status question in which all respondents 15 years of age and over are asked how they rate their health in general. While the wording of the questions is similar, the surveys provided respondents with a different number of categories in which to rate their overall health. The NHS 198990 allows for four possible responses (‘excellent’,‘good’, ‘fair’, and ‘poor‘), while the NHS 1995 includes an extra category ‘very good’ in addition to the other four. The latent variable method allows comparable scores to be derived despite these differences in the response categories. In the both surveys the equivalent income deciles are based on the distribution of income across the entire sample, but the analysis reported here included only persons who were 15 years of age or over and could be ranked according to their equivalent income decile. This sub-sample consisted of 36,498 and 35,325 persons in NHS 1989-90 and 1995 respectively. Results Using the methodology outlined in the previous section we calculated the latent health scores for each response category after applying direct age-sex standardization; we then assigned all individuals a latent health score that matched their stated response to the SRH question. The average of these scores for each income group is listed in Table 1. The numbers in each group differ due to lower equivalent income deciles containing a higher proportion of persons under 15 years of age who were excluded from the analysis. Using these data, the concentration index is estimated to be 0.1172 (95% C.I. -0.1299 to -0.1046) from the NHS 1989-90and -0.1094 (95% C.I. -0.1249 to -0.09387) from NHS 1995. Both Data and Method Data for the analysis were taken from Australia’s two m a t recent surveys of population health. While only the first survey conducted in 1989-90 falls within the period of the previous study3the inclusion of the second National Health Survey (NHS) conducted in 1995 allows us to examine whether the degree of income-related health inequality has changed over time. The Australian surveys have a larger sample size than most comparable population health surveys, with fully completed questionnaires obtained from 54,576 persons in NHS 1989-90, and from 53,828 persons in NHS 1W5.S19 The data files from Table 1: Mean of a latent health variable by income group ordered from lowest to highest Equivalent Income Decile 1 No Respondents nal Health Survev l$SS-eQ Mean Latent Mean Latent Health Score Health Score (Standardized) (Unstandardized) National Health Survev 1995 No. Mean Latent Mean Latent Respondents Health Score Health Score (Standardized) (Unstandardized) 2 3 4 5 6 7 8 9 10 -~ Note 1.95718 2640 2758 2.28234 2.13792 3594 3837 1.73134 3766 1.56381 351 1 1.45387 3862 1.41702 3889 1.39353 47 11 1.24260 4470 1.17193 ____.___. ~~~ 1.68967 2.20392 2.19210 1.77935 1.55336 1.48976 1.37880 1.34613 1.23240 1.1 -~~ 4081 - (a) The mean latent health score I S an index of health statuq denvedfrom the responsei to a self-reported health question (hinher values indicate lower overall health) _ _ ~ 2000 VOL 24 NO 4 ~~~ ~~~ AUSTRALIAN AND NEW ZEALAND JOURNALOF PUBLCHEALTH Clarke and Smith . indexes are negative and significantly different from zero, indicating the presence of significant income-related health inequality in Australia. To determine whether the degree of inequality has changed, a Student's t-test was applied to the standard errors. A t statistic of 0.887 was obtained, indicating no difference between the concentration indexes based on NHS 1989-90 and NHS 1995 at the 5% level of significance. The estimates of the concentration index for Australia can be compared with those previously reported for nine other developed countries? Figure 1 shows the estimated concentration indexes for the nine countries and 95% confidence intervals, alongside the estimated concentration index for Australia based on NHS 198990. The index for Australia lies between thae of the United Kingdom (-0.1148) and the United States (-0.1360),with all three anglophone countries having a higher degree of income-related n mequality than East Germany a d the other six Western European counties. This is confirmed by t-tests which show no signihcant difference at the 5% level between the concentration indexes of Australia, the United Kingdom and the United States, but n significant differences between these three countries a d the others. Figure 1: The Concentration index with 95% confidence intervals for Australia and nine other industrialized countries to the distribution of income. This hypothesis has been tested by regressing the concentration index on the Gini coefficients in each country. Based on data from the nine industrialized countries (excluding Australia) the regression equation is reported to be:3 Concentration index = 0.078-0.554 Gini Substituting the 1989 Gini coefficient of 0.336 for Australia'' in this equation yields a predicted concentration index of -0.108 which falls within the 95% CI for our estimated concentration index for 1989. Hence, our results strengthen the case for an association between health inequality and income distribution. It has recently been argued that social cohesion and the level of social capital within societies impacts on population health and health ineq~alities.'~~'~ raises the issue of whether the This anglophone countries are more individualistic, leading to lower accumulation of social capital than in other developed nations. An empirical examination of this issue will require a measure of a nation's social capital. While there has been some progress in the measurement of social ~ a p i t a l ' ~ ' ' ~ techniques have not these been sufficiently developed to permit their use in quantitative analysis of differences in the distribution of health between nations. It would also be useful to explore to what degree cultural factors are responsible for the distribution of SRH, since it has been argued17that several of the response categories used when asking SRH questionsmay not be conceptually equivalent across countries. Given the inclusion of SRH questions in the most recent national health survey in New Zealand it would be possible to estimate comparable concentration indexes for New Zealand to see where it falls in the inequality hierarchy. This would not only help to clarify whether there are differences in the distribution of SRH between Australia and New Zealand, it would also help resolve whether countries with an English heritage tend to have a more uneven distribution of health than other industrialized countries. Note: The concentration indexes for the nine other industdased countries are based on data from the ECuity project' Acknowledgements We would like to thank Steven Kunitz and Eddy van Doorslaer for providing helpful comments on previous drafts of this paper. Any remaining errors are our own. Conclusions and discussion Levels and trends in income-related health inequalities are attracting increasing attention, because of general concerns that health gaps between and within countries are widening in line with overall trends in economic inequalities.lo The measurement of the distribution of health also facilitates the testing of theories that attempt to account for the relationship between income distribution and health." The results presented here confirm the general picture of the hierarchy of income-related health inequalities, with English-speaking countries having higher levels of inequality than other developed nations. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

More or less equal? Comparing Australian income–related inequality in self–reported health with other industrialised countries

Download PDF
4 pages

Loading next page...
 
/lp/wiley/more-or-less-equal-comparing-australian-income-related-inequality-in-oVIWiGdIS2

References (18)

Publisher
Wiley
Copyright
Copyright © 2000 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.2000.tb01595.x
Publisher site
See Article on Publisher Site

Abstract

Abstract Objective: To measure the distribution o f self-reported health by income in order to compare the level o health inequality in f Australia wth other industrialized countries. Method: Using data from the two Len Smith National Centre for Epidemiology and Population Health, Australian National lJniversi& Australian Capital Territory National Health Surveys undertaken in 1989-90 and 1995, concentration indexes were calculated to quantify the distribution of self-reported heatth by equivalent income. The concentration index for Australia was compared with those reported for nine industrialized countris in Europe and North America. Results: The estimated incomerelated concentration indexes were 4 . 1172 in 1989-90and 4 1 0 9 4 in 1995. Conclusion: The level o heakh ihequslrity f is not significantly different from the US of the UK, but significantly greater than seven other European nations. lmplkations: Australia has significant income related health inequalities and the distribution o health appears to be m e f unequal than in many other industrialized nations. There is a need to further investigate and quantify those features of the anglophone societies that set them apart from some other industrialized nations. n recent years, several studies have compared health inequality across industrialized countries using selfreported health (SRH) as a measure of health status.lo4 One of the most comprehensive was undertaken by van Doorslaer et. a!. as part of the ECuity Project in which a concentration index was used to measure the distribution of SRH by income in East Germany, the United States and seven Western European countries, based on data from national health surveys conducted between 1987 and 1992.3 The purpose of this study is to use Australian data from the two most recent National Health Surveys conducted in 1989-90 and 1995 to calculate concentration indexes that would enable comparison of incomerelated health inequality in Australia with these nine other industrialized countries. can take values between +1 and -1. A negative concentration index signifies socioeconomic related health inequality; its minimum value is minus one when all morbidity is concentrated in those individuals in the lowest socioeconomic poUp? A typical SRH question asks respondents to rate their overall health as either: ( i ) Excellent; (ii) Very Good; (iii) Good; (iv) Fair; or (v) Poor. Because these responses provide only an ordinal ranking of health status, they are difficult to analyze. One approach is to combine response categories above and below an arbitrary cut-off point in order to transform SRH into a dichotomous variable.*t4For example, the proportion of respondents reporting their health as ‘good’or better could be compared with those reporting less than ‘good’. A limitation of this approach is that the chosen cut-off point affects the magnitude of the measure of inequality6 To overcome this problem van Doorslaer et. al. in their study3 employ a different approach of deriving a continuous latent health variable capable of generating the categorical responses to the SRH question. Based on the assumption that the distribution of health is skewed, they assume this latent variable has a standard lognormal distribution. Scores for each health category can be derived by calculating threshold values that ensure the standard normal distribution is divided up in proportion to the number of individuals in each category. The mean value of the latent variable for Method Background A concentration index is a commonly used measure of health inequality that is calculated from a concentration curve, which plots the cumulative proportion of a measure of morbidity against the cumulative proportion of the population ranked from the lowest to highest in terms of socioeconomic status. The concentration index is defined as twice the area between the concentration curve and the diagonal (which represents an equal distribution of health across the population), and the index t3ubrnttt.d: November 1Q9g Revidon roques- April 2000 Accqpted: June 2000 Correspondence to: Philip Clarke, Health Economics Research Centre, Institute of Health Science, Old Road University of Oxford, Oxford OX3 7LF, United Kingdom Fax: +44-1865-226842. Email: philip.clarke@dphpc.ox.ac.uk AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2000 VOL. 24 NO. 4 Comparing Australian income-related inequality in self-reported health with other industrialised countries each category can then be estimated as the exponent of each normal score multiplied by negative one (a worked example of this procedure is provided in a recent paper by Wagstaff and van Doorslaer6). All individuals are then assigned a latent health score that matches their response category and the mean health score is calculated for each income group. A concentration index can be calculated from individual level data or alternatively from grouped socioeconomic data such as income deciles. The latter approach facilitates the use of a convenient regression-based method to estimate the concentration index in a similar manner to the closely related regression-based relative index of ineqdq.’ However, if the concentration index is estimated using this approach,the standard error of the regression cannot be used to test whether there are significant differences in health inequality across countries, because the ranking of the data introduces serial correlation into the error term which violates one of the assumptions of least squares regression. Distribution free asymptotic estimators for the standard errors must used in hypothesis testing.’ To remove any confounding effects of demographic variables on the concentration index, van Doorslaer et aL3 use direct age-sex standardization. This method involves applying the age-sex specific illness rates of each income group to the agesex distribution of the whole country and produces a distribution of illness by income group that would apply if all income groups had the same age-sex composition. these surveys contain information on each individual’s equivalent income decile and SRH status. Equivalent income is derived by dividing the g r m income of each income unit (a person or a group of related persons who share a common income) by a standard Australian equivalence scale, to produce a measure that reflects each income unit’s relative standard of living.’ Both surveys contain a categorical SRH status question in which all respondents 15 years of age and over are asked how they rate their health in general. While the wording of the questions is similar, the surveys provided respondents with a different number of categories in which to rate their overall health. The NHS 198990 allows for four possible responses (‘excellent’,‘good’, ‘fair’, and ‘poor‘), while the NHS 1995 includes an extra category ‘very good’ in addition to the other four. The latent variable method allows comparable scores to be derived despite these differences in the response categories. In the both surveys the equivalent income deciles are based on the distribution of income across the entire sample, but the analysis reported here included only persons who were 15 years of age or over and could be ranked according to their equivalent income decile. This sub-sample consisted of 36,498 and 35,325 persons in NHS 1989-90 and 1995 respectively. Results Using the methodology outlined in the previous section we calculated the latent health scores for each response category after applying direct age-sex standardization; we then assigned all individuals a latent health score that matched their stated response to the SRH question. The average of these scores for each income group is listed in Table 1. The numbers in each group differ due to lower equivalent income deciles containing a higher proportion of persons under 15 years of age who were excluded from the analysis. Using these data, the concentration index is estimated to be 0.1172 (95% C.I. -0.1299 to -0.1046) from the NHS 1989-90and -0.1094 (95% C.I. -0.1249 to -0.09387) from NHS 1995. Both Data and Method Data for the analysis were taken from Australia’s two m a t recent surveys of population health. While only the first survey conducted in 1989-90 falls within the period of the previous study3the inclusion of the second National Health Survey (NHS) conducted in 1995 allows us to examine whether the degree of income-related health inequality has changed over time. The Australian surveys have a larger sample size than most comparable population health surveys, with fully completed questionnaires obtained from 54,576 persons in NHS 1989-90, and from 53,828 persons in NHS 1W5.S19 The data files from Table 1: Mean of a latent health variable by income group ordered from lowest to highest Equivalent Income Decile 1 No Respondents nal Health Survev l$SS-eQ Mean Latent Mean Latent Health Score Health Score (Standardized) (Unstandardized) National Health Survev 1995 No. Mean Latent Mean Latent Respondents Health Score Health Score (Standardized) (Unstandardized) 2 3 4 5 6 7 8 9 10 -~ Note 1.95718 2640 2758 2.28234 2.13792 3594 3837 1.73134 3766 1.56381 351 1 1.45387 3862 1.41702 3889 1.39353 47 11 1.24260 4470 1.17193 ____.___. ~~~ 1.68967 2.20392 2.19210 1.77935 1.55336 1.48976 1.37880 1.34613 1.23240 1.1 -~~ 4081 - (a) The mean latent health score I S an index of health statuq denvedfrom the responsei to a self-reported health question (hinher values indicate lower overall health) _ _ ~ 2000 VOL 24 NO 4 ~~~ ~~~ AUSTRALIAN AND NEW ZEALAND JOURNALOF PUBLCHEALTH Clarke and Smith . indexes are negative and significantly different from zero, indicating the presence of significant income-related health inequality in Australia. To determine whether the degree of inequality has changed, a Student's t-test was applied to the standard errors. A t statistic of 0.887 was obtained, indicating no difference between the concentration indexes based on NHS 1989-90 and NHS 1995 at the 5% level of significance. The estimates of the concentration index for Australia can be compared with those previously reported for nine other developed countries? Figure 1 shows the estimated concentration indexes for the nine countries and 95% confidence intervals, alongside the estimated concentration index for Australia based on NHS 198990. The index for Australia lies between thae of the United Kingdom (-0.1148) and the United States (-0.1360),with all three anglophone countries having a higher degree of income-related n mequality than East Germany a d the other six Western European counties. This is confirmed by t-tests which show no signihcant difference at the 5% level between the concentration indexes of Australia, the United Kingdom and the United States, but n significant differences between these three countries a d the others. Figure 1: The Concentration index with 95% confidence intervals for Australia and nine other industrialized countries to the distribution of income. This hypothesis has been tested by regressing the concentration index on the Gini coefficients in each country. Based on data from the nine industrialized countries (excluding Australia) the regression equation is reported to be:3 Concentration index = 0.078-0.554 Gini Substituting the 1989 Gini coefficient of 0.336 for Australia'' in this equation yields a predicted concentration index of -0.108 which falls within the 95% CI for our estimated concentration index for 1989. Hence, our results strengthen the case for an association between health inequality and income distribution. It has recently been argued that social cohesion and the level of social capital within societies impacts on population health and health ineq~alities.'~~'~ raises the issue of whether the This anglophone countries are more individualistic, leading to lower accumulation of social capital than in other developed nations. An empirical examination of this issue will require a measure of a nation's social capital. While there has been some progress in the measurement of social ~ a p i t a l ' ~ ' ' ~ techniques have not these been sufficiently developed to permit their use in quantitative analysis of differences in the distribution of health between nations. It would also be useful to explore to what degree cultural factors are responsible for the distribution of SRH, since it has been argued17that several of the response categories used when asking SRH questionsmay not be conceptually equivalent across countries. Given the inclusion of SRH questions in the most recent national health survey in New Zealand it would be possible to estimate comparable concentration indexes for New Zealand to see where it falls in the inequality hierarchy. This would not only help to clarify whether there are differences in the distribution of SRH between Australia and New Zealand, it would also help resolve whether countries with an English heritage tend to have a more uneven distribution of health than other industrialized countries. Note: The concentration indexes for the nine other industdased countries are based on data from the ECuity project' Acknowledgements We would like to thank Steven Kunitz and Eddy van Doorslaer for providing helpful comments on previous drafts of this paper. Any remaining errors are our own. Conclusions and discussion Levels and trends in income-related health inequalities are attracting increasing attention, because of general concerns that health gaps between and within countries are widening in line with overall trends in economic inequalities.lo The measurement of the distribution of health also facilitates the testing of theories that attempt to account for the relationship between income distribution and health." The results presented here confirm the general picture of the hierarchy of income-related health inequalities, with English-speaking countries having higher levels of inequality than other developed nations.

Journal

Australian and New Zealand Journal of Public Health – Wiley

Published: Aug 1, 2000

There are no references for this article.