Relative health performance in BRICS over the past 20 years: the winners and losers
Dennis Petrie a & Kam Ki Tang b
a. Melbourne School of Population and Global Health, University of Melbourne, Melbourne, Australia.
b. School of Economics, University of Queensland, St Lucia, Brisbane, QLD 4072, Australia.
Correspondence to Kam Ki Tang (email: email@example.com).
(Submitted: 30 October 2013 – Revised version received: 23 March 2014 – Accepted: 24 March 2014.)
Bulletin of the World Health Organization 2014;92:396-404. doi: http://dx.doi.org/10.2471/BLT.13.132480
In 2001, the acronym BRIC was coined to represent the fast-growing economies of Brazil, the Russian Federation, India, and China. In 2010 the acronym evolved into BRICS with the addition of South Africa. Since 2001 the five BRICS countries have all expanded in economic terms. China already has an economy that is 80% of the American economy.
As their economic power increases, BRICS are becoming increasingly influential in global health – via their domestic health and foreign assistance programmes and their activism in international fora.1,2 Together, these five countries account for over 40% of the world population and much of the global burden of disease.3 In India, which accounts for nearly 18% of the world’s population, 217 million individuals were estimated to be undernourished in 2012.4 In 2012 South Africa had more people infected with the human immunodeficiency virus (HIV) – 6.1 million – than any other country.5
In the light of the rising prominence of BRICS in the global health arena, we wished to evaluate the health progress that each of the five countries had made between 1990 and 2011. In this paper, we focus on the changes seen in mortality among males and females of different age groups.
We use years of life lost (YLL) – an estimate of the number of years a person has lost because of premature death – as a measure of the ill health of a population.6,7 Some premature deaths – and associated risk factors – are avoidable; others are not. The number of premature deaths and types of avoidable risk factors may vary over time. For example, the development of new technology may increase the percentage of deaths that are avoidable. If that technology becomes accessible to much of the world’s population, it may trigger a global reduction in avoidable mortality. In evaluating the changes in the health performance of a country over time, we may decide to compare the temporal trends in that country with the corresponding global trends or, at least, the corresponding trends seen in other comparable populations.8,9 In this way we can evaluate the country’s health performance relative to that seen in similar areas and so help to distinguish country-specific changes from more general global changes. As an added complication, the choice of “similar areas” may also have to change over time, especially when the countries being investigated are changing rapidly. For example, in 1990 we may have compared China and India with other countries that the Word Bank classified as low-income. In 2011, however, such comparisons would have been unjustified because China and India had then become upper-middle-income and lower-middle-income countries, respectively.
Years of life lost
YLL is generally calculated using the equation:
where i, a, and s are identifiers of the country, age group and sex involved, respectively,
is the number of deaths among a fixed number of people and
Avoidable years of life lost
In the calculation of YLL it is assumed that all deaths measured by
are premature and thus avoidable. However, national survival curves – even those for countries with exceptionally high life expectancies at birth – are far from rectangular, as some deaths at young ages are unavoidable. To account for this, we modified the equation for calculating YLL – subtracting the number of unavoidable deaths at each age from the total number of deaths
to give an estimate of avoidable years of life lost (AYLL):
is the estimated number of unavoidable deaths – based on the lowest observed mortality rate for the given age–sex group among all countries in the sample, and
is constructed from
– for the different age–sex groups – using standard life table methods. Since
takes account of unavoidable deaths caused by genetic factors, the estimates of AYLL are comparable across different age–sex groups and allow age and sex inequalities in health to be examined. The distinction between avoidable and unavoidable mortality is important in this context because health inequality represents the differences in health – or health losses – which are not only unnecessary and avoidable but, in addition, are considered unfair and unjust.10
we constructed a hypothetical reference country – with the lowest observed mortality in each age–sex group – and then used its mortality profile as a benchmark. We could then compare each study country’s mortality outcomes against this benchmark. This was consistent with using
as a measure of unavoidable deaths.
Using the mortality rates of particular population groups as proxies for unavoidable premature deaths is a long-established practice.11–16 It avoids the burden of having to decide which causes of death are preventable,9,17–20 and the need for an accurate determination of the cause of death. Identification of the cause of death can be particularly problematic in developing areas.21
Income-specific avoidable years of life lost
The reference mortality rates -
– used in the calculation of AYLL are often drawn from high-income countries such as Japan. However, the health services that are generally accessible in high-income countries are often not accessible in low- or even middle-income countries. The use of high-income countries as the reference group prohibits us from separating the loss of life years that results from lack of resources from the loss caused by other reasons, such as the inefficient use or unequal distribution of resources. To address this issue, we estimated income-specific avoidable years of life lost (IAYLL) as:
is the estimated level of unavoidable mortality among countries with the same income level as country i, and
is the corresponding reference life expectancy. By definition, IAYLL cannot be larger than AYLL.
Since no two countries have identical income, we used the data envelopment analysis method to estimate
This benchmarking technique allowed us to identify the lowest possible mortality rate for a given income level – based on the observed mortality rates of countries with similar income levels.22
We can justify the use of income as the sole benchmarking criterion in three ways. First, at any point in time – and thus at a given state of technology – there is a well-fitting logistic relationship between life expectancy at birth and income across countries.23 This relationship, which is known as the “Preston curve”, is extremely resilient – even with the current global trend of decreasing mortality. The existence of such a stable relationship means that we can compare a country’s health performance relative to its income over time. Second, income is strongly correlated with many non-income factors that impact on health, such as access to education, food, shelter and medicine. Income is, arguably, the best single proxy for these factors. Of course, economic growth can also bring about health problems, such as increasing air pollution. In northern China, for example, such pollution may have reduced mean life expectancy at birth by 5.5 years.24 Third, we did not wish to model the determinants of mortality. When controlling for income, we essentially grouped countries by income level. If we had grouped countries based on multiple factors, we would have produced more groups but have had fewer countries in each group. There had to be a trade-off between comparisons that were limited to countries that were very similar and comparisons that each included an “adequate” number of countries. We believed that – given the unique and important role of income in health – grouping countries by income alone struck a reasonable balance.
Besides resources, countries also differ in other factors that could contribute to mortality differences, such as geography and climate. The absence of malaria from most temperate areas is an obvious example. However, there is evidence that income can moderate the effect of climate on life expectancy and that – after controlling for socioeconomic factors – environmental factors have no discernible effect on mortality.15,25 Furthermore, a country’s geography and climate are unlikely to have changed dramatically within the 20-year time frame of our analysis.
Decomposing the change in relative health performance
For any age–sex group in a country, we can decompose the change in its IAYLL between period r and the later period t into two components. The equation we used –with the i, a, s indexes omitted for clarity – is:
To the right of the equal sign, the first term represents the country’s own health progress in terms of a lower death rate – weighted by the mean reference life expectancy over periods r and t. The second term represents the effect of technological improvement and economic growth on peer standards. This effect is measured in terms of higher remaining life expectancy – weighted by the mean avoidable death rate of the country over the two periods– and lower preventable death rates – weighted by the mean reference life expectancy over the two periods.
As N* and LE* are income-specific, rising standards could result from increases in the reference standards or – if the study country has grown faster than its original reference countries – the move of the study country to a higher income group with higher reference standards.
The mortality data that we used came from life tables published by the World Health Organization.26 The income data – purchasing power parities expressed as international dollars for the year 2005 – were drawn from World Development Indicators.27 Countries with populations of less than a million were excluded because they may show particularly large variability in their observed mortality – which could bias the estimation of the reference mortality rates. Data envelopment analysis was applied to the remaining 129 countries with available data, to construct the reference mortality profile for each BRICS country in 1990 and 2011.
Figures 1–5 show the age–sex-specific AYLL and IAYLL for each of the BRICS countries for 1990 and 2011. As we normalized the data used, by population, these figures indicate the years of life being lost by an average person. In Fig. 1, for example, we can see that an average Brazilian female born in 1990 was expected to lose an additional 3.26 years of life expectancy – compared with her global peers – due to excess mortality in this Brazilian birth cohort between birth and an age of 1 year. However, when compared with her peers in other countries with similar income, the same female was expected to lose 2.68 years due to such excess mortality. In other words, 0.58 of a year – or 17.8% of her life lost – could be associated with her country’s lack of resources. In 2011, this average female’s AYLL and IAYLL had dropped to 0.93 and 0.65 years, respectively, indicating substantial post-1990 improvement in one age–sex group’s relative health performance. In fact, almost all age–sex groups in Brazil experienced health improvement between 1990 and 2011.
Fig. 1. Avoidable years of life lost, Brazil, 1990 and 2011
In contrast, the Russian Federation experienced substantial declines in relative health performance – in all individuals older than 5 years – between 1990 and 2011 (Fig. 2). Men of working age appeared to have particularly poor health during this period. Within BRICS, only the Russian Federation saw much worse health among males than among females. Gender inequality in health is a particularly important issue in the Russian Federation – largely as a result of the rapid rises in unemployment, alcohol problems and suicide observed among adult males after the collapse of the former Soviet Union in the early 1990s.28–31
Fig. 2. Avoidable years of life lost, Russian Federation, 1990 and 2011
Within BRICS, over the study period, India’s children experienced the worst relative health performance – although they did also show substantial improvement over time (Fig. 3). Adult females in India – except those who were more than 70 years of age – also experienced steady improvements in relative health performance over time. The corresponding trends in the relative health performance of adult males were more mixed.
Fig. 3. Avoidable years of life lost, India, 1990 and 2011
For the majority of South Africa’s population, relative health performance worsened between 1990 and 2011 (Fig. 4). Only children younger than 5 years and some of the subgroups older than 70 years experienced improved relative health performance over this period and, even for these subgroups, the excess loss of life remained high in 2011. Adults aged between 20 and 60 years experienced particularly sharp declines in relative health performance after 1990. In South Africa in 1990, gender inequality in health was uniformly in favour of females, regardless of their age. In 2011, however, females between 20 and 35 years of age appeared to be experiencing much worse health performance relative to their male counterparts.
Fig. 4. Avoidable years of life lost, South Africa, 1990 and 2011
China’s health performance in and since 1990 appears to be particularly impressive (Fig. 5). In 1990, many of China’s age–sex groups had IAYLL values of zero – indicating that they were experiencing a health performance that was at least as good as that being experienced by their income peers elsewhere in the world. In 2011, all of the Chinese age–sex groups except one – girls younger than 1 year – appeared to be experiencing slightly poorer health performance than their income peers. However, this observation did not reflect an increase in mortality but a change in the income group used as a reference. In 1990, China was classified as a low-income country. By 2011 China had grown so much that it had become an upper-middle-income country. In our analysis of the data for 2011, China was therefore benchmarked not against low-income countries – as used for the 1990 data – but against other upper-middle-income countries. Upper-middle-income countries generally have better health performance than low-income countries.
Fig. 5. Avoidable years of life lost, China, 1990 and 2011
Fig. 6 shows how the changes in China’s IAYLL between 1990 and 2011 can be partially attributed to improvement in China’s mortality rates and partially to changes in the income group used as a benchmark. There were reductions in mortality in all age–sex groups. However, as China has moved up the “income ladder”, more of its deaths were considered avoidable and the mean loss of life years resulting from each avoidable death also increased.
Fig. 6. Decomposition of changes in income-specific avoidable years of life lost, China, 1990 and 2011
The concept of avoidable – or amendable – mortality has been used extensively as one indicator of the performance of a health system.8,9 However, there are often limitations to the application of this concept, such as difficulty in defining avoidable mortality – across areas with different resources and living habits and over time, as technology and disease prevalences change – and the frequent omission of data on quality of life.9,20 We attempted to avoid some of these issues by benchmarking against countries with similar mean incomes and allowing the benchmark to change over time. However, caution needs to be exercised when interpreting the results for any single year because income – although a useful proxy – does not capture all determinants of unavoidable mortality.
The method that we used to estimate relative health performance permits the identification of those subpopulations where large gains may be possible with the resources currently available. The social and political environment can clearly play a role in excess mortality, as reflected in the case of the “missing women” in India and China, the high prevalence of HIV infection in South Africa and the high prevalence of alcohol abuse among middle-aged men and women in the Russian Federation.32–35
The BRICS countries are far from homogenous. The Russian Federation has the highest mean income but among the worst relative health performance and – in terms of such performance – has not made any substantial gains over the past 20 years. India has the lowest mean income but managed to outshine South Africa in terms of relative health performance and health inequality. Brazil has a similar mean income to South Africa but exhibited much larger improvements in relative health performance across almost all age–sex groups. South Africa’s large level of avoidable mortality has been attributed to the country’s HIV epidemic, continuing rural poverty and inefficient implementation of high-impact interventions.36,37 Although China’s mean income is still less than half that of the Russian Federation’s, China has outperformed the Russian Federation in terms of health performance and health inequality by a large margin.
It remains unclear how much of China’s success can be attributed to its health-care system and how much to increases in its mean income.38 Since China changed to a market economy, out-of-pocket expenditure on health has increased substantially and this has contributed to widening health inequalities.39 Once account has been taken of mean income, our results also show that China’s elderly people are relatively disadvantaged compared with their younger counterparts – and this disparity has increased in magnitude since 1990. This observation is a concern, especially as China has a rapidly ageing population and increasing income inequality.40 Since 1990, various policy initiatives have been made in the country to improve health finance and access – such as Basic Medical Insurance, the New Rural Cooperative Medical Scheme and the Urban Resident Basic Medical Insurance Scheme – but evidence about the success of these initiatives are mixed.41,42 The 2003 epidemic of severe acute respiratory syndrome exposed other weaknesses in China’s health infrastructure and generated pressure for further reforms.43
In terms of the relative health performance within BRICS, Brazil is the second best performer after China. Brazil has undertaken various social and economic reforms since its democratic transition in the mid-1980s. These include the Unified Health System, which aims to provide universal health coverage and equitable access to health-care services free of charge, the Family Health Programme, which focuses on underserved communities, and the Bolsa Familia programme, which uses conditional cash payments to improve children’s school enrolment and nutrition.44 Although these initiatives have not been without their setbacks and shortfalls, they have contributed to substantial improvements in Brazil’s health performance and health equity.45–47
The application of IAYLL is not limited to emerging economies. It could be useful for monitoring the relative performance of any country’s health system at different stages of that country’s development. Comparison of IAYLL and AYLL can reveal to what extent a country’s poor health performance is the result of a lack of resources – as indicated by mean income. Donors of health aid would, presumably, like to know if other factors are as important as resources in limiting health performance – as appears to be the case in India. Improvements in national income and global health technology could mask a lack of progress or even regression in the efficiency of a country’s health-care system. A dynamic benchmarking approach such as the estimation of IAYLL could be used to focus on this efficiency issue and complement the use of other overall health indicators, such as life expectancy at birth.
- Shifting paradigm, how the BRICS are reshaping global health and development. New York: Global Health Strategies Initiatives; 2012.
- Key players in global health: how Brazil, Russia, India, China, and South Africa are influencing the game. Washington: Center for Strategic and International Studies; 2012.
- Global burden of disease. Geneva: World Health Organization; 2008.
- The state of food insecurity in the world: economic growth is necessary but not sufficient to accelerate reduction of hunger and malnutrition. Rome: Food and Agriculture Organization of the United Nations; 2012.
- Global Health Observatory Data Repository: number of people (all ages) living with HIV [Internet]. Geneva: World Health Organization; 2013. Available from: http://apps.who.int/gho/data/?theme=main&vid=22100 [cited 2014 Mar 27].
- O’Shea E. Social gradients in years of potential life lost in Ireland. Eur J Public Health. 2003;13:327–33. http://dx.doi.org/10.1093/eurpub/13.4.327 pmid: 14703319
- The global burden of disease: generating evidence, guiding policy. Seattle: Institute for Health Metrics and Evaluation; 2013.
- Gay J, Paris V, Devaux M, de Looper M. Mortality amendable to health care in 31 OECD countries: estimates and methodology issues. OECD Health Working Paper No.55. Paris: Organisation for European Economic Co-operation; 2011.
- Mackenbach JP, Hoffmann R, Khoshaba B, Plug I, Rey G, Westerling R, et al. Using ‘amenable mortality’ as indicator of healthcare effectiveness in international comparisons: results of a validation study. J Epidemiol Community Health. 2013;67:139–46. http://dx.doi.org/10.1136/jech-2012-201471 pmid: 23012400
- Whitehead M. The concepts and principles of equity and health. Int J Health Serv. 1992;22:429–45. http://dx.doi.org/10.2190/986L-LHQ6-2VTE-YRRN pmid: 1644507
- Farr W. Methods for comparing local with standard death-rates. London: Office of the Sanitary Institute; 1885.
- Woolsey TD. Toward an index of preventable mortality, no 85. Hyattsville: Department of Health and Human Services; 1981.
- Uemura K. Excess mortality ratio with reference to the lowest age-sex-specific death rates among countries. World Health Stat Q. 1989;42:26–41. pmid: 2711702
- McCracken K. What if New South Wales were more equal? NSW Public Health Bull. 2002;13:123–7. http://dx.doi.org/10.1071/NB02053 pmid: 12209197
- Tang KK, Chin JTC, Rao DSP. Avoidable mortality risks and measurement of wellbeing and inequality. J Health Econ. 2008;27:624–41. http://dx.doi.org/10.1016/j.jhealeco.2007.08.003 pmid: 18191488
- Tang KK, Petrie D, Rao DSP. Measuring health inequality with realization of potential life years (RePLY). Health Econ. 2009;18 Suppl 1:S55–75. http://dx.doi.org/10.1002/hec.1461 pmid: 19294636
- Rutstein DD, Berenberg W, Chalmers TC, Child CG 3rd, Fishman AP, Perrin EB, et al. Measuring the quality of medical care. A clinical method. N Engl J Med. 1976;294:582–8. http://dx.doi.org/10.1056/NEJM197603112941104 pmid: 942758
- Nolte E, McKee CM. Does health care save lives? Avoidable mortality revisited. London: The Nuffield Trust; 2004.
- Kossarova L, Holland W, Nolte E, McKee M. Measuring “avoidable” mortality: methodological note. Brussels: European Commission; 2009.
- Castelli A, Nizalova O. Avoidable mortality: what it means and how it is measured. CHE Research Paper No.63. York: University of York; 2011.
- Mathers CD, Fat DM, Inoue M, Rao C, Lopez AD. Counting the dead and what they died from: an assessment of the global status of cause of death data. Bull World Health Organ. 2005;83:171–7. pmid: 15798840
- Coelli TJ, Rao DSP, O’Donnell CJ, Battese GE. An introduction to efficiency and productivity analysis. 2nd ed. New York: Springer; 2005.
- Preston SH. The changing relation between mortality and level of economic development. Popul Stud (Camb). 1975;29:231–48. pmid: 11630494
- Chen Y, Ebenstein A, Greenstone M, Li H. Evidence on the impact of sustained exposure to air pollution on life expectancy from China’s Huai River policy. Proc Natl Acad Sci USA. 2013;110:12936–41. http://dx.doi.org/10.1073/pnas.1300018110 pmid: 23836630
- Tang KK, Petrie D, Rao DSP. The income-climate trap of health development: a comparative analysis of African and non-African countries. Soc Sci Med. 2009;69:1099–106. http://dx.doi.org/10.1016/j.socscimed.2009.07.016 pmid: 19666204
- Global Health Observatory Data Repository: life tables [Internet]. Geneva: World Health Organization; 2013. Available from: http://apps.who.int/gho/data/node.main.692 [cited 2013 Sep 6].
- World Development Indicators [Internet]. Washington: World Bank; 2013. Available from: http://data.worldbank.org/data-catalog/world-development-indicators [cited 2013 Sep 2].
- Brainerd E, Cutler DM. Autopsy on an empire: understanding mortality in Russia and the former Soviet Union. J Econ Perspect. 2005;19:107–30. http://dx.doi.org/10.1257/0895330053147921
- Bobadilla JL, Costello CA, Mitchell F, editors. Premature death in the New Independent States. Washington: National Academy Press; 1997.
- Minagawa Y. The social consequences of postcommunist structural change: an analysis of suicide trends in Eastern Europe. Soc Forces. 2013;91(3):1035–56. http://dx.doi.org/10.1093/sf/sos172
- Brainerd E. Economic reform and mortality in the former Soviet Union: a study of the suicide epidemic in the 1990s. Eur Econ Rev. 2001;45:1007–19. http://dx.doi.org/10.1016/S0014-2921(01)00108-8
- Sen A. Missing women. BMJ. 1992;304:587–8. http://dx.doi.org/10.1136/bmj.304.6827.587 pmid: 1559085
- Sen A. Missing women – revisited. BMJ. 2003;327:1297–8. http://dx.doi.org/10.1136/bmj.327.7427.1297 pmid: 14656808
- Klasen S, Wink C. Missing women: revisiting the debate. Fem Econ. 2003;9:263–300. http://dx.doi.org/10.1080/1354570022000077999
- Zaridze D, Lewington S, Boroda A, Scélo G, Karpov R, Lazarev A, et al. Alcohol and mortality in Russia: prospective observational study of 151 000 adults. Lancet. Epub 2014 Jan 31. http://dx.doi.org/10.1016/S0140-6736(13)62247-3 pmid: 24486187
- Chopra M, Daviaud E, Pattinson R, Fonn S, Lawn JE. Saving the lives of South Africa’s mothers, babies, and children: can the health system deliver? Lancet. 2009;374:835–46. http://dx.doi.org/10.1016/S0140-6736(09)61123-5 pmid: 19709729
- Mayosi BM, Lawn JE, van Niekerk A, Bradshaw D, Abdool Karim SS, Coovadia HM; Lancet South Africa team. Health in South Africa: changes and challenges since 2009. Lancet. 2012;380:2029–43. http://dx.doi.org/10.1016/S0140-6736(12)61814-5 pmid: 23201214
- Hsiao WC. The Chinese health care system: lessons for other nations. Soc Sci Med. 1995;41:1047–55. http://dx.doi.org/10.1016/0277-9536(94)00421-O pmid: 8578327
- Hu S, Tang S, Liu Y, Zhao Y, Escobar ML, de Ferranti D. Reform of how health care is paid for in China: challenges and opportunities. Lancet. 2008;372:1846–53. http://dx.doi.org/10.1016/S0140-6736(08)61368-9 pmid: 18930520
- Baeten S, Van Ourti T, van Doorslaer E. Rising inequalities in income and health in China: who is left behind? J Health Econ. 2013;32:1214–29. http://dx.doi.org/10.1016/j.jhealeco.2013.10.002 pmid: 24189450
- Wagstaff A, Yip W, Lindelow M, Hsiao WC. China’s health system and its reform: a review of recent studies. Health Econ. 2009;18 Suppl 2:S7–23. http://dx.doi.org/10.1002/hec.1518 pmid: 19551753
- Chen D, Tang KK, Zhao L, Zhang Y. Will China’s cooperative medical system fail again? Insight from farmer satisfaction survey. Health Promot Int. Epub 2013 May 12. http://dx.doi.org/10.1093/heapro/dat032 pmid: 23669160
- Dong Z, Phillips MR. Evolution of China’s health-care system. Lancet. 2008;372:1715–6. http://dx.doi.org/10.1016/S0140-6736(08)61351-3 pmid: 18930524
- Paim J, Travassos C, Almeida C, Bahia L, Macinko J. The Brazilian health system: history, advances, and challenges. Lancet. 2011;377:1778–97. http://dx.doi.org/10.1016/S0140-6736(11)60054-8 pmid: 21561655
- Guanais FC. Health equity in Brazil. BMJ. 2010;341:c6542. http://dx.doi.org/10.1136/bmj.c6542 pmid: 21115590
- Glewwe P, Kassouf A. The impact of the Bolsa Escola/Familia conditional cash transfer program on enrolment, dropout rates and grade promotion in Brazil. J Dev Econ. 2012;97:505–17. http://dx.doi.org/10.1016/j.jdeveco.2011.05.008
- Paes-Sousa R, Santos LM, Miazaki ÉS. Effects of a conditional cash transfer programme on child nutrition in Brazil. Bull World Health Organ. 2011;89:496–503. http://dx.doi.org/10.2471/BLT.10.084202 pmid: 21734763