Trends in infant/child mortality and life expectancy in Indigenous populations in Yunnan Province, China
The Centre for International Health & School of Public Health, Curtin University of Technology and Telethon Institute for Child Health Research, Centre for Child Health Research, The University of Western Australia
Correspondence to: Dr Jianghong Li, Telethon Institute for Child Health Research, PO Box 855, West Perth WA 6872, Australia, Tel: +61-8-9489-7800, Fax: +61-8-9489-7700, e-mail: firstname.lastname@example.org.
Objective: The 2000 Census in China registered 55 groups of Indigenous population, including 104.49 million people, making up 8.1% of China's total population. Yunnan Province, located in Southwest China, is the only province where all 55 Indigenous nationalities are represented (14.15 million), making up 33.4% of Yunnan's total population. This study aimed to examine trends in infant and child mortality and life expectancy at birth of the 22 largest Indigenous nationalities and compared these trends with those of the majority Han Chinese in Yunnan and China as a whole.
Methods: Data sources of mortality and socioeconomic status came from the population censuses of China (1953, 1964, 1982, 1990, and 2000) and Yunnan (1990–2000) and from the Provincial Health Department (1990, 1995, 1996 and 2000). Weighted linear regression analysis was used to examine the associations between infant/child mortality and life expectancy at birth, socioeconomic indicators and the use of preventive health services.
Results: In 2000, the infant mortality rate was 26.90 for China and 53.64 for Han Chinese in Yunnan per 1,000 live birth versus 77.75 for the 22 largest minority nationalities in Yunnan, despite improvements in health status indicators since 1990. The inequalities in life expectancy at birth between China as a whole and some minority nationalities remained striking in 2000 (57.18 versus 71.40). Literacy, prenatal examination, hospital deliveries, economic development were important predictors of these health indicators.
Implications: Efforts to continue to improve these intermediate proximate determinants and to target the most disadvantaged Indigenous groups are likely to further reduce health disparities between the Chinese and Indigenous populations.
The 2000 Census in China registered 55 Indigenous populations, including 104.49 million people (8.1% of China's total population). These 55 population groups are officially called minority nationalities and they reside in five Autonomous Regions (at the provincial level), 30 Autonomous Prefectures, and 72 Autonomous Counties (Banners) covering 62.5% of the nation's land area.1 The identification of China's minority nationalities corresponds to the definition of ‘Indigenous’ developed by the International Labour Organisation's Convention No. 169 of 1989 and the World Bank,2 which emphasises the social, cultural, economic conditions, customs, traditions, language and geographic concentrations that distinguish ‘Indigenous’ populations from other sections of the national community.3,4
Yunnan Province is situated in China's south west, bordering with Myanmar in the west and Laos and Vietnam in the south (Map 1). It is the only province where all 55 minority nationalities are represented. The total population of Yunnan's 55 minority nationalities is around 14.15 million, making up 13.4 % of China's total population of minority nationalities and 33.4% of Yunnan's total population (42.36 million).5 Among the 55 minority nationalities in Yunnan, 25 have a population of 5,000 or more, and 15 of these 25 groups are unique to Yunnan.6 Yunnan has the second largest Indigenous population after Guizhou Province. The areas inhabited by the 55 minority nationalities make up 70% of Yunnan's total land area and around 70% of the minority populations reside in the very remote hilly and mountainous regions (1,000-3,500 metres above sea level). Administratively speaking, the majority of the minority nationalities reside in 11 Autonomous Prefectures (Chuxiong, Honghe, Wenshan, Dali, Bangna, Dehong, Nujiang, Diqing, Lijiang, Simao, and Lincang), encompassing 29 Autonomous Counties (Map 2).7
Yunnan is also one of the poorest and most remote provinces in China. Ninety-four per cent of the total land area (394,000km2) in Yunnan is mountains, leaving about 6% suitable for cultivation. More than 75% of Yunnan's population is engaged in agriculture. The provincial average GNP per capita in 2001 was 4,872 yuan (US$609). However, of 42.36 million people, there were still 5.6 million people living with an average annual income below 300 yuan (US$36.1).
The minority nationalities have played an important role in the development of Yunnan.4 However, there is limited research on the health status of the various minority nationalities and much less has been published in international journals. A few studies have documented very high infant mortality in the Miao and Buyi peoples in Guizhou Province and in some of the minority nationalities in Yunnan Province in the 1980s and early 1990s.7–9 Large disparities in infant mortality were also observed between minority nationalities and the Han Chinese (the majority) in Guizhou Province: 73 deaths per 1,000 live births for the Han Chinese, 123 deaths per 1,000 live births for the Miao and 165 deaths per 1,000 live births for the Buyi.8 However, to date there is no research that has examined variations amongst minority nationalities in Yunnan. Neither has there been any research on the changes in the health status of these populations. In response to these knowledge gaps, the objective of this study was to examine recent changes in infant and child mortality and life expectancy at birth among the 22 largest minority nationalities and compare these changes with those shown by the Han Chinese in Yunnan Province. The study also aimed to examine some of the contributing factors to infant/child mortality at both nationality and the prefecture level which would inform future research on the causes of health disparities between the Indigenous and Han Chinese in Yunnan.
The analysis was based on data from the five population censuses of Yunnan.5,10–13 We also used health, economic and demographic data at the prefecture level, which was collected by Yunnan Provincial Maternal and Child Commission Office and Yunnan Provincial Bureau of Statistics.14 These are reliable resources of mortality data in China and it has been shown that the Census and the National Maternal and Child Health Surveillance system provide more complete data on infant death rates than other systems.15 For example, the 2000 Census reported 32 unadjusted infant deaths per 1,000 live births whereas the National Diseases Surveillance System recorded only 13.5 deaths per 1,000 births for China as a whole.15 Since underestimation is more likely than overestimation for infant mortality, particularly in rural and remote areas, we believe that the census report is closer to the reality than the Surveillance System.
Three key health indicators at the nationality level included infant and child mortality and life expectancy at birth. The infant mortality rate was defined as death under age 1 per 1,000 live births. Child mortality was defined as death per 1,000 children aged 1–4. Life expectancy at birth (in years) at the nationality level was calculated using life table methods and mortality data from the provincial censuses.16 Both 1990 and 2000 Censuses use a separate table to record the death of each household member of all ages and sociodemographic information. This allows for estimates of life expectancy based on provincial data. At the prefecture level, we also examined cu ulative child mortality for ages 0–4.
At the nationality level, illiteracy was examined as a predictor of mortality under age 1 and child mortality aged 1–4, with adjustment for population size and total fertility in concurrent and previous census years. The definition of literacy has undergone several changes in China but according to the most recent definition, a literate person is required to be able to use 2000 characters in urban areas and 1500 characters in rural areas. It is up to individuals to choose Chinese or a minority language.17 Illiteracy was defined as having knowledge of 300 or fewer characters and semi-illiteracy was defined as being able to use at least 500 characters. The illiteracy rate was defined as the proportion of the population ≥ 6 years of age that was illiterate or semi-illiterate, differentiated by sex where possible. At the prefecture level, however, only the proportion of the total population that was illiterate or semi-illiterate was known. Other prefecture level predictors included the proportion of minority nationality population in a prefecture/region, farmers’ net per capita income, gross domestic product, the proportion of pregnant women who had antenatal examinations, the proportion of births that were delivered at hospital, and the proportion of home birth deliveries attended by trained health workers using sterile instruments.
The units of analysis were the minority nationality and prefectures which include both autonomous and non-autonomous prefectures. An autonomous prefecture is an administrative region (within Yunnan province) that is governed by the representatives of the minority nationalities residing in that region, although it is still under the regional Communist Party Committee. The minority nationality population residing in autonomous prefectures ranges from 33% to 87%. The majority of the Han Chinese in these prefectures lives mostly in cities and towns. Differences were examined using linear regression analysis weighted by the population size of the prefecture.
In the first stage we examined the population change since 1950 and trends in infant/child mortality and life expectancy at birth since 1990 in the Han Chinese and the 22 minority nationalities. Three of the 25 largest minority nationalities were excluded from the analysis at the nationality level due to missing data in 2000. We then looked at the association between illiteracy and these health outcomes. Means of infant and child mortality and life expectancy were calculated for the 22 largest minority nationalities and were weighted by the total population of each nationality. In the second stage, we examined associations between infant/child mortality, use of preventive health services and socioeconomic factors, using the prefecture-level data instead, because of the lack of such information at the nationality level. Due to collinearity among predictors, only bi-variate linear regression models were used. Ethics approval was not necessary as the study was based on published aggregate data.
Over the last 50 or so years, the population size of all 26 nationalities (including the Han Chinese) in Yunnan has increased, and for some the increase was considerable (data not presented). For example, between 1953 and 2000, the population of the Buyi nationality increased from 2,119 in 1953 to 54,695 in 2000, a 24 fold increase. For the majority of them, including the Han, the increase peaked from 1982 to 1990.
Trends and disparities
The weighted average mortality rate under age 1 for the 22 minority nationalities was 80.1 deaths per 1,000 live births versus 60.3 for the Han Chinese in 1990 (Table 1). The corresponding rates for 2000 were 77.8 and 53.6 respectively. Many of the minority nationalities experienced considerable reduction in mortality under age 1, but in six groups the mortality rate increased (data not shown). This contributed to an increasing disparity ratio between the 22 minority nationalities and the Han Chinese: from 1.3 in 1990 to 1.5 in 2000. However, child mortality for ages 1–4 decreased from 8 deaths in 1990 to 4.3 deaths per 1,000 in 2000 for the minority populations. This decline resulted in a reduction in the minority/Han rate ratio from 1.7 in 1990 to 1.5 in 2000. The health disparities between the minority nationalities in Yunnan and China as a whole are much larger than those between the Han and minority populations in Yunnan owing to slower socioeconomic development in Yunnan compared with China as a whole.
Table 1. Trends in infant/child mortality and life expectancy at birth in Han Chinese and minority nationalities in Yunnan Province, China.a
Mortality under age 1 (death ‰ live births)b
Mortality ages 1-4 (death ‰)c
Life expectancy at birthb(years)
(a) Source: Mortality rates and life expectancy at birth were calculated based on census data.5,11,12,35
(b) Weighted average for minority nationalities was calculated for 22 minority nationalities for which data was available for both 1990 and 2000, using the total population as the weight.
(c) Weighted average life expectancy at birth for minority nationalities was calculated based on the data for 16 minority nationalities for which data was available for all censuses, using the total population as the weight.
Yunnan Han Chinese
Minority nationalities (22 groups)
Variation among minority nationalities
29.9 - 143.8
30.6 - 142.1
50.8 - 70.1
57.2 - 72.0
Yunnan minority-Han Disparity (rate ratio)
-5.0 (difference in years)
-4.4 (difference in years)
Yunnan minority-China disparity (rate ratio)
-8.7 (difference in years)
-6.9 (difference in years)
There was a great deal of heterogeneity between the different minority populations. For example, the under age 1 mortality rate ranged between 29.9 (Mongolian) and 143.8 (Wa) in 1990, and the same held true for 2000. Furthermore, in 2000 the mortality rate for under age 1 was lower in the Bai, Hui and Ah Cang nationalities than in the Han Chinese, whereas in the Lishu and Yao nationalities it more than doubled that of the Han. Amongst the 16 minority nationalities for which reliable data were available for both 1990 and 2000, the weighted average life expectancy at birth was 61.4 years versus 66.4 years for the Han population, with a difference of 5 years (Table 1). In 2000, corresponding figures were 64.5 versus 68.8, with a disparity of about 4.4 years, based on the same 16 nationalities to make the results comparable across years. Again, there was considerable variability between the different minority nationalities. However, the difference between China and the Yunnan minorities was greater in both years (8.7 and 7.0 respectively) than that between Yunnan's Han and minority nationalities.
The data at the prefecture level showed that on average, the minority nationality autonomous prefectures had higher illiteracy, and higher infant and child mortality, but a lower level of per capita net income, gross domestic product and lower utilisation of antenatal care (Table 2). In 1996 about 73% of pregnant women in non-autonomous prefectures had antenatal examinations, but only about 51% of pregnant women residing in autonomous prefectures made antenatal visits. Whilst the gaps between the two types of prefectures in illiteracy and use of antenatal care narrowed from 1996 to 2000, the disparity in gross domestic product and net per capita income increased from 1996 (3482.7 yuan, 233.4 yuan respectively) to 2000 (4135.2 yuan, 404.2 yuan respectively [1 US$= 8 yuan]). Infant mortality and cumulative child mortality (ages 0–4) declined in both predominantly minority and Han prefectures and the disparity between the two types of prefectures also decreased from 1996 to 2000.
Table 2. Key indicators of population health and socioeconomic development by prefecture, Yunnan 2000.a
% minority population
Gross domestic product (yuan)c
Farmers' average net per capital income (yuan)
% pregnant women had antenatal examination (%)
Infant mortality (‰)
Child mortality ages 0–4 (death ‰)
(a) Source: Yunnan Provincial Bureau of Statistics and Provincial Maternal/Child Commission Offce.14
(b) % of the total population that is illiterate or semi-illiterate.
(c) US $1 = 8 yuan.
(d) Rate ratio for % minority population, illiterate population, antenatal examination, infant and child mortality
(e) Absolute differences between autonomous and non-autonomous prefectures in gross domestic product and per capita income.
Non Minority Prefectures (N=5)′
Minority Autonomous Prefectures (N=11)
Illiteracy was a strong predictor of mortality under age 1 in 2000 (Figure 1): The nationalities that had high illiteracy also had high mortality. Both male and female illiteracy were strongly associated with mortality (R2= 0.77 and p < 0.001, R2= 0.86 and p< 0.001 respectively). This association remained largely unchanged after adjustment for population size in 2000 and total fertility in both 1990 and 2000. Illiteracy was also a strong predictor of child mortality ages 1- 4 (R2= .67 and p < 0.001 for male illiteracy and R2= .76 and p <0.001 for female illiteracy, data not shown). Again, the adjustment for the above demographic variables did not change the results. There was a strong association between life expectancy at birth and male and female illiteracy (R2=0.73 and p <0.001, 0.80 and p < 0.001 respectively) (Figure 2).
We examined the association between infant/child mortality and use of preventative health services and economic development indicators, using the prefecture level data for 2000. Bi-variate linear regression analysis showed that illiteracy was positively associated with infant mortality but use of prenatal care, immunisations against tetanus in children, hospital delivery, gross domestic product, and income were negatively associated with it (Table 3). A similar pattern was observed for the association between child mortality and these predictors, with the exception that use of modern home delivery was also significantly associated with child mortality.
Table 3. Association of infant/child mortality with health services use and economic development indicators 2000: bi-variate weighted linear regression estimates (n=16 Prefectures/Regions).
Source: Yunnan Provincial Bureau of Statistics and Provincial Maternal/Child Commission Offce.14
(a) IMR: Infant mortality rate
(b) CMR: Child mortality rate for ages 0–4.
% using prenatal examination
% of children immunised against tetanus
% using modern home delivery
% using hospital delivery
gross domestic product
farmers' net per capita income
1. 36, 4.08
% using prenatal examination
% of children immunised against tetanus
% using modern home delivery
% using hospital delivery
gross domestic product
farmers' net per capita income
There remained large disparities between the Han Chinese and minority nationalities in 2000 in all the health indicators that we examined in Yunnan. The inequalities between the Han and some small minority nationalities were striking: the mortality rate under age 1 was 53.6 per 1,000 live births in the Han Chinese compared with 142.1 in the Lahu Nationality in 2000. There is considerable heterogeneity within the minority nationalities in health status in Yunnan, and this may reflect the diversities in the level of economic development, language, religion, cultural beliefs and geographic locations.
At the prefecture level there was a reduction in infant and child mortality from 1996 to 2000 for both minority and non-minority prefectures and a decrease in disparities between the two types of prefectures but an increase in disparities in illiteracy, income and GDP. However, given substantial variability between the minority nationalities in infant and child mortality, the trend at the regional level may mask large differentials between the Han Chinese and the most disadvantaged minority groups. Literacy was a strong predictor of infant/child mortality at both the nationality and prefecture levels. Use of antenatal examination, immunisation against tetanus, hospital delivery, GDP and farmer's income were also significant predictors of infant and child mortality at the prefecture level. All the data here are observational and we are unable to draw firm conclusions about causality, however, there is already abundant evidence that improvements in maternal education in particular lead to improvement in many health indicators.18 Further, all our analyses are at the ‘ecological’ level and so we have refrained from estimating the size of any of the effect that possible interventions might have. However, we are sure that none of the many biases inherent in ecological studies19 could reverse the direction of our findings.
This study has some other limitations apart from lack of more detailed data at the nationality level and lack of individual level data. The under-reporting of infant death is still a possibility, particularly in remote areas where many minority nationalities reside, because the health surveillance system there may be less well-established. Such a bias would mask an even larger disparity between the Han and minority nationalities and possibly even greater variation within the latter in infant mortality. However, these limitations are not unique to China. Research in Australia and New Zealand (prior to September 1995) also suggests that the quality of Indigenous mortality data is a common problem mainly due to the undercounting of deaths which is likely to lead to an underestimation of the health disparities between Indigenous and non-Indigenous populations.20,21 After September 1995, the New Zealand Census Study has enabled the use of a close approximation to the census ethnicity question to define Māori and Pacific people on death certificates. As a result the numerator-denominator bias and hence undercount of mortality in these groups has been greatly reduced.21
Ample international research has shown large disparities between Indigenous and main-stream populations in developed countries. For instance, the Indigenous peoples in New Zealand, Australia, Canada and the United States also suffer from poorer health, higher early mortality and lower life expectancy when compared with the non-Indigenous population in these countries.20,22–25 These inequalities differ by country, with Australia falling significantly behind Canada and New Zealand,20,24 with some disparities in child health indicators apparently increasing.26
However, there is only limited research that has examined health disparities between Indigenous and other populations in developing countries.27,28 Our study contributes to international research on Indigenous health by adding information from China, a country with entirely different social and political systems from those found in Australia and New Zealand. For ideological, political and economic reasons, the Chinese Communist Party sees the policy of equality with the minority nationalities as a crucial component of their national policies aiming for national equality, unity, mutual prosperity and development. Such policies are reflected in employment, education, allocation of financial resources and specific measures taken to preserve and further develop the Indigenous languages, culture, religion and medicine.4
Based on mortality and census data for 1999–2001, Indigenous people can expect to live about 20 years (56 for males and 63 for females, with an overall average of 59.5 years) less than the total population in Australia (76.6 for males and 82.1 for females). In Canada and the United States the gap is less than eight years for both men and women.20 In New Zealand, the gap in life expectancy between Māori and non-Māori and non-Pacific people ranged from 9.8 to 12.1 in 1966–99, depending on how one conceptualises Māori.21 To add China to this comparison, we have shown that the overall disparity in life expectancy at birth between the Indigenous peoples and the non-Indigenous population in Yunnan (4.4 years in 2000) is far below those in Australia and still below those in the other three developed countries (6 to 9 years). When compared with China's total population in 2000, the Indigenous peoples in Yunnan can expect to live about seven years less, a gap that is still much narrower than in Australia. Given likely undercounting of Indigenous mortality in Yunnan, Indigenous life expectancy could be overestimated to some degree and this would imply that the ethnic gaps in Yunnan may be about equivalent to those in Canada and the US, and the gap between sole Māori and non-Māori and non-Pacific people in New Zealand. However, it would take an unusually large bias due to undercount for the gap to be as wide as in Australia.
One unique aspect of China's Indigenous peoples is a large variation in the key health indictors among the subgroups as we have shown with data for Yunnan. However, the disparity in life expectancy between even the most disadvantaged Indigenous groups (57.2 years) in Yunnan and China as a whole (71.4 years) in 2000 is still narrower than the same inequalities between Indigenous and non-Indigenous populations in Australia.
In terms of infant mortality rate, Yunnan's Indigenous peoples compare much less favourably to that of Australia, New Zealand and Canada both in absolute and relative terms. The infant mortality rate for Yunnan's Indigenous populations is more than three time as high as that for China as a whole in 2000 (77.7 deaths per 1,000 live births versus 26.9). The infant mortality rate for Indigenous populations is 15.1 deaths per 1,000 live births versus 5.7 for Australia as a whole,29 and this disparity has persisted into recent years.24,30 In New Zealand, the infant mortality rate for the Māori population is 8.6 per 1,000 live births versus 4 for the non-Māori population.24 In Canada it is 8 deaths per 1,000 live births for First Nations on reserve versus 5.3 deaths for Canada as a whole.22
Australian research suggests a number of factors that contribute to the large mortality and life expectancy disparities between Indigenous and non-Indigenous populations. Proximal factors include higher rates of infectious diseases, obesity, diabetes, heart and kidney diseases and cancer among Indigenous than in non-Indigenous Australians.23,30 Loss of culture, persistent racial discrimination and socioeconomic disadvantages, such as poor education, unemployment, and inadequate housing and infrastructure are also factors that may have contributed to the disparities.
In New Zealand, socioeconomic status as measured in income, education, employment, labour force status, car access, asset wealth, access to community resources and neighbourhood deprivation contribute significantly to overall mortality differences between Māori and non- Māori non-Pacific populations (between 32% to 37% in 1996–1999 for 45–75 year olds and 50% or more for working age adults),31,32 especially in infant and child mortality.33
What could account for the absolute and relative differences in life expectancy between Yunnan's minority nationalities and Indigenous Australians? The favourable Chinese government policies towards its Indigenous populations are markedly different from those in Australia. Possibly a higher level of proximal risk factors for premature mortality in Indigenous Australians than in Chinese Indigenous populations as described above and lower life expectancy of the Han Chinese relative to that of the mainstream population in Australia are also plausible contributing factors.
Based on our field observations, socioeconomic disadvantages and the lack of even basic preventive and curative services and affordability of such services are the predominant contributing factors to the Indigenous versus non-Indigenous health disparities in Yunnan but confirmation will require individual level data on socioeconomic status and use of preventive health services becoming available for more rigorous analysis.34 These factors are closely associated with the remote and mountainous location of many of the Indigenous groups. The geographic terrain, limited access to health facilities and high cost of health care remain significant barriers to improving Indigenous health in Yunnan and China. Most of China's Indigenous people live in rural and remote areas and are not covered by public health insurance. It remains to be seen how the newly implemented rural medical insurance system might break down some of these barriers.
This is the first study to examine the key population health indicators and their social, economic and demographic determinants for a large number of minority nationalities in Yunnan. It is also one of the few studies that have analysed temporal trends in the demographic and health profiles of a large number of individual minority nationalities in China. Thus, our findings may have wider policy implications. Our findings suggest that at the regional level an increase in the use of antenatal examinations, immunisations and home delivery, which are low-cost and much more amenable for interventions, and improvements in socioeconomic development (literacy and income) are likely to lead to further reduction in infant or child mortality in Yunnan's Indigenous populations. Future health policy must target minority nationalities whose health profile remains strikingly poor compared to both the Han Chinese and other minority groups.
Jianghong Li was supported by the Australian National Health and Medical Research Council Capacity Building Grant (#254545).