Bulletin of the World Health Organization

Breastfeeding rates in central and western China in 2010: implications for child and population health

Sufang Guo a, Xulan Fu a, Robert W Scherpbier a, Yan Wang b, Hong Zhou b, Xiaoli Wang b & David B Hipgrave a

a. United Nations Children’s Fund China Country Office, 12 Sanlitun Lu, Beijing, 100600, China.
b. School of Public Health, Peking University, Beijing, China.

Correspondence to: David B Hipgrave (e-mail: dhipgrave@gmail.com).

(Submitted: 08 August 2012 – Revised version received: 01 December 2012 – Accepted: 19 December 2012 – Published online: 23 March 2013.)

Bulletin of the World Health Organization 2013;91:322-331. doi: http://dx.doi.org/10.2471/BLT.12.111310


In 2010, China had the world’s fifth largest number of deaths among children younger than 5 years, despite child mortality rates in the country having fallen steadily over the last two decades.14 Consequently, the timing and causes of death in children in China are different from those in countries where large numbers of children perish early in life. Globally, around two thirds of deaths in children younger than 5 years are caused by infection5 and an estimated 35% are associated with poor nutrition.6,7 By contrast, only 21.4% of comparable deaths in China in 2008 were infection-related: 16.5% were due to pneumonia; 3%, to diarrhoea and 1.9%, to neonatal sepsis.2 However, child mortality rates are much higher in poor western provinces, where infectious diseases are more common:2 the rate in children younger than 5 years in the poorest counties is more than six times higher than in large cities.8 Children are more likely to die from pneumonia and diarrhoea in the western part of China.2,4

Breastfeeding has been shown to reduce child mortality and morbidity, especially from infectious diseases.6,911 It may be possible, then, to reduce infection-associated child mortality and morbidity in poor rural areas in China by promoting good infant and young child feeding practices, particularly breastfeeding, and by discouraging the inappropriate use of breast-milk substitutes. Moreover, infant and young child feeding practices may be associated with the long-term incidence of noncommunicable diseases in both developing and developed countries, regardless of socioeconomic status.12,13 Thus, improving breastfeeding practices may help reduce the future incidence of noncommunicable diseases both in China and elsewhere.

Currently, few data are available on infant and young child feeding practices in China. The aim of this study, therefore, was to describe these practices and associated variables by surveying a large geographical area in China’s central and western provinces. Our survey was the first in the country to include the full range of globally recommended breastfeeding indicators,11 enabling comparisons to be made between China and other countries. We also examined differences in infant and young child feeding practices between China and neighbouring countries in the context of increased consumption of breast-milk substitutes and rising noncommunicable disease rates.


We conducted a community-based, cross-sectional survey of infant and young child feeding practices in 26 counties, with a total population 11 000 000, in 12 central and western provinces of China: Chongqing, Gansu, Guangxi, Guizhou, Inner Mongolia, Jiangxi, Qinghai, Shaanxi, Shanxi, Sichuan, Tibet and Xinjiang. The counties were selected by staff at the United Nations Children’s Fund (UNICEF) and the Chinese Ministry of Health as being representative of poor rural counties on the basis of their level of socioeconomic development and measures of maternal and child health. Child mortality was as high as 65 per 1000 live births in these counties, compared with a national rural average of 34 per 1000 live births in 2009, and the per capita gross domestic product ranged from 150 to 2210 United States dollars (US$), as calculated using the 2009 conversion rate from renminbi, compared with the national average of US$ 3680. The survey was conducted by Peking University School of Public Health and UNICEF staff working with local health authorities and was approved by Peking University’s ethics committee.

Data collection

A multistage sampling technique was used to select townships and villages in the 26 counties. First, all townships in each county were ranked by hospital delivery rate in 2009 and divided into three approximately equal strata; one township was then randomly selected from each stratum, except in sparsely populated Tibet, in which two townships were selected from two equal strata in each county. Within the selected townships, villages were ranked by their distance from the nearest main town and were divided into three almost equal strata; one village was then randomly selected from each stratum. Finally, approximately 20 children younger than 5 years were selected in each village. In total, 4368 children were sampled from 209 villages in 75 townships.

The overall purpose of the survey was to assess child health, perinatal care and associated demographic and socioeconomic factors in households with a child younger than 5 years. This report focuses on breastfeeding indicators, variables associated with breastfeeding, and complementary feeding among children younger than 2 years. In families with more than one child younger than 5 years, only the youngest was surveyed. Consequently, the sample was weighted towards younger children: 2577 were less than 2 years old. After excluding 330 children left behind by mothers who migrated and hence could not breastfeed them and after weighting for sampling probability, the final data set comprised 2354 children younger than 2 years (Appendix A and Appendix B, both available at: http://www.unicefchina.org/en/index.php?m=content&c=index&a=show&catid=214&id=1512).

Interviews with mothers and caregivers were conducted between July and September 2010 by local health workers trained and supervised by UNICEF and Peking University staff. Interviewers used a structured questionnaire developed from published indicators of infant and young child feeding practices (Appendix C; available at: http://www.unicefchina.org/en/index.php?m=content&c=index&a=show&catid=214&id=1512) and UNICEF’s Multiple Indicator Cluster Survey.11,14 Child nutrition practices were assessed using 24-hour dietary recall in accordance with the relevant World Health Organization (WHO) module.15 The questionnaire was pretested and adjusted before data collection and oral informed consent was sought before each interview. The research team monitored data collection for quality control.

Survey data included demographic and socioeconomic variables for each child and mother or caregiver and information on care-seeking practices. Household wealth was categorized using the ownership of electrical appliances: poor households had 0 to 2 appliances; medium-wealth households had 3; and the least poor households had 4 or more.1618 We also surveyed the indicators of infant and young child feeding listed in Table 1,11,14 which include early initiation of breastfeeding (i.e. within 1 hour of birth), exclusive breastfeeding among children younger than 6 months, continued breastfeeding for 1 and 2 years after birth and age-appropriate breastfeeding, as defined in Appendix C. In addition, we surveyed the early introduction of complementary feeding, defined in Appendix C as giving any soft, semisolid or solid food to an infant before the age of 6 months, regardless of breastfeeding. The rates of five indicators of infant and young child feeding observed in the survey were compared with the most recent data available on corresponding national rates in most of China’s near neighbours, as well as Brazil and South Africa.19

Data analysis

Data analysis was performed using Stata 11 (StataCorp. LP, College Station, United States of America) and took into account any unequal sampling probabilities, clustering and stratification in the study design (Appendix A and Appendix B). Point estimates of indicators are reported with 95% confidence intervals (CIs). Design-based F testing was used to assess associations between the early initiation of breastfeeding, exclusive breastfeeding among children younger than 6 months and survey variables. Univariate analysis was performed to identify associations with these two breastfeeding indicators and crude odds ratios (ORs) and 95% CIs were derived using the Wald test. In addition, multivariate logistic regression analysis was carried out: the initial model included fixed “a priori” variables that were assumed to be associated with the early initiation of breastfeeding or exclusive breastfeeding among children younger than 6 months and other variables that were found to be associated with breastfeeding in the univariate analysis (i.e. P ≤ 0.1). The fixed variables were retained throughout the analysis but other variables were excluded one by one according to their P-values: only those with P ≤ 0.10 were retained in the final model. The strength of each association was expressed as an adjusted OR (aOR).

Since missing data and “don’t know” answers were relatively infrequent, except for maternal age, maternal education and antenatal clinic attendance in cases where maternal information was not available, we recoded them as null values and included them in both numerators and denominators, as recommended.15


Table 2 shows the demographic, socioeconomic and clinical characteristics of the 2354 children younger than 2 years included in the analysis. The male to female ratio was 1.4:1 and 53.3% were first-born children. More than a quarter were of minority ethnicity, the national minority proportion being 8.5% in 2010.20 The number of mothers younger and older than 25 years was nearly equal; only 5.7% were younger than 20 years. In addition, 71.4% had completed 9 years of education. Over 80% of mothers had attended an antenatal clinic five or more times. Among the children, 97.3% were delivered at a health-care facility and 23.8% were delivered by Caesarean section: the rate was 30.4% in county, referral-level facilities and 14.8% in first-level township hospitals (P < 0.01; data="" not="">

Breastfeeding indicators

Overall, 98.3% of the infants and young children surveyed had been breastfed at some time and 59.4% of mothers had initiated breastfeeding early (e.g. within 1 hour of birth) (Table 1). The rates of exclusive breastfeeding among children younger than 6 months, continued breastfeeding at 1 and 2 years of age and age-appropriate breastfeeding were all low. Fig. 1 shows infant and young child feeding practices during the first 2 years of life. The rate of exclusive breastfeeding was only 58.3% among newborn infants (aged 0 to 27 days); it declined further to 29.1% in those aged 3 to 4 months and to 13.6% in those aged 5 to 6 months. The most common reason for non-exclusive breastfeeding during the first 6 months, except during the first month, was giving water. The second most common reason was using breast-milk substitutes. Around 40% of newborn infants were given something other than breast milk: 27.1% received other milk; 9.4%, water; and 3.0%, soft, semisolid or solid food (Appendix D, available at: http://www.unicefchina.org/en/index.php?m=content&c=index&a=show&catid=214&id=1512).

Fig. 1. Feeding practices in children aged less than 2 years in 26 counties in central and western China, 2010
Fig. 1. Feeding practices in children aged less than 2 years in 26 counties in central and western China, 2010

Generally, the introduction of complementary feeding was timely: 89.7% of infants aged 6 to 8 months, the recommended age for starting complementary feeding (Appendix C),11 had consumed soft, semisolid or solid food in the 24 hours before the survey (Table 1). However, although only 18.7% of infants younger than 6 months had commenced complementary feeding, the sharp increase from 12.2% in infants aged 4 to 5 months to 41% among those aged 5 to 6 months suggests that many started earlier than recommended.

Variables associated with breastfeeding

Table 3 presents the results of the univariate and multivariate analyses of variables associated with the early initiation of breastfeeding and Table 4, the results for exclusive breastfeeding among children younger than 6 months. Univariate analysis indicated that the early initiation of breastfeeding was positively associated with poverty, minority ethnicity and maternal attendance at an antenatal clinic five or more times, and negatively associated with Caesarean section and delivery at a county-level or private hospital (rather than a township-level hospital). Early initiation was not associated with sex, the existence of elder siblings, maternal age or maternal education. After adjustment for other variables, multivariate analysis showed that the early initiation of breastfeeding remained negatively associated with delivery at a county or higher-level referral hospital (OR: 0.6) and delivery by Caesarean section (OR: 0.53) and positively associated with attending an antenatal clinic five or more times (OR: 3.48).

Univariate analysis indicated that exclusive breastfeeding among children younger than 6 months was associated with only three variables: there were positive associations with male sex and delivery at a county or higher-level referral hospital and a negative association with the child’s age. After adjustment for other variables, multivariate analysis showed that exclusive breastfeeding among children younger than 6 months remained positively associated with delivery at a county or higher-level referral hospital (OR: 2.22) and negatively associated with the child’s age (OR: 0.77).

Comparisons with selected countries

Table 5 shows the rates of five indicators of infant and young child feeding from our survey, from China’s most recent national survey21 and from the most recent, large-scale surveys in most of China’s neighbouring countries plus Brazil and South Africa. In our survey, breastfeeding was initiated early in 59% of infants, which is slightly above the median of 57% for Asia and 48% for the BRICS countries (i.e. Brazil, the Russian Federation, India, China and South Africa). However, both nationally and in the 12 provinces covered by this survey, exclusive breastfeeding was less common than in most other countries included in Table 5 and rates of continued breastfeeding at 12 to 15 months and 20 to 23 months were well below those in virtually all countries except Brazil, for which the rate at 12 to 15 months was lower than in our survey.


This study identified major concerns with infant and young child feeding practices in 26 poor rural counties in central and western China. The exclusive breastfeeding rate dropped precipitously during the first 6 months of life and became one of the lowest in the region. Breast-milk substitutes were frequently introduced early, followed soon after by water and non-milk liquids and, by around 4 months, various foods. The rate of continued breastfeeding at 1 year was much lower than in most neighbouring countries; the rate at 2 years was easily the lowest. Only the rate of early initiation of breastfeeding was encouraging.

We found that the rate of early initiation of breastfeeding in our rural sample was higher than in the 2008 national survey21 and we observed positive associations between early breastfeeding and attendance at an antenatal clinic and delivery in a township-level hospital, both of which suggest that the recent increase in use of these facilities in China has had additional benefits.22 The reason for the negative association between early breastfeeding and delivery at a county-level hospital may be that early breastfeeding was not prioritized in these hospitals or that the infants born in them were sicker and smaller than average or came from wealthier families who give formula early. However, our experience suggests that breast-milk substitutes are often promoted at county hospitals. The negative association between the early initiation of breastfeeding and Caesarean section, which is being performed ever more frequently in China,23 has been observed elsewhere:24 it may be related to the postoperative positioning of the mother, poor pain relief, slow lactogenesis25 or increased use of breast-milk substitutes. Spinal anaesthesia, which is in common use in China,26 should enable early breastfeeding if pain relief is adequate.

Our findings on exclusive breastfeeding are very similar to the most recent comparable data from rural China,21 particularly from Xinjiang in the far west in 2003 to 200427 and Zhejiang on the east coast in 2004 to 2005.28 Possibly because exclusive breastfeeding was infrequent after the first month, we found only one variable associated with exclusive breastfeeding among children younger than 6 months: delivery at a county-level hospital. Unlike other studies,29 ours showed no association with Caesarean section. In China, the most common reasons for ceasing exclusive breastfeeding are the perceived insufficiency of breast milk, traditional beliefs, maternal or child illness and maternal employment.27, 3032 Marketing of breast-milk substitutes may also be a factor.33 In addition, the early introduction of complementary feeding has an ancient history in China34 and related traditional beliefs are very influential in less developed areas and in areas with large minority ethnic populations.3537

The exclusive breastfeeding rates observed in our survey, in the 2008 national survey and in the Xinjiang and Zhejiang surveys indicate that there has been no improvement over time. This lack of progress has occurred even though, in 2007, the Chinese Ministry of Health adopted the WHO recommendation that exclusive breastfeeding should continue for 6 months38 and started using WHO’s definition of exclusive breastfeeding (previously it used the “predominant” breastfeeding indicator).11 Moreover, government policy that notionally banned the promotion of breast-milk substitutes39 has had little impact: consumption in China increased almost threefold between 2003 and 2008 and could double again by 2013,40,41 possibly due to laissez-faire regulation of breast-milk substitute advertising. There are signs, however, that educated urban women are becoming more aware of the benefits of breastfeeding.41

In 2008, it was estimated that, globally, suboptimal breastfeeding in the first 6 months of life accounted for 10 to 15% of deaths in children younger than 5 years and 10% of the disease burden.6 Moreover, the survey data we obtained show that breastfeeding practices are poor in many countries in Asia and in some large nations in other parts of the world, suggesting that there are widespread misconceptions about the benefits. Our survey showed that many infants and young children started complementary feeding early. In China, breast-milk substitutes are not of good quality and, in poor rural areas, they may be contaminated or diluted.42,43 Moreover, the weaning diet in rural China is known to be low in nutrient content.44 Poor-quality complementary feeding can increase the risk of undernutrition, infectious disease and death45 and a nutrient-poor, calorie-rich diet may be associated with a higher risk of obesity, hypertension, and diabetes in later life.12 Consequently, inappropriate infant and young child feeding in rural China places many infants at risk of early illness and death and may increase their risk of developing noncommunicable diseases. In addition to being more affordable, breastfeeding and appropriate complementary feeding could reduce the risk of noncommunicable diseases at least as effectively as mid-life dietary and lifestyle changes or drug treatment.12

Our analysis has several limitations. First, because the survey was cross-sectional, the observed associations with breastfeeding indicators may not be causal. Second, our survey participants lived in poor rural areas where the proportion of ethnic minorities was relatively high. Hence, our findings may not be generalizable to all of China. However, there is evidence that, globally, infant and young child feeding practices may be better among the poor than the rich46 and, in China, that breastfeeding is less common in urban than in rural areas.21,28 Third, since we excluded from the analysis children left behind by mothers who had migrated, our estimate of the breastfeeding rate may be higher than the true population rate. Fourth, the sampling strategy may have influenced the observed gender ratio. In rural China, families whose first born is a girl are permitted a second child and couples tend to continue having children until they have a boy. Thus, our selection of the youngest child in the household could have favoured the selection of males. Moreover, mothers may be more reluctant to register female children. The male to female ratio of 1.4:1 we observed is higher than the national average, which is 1.19:1 for infants and 1.24:1 for children aged 1 to 4 years, but it is not unusual for some rural areas.47 We considered using a life table to weight the sample but no appropriate data exist, particularly given the current high mobility of China’s rural population.

In the mid-1990s, China promoted breastfeeding by designating over 7000 hospitals “baby-friendly”. However, we recently observed that breast-milk substitutes are promoted in many of these facilities. It appears that this initiative and associated community measures had little effect on infant and young child feeding practices. In 2010, the Chinese Ministry of Health re-launched the baby-friendly hospital initiative but, unfortunately, some related activities are partly sponsored by, or involve, companies associated with breast-milk substitutes.

To prevent death and disability due to infectious diseases in childhood and to provide the best foundation for the prevention of noncommunicable diseases, China must focus on promoting breastfeeding throughout the nation. Although high- and mid-level government support is not yet widespread, some national champions are emerging.48 All sections of the community should be involved: appropriate infant and young child feeding practices should be encouraged, especially at antenatal clinics; health professionals and the community in general should be informed about the indications for, and the management of, Caesarean section; there should be strict limits on the use of breast-milk substitutes in health facilities; misconceptions about breast milk should be corrected; mothers should be advised against the early introduction of water, other liquids and complementary feeding; and exaggerated claims for the benefits of breast-milk substitutes and the promotion of substitutes by health-care workers should be actively challenged. In addition, the inappropriate marketing of breast-milk substitutes should be controlled with much stronger regulation and employers should support breastfeeding among women re-entering the workforce. Finally, the revitalised baby-friendly hospital initiative should be implemented consistently and appropriate performance indicators established.


The authors gratefully acknowledge the kind assistance of the provincial, county and township health bureaux in the surveyed locations. The opinions expressed in this article are those of the authors and are not intended to represent the perspectives of their parent institutions.


This research was funded by UNICEF China.

Competing interests:

None declared.