Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128·9 million children, adolescents, and adults

Summary

Introduction

Being underweight, overweight, or obese during childhood and adolescence is associated with adverse health consequences throughout the life-course. Underweight among children and adolescents is associated with higher risk of infectious diseases, and for girls of childbearing age, is associated with adverse pregnancy outcomes including maternal mortality, delivery complications, preterm birth, and intrauterine growth retardation.^{1, 2} Preventing and reversing excess weight in children and adolescents is also important for many reasons;^{3, 4} first, weight loss and maintenance after weight loss are hard to achieve,⁵ therefore gaining excess weight in childhood and adolescence is likely to lead to lifelong overweight and obesity.⁶ Second, being overweight in childhood and adolescence is associated with greater risk and earlier onset of chronic disorders such as type 2 diabetes.^{3, 4, 7, 8, 9} Third, childhood and adolescent obesity has adverse psychosocial consequences and lowers educational attainment.^{3, 4, 10, 11} Finally, children and adolescents are more susceptible to food marketing than adults, which makes reducing children's exposure to obesogenic foods necessary to protect them from harm.^{3, 12}

Although trends in children's and adolescents' weight status have been documented in individual countries, little comparable information exists on worldwide trends, and none for mean body-mass index (BMI) and underweight. We pooled population-based data to estimate trends from 1975 to 2016 in mean BMI and in the prevalence of a comprehensive set of BMI categories that cover the underweight to obese range among children and adolescents for all countries in the world. To compare the trajectory of obesity and underweight in children and adolescents with that of adults, we also generated updated estimates for adults.¹³

Methods

Results

The results can be explored using dynamic visualisations and downloaded from the NCD-RisC website. We pooled 2416 population-based data sources with measurement of height and weight on 128·9 million people aged 5 years and older from 1975 to 2016 (appendix). 1099 sources included data on 24·1 million participants aged 5–17 years, 848 sources included data on 7·4 million participants aged 18–19 years, and 1820 sources included data on 97·4 million participants aged 20 years and older (appendix). Additional information on the age distribution of data sources and participants is shown in the appendix. 1187 (49%) of 2416 data sources were from national samples, 390 (16%) covered one or more subnational regions, and the remaining 839 (35%) were from one or a few communities (appendix). 583 (24%) of 2416 data sources were from before 1995 and another 1833 (76%) were from 1995 and later (appendix). The number of data sources per country in different regions ranged from 3·4 in the Caribbean to 54·0 in the high-income Asia-Pacific region (appendix).

In 1975, the global age-standardised mean BMI of children and adolescents aged 5–19 years was 17·2 kg/m² (95% CrI 16·8–17·6) for girls (figure 1) and 16·8 kg/m² (16·3–17·2) for boys (figure 2). Mean BMI was lowest in south Asia, with an age-standardised mean of 15·8 kg/m² (15·2–16·3) for girls and 15·0 kg/m² (14·5–15·6) for boys, followed by east Africa (16·5 kg/m² [14·8–18·2] for girls and 15·5 kg/m² [13·6–17·4] for boys). Girls in Melanesia, Polynesia and Micronesia, and the high-income English-speaking region had the highest age-standardised mean BMI in 1975, all above 19·0 kg/m². The highest mean BMIs for boys were those in Polynesia and Micronesia (19·1 kg/m², 18·0–20·2), followed by the high-income English-speaking region.

Open in a separate window

Figure 1

Trends in age-standardised mean BMI by sex and region in females

Children and adolescents were aged 5–19 years and adults were aged 20 years and older. The lines show the posterior mean estimates and the shaded areas show the 95% credible intervals. See appendix for trends by country. BMI=body-mass index.

Open in a separate window

Figure 2

Trends in age-standardised mean BMI by sex and region in males

Children and adolescents were aged 5–19 years and adults were aged 20 years and older. The lines show the posterior mean estimates and the shaded areas show the 95% credible intervals. See appendix for trends by country. BMI=body-mass index.

From 1975 to 2016, children's and adolescents' age-standardised mean BMI increased globally and in most regions (Figure 1, Figure 2). The global increase was 0·32 kg/m² per decade (95% CrI 0·23–0·41, PP of the observed increase being a true increase>0·9999) for girls and 0·40 kg/m² per decade (0·30–0·50, PP>0·9999) for boys, leading to virtually identical age-standardised mean BMIs of 18·6 kg/m² (18·4–18·7) for girls and 18·5 kg/m² (18·3–18·7) for boys in 2016. The corresponding figures for adults were 24·8 kg/m² (24·6–25·0) in women and 24·5 kg/m² (24·3–24·6) in men.

Regional change in girls ranged from virtually no change (−0·01 kg/m² per decade [95% CrI −0·42 to 0·39; PP of the observed decrease being a true decrease=0·5098]) in eastern Europe to 1·00 kg/m² increase per decade (0·69–1·35, PP>0·9999) in central Latin America and 0·95 kg/m² per decade (0·64–1·25, PP>0·9999) in Polynesia and Micronesia. The range for boys was from 0·09 kg/m² per decade (−0·33 to 0·49; PP=0·6926) in eastern Europe to 0·77 kg/m² per decade (0·50–1·06, PP>0·9999) in Polynesia and Micronesia. In some regions, children's and adolescents' BMI increased gradually over the four decades of analysis (Figure 1, Figure 2). However, there has been a recent flattening of trends in northwestern Europe and the high-income English-speaking and Asia-Pacific regions for both sexes, southwestern Europe for boys, and central and Andean Latin America for girls. With the exception of women in the high-income Asia-Pacific region, adult mean BMI continues to increase in all of these regions and sexes (Figure 1, Figure 2). By contrast with this plateauing, the rise in mean BMI has accelerated since around 2000 in east and south Asia for both sexes, and in southeast Asia for boys.

The lowest mean child and adolescent BMIs in 2016 were still those in south Asia and east Africa, with age-standardised mean BMIs between 16·9 and 17·9 kg/m² for girls and boys; the highest were those in Polynesia and Micronesia for both sexes, followed by Melanesia and the high-income English-speaking region. Age-standardised mean BMIs of girls and boys in Polynesia and Micronesia, which were 23·1 kg/m² (95% CrI 22·4–23·8) and 22·4 kg/m² (21·6–23·1), respectively, were higher than those of adults in some regions. Children's and adolescents' age-standardised mean BMI was also more than 20 kg/m² in Melanesia and many parts of Latin America and the Caribbean.

The regional rankings in 2016 differed slightly between children aged 5–9 years and adolescents aged 10–19 years (appendix). For example, the lowest mean BMI in children aged 5–9 years was seen in east Africa in both sexes, whereas in those aged 10–19 years, south Asian girls and boys had lower mean BMI than their African peers. Polynesians and Micronesians had the highest mean BMI in those aged 5–9 and 10–19 years, with the subsequent spots held by the high-income English-speaking region, regions in Latin America and the Caribbean, and Melanesia. Among these regions, central Latin America had a poorer ranking (ie, higher BMI relative to other regions) at age 10–19 years than at age 5–9 years, as did boys in the high-income English-speaking region. By contrast, east Asia performed worse in ranking in 5–9 years of age than it did in 10–19 years.

The lowest age-standardised mean BMI over the 42 years of analysis among girls was in Bangladesh in 1975 (15·6 kg/m², 95% CrI 13·2–17·9), and among boys was in Ethiopia in 1975 (14·4 kg/m², 11·9–17·0; Figure 3, Figure 4). Age-standardised mean BMI in 1975 was less than 21 kg/m² in every country, except for girls in American Samoa, who had an age-standardised mean BMI of 21·2 kg/m² (20·6–21·9). From 1975 to 2016, age-standardised mean BMI increased by more than 0·25 kg/m² per decade in 155 countries in girls, with the rise more than 1·0 kg/m² per decade in some countries in Polynesia and in Mexico (PP of being a true rise>0·9999); in boys, the rise was more than 0·25 kg/m² per decade in 189 countries and more than 1·0 kg/m² per decade in the Cook Islands. When subsets of the analysis period are considered, before the year 2000, age-standardised mean BMI increased in almost every country. After 2000, there were non-significant declines in mean BMI in 29 countries for girls and 12 (mostly high-income) countries for boys.

Open in a separate window

Figure 3

Age-standardised mean BMI, prevalence of obesity, and prevalence of moderate and severe underweight by sex and country in 2016 in children and adolescents

Children and adolescents were aged 5–19 years. Obesity was defined as more than 2 SD above the median of the WHO growth reference. Moderate and severe underweight was defined as more than 2 SD below the median. See appendix for results for adults. BMI=body-mass index.

Open in a separate window

Figure 4

Age-standardised mean BMI in children and adolescents in 1975 and 2016

Each line shows one country. BMI=body-mass index.

In 2016, Ethiopia had the lowest age-standardised mean BMI for both sexes, 16·8 kg/m² (95% CrI 15·6–17·9) for girls and 15·5 kg/m² (14·4–16·6) for boys (Figure 3, Figure 4). Other countries with low BMI in both sexes in 2016 were Niger, Senegal, India, Bangladesh, Myanmar, and Cambodia. At the other extreme, age-standardised mean BMI was more than 24 kg/m² in girls and boys in the Cook Islands and Niue and girls in Samoa, which was greater than that for adults of the same sex in 36 countries for girls and 59 countries for boys. Age-standardised mean BMI was between 22 and 24 kg/m² in another 11 countries for girls and 10 countries for boys including in Polynesian and Micronesian islands, girls in the Bahamas and Chile, and boys in Qatar and Kuwait.

The age-standardised mean BMI for children and adolescents and for adults were correlated in 1975 and 2016 (correlation coefficients 0·80 and 0·85 for females and 0·92 and 0·89 for males; figure 5). Changes in age-standardised mean BMI were moderately correlated between the two age groups before 2000 (correlation coefficient 0·52 for females and 0·51 for males) but only weakly after 2000 (correlation coefficient 0·14 for females and 0·21 for males; figure 6). The decoupling of BMI trends in children and adolescents and those of adults is due to a set of distinct regional phenomena: adults continued to gain weight in most western countries, where children's and adolescents' mean BMI stopped rising. By contrast, the rise in adult BMI seems to have plateaued in Oceania, albeit at high levels, whereas children's and adolescents' BMI continues to rise. In Latin America and the Caribbean, there is more variation in the rate of BMI increase in children and adolescents than in adults.

Open in a separate window

Figure 5

Comparison of age-standardised mean BMI in children and adolescents and in adults

Children and adolescents were aged 5–19 years and adults were aged 20 years and older. Each point shows one country. The dotted line shows the linear association between the two outcomes. BMI=body-mass index.

Open in a separate window

Figure 6

Comparison of change in age-standardised mean BMI in children and adolescents and in adults

Children and adolescents were aged 5–19 years and adults were aged 20 years and older. Each point shows one country. The dotted line shows the linear association between the two outcomes. BMI=body-mass index.

In 1975, girls had higher age-standardised mean BMI than boys in most countries in sub-Saharan Africa, south Asia, and the Middle East and north Africa, and lower age-standardised mean BMI than boys in many countries in Europe and Latin America and the Caribbean (figure 7). Higher BMI in girls than boys was still seen in 2016 in many sub-Saharan African and south Asian countries. By contrast, the gap between sexes in BMI in the Middle East and north Africa shrank or reversed as boys gained more weight than girls. In Europe and Latin America, girls gained more weight than boys, closing the gap between sexes in BMI.

Open in a separate window

Figure 7

Comparison of age-standardised mean BMI in 1975 and 2016, and change per decade in age-standardised mean BMI from 1975 to 2016 by sex

Each point shows one country. BMI=body-mass index.

Over the 42 years of analysis, the global age-standardised prevalence of obesity in children and adolescents increased from 0·7% (95% CrI 0·4–1·2) in 1975 to 5·6% (4·8–6·5) in 2016 in girls (figure 8), and from 0·9% (0·5–1·3) in 1975 to 7·8% (6·7–9·1) in 2016 in boys (figure 9). Obesity increased in every region, with proportional rise being smallest in high-income regions (on average 30–50% per decade) and largest in southern Africa (about 400% per decade, albeit from very low levels).

Open in a separate window

Figure 8

Trends in age-standardised prevalence of BMI categories in female children and adolescents by region

Children and adolescents were aged 5–19 years. See appendix for results for adults. BMI=body-mass index.

Open in a separate window

Figure 9

Trends in age-standardised prevalence of BMI categories in male children and adolescents by region

Children and adolescents were aged 5–19 years. See appendix for results for adults. BMI=body-mass index.

Globally, the prevalence of moderate and severe underweight changed less than the rise in obesity, from 9·2% (95% CrI 6·0–12·9) in 1975 to 8·4% (6·8–10·1) in 2016 in girls and from 14·8% (10·4–19·5) in 1975 to 12·4% (10·3–14·5) in 2016 in boys. The relatively small change in moderate and severe underweight prevalence at the global level, however, was partly due to faster population growth in regions where underweight prevalence is higher (eg, the share of children and adolescents living in south Asia, where prevalence is highest, increased from 20·5% in 1975 to 26·4% in 2016 in girls, and 21·1% in 1975 to 27·1% in 2016 in boys) while prevalence declined in most regions. The largest proportional decline in the prevalence of moderate and severe underweight occurred in Polynesia and Micronesia and in southern Africa in both sexes, where prevalence declined by an average of up to one third per decade for girls and by about one fifth per decade for boys from 1975 to 2016 (Figure 8, Figure 9). There was a non-significant rise of about 6% per decade (PP=0·6630) in underweight in girls in southeast Asia. Nonetheless, in most regions, the increase in the prevalence of overweight and obesity was larger than the decline in the prevalence of underweight (Figure 8, Figure 9), ie, the width of the BMI distribution increased.

Regionally, moderate and severe underweight prevalence was highest in south Asia over the entire analysis period, at 20·3% (95% CrI 15·3–25·8) in girls and 28·6% (22·3–35·0) in boys in 2016, having decreased from 23·0% (13·9–33·6) in girls and 37·8% (26·6–49·2) in boys in 1975. Prevalence of obesity was highest in Polynesia and Micronesia in both sexes, 25·4% (16·8–35·2) in girls and 22·4% (13·4–32·9) in boys, followed by the high-income English-speaking region.

Nationally, the prevalence of moderate and severe underweight was less than 1% among girls in 45 countries and among boys in 29 countries in 2016 (figure 3). Prevalence of moderate and severe underweight was high throughout south Asia, reaching 22·7% (95% CrI 16·7–29·6) among girls and 30·7% (23·5–38·0) among boys in India. Obesity prevalence was between 1% and 2% among girls in Cambodia, Burkina Faso, Vietnam, Ethiopia, India, Madagascar, Republic of the Congo, Japan, Nepal, Niger, and Chad. Obesity prevalence was less than 1% among boys in Uganda, Rwanda, Niger, Burkina Faso, Ethiopia, Guinea, Chad, and Senegal and between 1% and 2% in another 24 countries.

Conversely, obesity prevalence was more than 30% in girls in Nauru, the Cook Islands, and Palau and boys in the Cook Islands, Nauru, Palau, Niue, and American Samoa in 2016, and was also high, around or above 20%, in some countries in Polynesia and Micronesia, the Middle East and north Africa (eg, Egypt, Kuwait, Qatar, and Saudi Arabia), the Caribbean (Bermuda and Puerto Rico), and in the USA. In 1975, obesity prevalence was less than 10% in every country except Nauru and Bermuda, where it was still less than 20%. From 1975 to 2016, obesity prevalence increased in every country, although the increase was not statistically significant in some high-income countries.

The number of moderately and severely underweight girls and boys worldwide peaked around the year 2000, and subsequently decreased to 75 (95% CrI 44–117) million girls and 117 (70–178) million boys in 2016, slightly higher than in 1975 (figure 10). In most regions, the number of moderately and severely underweight children and adolescents decreased despite population growth. The exceptions were south Asia; southeast Asia; and central, east, and west Africa, where population growth led to an increase in the absolute underweight burden, despite declining prevalence. 47·5 million (63%) of 75 million moderately and severely underweight girls and 73·6 million (63%) of 117 million underweight boys in the world lived in south Asia in 2016, substantially higher than its 27% share of the child and adolescent population.

Open in a separate window

Figure 10

Trends in the number of children and adolescents with obesity and with moderate and severe underweight by region

Children and adolescents were aged 5–19 years. See appendix for results for adults. BMI=body-mass index.

The number of girls with obesity increased from 5 (95% CrI 1–14) million in 1975 to 50 (24–89) million in 2016. The number of boys with obesity increased from 6 (1–19) million in 1975 to 74 (39–125) million in 2016. 73% of the increase in the number of children and adolescents with obesity was due to increase in prevalence of obesity, 3% due to population growth and changes in age structure of the child and adolescent population, and another 24% due to the interaction of the two (appendix). The regions with the largest absolute increase in the number of children and adolescents with obesity were east Asia, the Middle East and north Africa, south Asia, and the high-income English-speaking region. The worldwide number of adult women with obesity increased from 69 (57–83) million in 1975 to 390 (363–418) million in 2016; the number of men with obesity increased from 31 (24–39) million in 1975 to 281 (257–307) million in 2016. An additional 213 million children and adolescents and 1·30 billion adults were in the overweight range, but below the threshold for obesity.

Discussion

Mean BMI and prevalence of obesity increased worldwide in children and adolescents from 1975 to 2016, with the rate of change in mean BMI moderately correlated with that of adults until around 2000, but only weakly correlated afterwards. The trend in children's and adolescents' mean BMI has plateaued, albeit at high levels, in many high-income countries since around 2000, but has accelerated in east, south, and southeast Asia. Despite this rise, more children and adolescents worldwide are moderately or severely underweight than obese. However, if post-2000 trends continue, child and adolescent obesity is expected to surpass moderate and severe underweight by 2022.

No prior global analyses of mean BMI and underweight have been done in children aged over 5 years and adolescents. For overweight and obesity, our results are not directly comparable with those of Ng and colleagues³¹ because the two studies covered different age ranges (2–19 years in Ng and colleagues'³¹ study compared with 5–19 years in our study), used different classification systems for defining overweight and obesity (WHO in our study versus International Obesity Task Force [IOTF], by Ng and colleagues³¹), and differed in criteria for including data (only measured height and weight in our study versus measured and self-reported by Ng and colleagues³¹). Nonetheless, both studies concluded that the rise in excess weight in children and adolescents has plateaued in high-income countries but continues in low-income and middle-income countries. The plateau in children's and adolescents' overweight and obesity in high-income countries^{32, 33, 34} and the relatively rapid transition from underweight to overweight and obesity in low-income and middle-income countries^{35, 36} have also been noted in specific countries.

Our study is the first to make comparable estimates of mean BMI and the prevalence of a complete set of BMI categories with clinical and public health relevance—from underweight to obesity. We used an unprecedented amount of population-based data from almost all of the world's countries, while maintaining a high standard of data quality and using only measured height and weight data to avoid the bias in self-reported data. Characteristics of data sources were verified through repeated checks by NCD-RisC members, and data that could be systematically different from the general population were excluded, eg, those from samples of students or of ever-married women in ages and countries with low school enrolment or marriage rates. Data were analysed according to a common protocol to obtain the required mean and prevalence by age and sex. Finally, we used a statistical model that used all available data while giving more weight to national data than subnational and community studies, and took into account the epidemiological features of outcomes such as BMI by using non-linear time trends and age associations, and differences between rural and urban populations.

Despite using the most comprehensive global database of human anthropometry to date, some countries and regions had fewer data sources, especially those in the Caribbean, Polynesia and Micronesia, Melanesia, and central Asia. The scarcity of data is reflected in wider uncertainty intervals of our estimates for these countries and regions. Of sources with data on children and adolescents, 39·9% had data for 5–9 years of age, 50·3% for 10–14 years, and 78·9% for 15–19 years. Many sources with data on children aged 5–9 years were school-based measurement studies in high-income countries where school enrolment is almost universal. The relative paucity of data on children aged 5–9 years restricted our capacity to compare trends in this age group with those of adolescents, despite some evidence from high-income countries that trends are somewhat different before and after 10 years of age.³³ Finally, although the age-dependent cutoffs for defining overweight and obesity in children and adolescents reflect natural growth in these ages, they are based on BMI distributions in a reference population, and not explicitly selected to represent optimal BMI levels for health in prospective studies, as done for adults, or optimal nutritional status, as done for children younger than 5 years. The reference population used by WHO,¹⁶ and the cutoffs for defining overweight and obesity, differ from those used by IOTF^{37, 38} and the US Centers for Disease Control and Prevention (CDC). Specifically, in the WHO classification, a BMI of 30 kg/m² at ages 18–19 years corresponds to 2 SD (ie, about the 97·5th percentile) from the median of the reference population;¹⁶ in the IOTF classification, a BMI of 30 kg/m² at age 18 years corresponds to the 98·6th percentile for girls and the 98·9th percentile for boys.³⁸ While at 18 years the two systems classify the same children as obese, at younger ages a smaller proportion are classified as obese according to the IOTF definition compared with the WHO definition.³⁸ For this reason, comparisons of overweight and obesity prevalence based on the three definitions^{36, 39, 40, 41, 42} found that prevalence using the WHO classification was higher than those of IOTF and CDC, but that trends are similar.

The effectiveness of interventions for overweight and obesity in children and adolescents has been reviewed in several systematic reviews and modelling studies,^{3, 43, 44, 45, 46, 47} but how they are selected for implementation and their post-implementation effects at the population level are rarely investigated.⁴⁸ For this reason, there is no systematic information on the determinants of the divergent trends in BMI in children and adolescents and in adults, be it on food environments and behaviours or on policies that affect them. The plateauing of children's and adolescents' BMI in high-income countries as adult BMI continues to increase might be due to specific initiatives by governments, community groups, schools, and notable individuals that have increased public awareness about overweight and obesity in children, leading to changes in nutrition and activity that are sufficient to curb the rise in mean BMI.

A general feature of policies that target overweight and obesity in children and adolescents in high-income countries is a reluctance to use taxes and industry regulations to change eating and drinking behaviours.^{12, 49} Some middle-income countries are also adopting policies to combat overweight and obesity in children and adolescents, in some cases with a stronger emphasis on regulation and taxes than in high-income countries.⁴⁸ While momentum might be gathering to use taxes and regulations to reduce the consumption of energy-dense foods, few policies and programmes attempt to make healthy foods such as whole grains and fresh fruits and vegetables more affordable through targeted price subsidies, (conditional) cash transfers and food vouchers, or healthy school meals.⁵⁰ Unaffordability of healthy food options not only leads to social inequalities in overweight and obesity,^{51, 52} but might also limit the effect of policies that target unhealthy foods. Finally, efforts in population-based prevention of overweight and obesity in children and adolescents should be matched with enhancing access to health-care interventions for weight management and for reducing the adverse effects of obesity, including intensive behavioural therapy to change diet and exercise; screening for and management of hypertension, glucose intolerance, dyslipidaemia, and abnormal liver function in children and adolescents with obesity; and in extreme cases bariatric surgery.^{4, 53, 54}

Our finding that the number of children and adolescents aged 5–19 years in the world who are moderately or severely underweight remains larger than those who are obese shows the continued need for policies that enhance food security in low-income countries and households, especially in south Asia. Yet the experiences of east Asia and Latin America and the Caribbean show that the transition from underweight to overweight and obesity can be rapid, and overwhelm the national capacity needed to engender a healthy transition. More broadly, in an unhealthy nutritional transition, an increase in nutrient-poor, energy-dense foods can lead to stunted growth along with weight gain in children, adolescents, and adults, resulting in higher BMI and worse health outcomes throughout the life-course. Therefore, the findings from our comprehensive analysis of trends in underweight, as well as overweight and obesity highlight the disconnect between the global dialogue on overweight and obesity, which has largely overlooked the remaining undernutrition burden, and the initiatives and donors focusing on undernutrition that have paid little attention to the looming burden of overweight and obesity, itself a risk factor for adverse pregnancy outcomes.² The Sustainable Development Goals, which address poverty, education, nutrition, and universal health coverage, provide an opportunity for integrating policies that coherently address underweight and overweight in children and adolescents, and their health consequences, effectively and equitably. Doing so would require commitment from national and international agencies and donors for replacing the fragmented focus with an integrated approach.

Acknowledgments

Contributors

Declaration of interests

Contributor Information

NCD Risk Factor Collaboration (NCD-RisC):

References