Excess body weight is an important risk factor for mortality and morbidity from cardiovascular diseases, diabetes, cancers, and musculoskeletal disorders, causing nearly three million annual deaths worldwide [1–5]. In previous work, we used a systematic analysis of data from population-based health examination surveys and epidemiologic studies to estimate trends in mean body mass index (BMI) between 1980 and 2008 in 199 countries and territories in 21 regions [6]. The results demonstrated the extent of the rise in BMI, and its similarities and differences, across countries and regions. This work, however, provided limited information on the prevalences of overweight and obesity, at the regional level and for two years, 1980 and 2008, only. Data on the prevalence of overweight and obesity by country and their trends are needed for two reasons: first, although epidemiologic studies have shown that ischemic heart disease, ischemic stroke, and diabetes are linearly associated with measures of excess body weight from low levels, some other outcomes, including hemorrhagic stroke, may have thresholds at higher levels [1, 3, 5]. Second, prevalence estimates are commonly used for public and policy communication of the extent of the obesity epidemic; hence there is great demand for such information. In this paper, we advance our previous work by estimating trends in overweight and obesity for all countries and regions.

The fitted regression equations between prevalence and mean population BMI had an R² of 0.97 for overweight and 0.92 for obesity (Figure 1; Table 1). Overweight and obesity prevalences increased over time for any mean population BMI, indicating that above and beyond increasing mean BMI, there is a widening of the BMI distributions. After accounting for mean BMI and other predictors, women had higher prevalence of obesity, an effect that was stronger at low mean BMIs. Our predicted age-standardized prevalences of overweight ranged from 2.6% in Vietnamese women in 1980 to 93.6% in Nauru men in 2008, and of obesity from 0.3% in Vietnamese men in 1980 to 74.8% in Nauru women in 2008 (Additional file 1).

Globally, the age-standardized prevalence of overweight increased from 24.6% (95% uncertainty interval 22.7-26.7%) in 1980 to 34.4% (33.2-35.5%) in 2008. The prevalence of obesity increased from 6.4% (5.7-7.2%) in 1980 to 12.0% (11.5-12.5%) in 2008. Half of this rise (2.8 percentage points absolute increase in prevalence) occurred in the twenty years between 1980 and 2000, when the global prevalence of obesity was 9.3% (9.1-9.6%); the other half occurred in the 8 years between 2000 and 2008. In absolute numbers, this represents an increase from 572 (527–621) million overweight adults in 1980 to 1.46 (1.41-1.51) billion in 2008. Of these, 508 (486–530) million were obese in 2008.

In 1980, the age-standardized prevalence of overweight among men ranged from 2.8% (1.4-5.1%) in Vietnam to over 50% in 10 countries: Andorra, Cook Islands, Czech Republic, Ireland, Lithuania, Malta, Nauru, Samoa, Slovenia, and Tonga (Figure 2; Additional files 2 and 3). Among women, overweight prevalence was also lowest in Vietnam (2.6%; 1.1-5.0%) and was under 5% in seven other countries in South and Southeast Asia and Central and Eastern Africa. It was over 50% in 29 countries in Eastern, Central, and Western Europe, the Caribbean, Oceania, North Africa, the Middle East, and Southern Africa. By 1990, more than 50% of women were overweight in 47 countries, a number that increased to 74 in 2000 and to 101 in 2008. By 2008, the lowest prevalence of overweight among women was 8.0% (5.1-11.7%) in Bangladesh. Female overweight prevalence was also below 10% in Ethiopia, Madagascar, and Nepal; at the other extreme, it reached over 90% in Cook Islands, Nauru, and Tonga. Male overweight in 2008 ranged from 5.8% (1.7-13.3%) in the Democratic Republic of the Congo to over 90% in the Cook Islands and Nauru (these countries also had prevalences of overweight greater than 90% without age standardization). More than half of men were overweight in 100 countries in 2008. In both 1980 and 2008, the correlation between the male and female prevalence of overweight was 0.85.

In 1980, the age-standardized obesity prevalence in men ranged from 0.3% (0.1-0.5%) in Vietnam to 27.5% (10.3-44.5%) in Nauru (Figure 2; Additional files 2 and 4). Female obesity was as low as 0.3% (0.1-0.7%) in Vietnam, reaching 30.6% (17.2-43.5%) in Kuwait and 32.0% (11.2-52.3%) in Nauru. Male obesity was below 1% in 20 countries and female obesity was below 1% in 11 countries in South and Southeast Asia and Sub-Saharan Africa. By 1990, the prevalence of obesity among women was greater than 25% in 26 countries in Oceania, North Africa and Middle East, the Caribbean, and also in Moldova, the Netherlands, and South Africa; by 2008, this number increased to 80 countries. By 2000, obesity in women already exceeded 50% in four countries in Oceania; these four countries were joined by Kuwait and five more countries in Oceania by 2008. In 2008, female obesity ranged from 1.4% (0.7-2.2%) in Bangladesh and 1.5% (0.9-2.4%) in Madagascar to 70.4% (61.9-78.9%) in Tonga and 74.8% (66.7-82.1%) in Nauru. In 2008, the prevalence of male obesity was below 1% in three countries: Democratic Republic of the Congo, Ethiopia and Bangladesh. More than one half of men were obese in Cook Islands (60.1%, 52.6-67.6%) and Nauru (67.9%, 60.5-75%). Between 1980 and 2008, the correlation between the male and female prevalence of obesity increased slightly from 0.82 to 0.86.

In 1980, one half of the 572 million adults in the world with BMI ≥ 25 kg/m² lived in China (72 million), the United States (70 million), and five other countries (Russia, India, Germany, Italy, and Ukraine). In 2008, the countries with the most overweight people were China (241 million) and the United States (158 million). Half of the world's 1.46 billion overweight people were living in these two countries and seven others: India, Russia, Brazil, Mexico, Germany, Indonesia, and Turkey. The largest absolute increase in the number of overweight people was in China (169 million in 28 years) and the United States (88 million). The largest absolute increase in number of obese people occurred in these two countries (56 million in the United States and 42 million in China), followed by Brazil (20 million) and Mexico (18 million).

The regions with the lowest prevalence of obesity in 1980 were South and Southeast Asia, East, Central, and West Sub-Saharan Africa, with prevalences ranging from 1.1% to 1.9% (Figure 3). Central and Eastern Europe had the highest prevalence of obesity (16.3-18.2%) in 1980. Regional patterns of overweight were similar (Figure 4). By 1990, North America was also among the regions with the highest prevalence of obesity (17.6%, 15.9-19.4%), surpassing other regions by 2000 when obesity prevalence reached 24.9% (23.5-26.3%). Due to its large increase in obesity, Southern Latin America also joined regions with high prevalence of obesity by 2000 (22.2%, 19.8-24.6%). In 2008, the regions with the highest obesity prevalence were North Africa and Middle East, Central and Southern Latin America, Southern Sub-Saharan Africa, and high-income North America, with prevalences ranging from 27.4% to 31.1%. By 2008, South and Southeast Asia and Central and East Sub-Saharan Africa still had some of the lowest prevalences of obesity, as did high-income Asia Pacific, with prevalences ranging from 2.2% to 5.4%. On the other hand, Western Sub-Saharan Africa, which had remained a low-obesity region until 2000, experienced a relatively large rise in BMI in the next 8 years, and hence no longer was a region with low prevalence in 2008.

In 16 of 21 regions, the prevalence of female obesity was greater than that of men in every year from 1980 to 2008. In high-income Asia-Pacific, Australasia, Western Europe, Southern Latin America, and Central Sub-Saharan Africa, male obesity exceeded female obesity in some years, usually by a small margin. Male and female obesity prevalence differed most in Southern Sub-Saharan Africa, where 18.7% (15.6-21.5%) of men were obese vs. 36.7% (33.0%-40.2%) of women in 2008.

The prevalence of female obesity increased the most in some countries in Oceania, with the largest increase in Cook Islands (16.8 percentage points per decade, 9.4-24.1, pp > 0.99) and Tonga (16.0 percentage points, 8.8-23.1, pp > 0.99) (Figures 5, 6, 7, 8, 9). The prevalence of obesity increased more than 10 percentage points per decade in 16 countries in Oceania and the Caribbean, and in Egypt. A decline in the prevalence of obesity was estimated for women in 10 countries: Lithuania, Tajikistan, Estonia, Zimbabwe, Serbia, Democratic People's Republic of Korea, Montenegro, Singapore, Ukraine, and Romania; none were statistically distinguishable from no trend (pp ranged from 0.50-0.68). Male obesity increased more than 10 percentage points per decade in seven countries in Oceania, most notably in Nauru (14.5 percentage points per decade, 7.6-22.0, pp > 0.99) and Cook Islands (15.5 percentage points per decade, 9.9-20.7, pp > 0.99). A decline in the prevalence of obesity was estimated for men in seven countries: Tajikistan, Moldova, Democratic Republic of the Congo, Afghanistan, Central African Republic, Nepal, and Bangladesh; none were statistically distinguishable from no trend (pp ranged from 0.41-0.75).

The regions with the largest increase in the prevalence of female obesity were Central Latin America (8.5 percentage points per decade, 5.3-11.2, pp > 0.99), Oceania (8.2 percentage points per decade, 5.2-11.1, pp > 0.99), and Southern Latin America (8.2 percentage points per decade, 4.1-12.0, pp > 0.99); female obesity increased in every region (pp ranging from 0.79 in Central Asia to > 0.99 in 15 regions). The regions with the largest increase in male obesity were North America (6.9 percentage points per decade, 5.7-8.1, pp > 0.99) and Australasia (5.9 percentage points per decade, 4.6-7.2, pp > 0.99), while the regions with the smallest increases were Central Sub-Saharan Africa and South Asia, both of which had trends that were statistically indistinguishable from zero (pp ranged from 0.43-0.66). The absolute annual increase in obesity prevalence was greater during 2000 to 2008 than the 1980s or the 1990s in all 21 regions. In nine regions (Central and South Asia, Central and Eastern Europe, Tropical Latin America, and East, West, Central, and Southern Sub-Saharan Africa), more than half of the rise in obesity over the period 1980 to 2008 occurred in 2000 to 2008.

Diverging trends in obesity were observed among high-income regions, with larger increases in Australasia and North America and smaller increases in the high-income Asia Pacific and Western European regions, especially for females (Figure 3). Female obesity increased 7.6 (5.8-9.3) percentage points per decade in high-income North America and 6.8 (5.1-8.3) percentage points per decade in Australasia, vs. 2.7 (1.4-3.9) and 1.0 (0.4-1.5) percentage points per decade in Western Europe and high-income Asia-Pacific, respectively. Likewise, male obesity increased 6.9 (5.7-8.1) percentage points per decade in high-income North America and 5.9 (4.6-7.2) percentage points per decade in Australasia, vs. 4.0 (3.0-5.0) and 1.5 (1.1-1.9) percentage points per decade in Western Europe and high-income Asia-Pacific, respectively. Based on these trends and on 1980 levels, in 2008 the prevalence of obesity was higher in high-income North America (31.1%, 28.9-33.3%) and Australasia (25.3%, 23.1-27.6%), than in Western Europe (20.0%, 18.5%-21.6%) and high-income Asia-Pacific (5.4%, 4.6-6.2%).

While overweight and obesity prevalence generally increased worldwide, there were differences between men and women in some regions and countries (Figures 5-9). For example, the prevalence of female obesity increased more than the prevalence of male obesity in 46 of 48 countries in Sub-Saharan Africa and in 35 of 37 countries in Latin America and the Caribbean. In Central Sub-Saharan Africa and Southern Latin America, this resulted in women having a higher prevalence of obesity in 2008 than men, in contrast to their 1980 orders. In contrast, in Western, Central, and Eastern Europe, and in the high-income Asia Pacific countries, the prevalence of obesity increased more among men than women in 41 of 47 countries.

One in every 3 adults in the world was overweight and 1 in every 9 was obese in 2008. Beyond this global average, at least 1 in 5 women were obese in 117 countries and at least 1 in 5 men were obese in 73 countries. Notably, the increase in the prevalence of obesity has accelerated in the last decade compared to the 1980s and 1990s.

The rising obesity and overweight are clearly associated with increases in mean BMI. However, we found that the relationship between mean and prevalence was not linear: BMI increase at low BMI levels results in a smaller increase in obesity than an identical absolute increase at higher BMIs (Figure 1). The relationship between obesity and mean BMI also changed over time: in 2008, the prevalence of obesity was greater at all mean BMIs than it had been in earlier decades, indicating that the population BMI distributions is likely to have become wider over time. If recent trends in mean BMI continue [6], the prevalence of obesity will continue to rise at the recent alarmingly high rates.

This study is the first report of adult overweight and obesity prevalence by country, year, and sex. The strengths and innovations of this study include analysis of long-term trends; large amount of high-quality measured population-based data accessed and used; the use of a Bayesian hierarchical model for estimating mean BMI, incorporating non-linear time trends; systematic conversion of mean BMI to overweight and obesity prevalence, including adjusting for age, sex, region, and time; and systematic analysis of uncertainty. The main limitation of our study is that data gaps remained despite our extensive data seeking, especially in the 1980s and for men through the 1990s. We did not model the full population distribution of BMI by country, which should be a topic for future research. Our analysis did not consider trends in central adiposity due to a lack of population-based data, nor did it quantify within-country disparities by socioeconomic status or race [9–11]. Finally, we did not estimate childhood and adolescent adiposity. Child and adolescent overweight and obesity are important because childhood weight gain may have larger adverse effects than weight gain during adulthood due to the longer exposure, and because they are predictive of future trends in overweight and obesity in adults. Measuring overweight and obesity in school-aged children and adolescents is challenging for several reasons: first, there is an age gap in survey data as most surveys on maternal and child health include data up to 5 years of age and those related to noncommunicable disease risk factors and behaviors target adults; second, until the 2007 introduction of WHO growth standards, there was little consensus on definitions of overweight and obesity in children aged 5 years to 19 years. This has resulted in incomparable measures, available in the sparse published literature.

The United Nations General Assembly High-Level Meeting on the Prevention and Control of Noncommunicable Diseases in September 2011 requested that WHO develop targets for key noncommunicable disease indicators, which may include the prevalence of obesity. Data on historical trends allows setting ambitious, yet feasible targets [12]. Our estimates provide a consistent baseline for such empirical targets. The tenth best performing percentile of countries, ranked from lowest to highest absolute increase in obesity, experienced an increase of 0.6 percentage points per decade from 1980 to 2008. The relative change at the tenth percentile was around a 9% increase over 15 years for women and 25% for men. We estimated nonsignificant change (at the pp=0.75 level) in obesity prevalence in 15 countries, while the remainder experienced a statistically significant increase in obesity. Most of these countries maintained low levels of obesity prevalence through food scarcity rather than effective policy action, and thus their experience may not be the appropriate baseline for policies that aim to halt the rise in obesity. Therefore, given the widespread rise in obesity, a key question is how ambitious an obesity target should be relative to the historical experience.

Disentangling the contributions of declines in physical activity, changes in calorie consumption and composition, and other factors to the recent increases in mean BMI and overweight and obesity prevalence are currently active areas of research [13, 14]. The data presented in this paper can be used to generate and evaluate hypotheses about these trends via ecological analyses that explore the relationship between trends in overweight/obesity prevalence and in their possible causes. However, we must also carry out rigorous intervention studies to test programs and actions that may curb and reverse these trends [15].

Our systematic analysis of population-based data sources helped unpack the “global obesity pandemic” into its constituent trends by country, region, and sex. We found that, although the magnitude of the rise in obesity varies by region, country, and gender, stability in obesity prevalence was rare. Moreover, the rise in obesity has accelerated in the last decade. This documentation of country trends in obesity prevalence may be used to set targets as requested by the United Nations General Assembly's High-Level Meeting on the Prevention and Control of Noncommunicable Diseases.

Writing and Global Analysis Group: Gretchen A Stevens, Gitanjali M Singh, Yuan Lu, Goodarz Danaei, John Lin, Mariel M Finucane, Adil N Bahalim, Russell K McIntire, Hialy R Gutierrez, Melanie Cowan, Christopher J Paciorek, Leanne Riley, Majid Ezzati.

Country data group: Geir Aamodt; Ziad Abdeen; Nabila A Abdella; Hanan F Abdul-Rahim; Juliet Addo; Mohamed M Ali; Mohannad Al-Nsour; Ramachandran Ambady; Pertti Aro; Carlo M Barbagallo; Alberto Barceló; Henrique Barros; Leonelo E Bautista; Peter Bjerregaard; Enzo Bonora; Pascal Bovet; Grazyna Broda; Ian J Brown; Michael Bursztyn; Antonio Cabrera de León; Francesco P Cappuccio; Katia Castetbon; Somnath Chatterji; Zhengming Chen; Chien-Jen Chen; Lily Chua; Renata Cífková; Linda J Cobiac; Anna Maria Corsi; Cora L Craig; Saeed Dastgiri; Martha S de Sereday; Gonul Dinc; Yasufumi Doi; Eleonora Dorsi; Nico Dragano; Adam Drewnowski; Paul Elliott; Anders Engeland; Alireza Esteghamati; Jian-Gao Fan; Catterina Ferreccio; Nélida S Fornés; Flávio D Fuchs; Simona Giampaoli; Sidsel Graff-Iversen; Janet F Grant; Ramiro Guerrero Carvajal; Martin C Gulliford; Rajeev Gupta; Prakash C Gupta; Oye Gureje; Noor Heim; Joachim Heinrich; Tomas Hemmingsson; Victor M Herrera; Suzanne C Ho; Michelle Holdsworth; Wilma M Hopman; Abdullatif Husseini; Nayu Ikeda; Bjarne K Jacobsen; Tazeen H Jafar; Mohsen Janghorbani; Grazyna Jasienska; Michel R Joffres; Jost B Jonas; Ofra Kalter-Leibovici; Ioannis Karalis; Joanne Katz; Lital Keinan-Boker; Paul Kelly; Omid Khalilzadeh; Young-Ho Khang; Stefan Kiechl; Maressa P Krause; Yadlapalli S Kusuma; Arnulf Langhammer; Jeannette Lee; Claire Lévy-Marchal; Yanping Li; Yuqiu Li; Stephen Lim; Cheng-Chieh Lin; Lauren Lissner; Patricio Lopez-Jaramillo; Roberto Lorbeer; Guansheng Ma; Stefan Ma; Francesc Macià; Dianna J Magliano; Marcia Makdisse; Roberto Miccoli; Juhani Miettola; Jaime Miranda; Mostafa K Mohamed; V Mohan; Salim Mohanna; Ali Mokdad; Dante D Morales; Lorenza M Muiesan; Iraj Nabipour; Vinay Nangia; Barbara Nemesure; Martin Neovius; Kjersti A Nerhus; Flavio Nervi; Hannelore Neuhauser; Minh Nguyen; Ayse E Önal; Altan Onat; Myriam Orostegui; Hermann Ouedraogo; Demosthenes B Panagiotakos; Francesco Panza; Yongsoo Park; Mangesh S Pednekar; Marco A Peres; Rafael Pichardo; Hwee Pin Phua; Francesco Pistelli; Pedro Plans; Dorairaj Prabhakaran; Roaeid B Ragab; Olli T Raitkari; Sanjay Rampal; Finn Rasmussen; Josep Redon; Luis Revilla; Victoria Reyes-García; Fernando Rodriguez-Artalejo; Luis Rosero-Bixby; Harshpal S Sachdev; José R Sánchez; Selim Y Sanisoglu; Norberto Schapochnik; Lluís Serra-Majem; Rahman Shiri; Xiao Ou Shu; Leon A Simons; Margaret Smith; Vincenzo Solfrizzi; Emily Sonestedt; Pär Stattin; Aryeh D Stein; George S Stergiou; Jochanan Stessman; Akihiro Sudo; Valter Sundh; Kristina Sundquist; Johan Sundström; Martin Tobias; Liv E Torheim; Josep A Tur; Ana I Uhernik; Flora A Ukoli; Mark P Vanderpump; Jose Javier Varo; Marit B Veierød; Gustavo Velásquez-Meléndez; Monique Verschuren; Salvador Villalpando; Jesus Vioque; Peter Vollenweider; Mark Ward; Sarwono Waspadji; Johann Willeit; Mark Woodward; Liang Xu; Fei Xu; Gonghuan Yang; Li-Chia Yeh; Jin-Sang Yoon; Qisheng You; Wei Zheng.