Discussion

In this systematic review, we identified 151 studies reporting rates of excess weight in Brazilian adolescents, with data collected from 1974 to 2018. Prevalence of excess weight ranged from 2·2 % to 44·4 %, and an increase could be seen when comparing information from more recent to older studies. Populations were included from household surveys, birth cohorts and school-based samples, and only twelve studies included individuals from more than one macroregion in Brazil.

Individual cross-sectional studies have hinted the rising trends of overweight and obesity among this age group in different countries^{(Reference Ogden, Carroll and Lawman18,Reference Wang and Lobstein19)} , but representative data of trends about the prevalence of overweight and obesity among Brazilian adolescents were limited. To fill this gap, we performed a systematic review focused on adolescents from all regions in Brazil, and we show separate data by decades of data collection. In general, we observed an increase in the prevalence of overweight and obesity among Brazilian adolescents since the 70s. The high statistical heterogeneity found was expected, as previous meta-analyses of prevalence studies showed^{(Reference Maria Aiello, Marques de Mello and Souza Nunes12,Reference Wong, Huang and Wang20)} . In our analysis, the heterogeneity can be due to differences between samples of the studies in many characteristics, such as age (some included adolescents from all ages, others included only those with 10 to 13 years or 17 and older), ethnic background (some studies were made with indigenous populations), criteria for excess weight (five different classifications were observed) and type of study (school-based, household survey or cohort).

In the last decades, Brazil and other middle-income countries have experienced a quick transition in food availability and eating habits. Until recently, the most prevalent nutritional problem in Brazilian children and adolescents was underweight/undernutrition, especially in the poorer and less developed regions of the country^{(Reference Monteiro, Benicio and Konno21)}. However, due to a rapid industrial expansion and changes in lifestyle (such as increase in sedentary time^{(Reference Schaan, Cureau and Bloch22)} and consumption of ultra-processed foods^{(Reference Souza, Barufaldi and Abreu23)}), a double disease burden can be seen – undernutrition is still present in many areas, but obesity-related health problems are higher than ever^{(Reference Conde and Monteiro24)}. Countries like India and China, which also underwent significant socio-economic changes in a small amount of time, have similar growing trends in excess weight in youth. From the 80s to the last 10 years, for an example, overweight rates increased from 1·8 % to over 13 % in Chinese children and adolescents^{(Reference Yu, Han and Chu25)}.

In this systematic review, we present an overall prevalence of excess weight from studies that have adopted different criteria to classify adolescents with overweight and obesity, one of them specifically made using only Brazilian children^{(Reference Conde and Monteiro26)}. Classification based on the article by Must et al^{(Reference Must, Dallal and Dietz27)} was the least used. Classification based on WHO references^{(Reference de Onis, Onyango and Borghi28)} was the most commonly used, especially in studies published since 2009. Criteria by IOTF for overweight/obesity^{(Reference Cole, Bellizzi and Flegal29)} was the preferred before year 2000. It is possible to consider that adding prevalence from diverse cutoffs is a limitation and could lead to imprecise results; however, the prevalence for overweight/obesity still exponentially increased in last decades when we analyze separately studies that used WHO reference curves (4.4 % until year 2000 to 26.6 % in the 2010s) and IOTF criteria (9.2 % to 21.7 %), despite including a smaller number of articles.

In subgroup analysis, the prevalence of excess weight varied greatly among macroregions, reflecting the health inequalities between them. In the literature, many studies show that associations between socio-economic status (SES) and rates of obesity depend on local characteristics – people with higher SES are less likely to be overweight or obese in high-income countries, but more likely in lower-income countries^{(Reference McLaren30)}. In a continental sized country such as Brazil, it is important to address how significant differences among regions can influence nutritional patterns. In our study, we found a positive association between SES and rates of excess weight comparing regions – prevalence was lower in the Northeast region across all decades and higher in the South and Southeast regions in the last decade. States from the Northeast region have the lowest Human Development Index (HDI) in the country, and nutritional transition at these places is relatively recent or concurrent, with underweight still being an important public health concern. Conversely, at the South and Southeast regions, comprising states that have the highest HDI of the country, these changes were observed decades ago, and ultra-processed foods are more easily accessible^{(Reference Conde and Monteiro24,Reference Batista Filho and Rissin31)} . This dietary pattern has been previously reported in adolescents – in the ERICA study, higher SES was associated with greater consumption of unhealthy foods, such as sugary drinks and snacks^{(Reference de Alves, de Souza and Barufaldi32)}.

The prevalence of overweight/obesity in this study was similar and increased across decades in both sexes; however, girls had higher rates of excess weight before year 2000 (11·9 % v. 4·6 %). This can be partially explained due to the fact that, in this period, two big studies included only boys at older age (17–19 years) from the Brazilian Army database. Similar results by sex are observed in a recent country-wide survey in Brazil^{(Reference Bloch, Klein and Szklo33)}, supporting that prevention polices against obesity should be conducted independently of sex. Finally, analyses comparing group ages showed that younger adolescents had higher prevalence of excess weight compared to their older counterparts. This can be directly related to puberty and its effect on acquisition of fat-free mass, reaching the highest levels at peak height growth velocity^{(Reference de Oliveira, da Silva and Oliveira34)}.

Treatment of obesity in a health system usually involves recommendation of lifestyle modifications (increase physical activity, decrease sugar and fat consumption), pharmacological therapy and, in selected cases, bariatric surgery. Nevertheless, obesity rates continue to grow alarmingly worldwide. This can be attributed to the fact that the ‘pressure’ from the obesogenic environment is still the same, and therefore interventions should also try to change socio-economic, political and cultural context involving excess weight. Efforts such as clear nutritional labelling in packages or decrease in percentual of fats are commonly accepted by decision-makers and food industry. However, regulation of food advertising, taxing or banning energy-dense and nutrition-poor foods sold in schools find a barrier for implementation, mostly due to commercial interests^{(Reference Dias, Henriques and Anjos35)}. Actions for prevention of obesity during childhood also involve the same modifications in public policies. Only when obesity is approached in an interdisciplinary way beyond the health sector, we will have a real chance of altering its repercussions later in life.

Nonetheless, it is important to consider regional access to healthcare services and economical power when defining prevention strategies, as they vary greatly among Brazilian regions. Populations from the North and Northeast regions, especially those with lower SES, have more difficulty in reaching the healthcare system, and when they do, resources are scarce^{(Reference Uchimura, Felisberto and Fusaro36)}. Additionally, access to healthy foods, such as vegetables and fruits, can be limited, as they have higher prices compared to ultra-processed foods. In contrast, adolescents from the South and Southeast regions have higher prevalence of excessive screen^{(Reference Schaan, Cureau and Bloch22)} and sitting time^{(Reference Werneck, Oyeyemi and Fernandes37)}, as well as higher access to physical education classes. These inequity patterns show that there is no simple or single solution to tackle this health problem in Brazil. Public policies need to be specific for each region, considering their strengths and weaknesses, in order to be effective. In the South and Southeast, the higher accessibility to infrastructure (sports courts, tracks and swimming pools) can be used to diminish sedentary behaviour and to promote physical activity, while in the North and Northeast it is essential to improve nutritional composition and overall access to healthcare services.

Our study has some limitations. The number of studies from before the 2000s was much smaller than those from other decades, and more than half were composed by adolescents from the Southeast region of Brazil; therefore, the prevalence from this period may not be accurate. There was also a low representation of individuals from the North and Midwest regions, especially before 2010. This poor coverage of epidemiological changes in some Brazilian regions restricts the evaluation of national prevalence of excess weight over time. Further studies in these macroregions are required to correctly demonstrate racial, cultural and socio-economic diversity of this nation. Differences and changes in diagnosis criteria over time among the studies may also limit the interpretation of our results. High heterogeneity was found in all analyses; however, this is expected when gathering results from more than a hundred studies that used multiple criteria and included individuals from different regions and age groups.