Adolescence is one of the most important human life stages, characterized by rapid physical, intellectual, and psychological development. Globally, more than 1.2 billion or 16% of the population are adolescents1. However, adolescents often face many challenges. Interpersonal violence is one of the main challenges faced by adolescents and is a serious public health issue worldwide2. It is estimated that interpersonal violence in adolescents contributes to 15.8% and 30.7% of the total disability-adjusted life years and total years of life lost, respectively3.

More than one billion adolescents experience multiple forms of violence4. The high prevalence of violence during adolescence can cause short- and long-term physical, psychological, and behavioral problems, such as injury, mental health disorders, sleep disorders, self-harm and suicidal thoughts, problems with relationships, poor school performance and dropout, and noncommunicable diseases4. Physical fight and physical attack are two of the most common types of interpersonal violence among adolescents. Potential factors that contribute to physical fight and physical attack have attracted considerable attention from researchers5. For example, physical violence has been found to be associated with age, having no parental supervision, having no helpful classmates, having no close friends, and feelings of loneliness6,7,8, as well as sedentary behavior and physical inactivity6.

Food insecurity is a household-level social and economic condition characterized by limited or uncertain access to adequate food and is associated with adolescent developmental problems9. In a pooled study of 95 countries10, more than one in every four adolescents reported exposure to food insecurity in the past 30 days. Adolescents experiencing food shortages are likely to exhibit greater substance use10, poor educational attainment10, greater truancy10, mental health problems11,12, and suicidal behaviors13. However, the relationship between food insecurity and violent behaviors in adolescents is controversial. An association between food insecurity and physical fight has been observed in certain countries, such as Kuwait14, Chile15, and Namibia8, whereas no such association has been found in Egypt6, Pakistan7, or Paraguay16. A systematic review conducted in 2022 demonstrated that food insecurity was significantly associated with an increased likelihood of engaging in fights at school among high school-aged youth17. A study in 78 countries conducted by Probst et al. revealed a pooled significant association between each level increase in food insecurity and intentional injuries due to interpersonal violence, with an odds ratio of 1.30 for boys and 1.53 for girls18. However, the associations between food insecurity and violent behaviors in Probst’s study were adjusted for only age and many factors were still associated with violent behaviors among adolescents.

Although some previous studies have investigated this topic, gaps still exist in our understanding of the influence of food insecurity on violent behaviors among in-school adolescents. This study aimed to examine the adjusted associations between food insecurity and violent behaviors among in-school adolescents globally and to explore the differences in the associations by sex, income, and region.

Methods

Survey information

Data on the participating countries and adolescents were obtained from the Global School-Based Student Health Survey (GSHS), a multi-country, cross-sectional, school-based survey jointly developed by the World Health Organization (WHO) and the United States Centers for Disease Control and Prevention. The GSHS conducted at least one survey in each country. In each participating country, a standardized two-stage probability sampling design was used for the survey. First, schools were randomly selected via the probability proportionate to size method. Second, several classrooms in the school that included high proportions of students of the target age range were selected by systematic equal-probability sampling. All in-school students in the selected classes were included in the survey. The questionnaire data obtained from the GSHS can be directly compared between countries because of the use of standardized questionnaire modules across countries. The specific questionnaires used across the countries are available on the WHO website (http://www.who.int/chp/gshs/en). The questionnaire was translated into each country’s local language and completed by the included adolescents on a computer during regular class time. All surveys in the GSHS were approved by the Ministry of Health or Education and relevant ethics committees in each country. Adolescents voluntarily participated in the survey, and students and their parents or guardians provided verbal or written consent. As we used publicly available GSHS data, we did not need separate ethical approval for this study.

All of the nationally representative datasets from the GSHS that included the variables of interest for this study were extracted. The most recent survey data were chosen if a country had participated in more than two surveys. We excluded surveys conducted before 2009, as a new version of the questionnaire, which included a module on unintentional injuries and violence, was introduced in 2009. Consequently, many surveys conducted before this time did not incorporate this module. Thus, a total of 66 countries surveyed from 2009 to 2018 were included in this study (nine from the African region, 12 from the Eastern Mediterranean region, 22 from the Americas, 16 from the Western Pacific region, and seven from Southeast Asia) (Supplementary Fig. 1). The basic characteristics of each included country are shown in Table 1. Based on the World Bank classification, the included countries were divided into seven low-income countries, 20 lower-middle-income countries, 19 upper-middle-income countries, and 18 high-income countries; no income data were available for two countries.

Table 1 Survey characteristics by country.
Full size table

Dependent variables

The dependent variable, physical fight, was evaluated using the question “During the past 12 months, how many times were you in a physical fight?”, with response options of “0 times,” “1 time,” “2 or 3 times,” “4 or 5 times,” “6 or 7 times,” “8 or 9 times,” “10 or 11 times,” and “12 or more times.” We considered that students had not been in a physical fight if their responses were “0 times” and “1 time.” Students who reported being in two or more fights were classified as having participated in a physical fight6, to facilitate the identification of problematic fighting behavior. The variable physical attack was evaluated using the question “During the past 12 months, how many times were you physically attacked?”, and the response options were the same as those for the question for physical fight. We considered that students had not experienced a physical attack if their response was “0 times.” Students who reported being in one or more attacks were classified as having experienced a physical attack5.

Independent variables

The independent variable food insecurity was evaluated using the question “How often did you go hungry because there was not enough food in your home during the past 30 days?”, with response options of “never,” “rarely,” “sometimes,” “most of the time,” and “always.” We classified the response “never” as indicating no food insecurity, “rarely” and “sometimes” as indicating moderate food insecurity, and “most of the time” and “always” as indicating severe food insecurity13.

Covariates

In addition, 10 variables at the individual level (age, sex, bullying victimization, loneliness, physical activity, truancy, drinking status, sedentariness, fruit and vegetable intake, and soft drink intake)6,7,8 and two at the social level (having close friends and the presence of parental care)6,7,8 were investigated as other independent variables in our analysis. Definitions of these independent variables6,7,8 are provided in Supplementary Table 1.

Statistical analysis

Owing to the complex sampling design used in the GSHS, we added weights, strata, and a primary sampling unit to reflect the weighting process and the two-stage sampling design. We applied the SAS software PROC SURVEYMEANS procedure to calculate the weighted prevalence rates of food insecurity, physical fight, and physical attack. The pooled estimates of prevalence between countries were calculated via a random-effects meta-analysis using STATA. The PROC SURVEYLOGISTIC procedure was used to explore the country-specific associations of food insecurity with physical fights and physical attacks using odds ratios (ORs) and their 95% confidence intervals (CIs), with adjustments for age, sex, bullying victimization, loneliness, physical activity, truancy, drinking status, sedentariness, fruit and vegetable intake, soft drink intake, having close friends, and the presence of parental care. Then, we pooled these associations by regions, World Bank income levels and overall via a random/fixed-effects meta-analysis. Also, the country-specifc adjusted associations in boys and girls were calculated, and then pooled associations were generated using a random-effect meta-analysis according to sex, the Z test was used to determine the differences by sex. The Higgins I2 statistic was used to estimate between-country heterogeneity. An I2 value > 25% was considered to indicate moderate to severe heterogeneity, and random-effects models were used for the meta-analysis. Sensitivity analyses were conducted to determine the stability of the pooled estimates after excluding each country. SAS version 9.4 (SAS Institute, Cary, NC, USA) and STATA version 12.0 (Stata Corporation, College Station, TX) were used to perform statistical analyses. A two-sided p value < 0.05 was considered to indicate a statistically significant difference.

Results

A total of 234,216 adolescents aged 12–17 years from 66 countries in five regions were included in our study, and the overall response rate was 98.98% (Table 1). The overall prevalence of exposure to moderate food insecurity in the past 30 days was 40.2%, ranging from 18.8% in Costa Rica to 72.6% in the Solomon Islands (Supplementary Tables 2 and Fig. 1). The overall prevalence of exposure to severe food insecurity in the past 30 days was 7.1%, ranging from 1.0% in Viet Nam to 19.8% in Afghanistan (Supplementary Tables 2 and Fig. 1). The overall prevalence of physical fight and physical attack among adolescents was 18.9% and 31.8%, respectively (Supplementary Table 3). The prevalence of moderate and severe food insecurity, physical fights, and physical attacks also varied by gender, region, and income (Table 2).

Table 2 Prevalence of food insecurity, psychical fighting, and psychical attack by sex, region and income level.
Full size table
Fig. 1
figure 1

The prevalence of food insecurity (FI), physical fighting (PF) and physical attack (PA) by country.

Full size image

According to the results of the multivariable logistic regression models, after adjusting for covariates, exposure to moderate food insecurity in the past 30 days was associated with physical fights and physical attacks, with odds ratios (ORs) and 95% confidence intervals (CIs) of 1.18 (1.12–1.23) and 1.49 (1.43–1.56), respectively, and with I2 values of 43.2% and 59.6%, respectively (Figs. 2 and 3 and Supplementary Figs. 2–3). Severe food insecurity was associated with physical fights and physical attacks, with ORs and 95% CIs of 1.52 (1.39–1.66) and 1.88 (1.74–2.03), respectively, with moderate heterogeneity (Figs. 2 and 3 and Supplementary Figs. 4–5).

Fig. 2
figure 2

The adjusted association of food insecurity with physical fighting by sex, region and income level. Association adjusted for age, sex, bullying victimization, loneliness, physical activity, truancy, drinking status, sedentariness, fruit and vegetable intake, soft drink intake, having close friends, and the presence of parental care.

Full size image
Fig. 3
figure 3

The adjusted association of food insecurity with physical attack by sex, region and income level.

Full size image

Similarly, the associations between food insecurity and violent behaviors were significant in both boys and girls, but the strength of the association was greater in girls than in boys (all p < 0.05) (Figs. 2 and 3). The associations between food insecurity and violent behaviors varied across regions and income subgroups. The sensitivity analysis revealed robust results for the associations (data not shown).

Discussion

Given the contribution of food insecurity to violent behaviors in adolescents, this study is the first to use global data from the GSHS to investigate the associations of food insecurity with physical fights and physical attacks among adolescents. Overall, the study revealed that many adolescents worldwide had experienced starvation in the past 30 days and that 40.2% and 7.1% of adolescents were exposed to moderate and severe food insecurity, respectively. Those adolescents who reported ever being exposed to moderate food insecurity in the past 30 days had significantly higher odds of being in a physical fight (1.18 times) and experiencing a physical attack (1.49 times) than those who reported never being exposed to food insecurity, whereas severe food insecurity exposure was associated with 1.52 times and 1.88 times higher odds of being in a physical fight and experiencing a physical attack, respectively.

Approximately one in two adolescents reported moderate to severe food insecurity, which suggests a great challenge in attaining the United Nations’ Sustainable Development Goals, especially Goal 2 (“By 2030, end hunger and ensure access by all people, in particular the poor and people in vulnerable situations, including infants, to safe, nutritious and sufficient food all year round”)19. Thus, urgent action and policy solutions are imperative to reduce food insecurity in adolescents worldwide, especially in response to the sequential impacts of regional conflict and the pandemic on global food security. In addition, general strategies (e.g., investing in rural infrastructure and increasing agricultural productivity) can be used to reduce food insecurity in the population19. Governments are also advised to enact the School Feeding Law and set up a sustainable School Feeding Network, such as that established in Brazil (Brazil’s National School Feeding Program), to mandate that public schools nationwide provide all students with nutritious meals.

A large transnational variation in the prevalence of violent behavior in adolescents was observed in our study (I2 > 95%). Although the reasons for the difference between countries are unknown, they may be related to differences in socioeconomic, cultural, and religious factors and substance use20,21. For example, the prevalence of risky behaviors, such as alcohol consumption, may be relatively low in Muslim countries (e.g., Indonesia and Bangladesh), where such behaviors are discouraged, and thus, violence resulting from alcohol consumption may also occur less in such countries than in other countries22. Consistent with the findings of previous studies8,20, the prevalence of physical violence was higher in boys than in girls. This sex difference may be due to greater societal tolerance for male physical violence23. In general, physical violence is considered part of male power and status24. In addition, differences in the levels of sex hormones, such as testosterone, may account for the greater levels of physical violence among men than women. Finally, previous studies have suggested that boys are also more likely than girls to be exposed to risky behaviors linked with violence25, such as alcohol consumption26.

Consistent with the literature on high-income countries20,27, food insecurity, as an indicator of low socioeconomic status, was associated with physical fights and physical attacks in adolescents in 66 countries. Semahegn et al. suggested that the association between food insecurity and physical fight may be related to the social status of adolescents with food insecurity among their peers28. In addition, the stigma and shame associated with food insecurity may lead adolescents to engage in physical violence with their peers28. Potential biological effects of food insecurity on adolescents may lead to fluctuations in their cortisol levels, which are associated with emotional stress and, in turn, with violence29. Besides, the low glucose levels as a consequence of food insecurity might also lead to impaired impulse control30, and increased the risk of physical violence. The exploration of the association between physical fight and country wealth in 79 countries revealed that this association was modified by expenditure on education and inequality in income25. This also explains the varied associations across regions and income-levels subgroups between food insecurity and physical violence in our results.

The positive association between food insecurity and violent behavior was evident in in both boys and girls, but demonstrating a stronger correlation magnitude among female adolescents compared to their male counterparts. The stronger association identified in girls was not previously known and may be attributable to the different attitudes toward and responses to food insecurity in boys and girls31. Thus, we are motivated to seek further explanation for such a marked gender difference. Such explanation would guide us in developing targeted interventions to address food-related challenges and situations of violent behaviors associated with food insecurity. Given the results of this study, it may be necessary to adjust for gender-specific adolescent outcomes, vulnerability, and protective measures.

Strengths and limitations

The strengths of the study are its large sample from 66 countries and the standardized assessments used in the GSHS, which allowed standardized prevalence rates of violent behaviors to be obtained across countries. To the best of our knowledge, this is the first cross-national analysis of the association between food insecurity and violent behaviors in adolescents. However, some limitations exist in our study. First, the measurements of food insecurity, physical fights and physical attacks were the primary limitation as these were assessed using only one question, which may not provide a comprehensive evaluation. Second, the cross-sectional design could not reveal the causal nature of the association. Third, recall bias and desirability bias were unavoidable due to the self-reported nature of the data collected in the GSHS. Fourth, the survey did not cover students who were absent on the day of the survey or those who dropped out for other reasons. These adolescents may belong to disadvantaged groups and may be more likely to exhibit risky behaviors than other groups; thus, their exclusion may not accurately represent the general population.

Conclusions

Food insecurity is prevalent among adolescents worldwide and is positively correlated with violent behaviors. As food insecurity tends to become more severe globally32, especially with the development of regional conflict and global epidemics, it is not difficult to foresee that food insecurity and violent behaviors in adolescents will become international public health problems. Thus, it is urgently necessary to develop targeted policies and measures at the government, society, and family levels, together with international cooperation, to decrease food insecurity and violent behaviors in adolescents.