Prevalence of and factors associated with overweight among university students from Rio Branco , Acre – Brazil

The aim of this study was to determine the prevalence of and factors associated with overweight among university students from the city of Rio Branco, Acre, Brazil. A cross-sectional study was conducted on 1,104 students (mean age: 24 years) from Rio Branco, Acre. Overweight was used as the dependent variable. The following independent variables were studied: socioeconomic variables (gender, age, teaching institution, period of study, marital status, income, household members, living with other people, parent’s education level, and working or not); self-reported health and morbidity variables (health perception, hypertension, diabetes, and altered cholesterol); lifestyle variables (physical activity, watching television, computer use, smoking, and alcohol consumption). The nutritional status was classified based on body mass index. Univariate analysis was performed using Pearson’s chi-squared test or Fisher’s exact test. Variables with a p value < 20% on univariate analysis were included in the multiple Poisson regression model. The prevalence of overweight was 35.6%. In the multiple regression model for male students adjusted for age, overweight was associated with marital status, household income, hypertension, and watching television every day. In the model for female students, the variables that remained associated were self-perceived health status, hypertension, and alcohol consumption in the last month. Overweight is determined by a combination of different factors which differed in the present study between men and women.


INTRODUCTION
Several factors are associated with the development of overweight, including genetic, physiological and metabolic factors 1 .However, lifestyle changes such as a reduction in physical activity levels and an increase in the time spent in sedentary behaviors, in the consumption of simple carbohydrates, high-fat foods and soft drinks and in food portion size explain the increasing number of overweight adults 2 .
Overweight is associated with different chronic diseases such as type 2 diabetes mellitus, cardiovascular diseases, hypertension, dyslipidemia, metabolic syndrome, osteoarthritis, insulin resistance 3 , psychosocial problems (stress, depression, anxiety, and low self-esteem), decreased academic and professional performance, and low overall quality of life 4 .In view of the associated health problems and the low success rate of treatment of this important condition, as well as the health expenditures with hospitalizations, medications and surgeries, overweight is now a leading public health problem.
Studies have shown a high and increasing prevalence of overweight among university students 5,6 .National surveys on overweight among adults indicate that this prevalence has more than doubled over 35 years 7 .Similar results have been reported for adults from other countries where this increase is even greater 8,9 .
College life is a critical period in terms of overweight since it is generally accompanied by a reduction in physical activity levels, increased stress, and the adoption of unhealthy eating habits such as increased consumption of soft drinks, sweets, fried and processed foods and skipping meals, factors that favor an increase in the amount of body fat 10 .Therefore, assessment of the nutritional status of university students has been recommended in order to identify subgroups that are at risk of being overweight and factors associated with this outcome.In this respect, the objective of the present study was to evaluate the prevalence of and factors associated with overweight among university students from Rio Branco, Acre, Brazil.

METHODOLOGICAL PROCEDURES
A cross-sectional study was conducted in the city of Rio Branco, Acre, Brazil, between May and September 2011, which involved university students of both genders from three higher education institutions of the city, including one public and two private institutions.
For calculation of the sample size, a prevalence of overweight of 46.9% 11 , an estimated maximum sampling error of 3% and a 95% confidence interval were adopted, resulting in a sample of 986 students.The sample size was increased by 14% to account for possible losses and refusals.The final sample consisted of 1,125 university students.Two-stage cluster sampling was used in this study.In the first stage, 38 courses at the three institutions were randomly selected.In the second stage, one class per course was randomly selected at each institution participating in the study.Finally, all students of the selected classes who were present on the day of application of the questionnaire were invited to participate in the study.Students younger than 18 years, subjects with some disability impairing the measurement, and pregnant students were excluded.
The data were collected by a trained team consisting of teachers from the physiotherapy, physical education, nutrition and medicine courses using the protocol for anthropometric measurement of the Food and Nutritional Surveillance System (Sistema de Vigilância Alimentar e Nutricional -SIS-VAN) 12 .After receiving detailed information about the objectives of the study and signing the free informed consent form, the students received the questionnaire and instructions on how to fill it out.Next, the body weight and height of the participants were measured.
The instrument used in this study was an individual self-completion questionnaire elaborated based on the Telephone-based Surveillance of Risk and Protective Factors for Chronic Diseases (Vigilância de Fatores de Risco e Proteção para Doenças Crônicas por Inquérito Telefônico -VIGI-TEL) 13 and the International Physical Activity Questionnaire (IPAQ) 14 .The questionnaire was divided into modules containing the following sections: socioeconomic data; evaluation of individual health status; lifestyle assessment, and evaluation of nutritional status.
Body weight and height were measured in duplicate by the same examiner and the mean of the two measurements was used for analysis.Body weight was measured with a Tanita UM080 electronic scale (capacity of 130 kg) to the nearest 100 g, and height was measured in millimeter with a Seca portable stadiometer.A body mass index (BMI) ≥ 25 kg/m² was defined as overweight according to World Health Organization recommendations 15 .
The following socioeconomic variables were studied: gender (male or female), age (divided into three strata: < 20 years, 20 to 29 years, or ≥ 30 years), institution (public or private), period of study (morning, afternoon, night, or full-time), marital status (single, stable union or married, or widowed, separated or divorced), household income in minimal wages (MW) (< 3 MW, 3 to 10 MW, or > 10 MW), household members (1 person, 2 to 5 persons, or 6 or more persons), with whom the respondent lives (father, mother or both, spouse, or alone/others), parent's education level (stratified into the following categories: 0 to 8 years of schooling, 9 to 11 years, or 12 or more years), and whether the respondent works (yes or no).
For health status assessment, the participants were asked about health perception (very good or good, regular, and poor or very poor), hypertension (yes or no), diabetes (yes or no), and altered cholesterol (yes or no).
For lifestyle evaluation, the students were asked about leisure-time physical activities performed for at least 10 minutes during the week prior to data collection (IPAQ, short version).The participants were classified as inactive or irregularly active (< 150 minutes per week) and active or very active (≥ 150 minutes per week).The participants were also asked about the use of television and computer on each day of the week (yes or no), smoking (Do you smoke?Response options: yes; yes, occasionally, and no), and alcohol consumption (Do you usually drink alcoholic beverages?Response options: yes or no).
Proportions were calculated for descriptive analysis.Differences in proportions between the overweight and non-overweight groups were compared individually for each factor using Pearson's chi-squared test or Fisher's exact test.
Variables presenting a p value < 20% on univariate analysis were included in the Poisson multiple regression model adjusted by robust estimation of variance.During this phase, the data were analyzed hierarchically: behavioral or lifestyle variables at the proximal level (physical activity, television, computer, smoking, and alcohol), health-related variables at the intermediate level (health perception, hypertension, diabetes, and altered cholesterol), and socioeconomic variables at the distal level (gender, age, institution, period of study, marital status, income, household member, with whom the respondent lives, father's and mother's education level, and working or not).The exhibiting variables that presented a p value < 5% or adjusted the prevalence ratios by at least 10% were maintained in the multiple models.Thus, the exhibiting variable was considered to be associated with the outcome when it remained in the final multiple model at a level of significance of 5% by the Wald test.The Microsoft Office Access 2007 program was used for construction of the database and the STATA/ SE 11.0 program for data analysis.
The participants received detailed information about the risks and benefits of the study and provided free informed consent before the beginning of the study.The methodological procedures were approved by the Research Ethics Committee of the Federal University of Acre (Permit No. 23107.004753/2011-62).

RESULTS
The final sample consisted of 1,104 students, with losses accounting for 1.5% of cases (pregnant women, students with disabilities, incorrect response to the questionnaire, and lack of weight and height measurements).The mean age was 24 years (± 7.2); 42.6% of the students were men and 57.4% were women.
Most students were in the age group of 20 to 29 years, were enrolled in a private institution, studied at night, were single, had a household income of 3 to 10 MW, and lived with 2 to 5 household members and with the parents who had up to 8 years of schooling.The health perception of the students was good or very good and none of them reported hypertension, diabetes or altered cholesterol.The students were active or very active, watched television every day, used the computer every day, did not smoke, and had not consumed alcohol in the last month.
The overall prevalence of overweight was 35.6%.There was a significant difference in the prevalence of overweight between male and female students (p < 0.001), which was 43.6% among men and 29.7% among women.The variables associated with overweight in male students were age, marital status, household income, with whom the respondent lives, working or not, self-reported hypertension, and self-reported altered cholesterol.For female students, the variables associated with overweight were age, marital status, with whom the respondent lives, parent's education level, working or not, self-reported health perception, and self-reported hypertension, diabetes and altered cholesterol.
For male students, the following factors continued to be associated with overweight in the crude multiple model and in the model adjusted for age: being married or in a stable union, being widowed or separated or divorced, having a household income of up to 3 MW, being hypertensive, and watching television every day.A household income of up to 3 MW was negatively associated with overweight.For female students, the variables that continued to be associated with overweight in the crude multiple model and in the age-adjusted model were: having a regular, poor or very poor health perception, being hypertensive, and having consumed alcohol in the last month (Tables 2 and 3).

DISCUSSION
The present study found a high prevalence of overweight (35.6%) among university students.As reported in other studies 9,10,13 , the prevalence of overweight was higher among male students than female students (43.6% versus 29.7%).This higher prevalence of overweight among men might be explained by the greater care of women with body esthetics, valuing a slim and fit body in order to meet beauty standards.The prevalence of overweight observed in this study was higher than that reported in a population-based survey conducted in Rio Branco in 2008, in which 23.6% of adults aged 18 to 24 years were overweight 11 , and in the 2011 VIGITEL survey comprising 26 Brazilian capitals and the Distrito Federal, in which the prevalence was 27.3% among adults aged 18 to 24 years from the city of Rio Branco, Acre.Differences in the prevalence of overweight between studies can be explained by differences in the composition of the sample (adult population in the population-based survey) and by methodological differences (self-reported measures in the VIGITEL survey).However, the proportion of overweight students was high in the present study.This finding is a matter of concern since excess weight is associated with a variety of health problems, is difficult to reverse (persisting throughout life), and is associated with high treatment costs 16 .
The prevalence ratios of overweight were higher among married students or those living in a stable union, as well as among widowed, separated or divorced students, when compared to singles.Similar results have been reported in studies conducted in other countries 17,18 .Silva et al. 2 suggested this finding to be related to the adoption of new lifestyle habits after marriage, such as a reduction in physical activity levels and dietary changes.In addition, this higher prevalence may also be related to age.It is natural that individuals gain weight with increasing age as a result of a decrease in basal metabolic rates and physical activity levels.However, Oliveira et al. 19 found no association between overweight and marital status in an adult population from Salvador.
A household income of up to 3 MW was inversely associated with overweight in male students.A study conducted in Rio Branco demonstrated an increasing linear trend in the percentage of overweight adults with increasing household income 11 .Suleiman et al. 20 found a higher prevalence of overweight among low-income university students compared to highincome students.These data agree with the inverse relationship between overweight and income in developed countries proposed by Ljungvall and Zimmerman 21 .In Brazil, this relationship is different due to the period of transition during which income tends to be a risk factor for an increasing prevalence of overweight, characterized by changes in the patterns of consumption of the population.This pattern has been demonstrated among men, but not among women.
The prevalence of overweight was 64.6% in the population aged 30 years or older, with a prevalence of 75.6% among male students and of 57.6% among female students.According to Burton et al. 22 , overweight tends to increase with age.Economos et al. 23 suggested this increase to be more accelerated in men and slower in women, with the observation of higher prevalences among younger male students and among older female students.
The present study showed that male students watching television every day of the week had a 35% higher chance of being overweight than those who did not.Although the results of cross-sectional studies associating the time spent watching television with indicators of overweight and obesity are controversial, spending excessive time in sedentary behavior may contribute both directly and indirectly to overweight.Directly, the time spent in sedentary behaviors reduces physical activity levels and favors an increase in the consumption of high energy density foods 24,25 .Indirectly, excessive television watching increases the exposure to advertisements that encourage the consumption of high-calorie foods of low nutritional value 25 .Taken together, these factors contribute to increase the positive energy balance and, consequently, body weight.
In the present study, female students with a negative perception of health (regular, poor, very poor) had a higher chance of being overweight.Similar results have been reported in other studies demonstrating a higher frequency of negative health perception among overweight and obese individuals compared to their non-obese or non-overweight peers 11,26 .Self-perceived health status has been used in epidemiological studies for health status assessment of adults and is an important health indicator of populations 27 .Taken together, these results suggest that individuals with excess weight perceive this condition to be affecting their health status.
Overweight university students were more likely to develop arterial hypertension.Similar results have been reported in other studies 23,28 which showed that overweight is associated with systemic arterial hypertension.These results indicate the need for continuous prevention and education programs as well as activities designed to promote a more active lifestyle among university students based on regular physical activity and the adoption of healthy eating habits.
The prevalence ratios of overweight in female students were higher among those reporting the consumption of alcohol in the last month compared to those who did not.Different results have been reported by Mataix et al. 29 who observed a higher prevalence of overweight among women who did not consume alcohol.
The role of alcohol in the mechanism underlying the development of overweight is unclear, mainly because of differences in the methods used to determine the frequency and quantity of alcohol consumed.The responses to alcohol vary depending on individual characteristics and the same trend is observed for weight gain related to alcohol consumption.Furthermore, the method selected for the evaluation of alcohol consumption in the present study (consumption in the last 30 days) is likely to be a limiting factor for analysis of the association with overweight, especially because this definition does not reflect the frequency or quantity of alcohol consumed 30 .
One limitation of this study is its cross-sectional design which does not permit to determine whether the association found is causal; thus, reverse causality cannot be excluded.The strengths of the study include the small percentage of losses and refusals, the use of a representative and probability sample of university students, and the fact that the students were recruited from public and private institutions.

CONCLUSION
The prevalence of overweight was high, especially among male students older than 30 years, who were married or in a stable union, had a household income of 3 to 10 MW, and worked.Overweight was associated with marital status, household income, hypertension and watching television every day in men, and with health perception, hypertension and alcohol consumption in the last month in women.It can be concluded that overweight is determined by a combination of different factors which differed in the present study between men and women.These results highlight the need of intervention programs for this important public health problem.
* Income reported as Brazilian minimum wage (MW) in 2011: R$ 545.00.° p < 0.05 (Pearson's chi-squared test or Fisher's exact test).Covariates that did not show a p value < 0.20 on univariate analysis are not reported in the table.Male students: household members, parent's education level, self-reported diabetes, and computer use every day.Female students: family income, household members, physical activity classification, and watching television every day.

Table 2 .
Poisson regression estimates for overweight among male students from the city of Rio Branco, Acre, Brazil (2011).
Exclusion of all covariates with p > 20% in Table1(household members, parent's education level, computer use every day, and self-reported diabetes).‡ Model adjusted for age.PR: prevalence ratio; 95% CI: 95% confidence interval.The dotted line indicates variables that did not remain in the multiple model.