Determinants of overweight and obesity in Thai adolescent girls.
Abstract: Background

Obesity during childhood and adolescence is a serious public health problem as it increases the risk of chronic diseases throughout life. Obesity has recently been recognized as a public health threat within low and middle income nations such that research from Thailand has documented an increase in obesity in the past 10 years. While there are several descriptive studies reporting the presence of obesity among adults and children, few have examined relationships between social and dietary factors and obesity in this country. Thus the objective of this study is to examine predictors of overweight/ obesity among Thai adolescent girls.


This cross-sectional exploratory study uses logistic regression to examine predictors of overweight/obesity among Thai adolescent girls. Anthropometric, dietary, economic, demographic, and activity data were collected from 342 adolescent girls ages 9 to 18 years living in suburban Bangkok, Thailand in 2004 and in 2005.


The most important predictors of overweight/obese girls included early menarche (p < 0.01), birth order (p < 0.01), sports after school (p < 0.05), and moderate consumption of potato chips (P < 0.05). For instance, girls who were at most 10 years old at menarche were 4.24 (95% CI = 1.42, 12.72) times more likely to be overweight/obese than girls who were 12 years old.


The predictive nature of early menarche has been well documented in studies around the world. However, the other major predictors may be specific to this population in Thailand, and thus may assist in better understanding social determinants of obesity.

Keywords: Obesity, Overweight, Thailand, Adolescents
Article Type: Report
Subject: Chronic diseases (Research)
Obesity in children (Research)
Obesity in adolescence (Research)
Public health
Teenage girls
Authors: Pawloski, Lisa R.
Kitsantas, Panagiota
Ruchiwit, Manyat
Pub Date: 04/01/2010
Publication: Name: Archives: The International Journal of Medicine Publisher: Renaissance Medical Publishing Audience: Academic Format: Magazine/Journal Subject: Health Copyright: COPYRIGHT 2010 Renaissance Medical Publishing ISSN: 1791-4000
Issue: Date: April-June, 2010 Source Volume: 3 Source Issue: 2
Topic: Event Code: 310 Science & research
Product: Product Code: 8000120 Public Health Care; 9005200 Health Programs-Total Govt; 9105200 Health Programs NAICS Code: 62 Health Care and Social Assistance; 923 Administration of Human Resource Programs; 92312 Administration of Public Health Programs
Accession Number: 263251117

As undernutrition continues to be a problem in low and middle income countries, a new problem of overnutrition has been documented in many parts of the world. Also referred to as the "nutrition transition", this epidemiological trend is defined as changes in the nutritional status and dietary intake among populations that are caused by economic, environmental, demographic, and cultural shifts. (1)

While the nutrition transition was first recognized in South America, it has also been documented in Southeast Asia, and more specifically in Thailand. Most researchers have attributed this to the improved economic and demographic trends over the last three decades in Thailand. (2) Litimaskul reported a rise in the prevalence of adult obesity in Thailand to increase from 5% in 1986 to 17.9% in 1999. (3) Researchers have also reported greater intakes in fat, animal proteins, and processed foods, contributing to increased caloric consumption, thereby increasing the risk for overweight and obesity. (4) Langendijk et al. found the prevalence of obesity among Thai children to be 10.8% in the Northeastern region. (5) They also found obesity to be greater among boys when compared to girls. Further, Sakamoto et al. found a high prevalence of obesity among pre-school children in the Central Region of Thailand and showed a marked correlation between socioeconomic status and obesity in children, such that there was greater obesity among wealthier and better-educated parents. (6)

While several studies have reported the nutritional state of children and adults in Thailand, very few studies have specifically examined adolescent girls living in this country. Most recently, Pawloski et al. published a descriptive study showing a greater prevalence of obesity among primary school girls compared to secondary school girls. (7) This finding suggests younger adolescents may be more greatly affected by obesity than their older counterparts. This trend was thought to be influenced by the nutritional transition occurring in Thailand and it is possible that younger adolescents are more likely to be influenced by a Western diet. This study further examines the data collected by Pawloski et al. by exploring determinants of overweight and obesity among Thai adolescent girls. (7) Thus the purpose of this study is to examine determinants of obesity including dietary intake, activity patterns, household structure and socioeconomic status in Thai adolescent girls.



The data, including anthropometric, age at menarche, wealth, demographic, dietary, and activity information were collected in Rangsit, Thailand, a peri-urban community 40 km north of Bangkok. A total of 500 girls were invited to participate in the study. Data were collected from a convenience sample of 410 adolescent girls ages nine to eighteen years who agreed to participate and signed informed consent forms, thus a response rate of 82%. However, only 344 girls had complete data for the variables of age, weight and height which in this study were used in the calculation of the dependent variable, body mass index (BMI).

Sampling Process

Participants were recruited from two local public primary and secondary schools. Random sampling was not possible, as the school was not large enough to allow for it. Further, school officials required all eligible girls to be able to participate. Informed consents were sent home to 500 girls a week before data collection took place. Only students who had informed consent forms signed and returned were eligible to participate. Human subjects' approval was made by the George Mason University Human Subjects Review Board and the Thammasat University Faculty of Medicine Human Subjects Review Board. Participants and their parents were given assent and consent forms translated into Thai.

Anthropometric Indicators

Height and weight measurements were taken by the researcher in a clinical setting and followed the methods described by Lohman, Roche, and Martorell. Weight was collected using a digital body fat scale (Tanita). (8) Height measurements were done using a field portable anthropometer (GPM Seritex Brand).

To determine overweight and obese girls, comparison data were chosen to serve as a reference rather than a standard. Because no large reference sample was available from Thailand with which to make a comparison, these data were compared with international cut-off points for BMI-for-age developed by Cole et al. (9)

Due to the relatively small number of overweight and obese adolescents in this sample, overweight and obese individuals were collapsed into one category called overweight/obese. Thus, the dependent variable is dichotomous, comprised of normal weight and overweight/obese adolescents.

Age at menarche

To determine age at menarche, a retrospective method was used such that girls were asked if they had begun to menstruate and at what age it began.

Dietary behaviour and activity questionnaires

To determine dietary behaviour, a dietary behaviour instrument was developed by Thai and U.S. researchers, specific to this population. The behaviour instrument was modelled after a dietary instrument which was developed by the researchers in previous studies in Nicaragua and the U.S. (10,11) and content validity was established by Thai researchers at Thammasat University as well as researchers from George Mason University. The questionnaires were translated into Thai and back-translated to ensure the meaning of the instrument. Questions concerned general dietary behaviours included a food frequency component in which participants were asked how many servings per week they consumed specific types of foods, i.e. chips, vegetables, sodas, fruits, etc. These dietary variables were categorized into serving amounts (i.e., none, one, two, etc) based on the individual distributions of servings for both normal weight and overweight/obese girls. To better understand activity patterns, girls were asked about how often they exercised and participated in sports during and after school. They were also asked about how many minutes they spent doing sedentary activities such as television viewing and computer usage. In the data analysis, the variables of television viewing and computer usage were combined to create a "screen time" variable that indicates hours spent in these sedentary activities.

Sociodemographic questionnaire

Sociodemographic questions assessed the participants' birthdates, birth order, and household income. Birthdates were verified by each school using student records. Income was recorded as the amount of Thai Baht earned per month. The classification of the income variable into three categories ([less than or equal to] 10,000, 11,000-24,000 and [greater than or equal to] 25,000) was determined based on the distribution of the individual incomes and using measures such as quartiles to identify their economic status relative to the median.

Statistical Analysis

Descriptive statistics were conducted for both groups. The chi-square test was used to assess differences between the normal and overweight/obese adolescents for all predictor variables. We performed stepwise logistic regression analysis to assess the impact of various predictor variables on the risk of overweight/obesity among these adolescent girls. Furthermore, decision tree analysis using the CART software 5.012 was carried out to confirm the logistic regression findings.

CART, which was firmly established by Breiman et al., constitutes a nonparametric tool that makes no distributional assumptions and is not affected by outliers or collinearities. (13) It can be used to classify data that involve both continuous and categorical variables and it can effectively handle a large number of variables submitted for analyses. The accuracy of CART is comparable with linear and logistic regression. These important features of CART have made this technique popular in various fields including medicine, public health and biological sciences. (14,15)


The sample consisted of 344 girls of whom 82.5% had normal weight and 17.5% were overweight/obese. Table 1 provides descriptive information about the variables used in the analyses and adjusted odds ratio (OR) with 95% confidence intervals. Chi-square analyses revealed that a significantly higher percent of overweight/obese girls experienced menarche at a younger age in comparison to normal weight girls (p < 0.01). Furthermore, normal and overweight/obese girls differed significantly in regards to their order of birth (p < 0.01) with more overweight/obese girls being born first (61.7%). The two groups also differed according to specific aspects of their diet. For instance, 49.1% of the overweight/obese girls tended to consume one serving of potato chips per week while only 28.1% of normal-weight individuals did so. Normal-weight girls, however, had a higher consumption of at least two and three servings of chips per week than overweight/obese ones (p < 0.05). They were also significantly (p < 0.05) more likely to eat sweets on a daily basis (23.5%) compared to overweight/obese girls (11.9%).

Significant differences were also observed in certain activities outside of school. In particular, 30.0% of overweight/obese girls held a job while only 17.1% of normal-weight girls did so (p < 0.05). They were also more likely to participate in sports after school (41.7%) in comparison to 27.0% participation among the normal-weight individuals (p < 0.05). No significant differences were observed between these two groups and the other predictor variables listed in Table 1.

The logistic regression analysis indicates that girls who were at most 10 years old at menarche were 4.24 (95% CI = 1.42, 12.72) times more likely to be overweight/obese than girls who were 12 years old. On the other hand, girls who were at least 13 years old at menarche were less likely to be overweight /obese than girls who were 12 years old (OR = 0.21; 95% CI = 0.06, 0.76). Also, those adolescents born second were less likely to be overweight/obese than firstborn ones (OR = 0.37, 95% CI = 0.15, 0.90). In addition, adolescents whose families earned an income of at least 25,000 Thai Baht per month were 3.11 (95% CI = 1.22, 7.97) times more likely to be overweight/obese than those earning an income between 11,000 and 24,000 Thai Baht per month. Girls who consumed one serving of chips/week were 2.34 (95% CI = 1.01, 5.54) times more likely to be overweight/obese than girls who did not consume any. None of the other variables were found to be significant in predicting overweight/obesity in this sample.

The classification tree analysis (Figure 1) confirmed the findings of the logistic regression methodology as the same variables were selected in predicting overweight/obese adolescents. Specifically, age at menarche, servings of chips per week, and income were selected in that order as the most important variables in discriminating between the two classes. Overall, overweight/obese girls were more likely to have their period start before the age of 10 years old. Individuals with menarche occurring later in their adolescent years were also likely to be overweight/obese if they consumed one serving of potato chips per week and their family's income was [greater than or equal to] 25,000 Thai Baht per month.


The results suggest that girls who are economically advantaged are more likely to be overweight/obese, such that girls from families of higher incomes, those who hold jobs, those who participate in sports, and girls who are first born are more likely to be overweight/obese. These findings from Thailand are striking because they are somewhat the opposite of what is occurring in the U.S. and other developed countries. In the U.S., those who are at greater risk of obesity include more disadvantaged populations as well as those of lower socioeconomic status. For example, Singh et al. conducted a multilevel analysis of state and regional disparities regarding childhood and adolescent obesity in the U.S. and found low socioeconomic status and geographic location to be predictors of obesity. (16) Further, O'Dea et al. revealed obesity in Australian adolescents to be higher among those of lower socioeconomic status and among disadvantaged minorities. (17) Additionally, Wardle et al. conducted a longitudinal analysis in Great Britain among adolescents and also concluded those of low socioeconomic status to be at greater risk for developing obesity. (18) These findings also support earlier work by Sobal and Stunkard showed that there is a positive correlation between obesity and socioeconomic status in developing countries while this relationship is negative in developed countries. (20) Similar results have also been reported in China depicting a greater evidence of obesity among those of high socioeconomic status. (21)


With the exception of consuming a greater quantity of chips, these data also reveal little evidence to suggest that overweight/obese girls are consuming greater quantities of snacks or sodas, have longer periods of screen time, or are eating fewer fruits and vegetables compared to girls of average weight. Again, these findings are quite different from reports on obese children and adolescents from the U.S. and other countries which suggest strong correlations between obesity and low consumption of fruits and vegetables and high consumption of sugary drinks, and screen time. For example, Shields reviewed Canadian and U.S. Health surveys to examine determinants of overweight and obesity among children and adolescents and found that those who consumed fewer fruits and vegetables were at significantly greater risk of being overweight or obese. (22) Burke et al. found similar results to Shields among Australian adolescents. (23) Furthermore, Mark and Janssen reported that increased screen time increases the risk of metabolic syndrome among Canadian adolescents. (24)

Thus, among these girls in Thailand, the findings may ultimately suggest a "privilege" factor, for which girls who are first born or of higher socioeconomic status may have a greater risk of being overweight/obese than those who are not as privileged. Consumption of chips would also suggest a "privilege" factor. Even though chips are widely distributed in Thailand, they are typically more available at larger and more western style stores. Informal interviews with parents supported these results such that parents expressed that as their wealth increased they were more apt to succumb to the child's demands of western style foods and snacks. Parents expressed that they would choose restaurants which would provide traditional Thai foods for them and more western style cuisine for their children.

In addition to demographic and dietary factors, those who had an early age of menarche were more likely to be overweight/obese. These results are supported by a variety of literature worldwide, which suggest that overweight and obesity during early growth and development may influence the timing of maturation and age at menarche. (26) Thus, these data may suggest the need to intervene significantly earlier than the onset of the pubertal growth spurt.

Limitations of this study include the relatively small sample size and that it is a convenience sample. Also, the cross-sectional nature of the data does not allow for a longitudinal assessment of participants' BMI throughout adolescence. Future studies should collect data longitudinally to determine how changes in the socioeconomic status and diet impact their weight status at different stages during adolescence. Qualitative studies that assess cultural related beliefs about weight, wealth, and diet would be helpful in understanding the dynamics of the Thai culture and its impact on weight gain among young girls.

This is one of the first studies to examine the impact of dietary and socioeconomic factors on the risk of overweight/ obesity in Thai adolescent girls. The findings reveal that a "privilege" factor, which refers to families with high incomes, increased risk of overweight/obesity in comparison to those who were not as privileged. Overall, dietary indicators suggested that overweight/obese girls did not necessarily report having poorer diets than those who were not overweight/obese. Further, girls with early age at menarche were more likely to be overweight/obese. These data represent a specific population in Thailand and have potential to assist in developing nutrition intervention programs focusing on obesity prevention in high risk communities, such as those of higher socioeconomic status.


Lisa Pawloski and Manyat Ruchiwit conceived the study and collected the data in Thailand. Manyat Ruchiwit facilitated and obtained permissions from the schools and Human Subjects Review Board in Thailand. Panagiota Kitsantas assisted in the design and conducted the statistical data analyses. All three authors contributed to the drafting and writing of the manuscript. All authors have seen and approved the final version of this paper.

Conflict of interest: None declared.


The authors would like to acknowledge the Thailand-U.S. Fulbright Foundation and the Faculty of Nursing at Thammasat University.


(1.) Popkin BM. Nutritional patterns and transitions. Popul Dev Rev 1993; 19:138-57.

(2.) Kosulwat V. The nutrition and health transition in Thailand. Public Health Nutr 2002; 5:183-89.

(3.) Likitmaskul S. Increasing prevalence of type 2 diabetes mellitus in Thai children and adolescents associated with increasing prevalence of obesity. J Pediatr Endocrinol Metab 2003; 16:71-77.

(4.) Areekul W, Viravathana N, Aimpun P, Watthanakijthavongkul K, Khruacharooen J, Awaiwanont A, et al. Diet behaviors and nutritional status of adolescents in a remote rural area of Thailand. J Med Assoc Thai 2005; 88(Suppl):240-46.

(5.) Langendijk G, Wellings S, Van Wyke M, Thompson SJ, McComb J, Chulsilp K. The prevalence of childhood obesity in primary school children in urban Kohn Kaen, northeast Thailand. Asia Pac J Clin Nutr 2003; 12:66-72.

(6.) Sakamoto N, Wansorn S, Tontisirin K, Marui E. A social epidemiologic study of obesity among preschool children in Thailand. Int J Obes Relat Metab Disord 2001; 25:389-94.

(7.) Pawloski, LR, Ruchiwit M, Pakapong Y. Growth data from Thai adolescent girls: A transition from under to overnutrition? Ann Hum Biol 2008; 35:378-385.

(8.) Lohman TG, Roche AF, Martorell R. Anthropometric Standardization Reference Manual. Chicago, IL, Human Kinetics, 1988.

(9.) Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ 2000; 320:1240-43.

(10.) Pawloski LR. The growth, development, and nutritional status of adolescent girls from the Segou Region in Mali: A biocultural approach to the analysis of health. PhD dissertation . United States: Indiana: University Microfilms, Indiana University, Department of Anthropology 1999.

(11.) Moore JB, Pawloski LR, Rodriguez C, Baghi H, Lumbi L, Zamora L. The effects of a nutrition education program on the nutrition self-care practices of Nicaraguan Adolescent girls and the nutrition dependent practices of their mothers. Self-care and Dependent Care Nursing. 2007; 15:6-11.

(12.) Salford Systems, CART. Tree-Structured Non-Parametric Data Analysis Version 4.0. San Diego, CA, 2000.

(13.) Breiman L, Friedman JH, Olshen RA, Stone CJ. Classification and regression trees. Boca Raton, CA, Chapman & Hall, 1984.

(14.) Kitsantas P, Moore T, Sly DF. Using classification trees to profile adolescent smoking behaviors. Addict Behav 2007; 32:9-23.

(15.) Pawloski LR, Kitsantas P. Classification tree analysis of stunting in Malian adolescent girls. Am J Hum Biol 2008; 20:285-92.

(16.) Singh GK, Kogan MD, van Dyke PC. A multilevel analysis of state and regional disparities in childhood and adolescent obesity in the United States. J Community Health 2008; 33:90-102.

(17.) O'Dea JA. Gender, ethnicity, culture, and social class influences on childhood obesity among Australian schoolchildren: implications for treatment, prevention, and community education. Health Soc Care Community 2008; 16:282-90.

(18.) Wardle J, Brodersen NH, Cole TJ, Jarvis MJ, Boniface DR. Development of adiposity in adolescence: five year longitudinal study of an ethnically and socioeconomically diverse sample of young people in Britain. BMJ 2006; 332:1130-35.

(19.) Gordon-Larsen P, Adair LS, Popkin BM. Ethnic differences in physical activity and inactivity patterns and overweight status. Obes Res 2002; 10:141-49.

(20.) Sobal J, Stunkard AJ. Socioeconomic status and obesity: a review of the literature. Psychol Bull 1989; 105:260-75.

(21.) Shi Z, Lien N, Kumar BN, Holmbe-Ottesen G. Socio-demographic differences in food habits and preferences of school adolescents in Jiangsu Province, China. Eur J Clin Nutr 2005; 59:1439-48.

(22.) Shields M. Overweight and obesity among children and youth. Health Rep 2006; 17:27-42.

(23.) Burke V, Lawrie JB, Durkin K, Werner GK, Stritzke SH, Cameron CA. Television, computer use, physical activity, diet and fatness in Australian adolescents. Int J Pediatr Obes 2006; 1:248-55.

(24.) Mark AE, Janssen I. Relationship between screen time and metabolic syndrome in adolescents. J Public Health 2008; 30:153-60.

(25.) Rey-Lopez JP, Vicente-Rodriquez G, Biosca M, Moreno LA. Sedentary behaviour and obesity development in children and adolescents. Nutr Metab Cardiovasc Dis 2008; 18:242-51.

(26.) Bratberg GH, Nilsen TI, Holmen TL, Vatten LJ. Early sexual maturation, central adiposity and subsequent overweight in late adolescence, a four year follow-up of 1605 adolescent Norwegian boys and girls: the Young HUNT study. BMC Public Health 2007; 7:1471-245

Lisa R. Pawloski [1], Panagiota Kitsantas [2], Manyat Ruchiwit [3]

[1] Department of Global and Community Health, College of Health and Human Services, George Mason University, Fairfax, VA 22030, USA

[2] Department of Health Administration and Policy, College of Health and Human Services, George Mason University, Fairfax, VA 22030, USA

[3] Department of Mental Health and Psychiatric Nursing, Thammasat University, Klong Luang, Pathumthani, 12121 Thailand

Corresponding author: Lisa R. Pawloski, PhD

Department of Global and Community Health

College of Health and Human Services

George Mason University

4400 University Drive, MSN 5B7

Fairfax, VA 22030, USA

Tel.: +1 703 993 4628

Fax: +1 703 993 190E-mail:
Table 1. Sociodemographic and Dietary Characteristics (%)
among Normal and Overweight/Obese Thai Adolescents

                                      Normal     Overweight/   P-value
                                      (n=284)%   obese

Age at menarche                                                <0.001
At most 10 yrs old                    4.2        21.7
11 yrs old                            24.3       21.7
12 yrs old                            54.2       48.3
At least 13 yrs old                   17.3       8.3

Order of birth                                                 0.004
First                                 54.3       61.7
Second                                36.9       18.3
Third and higher                      8.9        20.0

Number of siblings                                             0.210
None                                  31.9       40.0
One                                   45.7       33.3
At least two                          22.3       26.7

Income                                                         0.135
[less than or equal to]  10,000       39.0       43.3
11,000-24,000                         36.2       23.3
[greater than or equal to] 25,000     24.8       33.3

Number of meals eaten/day                                      0.135
Two                                   16.3       16.7
At least three                        83.7       83.3

Meals eaten at home                                            0.906
One                                   22.4       20.0
Two                                   59.1       60.0
At least three                        18.5       20.0

Snacks/day                                                     0.823
One                                   45.2       43.3
Two                                   35.9       40.0
[greater than or equal to] 3          18.9       16.7

Servings of chips/week                                         0.014
None                                  24.1       17.5
One                                   28.1       49.1
Two                                   21.6       19.3
[greater than or equal to] 3          26.3       14.0

Number of sodas/week                                           0.501
None                                  17.1       25.0
One                                   21.5       19.6
Two                                   21.8       16.1
Three                                 17.1       21.4
[greater than or equal to] 4          22.5       17.9

Servings of vegetables/day                                     0.467
None                                  3.9        6.9
One                                   18.9       13.8
Two                                   32.9       39.7
[greater than or equal to] 3          44.3       39.7

Number of fruit servings/day                                   0.235
One                                   57.1       47.5
Two                                   26.4       37.3
[greater than or equal to] 3          16.4       15.3

Sweets/everyday                                                0.048
Yes                                   23.5       11.9
No                                    76.5       88.1

Number of servings of milk/day                                 0.655
None                                  6.5        3.4
One                                   44.6       46.6
Two                                   33.0       37.9
[greater than or equal to] 3          15.9       12.1

Exercise                                                       0.486
Always                                46.7       40.4
Usually                               18.1       24.6
Sometimes                             35.1       35.1

Walk to school                                                 0.766
No                                    94.3       93.3
Yes                                   5.7        6.7

Work                                                           0.021
No                                    82.9       70.0
Yes                                   17.1       30.0

Sports in school                                               0.495
No                                    27.3       31.7
Yes                                   72.7       68.3

Sports after school                                            0.023
No                                    73.0       58.3
Yes                                   27.0       41.7

Screen time (TV & computer hrs/day)                            0.023
<4 hours                              37.5       33.9
4-5 hours                             20.7       22.0
6-7 hours                             25.7       25.4
[greater than or equal to] 8 hours    16.1       18.6

Table 2. Adjusted Odds Ratios (95% CI) of Sociodemographic
and Dietary Characteristics among Normal and Overweight/
Obese Thai Adolescents

                                    OR (95% CI)          P-value

Age at menarche
At most 10 yrs old                  4.24 (1.42, 12.72)   0.040
11 yrs old                          1.05 (0.46, 2.38)    0.615
12 yrs old                          1.00
At least 13 yrs old                 0.21 (0.06, 0.76)    0.050

Order of birth
First                               1.00
Second                              0.37 (0.15, 0.90)    0.050
Third and higher                    2.11 (0.82, 5.39)    0.457
[less than or equal to] 10,000      2.21 (0.91, 5.35)    0.094
11,000-24,000                       1.00
[greater than or equal to] 25,000   3.11 (1.22, 7.97)    0.025

Servings of chips/week
None                                1.00
One                                 2.34 (1.01, 5.54)    0.050
Two                                 1.15 (0.41, 3.23)    0.790
[greater than or equal to] 3        0.61 (0.18, 1.96)    0.405
Gale Copyright: Copyright 2010 Gale, Cengage Learning. All rights reserved.