International Journal of Obesity (2001) 25, 389±394 ß 2001 Nature Publishing Group All rights reserved 0307±0565/01 $15.00 www.nature.com/ijo PAPER A social epidemiologic study of obesity among preschool children in

N Sakamoto1*, S Wansorn2, K Tontisirin3 and E Marui1

1Department of Community Health and Medicine, Research Institute, International Medical Center of Japan, Tokyo, Japan; 2Department of Education, Faculty of Social Sciences and Humanities, Mahidol University, Salaya, Nakorn-Pathom, Thailand; and 3Institute of Nutrition, Mahidol University, Mahidol University, Salaya, Nakorn-Pathom, Thailand

OBJECTIVE: To describe the prevalence of childhood obesity in a moderately industrialized province in Thailand and examine the in¯uence of socioeconomic status (SES) on childhood obesity. DESIGN: Cross-sectional study. SUBJECTS: One thousand one hundred and ®fty seven children in the second or third grade of kindergartens in Province, Thailand. MEASUREMENTS: Height and weight were measured and the weight-for-height index with the Thai national standard was used for assessing nutritional status. A questionnaire was used for measuring parents' socioeconomic status.

RESULTS: The prevalence of childhood obesity over 97th percentile for weight-for-height (>p97) was 22.7% in urban and 7.4% in rural areas. There were marked relationships between childhood obesity and parents' educational level and household income. CONCLUSION: Childhood obesity is an emerging health problem in developing countries, especially in urban areas. This study demonstrates a marked correlation between SES and prevalence of childhood obesity in an Asian developing country. International Journal of Obesity (2001) 25, 389 ± 394

Keywords: childhood obesity; preschool children; socioeconomic status; social epidemiology; Thailand

Introduction obesity for the successful prevention of adulthood obesity.6 Childhood obesity is an increasing health problem in devel- The major grounds of this recommendation are the tracking oped countries, and is also a growing concern in developing to adulthood obesity and the forming of a lifestyle including countries. An increase in childhood obesity has been a dietary style.7±10 reported sporadically in some Asian countries, such as Among the South East Asian countries, Thailand has the Korea, Singapore, Thailand, and Indonesia, following their highest literacy rate due to its highly organized educational economic development.1±4 Most of the studies were con- system.11 The duration of compulsory education is only ducted in only capital or big cities where socioeconomic primary school at the present, but it will be extended in status (SES) was higher than in any other areas in those the near future. Though a few studies on childhood obesity countries. of school age children were conducted by using the school- Childhood obesity is positively correlated with eventual ing system, no valid study of preschool children with a large adult obesity. It is related to many of the same risk factors as sample size has been done yet.12 ± 14 It is necessary to describe adult obesity, most notably cardiovascular and psychosocial the present situation of childhood obesity among preschool factors.5 It is widely accepted that prevention is the best way children in Thailand and examine the factors related to SES, to control childhood obesity. The preschool age has been with the valid data of a large sample size. recognized as the most effective period to start to care about

Objectives and methods *Correspondence: N Sakamoto, TAKEMI Program, Harvard School of The aim of this study was to describe the prevalence of Public Health, 665 Huntington Avenue, Boston, MA 02115, USA. childhood obesity and to provide valid data in an Asian E-mail: [email protected] Received 8 September 1999; revised 3 July 2000; developing country. Moreover, by comparing two areas accepted 4 August 2000 having different SES, but similar cultural and geographical Obesity among preschool children in Thailand N Sakamoto et al 390 backgrounds, we were aiming to examine the in¯uence of was 5.8 y. There is no difference in age between the urban SES on childhood obesity. and rural subjects. The nutritional status was assessed by using the weight- for-height index and division into percentiles. The results were classi®ed into ®ve categories by percentile; <3, 3 ± <10, Subjects 10 ± 90, <90 ± 97, or >97 of the Thai national standards.15 The survey was conducted in Saraburi Province, located The standards were developed for boys and girls separately. 100 km north of , in the Central Region of Thailand. The statistics in 1996 showed that 292 kindergartens and nursery schools existed in the province and the number of Results preschool children in the second and third grade of the Anthropometry kindergartens was 10 640. Table 1 shows the weight-for-height index distributions for Two groups of study children were selected by two-stage the children at each percentile category in the urban and random sampling. Three districts were randomly selected rural areas. Signi®cant differences were recorded between from the 13 districts in the province for representatives of them. The prevalence of overweight (p ) was 16.1% in children in rural areas, and the Saraburi municipality was 90 ± 97 urban group, and 8.7% in rural group. Moreover the pre- chosen for representatives of children in urban areas. Two valence of childhood obesity (>p ) was over 22% in the kindergartens were randomly selected from each district for 97 urban group, which was three times that in the rural group. the rural group and two from the municipality for the urban group. All the children in the second or third grade in the selected kindergartens who attended their kindergartens on the day of the anthropometry, a total of 1441 children, were Socioeconomic status of parents included in the study. SES of the parents was compared between urban and rural measured by educational level of parents and household income. Table 2 shows the distribution of educational Methods levels attained by the urban and rural parents. Urban parents The study was designed as a cross sectional survey and attained a higher educational level than rural parents conducted in March 1997. The anthropometry was carried (P<0.001). More than 90% of the fathers and 80% of the out at each kindergarten in collaboration with two commu- mothers ®nished junior high school in the urban area, while nity nurses. Height and weight were measured using a tape measure and Tanita digital scale to the nearest 0.1 cm and Table 1 Distribution of children in each percentile category of weight- 0.2 kg respectively. The tape measure was used ®xed onto a for-height index by urban=rural pillar. Rural Urban Other data were obtained by questionnaires completed by (n ˆ 554) (n ˆ 603) parents at home. The questionnaires were distributed to the children who underwent anthropometry. The number of the No. % No. % distributed questionnaires was 1441, and the number of P<3 23 4.2 8 1.3 returned questionnaires was 1208 with a return rate of P3±10 32 5.8 18 3.0 83.8%. The cases with many blanks in the returned ques- P10 ± 90 410 74.0 343 56.9 P 48 8.7 97 16.1 tionnaires were eliminated from the study, and the data of 90 ± 97 P 41 7.4 137 22.7 the remaining 1157 cases (rural 554; urban 603) were used >97 for analysis. The age ranged between 4 and 6 y, and mean age P<0.001, chi-square test between urban and rural.

Table 2 Parents' educational level by urban=rural

Father* Mother*

Rural Urban Rural Urban (n ˆ 533) (n ˆ 596) (n ˆ 538) (n ˆ 599)

No. % No. % No. % No. %

>University 12 2.3 186 31.2 8 1.5 194 32.4 High school 106 19.9 275 46.1 68 12.6 220 36.7 Junior school 100 18.8 82 13.8 86 16.0 76 12.7 Primary school 298 55.9 51 8.6 341 63.4 106 17.7

*P<0.001, chi-square test between urban and rural.

International Journal of Obesity Obesity among preschool children in Thailand N Sakamoto et al 391 Table 3 Monthly household income level by urban=rural 282 Baht; n ˆ 554), while that of the rural was 9895 Baht (s.d. ˆ 10 344 Baht; n ˆ 482). Income was categorized into Rural Urban (n ˆ 482) (n ˆ 554) four levels, <5 000, 5 000 ± <10 000, 10 000 ± <20 000, and 20 000 as Mo-suwan did in his study in Hat Yai.16 Table 3 Baht=month No. % No. % shows the distribution of income levels of the urban and

<5,000 165 34.2 21 3.8 rural households. The income level of the urban households 5,000-10,000 201 41.7 96 17.3 was signi®cantly higher than that of the rural (P<0.001). 10,000-20,000 81 16.8 206 37.2 Nearly 80% of the urban households earn more than 10 000 20,000 35 7.3 231 41.7 Baht, while 76% of the rural earn less than 10 000 Baht per P<0.001, chi-square test between urban and rural. month. more than half of the rural parents had an educational level Socioeconomic status and body physique of less than primary school. The relationship between socioeconomic status of parents Total monthly household income was investigated. The and body physique of children was examined. The weight- mean income of the urban group was 29 797 Baht (s.d. ˆ 55 for-height index distributions of the children were compared

Figure 1 Proportion of weight-for-height index percentile categories of children by mothers' educational level (All, Urban, Rural).

International Journal of Obesity Obesity among preschool children in Thailand N Sakamoto et al 392

Figure 2 Proportion of weight-for-height index percentile categories of children by fathers' educational level (All, Urban, Rural).

by educational level of parents and monthly household level in all cases and the rural group. In the urban group, the

income. To avoid null cells, the comparisons were done proportion of p90 ±p97 of the lower income level was higher between two levels. Higher educational level was de®ned as by one percent than that of the higher income level. high school or more and lower educational level was de®ned as junior high school or less. Higher income level was de®ned as 10 000 Baht and lower income level was de®ned Discussion as <10 000 Baht. Though the etiology of childhood obesity is common in the Figures 1 and 2 present percentage distributions of the world, the processes of the development of childhood obe- weight-for-height index of the children by educational levels sity could vary in different populations, which have cultu- of the father and mother, by all, urban, and rural. The rally and socially different backgrounds. Although the

proportion of the percentile categories of p90 ±p97 and >p97 biological factors in Asian developing countries would be became higher, in higher parental educational levels, in all the same, the social backgrounds differ from those of the cases as well as sub-groups. Western developed countries. Therefore, it is important to As shown in Figure 3, the proportion of the categories of examine the social factors in¯uencing the development of

p90 ±p97 and >p97 became clearly higher in higher income childhood obesity in Asian developing countries.

International Journal of Obesity Obesity among preschool children in Thailand N Sakamoto et al 393

Figure 3 Proportion of weight-for-height index percentile categories of children by monthly household income level (All, Urban, Rural).

Sobal and Stunkard reviewed the relationship between SES ables directly relate to adults, therefore mechanisms that SES and childhood obesity, and they described a relationship in in¯uence childhood obesity need to be investigated. developing countries that appeared robust whereas that in Gerald et al gave a clue for clarifying the mechanism.19 developed societies was weak.17 Higher SES correlates with They studied selected characteristics of social environment childhood obesity in developing countries, whereas lower to be predictors of children's risk for obesity in the City of SES correlates with childhood obesity in developed Mobile, AL, USA. Obesity risk was measured by the child's countries. They concluded that the relationship between weight-for-height score and energy intake. Caretaker's SES SES and childhood obesity differed between developed and marital status, and social support predicted children's obe- developing countries. sity. Lower social class position, lower expressive social sup- Different mechanisms seem to be involved in the causal port, and unmarried status of the caretaker are associated pathways where SES in¯uences childhood obesity, between with a higher energy intake and a higher weight-for-height developed and developing countries. The social variables of score. This study suggested that the lower SES promoted SES that in¯uence obesity are education, income and occu- high-energy intake that resulted in obesity in a developed pation causing variations in behavior, which change energy country. consumption, energy expenditure and metabolism.18 This is In contrast to a developed country, a positive relation widely mentioned about adult obesity and those social vari- between SES and childhood obesity would be seen in

International Journal of Obesity Obesity among preschool children in Thailand N Sakamoto et al 394 developing countries. However, very few studies were con- References ducted, and it is the general understanding that lower SES 1 Cho KB, Park SB, Park SC, Lee DH, Lee SJ. The prevalence and causes lower energy intake and higher SES enables a higher trend of obesity in children and adolescents. J Korean Pediatr Assoc 1989; 32: 597 ± 605. energy intake. This has to be veri®ed by some studies at ®rst. 2 Ho TF. Childhood obesity in Singapore primary school children: Then the mechanism that SES in¯uence childhood obesity epidemiological review and anthropometric evaluation. J Singa- has to be explained, if it were proved correct. pore Paediatr Soc 1985; 27: 5 ± 40. As mentioned in the introduction, few studies on child- 3 Mo-suwan L, Junjana C, Puetpaiboon A. Increasing obesity in school children in a transitional society and the effect of the hood obesity have been reported in the Asian developing weight control program. Southeast Asian J Trop Public Health 1993; countries. Most of the studies were conducted in only eco- 24: 590 ± 594. nomically developed urban areas where af¯uent segments 4 Droomers M, Gross R, Schultink W, Sastroamidjojo S. High socio- are living, and high a prevalence of childhood obesity was economic class preschool children from Jakarta, Indonesia are taller and heavier than NCHS reference population. Eur J Clin Nutr found there. The conclusions of the studies suggested that 1995; 49: 740 ± 744. high prevalence of childhood obesity in urban areas is due to 5 Jonides LK. Childhood obesity: an update. J Pediatr Health Care urban high SES, without looking at the other side of the 1990; 4: 244 ± 251. evidence. Therefore a well-balanced perspective in study 6 Mossberg H-O. 40-year follow-up of overweight children. Lancet 1989; 2: 491 ± 493. design is still needed in developing countries. 7 Charney E, Goodman HC, McBride M, Lyon B, Pratt R. Childhood The present study, aiming at the comparison between antecedents of adult obesity. Do chubby infants become obese urban and rural areas in the same province, showed the adults? New Engl J Med 1976; 295:6±9. same correlation between SES and childhood obesity as the 8 Epstein LH. Childhood obesity. Pediatr Clin North Am 1985; 32: 363 ± 379. previous review. There were marked relationships between 9 Abraham S, Nordsieck M. Relationship of excess weight in chil- childhood obesity and parents' educational level and house- dren and adults. Public Health Rep 1960; 75: 263 ± 273. hold income. The correlation would explain the difference 10 Gasser T, Ziegler P, Seifert B, Molinari L, Largo RH, Prader A. in prevalence of childhood obesity between urban and rural Prediction of adult skinfolds and body mass from infancy through adolescence. Ann Human Biology 1995; 22: 217 ± 233. areas, where there was an obvious economic gap. Further 11 SEAMIC. Health statistics 1996. IMFJ: Tokyo; 1997. p 108. investigation is needed to solve the mechanism of which 12 Suttapreyasri D, Suthontan N, Kanpoem J, Krainam J, Boonsuya parents' SES in¯uences childhood obesity in developing C. Weight-control training models for obese pupils in Bangkok. J Med Assoc Thai 1990; 73: 394 ± 400. countries. 13 Yamborisut U, Kittidilokkul W, Chittchang U, Rojroongwasinkul In most of the developing countries, childhood obesity N, Tontisirin K. Factors in¯uencing obesity in school children in was thought not to be an important health problem until Bangkok metropolitan area. Siriraj Hosp Gaz 1993; 45: 759 ± 769. now, since it does not directly connect with death and its 14 Mo-suwan L, Junjana C, Puetpaiboon A. Increasing obesity in school children in a transitional society and the effect of the prevalence has been recognized as very low. Now childhood weight control program. Southeast Asian J Trop Med Public Health obesity is emerging in certain af¯uent segments of society. It 1993; 24: 590 ± 594. will lead to serious health problems in the near future, since 15 Nutrition Division, Ministry of Public Health, Thailand. Body the af¯uent segments could expand along with economic weight and height references and nutritional status indicators of Thai population, aged 1 day ± 19 years. 1994. p 1 ± 20. development. Japanese experience suggests that the preva- 16 Mo-suwan L, Geater AF. Risk factors for childhood obesity in a lence of childhood obesity rises following an increase in transitional society in Thailand. Int J Obes Relat Metab Disord adult obesity.20 The increase in adult obesity has been 1996; 20: 697 ± 703. reported in developing countries already.21 To improve the 17 Sobal J, Stunkard AJ. Socioeconomic status and obesity: a review of the literature. Psychol Bull 1989; 105: 260 ± 275. future health situation, particularly for controlling adult 18 Sobal J. Obesity and socioeconomic status: a framework for obesity in the future, measures for controlling childhood examining relationships between physical and social variables. obesity need to begin in developing countries, since child- Medical Anthropology 1991; 13: 231 ± 247. hood obesity is positively correlated with eventual adult 19 Gerald LB, Anderson A, Johnson GD, Hoff C, Trimm RF. Social class, social support and obesity risk in children. Child: Care obesity. Health & Development 1994; 20: 145 ± 163. 20 Ministry of Education, Science, Sports and Culture. School Health Statistics. Printing Bureau, Ministry of Finance: Tokyo; Acknowledgements 1970 ± 1996. 21 Khor GL, Gan CY. Trends and dietary implications of some This study was supported by a grant from the Ajinomoto chronic non-communicable diseases in Peninsular Malaysia. Foundation for Dietary Culture. Asia Pasi®c J Clin Nutr 1992; 1: 159 ± 168.

International Journal of Obesity