Lifestyle factors associated with underweight among Japanese adolescents: a cross-sectional study

Underweight in children and adolescents poses a considerable public health problem internationally [1]. For example, an underweight status in adolescents was shown to be associated with scoliosis, osteoporosis, pubertal delay, and psychiatric disorders [2]. In addition, it was recently reported that underweight in adolescents is associated with poor perceived health [3]. Thus, underweight in children and adolescents is a serious health condition with substantial consequences for development, health, and well-being [4].

A previous study showed that the tracking correlation for body mass index (BMI) between ages 13 and 19 was 0.85 [5]. This finding suggests that underweight adolescents are likely to be underweight in young adulthood. Therefore, it is important to prevent underweight in adolescents for their health during both adolescence and adulthood, particularly in Japan, where the prevalence of underweight has increased among adolescents during recent years [6, 7].

For prevention of underweight, it is important to investigate risk factors, especially lifestyles, because these can be modified. Lifestyle factors are associated with BMI; for instance, physical activity and eating behaviors were reported to be associated with BMI [8‐10]. Therefore, it is necessary to consider both physical activity and eating behaviors when examining lifestyle factors associated with underweight. Although the relationship between physical activity and underweight among adolescents has been investigated [11, 12], eating behaviors such as eating speed, snacking, and breakfast were not considered in these studies. Because previous studies have reported that eating behaviors such as snacking, skipping breakfast, and eating speed were associated with overweight/obesity [13‐15], we hypothesized that underweight adolescents were less likely to snack and skip breakfast and were more likely to eat slowly.

Accordingly, the aim of the present study was to investigate the association of lifestyle factors including physical activity and eating behaviors with underweight among Japanese adolescents.

Among all 2641 subjects, 28 refused to participate in the questionnaire survey and physical examination (participation rate: 98.9%), and 263 were excluded due to missing data about variables in the present study. Thus, data from 2350 subjects were analyzed.

Table 1 shows characteristics of study participants by sex. Boys were significantly higher and heavier than girls. BMI-SDS was significantly higher in boys than in girls. The prevalence of overweight/obesity was higher in boys than in girls, while the prevalence of underweight was higher in girls than in boys. There was a statistically significant association between boys and girls in exercise. A statistically significant association between boys and girls was found in the proportion of those who answered “Yes” to the questions about skipping breakfast or snacking while watching television or reading books. A significantly higher proportion of girls reported eating slowly compared with boys.

Comparisons of characteristics between the normal weight group and the underweight group in boys are shown in Table 2. Anthropometric variables were significantly lower in the underweight group than in the normal weight group. The proportion of those who ate slowly in the underweight group was significantly higher than that in the normal weight group.

Table 3 shows comparisons of characteristics between the normal weight group and the underweight group in girls. Anthropometric variables in the underweight group were significantly lower than those in the normal weight group. There was a statistically significant association between the normal weight and the underweight group in exercise. The proportion of those who ate slowly was significantly higher in the underweight group compared with the normal weight group.

Adjusted ORs of lifestyle factors for underweight and their 95% CIs were calculated in each sex (Table 4 in boys and Table 5 in girls). Boys who ate slowly were more likely to be underweight than those who did not eat slowly (OR: 2.78, 95% CI: 1.77–4.39). Girls who did not exercise were more likely than those who exercised to be underweight (OR: 1.64, 95% CI: 1.07–2.51), and there was a significant dose-response relationship between the frequency of exercise and underweight (P for trend = 0.021). Girls who ate slowly were more likely to be underweight than those who did not eat slowly (OR: 2.69, 95% CI: 1.81–3.98).

Table 6 showed the adjusted ORs of lifestyle factors for underweight and their 95% CIs in total participants. Adolescents who did not exercise were more likely than those who exercised to be underweight (OR: 1.59, 95% CI: 1.11–2.27), and there was a significant dose-response relationship between the frequency of exercise and underweight (P for trend = 0.012). Adolescents who ate slowly were more likely to be underweight than those who did not eat slowly (OR: 2.67, 95% CI: 1.99–3.59). Girls were not more likely than boys to be underweight (OR: 1.16, 95% CI: 0.86–1.56). There were no effect modifications of sex and lifestyle factors on underweight.

The present study investigated the relationship between lifestyle factors and underweight among Japanese adolescents. In this study, eating slowly and exercise were associated with underweight. To the best of our knowledge, this is the first study regarding the association of lifestyle factors (eating speed and exercise) with underweight among adolescents in Japan. However, the causality between lifestyle factors and underweight should be carefully interpreted because this was a cross-sectional study.

In this study, there were statistically significant differences between boys and girls in some anthropometric variables and lifestyles. A previous study reported that adolescence is characterized by a global acceleration of growth and maturation, with differential changes between both sexes [26]. In addition, a recent study showed that there were statistically significant differences between boys and girls in lifestyle factors such as breakfast, eating speed, and physical activity [14]. Therefore, we analyzed the data separately for each sex to examine the association between lifestyle factors and underweight among adolescents.

In our study, the prevalence of underweight was 8.2% in boys and 11.2% in girls. According to national statistics from Ministry of Education, Culture, Sports, Science and Technology, the prevalence of underweight (defined by percentage of overweight, which is the modified weight-for-height method [27]) in 2008 was 2.3% among boys (ages 12) and 3.9% among girls (ages 12) in Japan [28]. When the underweight in our study was defined by the same criteria using percentage of overweight, the prevalence of underweight in 2008 was 2.6% in boys aged 12 years and 3.4% in girls aged 12 years, which were similar to the results of the national statistics in Japan. A recent study reported that the prevalence of underweight among Dutch boys and girls (ages 12–18) in 2009 was 9.1% and 10.4%, respectively [4]. Elinder et al. reported that the prevalence in a Swedish cohort of boys and girls (mean age: 15.6) in 2002 was 3.5% and 11.1%, respectively [11]. A previous study among German adolescents reported that the prevalence was 12.6% among boys and 19.1% among girls in 2001/2002 [29]. Because the definition of underweight, the year of the data collection, and the age of study subjects differed by studies, future studies will be needed to compare the prevalence of underweight between Japan and other countries.

Eating slowly was significantly associated with underweight in this study. One of the reasons could be due to total energy intake. Murakami et al. showed that rate of eating was positively associated with energy intake, BMI, and the risk of overweight [30]. Otsuka et al. found a positive association between “the rate of eating” and “energy intake and current BMI”, suggesting that eating fast would lead to obesity [31]. Therefore, we hypothesized that total energy intake in the eating slowly group was lower than that in the not eating slowly group, which could lead to lower BMI in the eating slowly group compared with the not eating slowly group in the present study. In fact, BMI among the eating slowly group was lower than that among the not eating slowly group in our study (data not shown). Another possible explanation might be due to appetite. A previous study reported that higher BMI SD scores were associated with lower satiety responsiveness and higher food cue responsiveness [32]. In addition, Webber et al. showed that there was a significant negative association between “satiety responsiveness/slowness in eating” and “physique (underweight, lower healthy weight, higher healthy weight, overweight, and obese)”; the score of satiety responsiveness/slowness in eating among underweight children was higher than that among healthy weight children [33]. These studies [32, 33] suggested that underweight adolescents could have a poor appetite, which leads to eating slowly. However, it is difficult to determine the temporal sequence of eating slowly and underweight in our study because the present study was cross-sectional. Therefore, further studies will be needed to evaluate the causal relationship between eating slowly and underweight.

In the present study, girls who did not exercise were more likely to be underweight than those who exercised. These results persisted even after adjustment for age, birthweight, and all lifestyle variables. Recent studies have shown that low physical activity was associated with underweight [11, 12], which is consistent with the results in this study. One of the reasons for the association between exercise and underweight could be due to the influence of exercise on lean body mass. Elinder et al. showed that physical activity is a major determinant of skeletal muscle growth leading to an increase in lean body mass [11]. In addition, a previous study reported that exercise exerts a positive effect on bone mass [34]. In fact, it was recently shown that total body lean mass among an inactive group was lower than that among average or active groups [35]. Another reason is due to health problems related to underweight. A recent study reported that an underweight status in adolescents was associated with several health problems [2]. These health problems in underweight adolescents could lead to physical inactivity.

There are some limitations to this study. First, information regarding lifestyle factors such as eating speed and exercise was collected by a self-administered questionnaire, which might be not objective. However, a previous study showed a high level of concordance between self-reported and friend-reported rate of eating [10]. In addition, Petty et al. indicated that self-reported eating rate aligned with laboratory-measured eating rate [36]. Moreover, Chen et al. reported that schoolchildren’s self-reported physical activity is in accordance with objective data [37]. Second, the present study results could be affected by some potential confounders. For example, socioeconomic factors were reported to be associated with underweight [29, 38]. Moreover, the social pressure to be thin, body image perception, and gestational age could affect our study findings. Because this information was not considered in the present study, residual confounding cannot be ruled out. However, a recent study showed that Japan is still one of the most egalitarian nations in the world, and social inequalities within the population are less expressed [39]. Therefore, the influence of socioeconomic factors on this study might not be substantial. Third, the age of each subject was indicated as “12 or 13” (years) in this study. Therefore, the age- and sex-specific BMI cut-off points for adolescents aged 12 or 13 years [19, 20] were applied, whereas the cut-off points for those aged 12.5 years were not applied. Finally, our study subjects (the number of children: 298 in 2001, 317 in 2002, 293 in 2003, 315 in 2004, 312 in 2005, 354 in 2006, 380 in 2007, and 372 in 2008) were from one town, which might limit generalizability to other populations.

The present study showed that eating slowly and exercise were associated with underweight among Japanese adolescents. The results suggest that eating speed and exercise need to be considered in school health programs for healthy body weight, although future studies are needed to verify our findings.