Ethics statement

The survey and data analysis was approved by the Research Subcommittee of the Family Planning Association of Hong Kong. This study was supported by the University Research Committee Strategic Research Theme (SRT) of Public Health, The University of Hong Kong.

Sample

The Youth Sexuality Study is the longest running community-based sexuality survey in Hong Kong. The objective of the study is to assess sexual knowledge and attitude among young adults in the community. The survey was first conducted in 1986 and from then on once every 5 years. The survey data obtained in 2006 was used for data analysis of the current study. Our sample was obtained from the Census and Statistics Department under the Hong Kong SAR government who had a complete sampling frame of all living quarters in Hong Kong. Upon our request for about 10,000 living quarters, that department used a systematic sampling method to draw samples from the sampling frame. For each selected living quarter, we used Kish grid method to randomly select one household member who was aged 18 to 27 as respondent to fill in a self-administered questionnaire at his/her own time and return it before a specific time. Among the 10,001 living quarters being selected and visited, 2,369 had no one at home at the time of visit, were rejected by the first contact person of the living quarter, or had invalid addresses. Of the remaining 7,632 living quarters, 1,356 households contained eligible respondents. Of the eligible respondents, 41 (3%) chose not to participate, 78 (5.8%) dropped out half-way, and 32 (2.4%) did not return the questionnaire, leaving 1,205 young adults (88.9%) who successfully completed the questionnaire.

Measures

Body mass index (BMI) and weight status

Body shape dissatisfaction

BSD was measured with a rating scale from Stunkard visual figures which consists of nine gender-specific body silhouettes ranging from very thin (value = 1) to very big (value = 9) (Figure 1). Validity in this figure rating scale was affirmed by a previous research study [26], and was found to have good test-retest reliability [27]. Respondents were asked to choose the figure that best represented their "current shape" and then their "ideal shape". The absolute value of BSD was derived from the discrepancies between the chosen current and ideal figures, regardless of direction of thinness or fatness preference. Respondents were then classified into three groups: (1) desired to maintain the current body shape (current = ideal), (2) desired to have a slimmer body shape (current > ideal), and (3) desired to have a bigger body shape (ideal > current).

Statistical analysis

Descriptive statistics were used to illustrate the distribution of BMI, weight status, body shape preferences and BSD. To examine sex differences, chi-square tests were conducted to test the differences for categorical variables, and independent samples t-tests were used for continuous variables. Furthermore, the relations of BSD with education level and employment status were examined for each weight status with chi-square tests.

Three multiple logistic regressions were conducted to calculate odds ratios of desire to maintain the current body shape, desire to have a slimmer body shape, and desire to have a bigger body shape for weight statuses based on BMI, age, education level and employment status for each sex. For each regression model, other respondents not in the response category were treated as the reference category. Adjusted odds ratios were obtained to estimate the predictive power of each variable. Infinite or zero odds ratios, with very high p-values, were obtained for overweight and obese in the women's regression models because all overweight and obese women respondents desired a slimmer body shape. In the final model, therefore, we aggregated women of normal weight, overweight and obese into one category for a more robust estimation. The Statistical Package for the Social Science (SPSS version 16.0) was used for all the analyses.

Distribution of actual weight status and BSD

Overall, 25.7% of the young adults were classified as underweight, 9.3% were overweight, and 10.7% were obese. Underweight was more common in women while overweight or obesity were more common in men. More men were obese (15.4%) than were overweight (11.3%). Significant sex differences were observed for BMI, current body shape and ideal body shape. BSD was found to be significantly greater for women than men (t = -11.065, p > 0.01). When the absolute value of BSD was compared, the sex difference was no longer statistically significant (t = 1.764, p = 0.078).

For women, current body shape ratings of 3 and 4 were the most common (Figure 2). For men, the ratings were more evenly distributed from 3 to 6. Most women had an ideal body shape rating of 2, while most men desired higher ratings of 4 or 5. No subjects chose ratings of 6 or above as their ideal body shape.

Mismatch between weight status and body shape dissatisfaction

Among men who had BSD, about half of them desired to have a slimmer body and half of them desired to have a bigger one (Table 2). We observed that (i) 23.3% of those underweight desired to maintain the current body shape, (ii) 12.2% of underweight and 27.2% of normal weight desired a slimmer body, and (iii) 45.3% of men with normal weight desired a bigger body shape. For women, (i) 40.3% of those underweight expressed no BSD and 30.9% desired an even slimmer body shape despite being underweight. In other words, 71.2% of those underweight did not desire a bigger body shape, and (ii) 75.5% of those normal weight desired a slimmer body shape.

Determinants of desiring slimmer body shape

Figures 3 and 4 depict the distribution of BSD by education level for men and women respectively. Men who were overweight and with lower education level were less likely to desire a slimmer body shape than those overweight men with higher education level (Figure 3). Underweight women with lower education level were more likely to be, with marginally statistical significance, satisfied with their current body shape (Figure 4). Most underweight women with junior secondary education level or below were satisfied with their body (83.3%), compared with only 34% among their counterparts with educational attainment at undergraduate level or above. On the other hand, underweight women who were full-time students were more likely to desire a bigger body shape (37.7% vs 24.3%), and less likely to desire to maintain their current body shape than full-time working respondents (28.6% vs 45.6%) (${\chi }_4^2=9.2$, p = 0.057). Such difference by employment status was not apparent among female respondents who were normal weight, overweight, and obese, nor among male respondents of any weight category.

Mismatch between actual weight status and body shape satisfaction

There are two competing views on the relationship between actual physical status (e.g. BMI) and subjective body image evaluations. Some studies have found that BMI is an important determinant of body satisfaction [10, 21, 30], while others have argued that objective indices such as BMI are less important than psychological constructs such as self-esteem in affecting body dissatisfaction [2, 8, 10]. Our findings supported the association, though we also identified the existence of "mismatch" between actual weight status and body shape dissatisfaction [31, 32]. The majority of the female respondents with normal BMI weight status still desired for a slimmer body. Those women who had achieved healthy body mass were not necessarily more satisfied with their body shapes than those who were underweight, while underweight women were even more likely to consider their body weight as "about right" and expressed the smallest disagreement between current and desired body shape [10, 12]. This mismatch is coherent to the fact that desire for ideal body shape is highly biased by sociocultural norms, environmental neighborhood affluence and media impact for women [7, 9, 20, 33]. On the other hand, men also had the mismatch but to a less extent than women. Similar to women, men with normal weight did not have higher intention than underweight and overweight men to maintain their current body shapes. Although nowadays having BSD is common in developed and affluent societies, the actual weight status should be taken into account in making any choice of actions to acquire the desired body shape. The emphasis on subjective judgment rather than objective indicators on current body shape desire might lead to unnecessary body weight reduction strategies and adverse psychological distress among the general public. This misperception of BMI and current body shape could be a major problem and further studies on this are warranted. To avoid such public health crisis arising from current and ideal body shape misconception, public health education aiming to raise awareness on actual weight status in desiring body shape is necessary.

Education level and body shape dissatisfaction

One previous study has found that poor body satisfaction was more prevalent among women with higher education attainment than women with lower education attainment [34]. On the other hand, the impact of obesity on women's general health was greater among those with a lower education level [35]. Our study attempted to explore the role of education attainment and employment status on the mismatch between weight status and body dissatisfaction, with an extension to male sample. Our findings showed that overweight men with lower education level had less desire for a slimmer body shape than their counterparts with higher education level. For women, the desire to stay in the current body shape was more common among the underweight with lower education level, compared to their counterparts with higher education level. These findings supported that the role of education level on BSD is not only influential to women but also to men. Both overweight men and underweight women with low education level tended to neglect actual weight status in pursuit of an ideal body shape. The possible explanation is that these people with lower education level lack awareness about their actual weight status due to insufficient knowledge of determining fatness and thinness. Both overweight and underweight population with a low education level should be targeted in educational programs regarding body weight control.

Underweight problem among young adults

Since obesity and underestimation of weight status have been very much emphasized within the community and public health constituencies globally [16, 36], relatively less attention has been paid to the underweight population and its health consequences in the territory. Our study showed that one-fourth of the young adults were underweight (BMI<18.5), compared with 9.3% and 10.7% who were overweight and obese respectively. Our findings are coherent with the recent territory-wide based Population Health Survey which found similar underweight prevalence in the territory [25]. In that study, overall 10.3% of the community were underweight based on BMI, and 28.8% were overweight or obese. However, among the age group of 15-24, the corresponding figures were 25.1% and 13.8%, respectively. These figures supported that underweight is as prevalent as overweight and obese among young adults in the community. Media induced perception on the myth of a slimmer self-body image through TV-ads, fashion magazines, beauty contests, and reinforcement of culturally idealizing slimness by diet and fashion industries could also be the possible factors underlying the desire of the slim ideal image and weight-loss behaviors [37]. Indeed, underweight among young adults is associated with adverse health outcomes such as maladaptive weight loss behaviors, disordered eating and other mental health consequences [8, 38]. Those underweight who desired to have a slimmer body shape could be a high-risk group for these adverse outcomes.

Limitations of the study

Although our study yielded a representative youth sample, the small sample size may limit the reliability of our statistical estimate. For instance, only 8.3% of the respondents reported an education level below junior secondary school. Hence, our conclusion about the impact of low education level on BSD can only be drawn on this group with a small sample size. We suggest that a larger sample size and proportion of people having a low education level would be needed for more in-depth analysis to understand how education level is related to BSD.

Methodological issues about the BMI cut-off and the use of figure rating scale of this study should also be acknowledged. BMI cut-off for obesity is directly related to various health risks and risk factors of morbidity and mortality [39], however, there are fewer consensus on cut-offs for underweight in adults [40]. Using the cutoff value of 18.5, the WHO-defined thinness grade I (mild thinness), as the cut-off for underweight, 25.7% of our subjects were underweight. The WHO expert committee, however, has pointed out that though the single cut-off point of 18.5 for specified mild deficiency was a reasonable value to define underweight, it has little experimental support [41]. Especially for men aged 18 to 25, their average BMI were lower than that of men aged 26 to 40. In a recent study on definition of thinness among young adolescents, the universal cut-off BMI for underweight children at age 18 was more suitably set at 17 [40]. Using the value of 17 as the cut-off, only 6.3% of men and 10.4% of women of our respondents were underweight. Such substantial difference owing to the use of different cut-offs of underweight would result in large discrepancies in assessing prevalence of underweight and its associated health risk. Thus, further work need to be conducted to develop an empirically based cut-off for underweight among young adults in Asia.

There are two main measurement tools to assess body image satisfaction, (i) assessment of satisfaction with specific body parts [2, 8, 21, 42, 43].; and (ii) Figure rating scale (FRS) [10, 44]. Specifically on body shape dissatisfaction, Neighbors & Sobal (2007) applied a figure rating scale (FRS) which requires respondents to choose their actual perceived and ideal body shape from a series of body silhouettes of increasing body size and shape, with body shape dissatisfaction being operationalized as the discrepancies between actual and ideal body shapes shown on the questionnaire [26]. Measurement tool (ii), which is the tool being applied in our study, is more relevant to our research questions to assess overall body shape than (i). However, it should be noted that the body silhouettes were based on Western norms and its suitability for Asian Chinese is yet tested. Since a standard weight status according to BMI for Asians is already in use, developing an ethnically specific FRS would be particularly worthwhile.