The main findings in this population-based longitudinal study are that the subclinical groups of individuals with different combinations of ED symptoms displayed distinct age- and gender-related trends. Females showed a significantly higher risk than males for purging behaviours. The prevalence of binge eating and CB in females and CB in males gradually declined with the transition from adolescence to young adulthood. Furthermore, individuals with subclinical forms of ED differed significantly on measures of general eating and psychosocial problems. Specifically, in terms of our psychosocial measures, purging emerged as the most serious type of behaviour when compared with binging without CB, binging-CB and non-purging CB. On the other hand, individuals in the binging without CB group were the least disturbed and in most aspects comparable to the reference group, i.e. those without inappropriate eating behaviours.
To our knowledge, this is the first population-based study documenting the relationship between distinct types of ED behaviour (binging without CB, binging-CB, purging and non-purging) and different measures of general eating and psychosocial problems. The study shows a particularly high symptom load in individuals who purged, since those who purged, largely by self-induced vomiting, had significantly higher levels of appearance dissatisfaction, anxiety and depressive symptoms, alcohol consumption, self-concept instability and loneliness. A national study of high school students in the USA also reported vomiting for weight control as the most clinically significant behaviour and one that may be a particularly deleterious component of ED [34
]. In this respect, individuals engaging in purging behaviours should probably be targeted as a high-risk group.
In contrast, Mond and colleagues reported that the combination of routine binge eating and CB was associated with higher ED psychopathology and impairment in mental and physical functioning than was the occurrence of CB alone, regardless of the type, i.e. purging or non-purging [8
]. These divergent findings may be the result of different definitions for binge eating and different scales for measuring psychopathology. We defined binge eating only in terms of the amount of food consumed. Mond et al., on the other hand, integrated loss of control in their definition, and this could add an important dimension to the relationship between binge eating and psychological disturbance. As well, Mond et al. used the Eating Disorder Examination Questionnaire and the Short-Form Disability Scale, whereas we used a wider range of eating and psychosocial measurements, providing in-depth information about the differences among the groups. In general, both community- and clinical-based studies indicate that individuals with purging subtype of bulimia nervosa have more severe psychopathology than those with non-purging bulimia nervosa and binge eating disorder, thereby demonstrating that purging behaviors may increase the severity level of psychopathology among those with the full-threshold ED [35
]. These studies support our finding that purging behaviors are related to particularly serious levels of psychosocial problems. Due to few participants with a combination of binging and purging in our study, we could not investigate how individuals with purging only symptoms differed from those with a combination of purging and binging. More research would help to clarify distinctions between these groups, which, in turn, could have important implications for the future classification of subclinical ED.
To an extent, findings from this study parallel those of earlier community-based studies [16
]. The prevalence of binge eating and several CB behaviours appears to be higher among females than males. The gender difference in the prevalence of symptoms was especially notable for purging, with females at a significantly higher risk. A possible explanation is that females have a stronger desire to lose weight and a higher drive for thinness than males, leading them to engage in more purging behaviours [16
]. Alternatively, males may be more uncomfortable reporting purging behaviours than females [38
In both genders, the age-related decline in prevalence rates for binge eating and CB may be partially explainable by roles associated with the transition to adulthood, specifically partnering and motherhood (for females) [17
]. However, it may be possible that psychological problems which in many cases are associated with disordered eating may not be reduced to a comparable degree as disordered eating symptoms in the transition to adulthood. For instance, even though Patton et al. reported most part of adolescent eating disorders to be limited to the teens, co-morbid conditions, such as depression, anxiety and binge drinking persisted to a much larger degree into adulthood [14
]. Such findings may indicate that the decline of symptoms of disordered eating not necessarily is related to symptom alleviation for other mental health problems.
Moreover, even though we saw this declining trend for both males and females, binge eating among females stands out in terms of the magnitude of the reduction over age. This may be related to a greater perception on the part of females that binge eating is a problematic behaviour [40
]. Along the same line of thought, males report a more positive affect (feeling happy) than females after binging [41
Overall, the prevalence rates of symptoms in this study are lower when compared with the rates from two studies of US college men and women [43
]. They are, however, similar to rates found in two community-based studies conducted in Australia and the USA, respectively [5
]. Nonetheless, factors such as the use of different assessment measures, lack of standard definitions for syndromes, differences in methodological approaches and characteristics of sample populations may limit the comparability of the findings.
This study has a number of limitations that warrant consideration. First, the BITE does not specify a time frame for the occurrence of symptoms, thereby providing somewhat limited information about the duration of a respondent's ED. Second, binge eating determined solely by the amount of food consumed may be a somewhat limited indicator for binge eating psychopathology. The inclusion of additional BITE items such as feeling distress after binging, binging alone, and urge to binge or loss of control might better distinguish binging from normal eating. This matter bears further investigation. Third, even though it has been argued that full-syndrome ED may not differ qualitatively from sub-threshold levels [45
], it remains to be seen whether the findings from this study will be supported in studies using diagnostic categories of ED and in those using diagnostic interviews. Fourth, age-related differences in prevalence rates of ED behaviours may reflect time of measurement effects and differences in exposure to risk factors rather than effects associated with growing into adulthood. Other studies suggest that a wide range of ED behaviours remain relatively stable over time, supporting the negligible effect of time on the trend of prevalence [5
]. Furthermore, separating the effect of age from period and cohort through statistical model estimation has proven problematic and has led to incorrect conclusions [48
]. Fifth, limited statistical power resulting from the small sample sizes should be considered. More specifically, because the number of participants in the purging CB and binging-CB groups was rather small, potential differences among the groups might not have been detected. Sixth, to obtain more differentiated information, the binging-CB group should ideally have been divided into those who binged and purged and those who binged and engaged in non-purging CB. Unfortunately, the small number of participants reporting both binging and purging behaviours precluded such analyses. Finally, we only followed about 25% of the representative sample at T0. Even though most of the attrition was planned, and that attrition analyses showed no significant differences between those who dropped out and those who completed the study, the large proportion of drop out at the follow-up could be a source of bias.