Do sex differences in reported weight loss intentions and behaviours persist across demographic characteristics and weight status in youth? A systematic review

Existing literature published between January 1990 and May 2017 was searched electronically using three different databases: PubMED, Web of Science, and PsycInfo. The full search strategy is available in Additional file 1. The first search was conducted in May of 2015, during which 3355 articles were identified (Fig. 1). A second search was conducted in May of 2017 and yielded no new results.

The criteria for inclusion in the review were: (1) participants were 18 years of age or younger, (2) past or current weight loss intention alongside use of weight loss strategies were measured (i.e., either a subsample of those reporting weight loss intentions was analyzed, or question wording was such that weight loss intention could be assumed, such as “do you exercise to lose weight”), (3) the study used a Canadian or American sample, (4) the article was written in English, and (5) the study reported data on at least one of the characteristics of interest to the review. Articles were excluded if: (1) the study involved an intervention (e.g., behavioural counseling for weight loss with no baseline data on behaviours provided), (2) the study sample was an exclusively at-risk or other special population (e.g., diagnosed with an eating disorder), (3) weight loss strategies were too broadly defined without detail (e.g., “Do you diet to lose weight?”), or (4) the study was qualitative in nature. Of note, studies that assessed weight intention by combining weight loss and weight maintenance objectives (e.g., “do you exercise to lose weight or keep from gaining weight”) were excluded, as research suggests that weight loss and weight maintenance are cognitively different objectives that may lead to different health outcomes [32].

This review focuses on the United States and Canada, given that weight loss intentions are notably highest among this population of youth [33], more data is available for this region than in other countries, and the makeup of characteristics of interest to this review (e.g., race/ethnicity) likely differ substantially in other regions. Further, as the intent of this research is to highlight areas where further research is needed to better inform decisions made by public health agencies, relative homogeneity amidst study samples should allow for more generalizable findings. Studies excluded from other countries include those conducted in Australia and New Zealand, Eastern Europe, China, Iran, India, Ireland, Mauritius, Spain, Sweden, Switzerland, and Taiwan.

All records identified in the search were retrieved. The software EndNote X7 [34] was used to filter duplicate records and organize references. Both authors independently reviewed the titles and abstracts for articles considered for inclusion. Disagreements were resolved through discussion. Reference lists of studies deemed relevant for the review were examined to identify additional studies for inclusion. The reference lists of those identified were then consulted, and so on, until no additional articles could be identified as relevant for inclusion. During this process, 51 additional articles met the necessary criteria, and were deemed relevant for consideration in the review.

Descriptive data across studies on weight loss intentions and strategies were extracted in order to report overall prevalence, and prevalence according to the following sub-categories: (1) weight status (underweight, normal-weight, and overweight/obese), (2) grade level (elementary school [approximately ages 5–10], middle school [ages 11–13], and high school [ages 14–18]), and (3) race/ethnicity (African American, Asian American, Caucasian, Hispanic, Native American). Additional details on the coding of these sub-categories are provided in their relevant results sections. Sex was taken into consideration within each of these subsections. Both authors were involved in the data extraction process – the first author conducted the primary data extraction, which was then verified by the second author. Weight loss strategies were broadly grouped into three categories based on those proposed by Neumark-Sztainer and colleagues (healthy, unhealthy, and extreme; see Table 1) [35]. In general, healthy behaviours are those that are recommended for weight management (e.g., exercise, dietary changes), while unhealthy and extreme behaviours (e.g., laxative use, fasting) are those that do not provide adequate nutrient intake, with extreme behaviours representing those that are most dangerous and counterproductive to health [35, 36].

Both authors independently assessed the quality (pertaining to all aspects of the publication including introduction, methods, results and discussion) and potential bias of each study using the AXIS tool. Disagreements were resolved through discussion. The AXIS tool was specifically designed to be utilized across disciplines to critically appraise cross-sectional studies. With an international panel of 18 experts, the AXIS tool was developed through several iterative rounds of a Delphi process in order to reach consensus [37].

The demographic characteristics of each study sample are presented (Table 1). All but one study [20] was conducted in the United States; three used data from large, nationally representative datasets (Add Health [19, 38], and the Youth Risk Behavior Survey) [39]. Eighteen of the nineteen studies were cross-sectional. For the one longitudinal study, only data from the last year of study is reported here [40].

Exercise was the most commonly assessed healthy weight loss strategy (13/19 studies), while the prevalence of using dietary changes was the least frequently assessed (8/19 studies). Over half the studies assessed the use of unhealthy strategies (12/19), and three-quarters assessed the use of extreme strategies (14/19 studies). Skipping meals and fasting were the most commonly assessed unhealthy strategies (8/19) and using diet pills/aids was the most commonly assessed extreme strategy (14/19). No studies evaluated the use of multiple strategies simultaneously.

Results of the quality assessment are shown (Table 2). None of the studies provided a power calculation or sample size justification. Non-response bias was not a concern for seven studies, while sampling procedures from two studies raised concern about non-response bias. There was not enough information to determine a potential non-response bias for the remaining ten studies. Only one study presented information on the non-responders in order for the reader to determine issues of generalizability. Approximately 40% (8/19 studies) did not report psychometric properties nor referenced previous work demonstrating the reliability and validity of their weight intentions and weight loss strategy measures. Details pertaining to ethical approval or participant consent were not provided for the majority of studies (12/19). As studies reporting no sex difference, or a higher endorsement of weight loss strategies by males compared with females were published and reported in this systematic review, publication bias is suspected to be minimal.

For easier readability, overall prevalence information is presented first, followed by summary information organized by the sub-categories of weight status (separately for measured weight, and perceived weight), grade level, and race/ethnicity. A summary of all results presented is shown (Table 3).

Based on measured weight status, the total prevalence of overweight ranged from 22 to 42% [18, 19, 21, 25], from 21 to 43% in females [18, 21, 25, 38, 44], and 35–41% in males [18, 21, 25]. Based on perceived weight status, the prevalence of overweight ranged from 12 to 54% [18, 19, 21, 25], and ranged from 14 to 53% in females [18, 21, 25, 38, 44] and 9–58% in males [18, 21, 25].

While difficult to determine the exact age students attend each grade level, generally speaking in the United States and Canada the school system is such that the ages are approximately 5–10 years of age during elementary school, 11–13 years of age during middle-school, and 14–18 years of age during high-school.

Of note, the terms used here to describe race/ethnicity have been chosen to represent a set of interchangeable terms provided in the original articles, and are grouped as follows: Study participants referred to in the original articles as Caucasian, White or non-Hispanic white are referred to here as “Caucasian.” Those referred to as African-American, African-Canadian, or Black are referred to here as “African-American.” Those referred to as Hispanic, Mexican-American, or Puerto Rican are referred to here as “Hispanic”. Study participants referred to as Native-American, American-Indian, or Alaska Native in the original articles are referred to here as “Native-American.” Those referred to as Asian-American, Asian, Pacific Islander, or South Asian are referred to here as “Asian-American”.

Our results demonstrate that up to two-thirds of youth report trying to lose weight. The prevalence of trying to lose weight reached 74% in females and 63% in males, however results from statistical comparisons show mixed findings, with some studies reporting no sex difference in weight loss intention, and some reporting higher endorsement among females. Mixed findings may be observed due to differences in demographic characteristics, such as age and race/ethnicity. For instance, several studies report no sex difference in weight loss intentions among elementary school children, while studies conducted among middle and high school students consistently report that females are more likely than males of the same age to be trying to lose weight. Indeed, research has demonstrated that children as young as 8 years old express concerns about their weight and body shape [50], and that these concerns, as well as actions taken to control body weight, increase substantially as youth move from elementary to middle school [48]. This increase in attention to body weight in young females is likely due to a combination of factors, including onset of menarche, dating, and increased peer pressure [48], as well as increased exposure to societal ideals of beauty through television and social media during childhood and emerging adolescence [51]. While the overall finding that young females are more likely to be trying to lose weight compared to males is consistent with findings among adults [52], the reasons for wanting to lose weight are likely qualitatively different. Indeed, while both adults and adolescents cite self-esteem as an important factor in wanting to lose weight, research suggests that children are more motivated by a desired appearance when trying to lose weight compared to adults, who cite health as a primary concern [53, 54].

In addition, parent/peer influences have been shown to shape weight loss intentions [55‐58]. This is likely a confluence of behaviour modeling, social support, and parental/peer pressure. Although research is limited, it does appear that the importance of these influences varies by age and sex. Further research into the influence of these social networks is needed, as the importance of these social networks on body image, self-esteem, and weight intention may be critical to preventing maladaptive weight loss habits. In particular, as it appears that once young females reach middle and high school their intentions of maintaining a particular body shape are already exaggerated compared to boys, understanding the influence of these social networks on younger females may be especially critical from a public health perspective. Thus, public health efforts should consider targeting elementary school children to encourage the development of a healthy body image and lifestyle. Given that sex differences in weight loss intentions persist into adulthood [52], a more comprehensive evaluation of which concerns and attitudes about weight loss persist from childhood into adulthood is certainly necessary.

With regards to race/ethnicity, Caucasian, African-American, and Hispanic females appear to report consistently higher prevalence of trying to lose weight compared to males in the same group, while rates were similar among Native-American females and males. Of note, Native-American males demonstrated a higher pattern of endorsement of weight loss intentions compared to all other ethnicities examined. Such observations are important given that previous research indicates important distinctions between race/ethnicity with regard to weight perceptions and management, largely due to cultural differences. For instance, one study suggests that body weight concerns in young females across cultures are influenced by media messaging [59], while another implies that consumption of media targeted to African-Americans is unrelated to body weight concerns [60]. A better understanding of factors associated with race/ethnicity and weight intentions is warranted, especially given the disparity in the prevalence of overweight among African-American, Native-American, and Caucasian youth in the United States [61] and Canada [62].

Unsurprisingly, it was observed that a larger proportion of overweight/obese participants reported trying to lose weight compared to normal-weight and underweight individuals (based on measured weight status), but sex differences were not reported. The lack of data on sex differences across these categories represents a major limitation as research shows that young females are more likely to perceive themselves as overweight compared to young males [19]. Further, the literature suggests that perceived weight status is a stronger predictor of weight intentions and strategies than measured weight status. For instance, preliminary evidence among adults suggests that perceived weight status may fully mediate the association between measured weight and weight intentions [63, 64] but differs by sex and race/ethnicity. Whether this mediation occurs across sex and race/ethnicity among adolescents should be further evaluated.

The proportion of youth using unhealthy or extreme strategies is as high as 44 and 13%, respectively. Overall, a similar proportion of males and females endorsed the use of each category of weight loss strategies across studies. This similarity is important to note given the tendency for the literature to focus on females when examining weight loss behaviours. This phenomenon is reflected in the studies included in this review, of which 32% had all-female samples, and no studies examining all-male samples. While trends do suggest that females are more concerned with weight loss while males are more concerned with increasing muscle [65], an overemphasis on young females may obscure the rate of risky behaviours engaged in by young males who are trying to lose weight, including the use of unhealthy and extreme strategies.

Still, given the differences in motivating factors, young females may require special attention with regards to education about weight loss behaviours. As noted, research suggests that appearance is a primary motivator for young females to lose weight, rather than health. Thus, additional efforts are needed to promote healthy body shape and body image in young females, to supplement the promotion of health. Doing so is likely to decrease the frequency with which youth are engaging in these dangerous behaviours. Indeed, research suggests that the use of unhealthy and extreme weight loss strategies is linked to other risky behaviours (e.g., using indoor tanning beds, using drugs) [66], some of which are likely also motivated by appearance concerns.

Results of this review demonstrate the prevalence of using extreme weight loss strategies was as high as 27% in Native-American young females - the highest of any race/ethnicity examined for that category. Four studies included in the review examined samples of Native-Americans exclusively, but demonstrated no clear pattern regarding the preference to use specific strategies by sex. The use of unhealthy strategies was observed to be as high as 69% in young African-American females, 73% in young Caucasian females, and 75% in young Hispanic females. Given the observational nature of this study, however, no conclusions can be drawn regarding risk for females in these groups at this time, and further research is needed.

Surprisingly, very little data was available regarding the use of particular weight loss strategies across weight status categories based on measured weight status. Of note is the contrast among endorsement rates for the use of extreme strategies between measured and perceived weight status, where 1–3% of females across weight categories endorsed their use based on measured weight status, but that up to 44 and 70% of normal weight and overweight/obese females endorse the use of extreme strategies based on perceived size. As noted above, this misperception of weight in youth appears to play a role in the weight loss strategies favoured by youth, and may lead to the use of maladaptive behaviours. Again, while sex differences for the use of particular weight loss strategies were not examined in any study, it appears as though young females are using extreme strategies to a much greater extent than male counterparts in the same weight category, a difference that does not appear across the use of healthy or unhealthy strategies. Given the significant lack of information in this area, further research is urgently needed to examine how weight status, body image perceptions, and sex are related to weight control behaviours.

This systematic review focused on Canadian and American youth, however variability in study methodology and quality precluded the carrying out of a meta-analysis. Insights into sex differences across demographic characteristics and weight status were thus weakened by the limited literature and more studies focusing on weight loss intentions and strategies specifically are needed. Further, results from this review cannot be generalized beyond the United States and Canada.

Study quality and rigour was decreased primarily due to a lack of methodological information (such as a sample size calculation, psychometric properties of measures, and response rates). Although some of these details may be more pertinent to cohort studies, the quality assessment tool used was specifically designed to evaluate cross-sectional studies. Studies were excluded if they did not provide sufficient data necessary for inclusion in this systematic review (such as weight loss strategies which were too broadly defined). Although it is possible that these studies did not collect greater detailed information on weight intentions and weight loss strategy use, it is also possible that several categories were collapsed due to insufficient numbers. This may impact the lower bounds of our estimates such that they are overestimated.

One major limitation of the existing literature is our observation that no study assessing weight loss intentions and strategy use examined the use of multiple strategies in tandem with one another. This is an important concern, as while the high rates of endorsement for the use of healthy weight loss strategies is encouraging, it is unclear the extent to which these are being used alongside more deleterious methods. Another important limitation to note is that no study that met criteria for inclusion in this review reported prevalence information based on socio-economic characteristics. Given the well documented relationship among race/ethnicity, socio-economic status, and health [67, 68], researchers need to report basic data for such characteristics when examining health and weight loss behaviours, and should consider these factors in their analyses.

Several important implications from this systematic review are apparent. As considerable methodological variation in study measures were found, further research on harmonizing these measures are needed in order for meta-analyses to be conducted and more reliable conclusions to be drawn. For instance, weight loss strategies were commonly described in very broad terms (e.g., “dieting”), and effort in creating stricter guidelines with regards to measurement methodology for research on weight loss behaviours is needed. Better defining strategies may be particularly important as research indicates that dietary changes are reportedly stronger contributors to weight loss than exercise [69, 70]. Similarly, there was considerable variability in the measurement of weight loss intentions, which is commonly measured as either ‘current’ intentions, or lifetime intentions. Thus there is a concern for recall bias, particularly from the studies which use ‘lifetime’ intentions.

Further, weight loss and weight maintenance intentions are commonly grouped together into a single survey question when examining weight change strategies (e.g., which of the following strategies did you use to “to lose weight or keep from gaining weight”). Very few studies reported weight loss strategy use separately from weight maintenance, resulting in the relatively small sample of studies included in this review. This is an important distinction given that weight loss and weight maintenance are cognitively different goals, and lead to differences in the use of particular weight change strategies [32, 71]. Thus, grouping them together may obscure these important differences.

The literature is limited by an over-emphasis on cross-sectional studies, and the longitudinal implications of using healthy, unhealthy or extreme strategies remain unclear. In particular, it is unknown whether the age at which youth first attempt weight loss is associated with the types of strategies used. Additional research is needed to refine our understanding of how weight loss behaviours are developed and maintained in youth.

In this systematic review, several patterns emerged regarding the prevalence of weight loss intentions and strategy use between young females and males across demographic groups and by weight status. However, insufficient data based on race/ethnicity and weight status limit our abilities to conclude whether sex differences based on these characteristics exist, and further research is needed. The development of standardized measures and guidelines for research methodology related to weight loss intentions and strategies in youth is also warranted to reliably assess trends in future research.