Identification and Correlates of Weight Loss in Adolescents in a National Sample
Division of General Pediatrics and Adolescent Health, University of Minnesota, 200 Oak Street SE, Suite 160, Minneapolis, MN 55455. E-mail: email@example.com
Objective: Little is known about behaviors associated with successful weight loss during adolescence. The first objective of the current study was to identify meaningful weight loss, weight maintenance, and weight gain in male and female adolescents. The second objective of this study was to apply these methods to U.S. adolescents from the National Health and Nutrition Survey 1999 to 2002 data and to identify factors associated with these weight change outcomes.
Research Methods and Procedures: The current analyses include 1726 (female, 836; male, 890) 16- to 18-year-old adolescents who completed the questionnaire components and interview for either the 1999–2000 or the 2001–2002 National Health and Nutrition Survey study. Dietary intake, physical activity, and dieting attitudes were compared across the weight loss (L), maintain (M), and gain (G) groups in the entire sample and in a subset of adolescents who are overweight and at-risk-for-overweight (≥85th percentile).
Results: The tested method for identifying weight L, M, and G groups has both theoretical and statistical validity and, when applied to the sample, showed the expected direction of changes in weight. Results suggest that more overall physical activity, more vigorous exercise, and less sedentary activity are associated with being in the L group in both the full sample and the overweight and at-risk-for-overweight sample. In addition, fewer teens in the L groups endorsed efforts at trying to lose weight, compared with the M and G groups.
Discussion: This study provides a method to determine successful adolescent weight loss for researchers and provides useful concrete information about successful weight loss for clinicians and others who work with adolescents.
The prevalence of obesity continues to rise, and 31% of all adolescents are considered overweight or at-risk-for-overweight (1). Obesity is associated with significant concomitant and subsequent morbidity and mortality (2) and psychosocial consequences (3). An estimated 4 to 5 million children between the ages of 6 and 17 years in the United States are overweight (4), and hospital costs for obesity-related conditions in youth increased from 35 million dollars in 1979 to 127 million dollars in the year 2000 (5). Obesity in adolescence is associated with a number of chronic diseases, including diabetes, colon cancer, arthritis, and hip fractures in adulthood (6, 7, 8). In addition, research shows that obesity is associated with a number of psychosocial consequences in childhood and adolescence, including poor self-esteem, teasing, and verbal abuse (9, 10, 11), and overweight adolescents are more likely to be socially isolated and to be peripheral to social networks (12) than their lighter peers. These psychosocial consequences are considered to be more important than those of other chronic physical conditions (13). Obesity tracks from adolescence into adulthood as 70% of overweight adolescents will be obese in adulthood (14, 15). Thus, adolescence is an important developmental period and is identified as one of the critical periods for the development of adult obesity (16).
Little is known about behaviors associated with successful weight management in adolescence. One method to learn about adolescent behaviors associated with weight loss is by investigating adolescents who have been successful in losing weight. This strategy has been employed with adults in the National Weight Control Registry (NWCR).1 The NWCR is composed of over 5000 adults nationwide who reported that they have lost a minimum of 30 pounds and kept it off at least 1 year. Individuals in the NWCR have lost an average of 33 kg and maintained the loss for >5 years (17). Results from the NWCR show that members report engaging in a number of behavioral strategies, including high levels of physical activity (PA); eating a low-calorie, low-fat diet; eating breakfast regularly; self-monitoring weight; and maintaining a consistent eating pattern across weekdays and weekends as a way to maintain their weight loss (17, 18, 19, 20, 21, 22, 23). NWCR members report that weight loss maintenance becomes easier over time; after individuals have successfully maintained their weight loss for 2 to 5 years, the chance of longer term success greatly increases (18, 19). In addition, findings from the NWCR have provided a greater understanding of triggering events for weight loss and improvements in psychological and physical quality of life (24). Similar knowledge about adolescent weight loss could be useful in creating and disseminating possible strategies and interventions for adolescents, with the goal to ultimately prevent adult obesity.
Identifying and defining adolescent weight loss presents a challenge to researchers. Healthy adolescents are expected to grow and gain body fat and weight, and the concept of weight loss and its study is different in adolescents compared with adults. The pubertal growth spurt, occurring over an average of 3 years, accounts for 15% of adult height and for 50% of eventual weight (25, 26), and fat-free mass increases differently for boys and girls (27). Although fixed categories of BMI (kilograms per meter squared) are used to describe adult obesity status, adolescent BMI status is evaluated relative to growth curves (28) where age and gender are used to determine an individual adolescent's percentiles for BMI. This BMI-for-age percentile is customarily used to identify adolescents who are at-risk-for-overweight (85th to <95th percentile) and those who are considered overweight (≥95th percentile) (29).
Because adolescents are expected to grow during this dynamic developmental stage, a decrease in body weight, maintenance of body weight, and even a small gain in body weight could result in changes in the individual's weight and obesity trajectory. Nonetheless, weight alone is inadequate to meaningfully identify weight loss as it relates to obesity status because it does not consider concomitant adolescent changes in height. Because adolescents are growing in height and weight, a measure like BMI percentile that integrates both measures is required. A clear, meaningful definition of adolescent weight loss that accommodates the adolescent weight-height relationship has not been published to date.
The current study addresses the methodological and contextual gaps in the literature examining weight loss in adolescents. The first objective of this study was to present a method for identifying weight loss, weight maintenance, and weight gain in male and female adolescents. The second objective of this paper was to apply these methods to U.S. adolescents included in the National Health and Nutrition Survey (NHANES) 1999–2002 data and to identify factors associated with these weight change outcomes. The overarching purpose of this line of research is to ultimately contribute to an understanding of the factors associated with weight change in adolescents and to inform the development of effective weight management interventions for youth.
Research Methods and Procedures
Procedures and Subjects
This study used data from the NHANES 1999–2002 surveys and interviews. NHANES previously used a stratified multistage probability sample designed to give an annual sample that was nationally representative of the U.S. civilian non-institutionalized population. Beginning in 1999, the NHANES became a continuous, annual survey with oversampling of minorities, including African Americans, Mexican Americans, adolescents, and elderly persons. The current analysis includes 1726 (female, 836; male, 890) 16- to 18-year-old youth who completed the questionnaire components and interview for either the 1999 to 2000 or the 2001 to 2002 NHANES. Sample weights for the individual surveys were not utilized in this study because we were comparing behaviors and attitudes between groups. National Health and Nutrition Examination Survey (NHANES 1999 to 2004) Protocol 9812 was reviewed and approved by the National Center for Health Statistics Institutional Review Board.
NHANES data included in this study were derived from the Diet Behavior and Nutrition Questionnaire, PA and Physical Fitness Questionnaire, Weight History Questionnaire, and the dietary intake interview.
Weight Change Groups
Self-reported current weight, self-reported height, and previous year's weight were identified from the Weight History Questionnaire. NHANES data include measured height and weight, but the previous year's weight is available only by self-report. Our analyses showed that measured weight was highly correlated to self-reported weight in this sample (0.96 in both genders) and that the regression equations estimating bias found a difference in measured and self-reported weight of <3%. Because the data included self-reported weight from 1 year ago and self-reported current weight and measured current weight, we used self-reported current weight and self-reported previous year's weight in the determination of the weight change groups. The assumption is that the reporting bias within individuals would be similar for current weight and previous year's weight. Factors related to bias in reporting weight are tied to individual characteristics, such as race, socioeconomic status, and gender (30) that do not change within individuals.
To allow for expected growth, the conversion parameters available from Centers for Disease Control and Prevention (http:www.cdc.govgrowthcharts) were used to transform gender- and age-specific current weight and previous year's weight to Z scores for each participant [weight (WT)_Z, current and previous]. The difference between these two, ΔWT_Z, is approximately normally distributed. From examination of the gender-specific distributions of the change, cut-off points ±0.35Z were chosen for each gender to define groups [loss (L), maintain (M), and gain (G)]; these cut-off points also approximately corresponded to 1SD (standard deviation) of the observed ΔWT_Z. The intent was to separate youth whose annual change in BMI_Z score was sufficiently large to overcome the expected year-to-year variation for individuals (6). Thus, losers are defined as those for whom ΔWT_Z < −1SD; correspondingly, gainers are those for whom ΔWT_Z > 1SD, and maintainers did not change WT_Z by > 1SD (−1SD ≥ ΔWT_Z ≥ +1SD). This definition resulted in 14% of the sample defined as losers and 16% as gainers over the prior 1-year period. After reviewing the data and comparing measured weight and self-reported weight, we eliminated outliers whose ΔWT_Z score changed by >2SD over the 1-year time period (n = 14) because they appeared to be probable errors or members of a different population distribution.
Weight-Related Attitudes and Behaviors
Weight-related attitudes and behaviors were measured with items from the Weight History Questionnaire. Questions regarding weight-related attitudes included a desire to weigh more or less and perceptions of weight status (underweight, overweight). Weight-related behaviors included endorsement to lose weight and/or maintain weight over the past 12 months.
The dietary intake data were collected in person through an interview. One 24-hour dietary recall was used to estimate total intake of energy, nutrients, and non-nutrient food components from foods and beverages that were consumed during the 24-hour period before the interview (midnight to midnight), using the multiple-pass technique. The NHANES computer-assisted dietary interview system is an automated data collection form that was developed to provide a standardized dietary intake data. We report total energy (kilocalories) and percentage of calories from protein, carbohydrate, and total fat. In addition, average number of meals eaten at a restaurant each week was derived from the Diet Behavior and Nutrition Questionnaire.
PA and television watching were assessed with the Physical Activity and Physical Fitness Questionnaire. Participants responded to questions regarding the frequency of 10-minute bouts of vigorous and moderate activity, duration of vigorous and moderate activity, engagement in strength training, and frequency of strength training. From these variables, we calculated and report frequency of vigorous activity bouts per week, hours of vigorous activity per week, frequency of moderate activity bouts per week, hours of moderate activity per week, total hours of PA per week, any strength training activity per week, frequency of strength training activities per week, and hours of television watching per week. In addition, we calculated the percentage of adolescents who met the PA requirements from the 2010 Healthy People Objectives (three vigorous or five moderate activities per week) (31).
All tables are gender specific and have the categories L, M, or G defined a priori according to procedures previously described. Demographics are presented by the LMG categories. The first analyses compared the LMG groups on diet, PA, and attitudes regarding weight by gender. Using general linear models, we generated adjusted means by LMG category for continuous measures and adjusted probabilities for categorical measures. Tests of significance are based on the contrast of L and G groups. To capture behaviors specific to overweight adolescents, we compared the L vs. the M and G groups (L vs. M + G) in the overweight and at-risk-for-overweight (≥85th percentile) adolescents by gender. All analyses used SAS (32) and are adjusted for race, family income, and previous year's weight except when previous year's weight is the outcome (Table 2).
Table 2. Current BMI, current weight, previous year's weight, change in weight, and proportion at-risk-for-overweight in the L, M, and G groups by gender [means (SE)]; lower panel is restricted to those who are overweight and at-risk-for-overweight
|Current BMI||22.1 (0.5)||24.1 (0.2)||24.8 (0.4)||<0.001||24.3 (0.4)||25.0 (0.2)||24.6 (0.4)||0.55|
|Current weight (pounds)||125.0 (3.0)||140.3 (1.4)||141.1 (2.3)||<0.001||163.5 (3.3)||168.4 (1.6)||165.0 (2.8)||0.73|
|Weight 1 year ago (pounds)||141.5 (3.1)||137.0 (1.4)||120.3 (2.4)||<0.001||178.1 (3.4)||160.8 (1.7)||138.8 (2.9)||<0.001|
|Change in weight (pounds)||−16.5 (0.9)||3.2 (0.4)||20.8 (0.7)||<0.001||−14.7 (0.8)||7.6 (0.4)||26.2 (0.7)||<0.001|
|Proportion ≥85th percentile for weight 1 year ago||32.1%||27.0%||14.2%||<0.001||55.8%||33.0%||11.0%||<0.001|
| || || || || || || || || |
|Adolescents who are overweight and at-risk-for-overweight (≥85th percentile)|| || || ||L vs. M + G†|| || || ||L vs. M + G†|
|Current BMI||24.8 (0.9)||31.0 (0.4)||32.8 (1.0)||<0.001||27.0 (0.6)||30.7 (0.4)||32.6 (1.1)||<0.001|
|Current weight (pounds)||144.9 (5.0)||185.9 (2.5)||194.7 (5.7)||<0.001||188.5 (4.1)||214.1 (2.7)||225.0 (7.9)||<0.001|
|Weight 1 year ago (pounds)||175.3 (5.5)||183.0 (2.8)||161.9 (6.3)||0.66||210.1 (4.3)||206.3 (2.8)||188.4 (8.2)||0.04|
|Change in weight||−30.4 (2.5)||2.8 (1.2)||32.9 (2.9)||<0.001||−21.5 (1.4)||7.8 (0.9)||36.7 (2.6)||<0.001|
Weight changes in female and male adolescents were examined by age, race, and socioeconomic status (Table 1). Among female adolescents, there were statistically significant differences in weight changes across race; black female adolescents were most likely to be weight gainers and the least likely to be weight losers, compared with other racial groups. White female adolescents were least likely to be gainers. In male adolescents, there were differences in weight changes by age group; the older boys were the most likely to gain weight and the least likely to be in the L group.
Table 1. Proportion (percentage) of female and male adolescents in weight L, M, and G groups by sociodemographic characteristics
|Age (years)|| || || || || || || || || || |
| 17||288||13.2||61.8||25.0|| ||297||17.2||60.6||22.2|| |
| 18||275||13.5||65.5||21.1|| ||285||10.5||62.5||27.0|| |
|Race|| || || || || || || || || || |
| Black||224||8.5||61.2||30.4|| ||266||15.8||63.2||21.1|| |
| Mexican/Hispanic||363||15.2||61.7||23.1|| ||350||16.6||59.7||23.7|| |
| Other||34||14.7||64.7||20.6|| ||46||8.7||73.9||17.4|| |
|Income ($/yr)*|| || || || || || || || || || |
| 20,000 to <55,000||309||13.6||62.5||23.9|| ||328||16.7||63.6||19.7|| |
| ≥55,000||223||14.0||70.5||15.5|| ||242||12.9||67.3||19.8|| |
Consistency of L, M, and G Groups with Weight Changes
Table 2 shows weight and BMI variables by gender for LMG groups for the entire sample and for the overweight and at-risk-for-overweight adolescents (≥85th percentile). The statistical determination of the LMG groups as described in “Research Methods and Procedures” is consistent in both genders because the L group for both female and male adolescents in the total sample showed a decrease in weight from the previous year to current (female, −16.5 pounds; males, −14.7 pounds), and the G group showed an increase in weight (females, 20.8 pounds; males, 26.2 pounds). The overweight and at-risk-for-overweight adolescents demonstrated larger changes in weight from the previous year to current (L group, females, −30.4 pounds; males, −21.5 pounds; G group, females, 32.9 pounds; males, 36.7 pounds).
Total Sample: Correlates of the L, M, and G Groups
When comparing the three weight change groups on weight control behaviors and perceptions, dietary intake, and PA, a number of differences emerge for both the female and male adolescents (Table 3). Compared with the M and G groups, the L group had a larger proportion of participants who indicated that they would like to weigh more, relatively fewer participants who would like to weigh less, relatively fewer participants who perceived themselves as overweight, and relatively more participants who perceived themselves as underweight. Relatively fewer of the male adolescents in the L group reported that they had tried to lose weight, and in female adolescents, the data exhibit a trend of marginal significance (p = 0.08), suggesting that relatively fewer female adolescents in the L group had tried to lose weight. There were no significant differences for female or male adolescents among the weight change groups for diet variables, for frequency of weekly restaurant usage, or for moderate PA measures.
Table 3. Weight-related attitudes and behaviors, diet, and PA in the L, M, and G groups by gender [means (SE)] and test of linear trend, adjusted for race, family income, and weight 1 year ago
|Weight|| || || || || || || || |
| Like to weigh more (%)||4.9||1.9||0.7||<0.001||27.7||22.2||17.9||0.05|
| Like to weigh less (%)||31.8||58.9||76.3||<0.001||13.2||18.5||36.6||<0.001|
| Consider self overweight (%)||13.8||35.8||61.6||<0.001||5.6||12.6||31.7||<0.001|
| Consider self underweight (%)||7.5||2.2||0.9||<0.001||14.7||8.3||4.6||<0.001|
| Tried to lose weight (%)||31.1||35.2||44.3||0.08||8.3||12.0||25.2||0.002|
| Tried not to gain (%)||43.9||31.9||37.4||0.50||22.6||15.8||20.3||0.75|
|Diet|| || || || || || || || |
| Energy (kcal)||1944 (89)||2020 (42)||2025 (74)||0.49||2660 (105)||2701 (52)||2649 (91)||0.94|
| Protein (%)||13.1 (0.4)||13.7 (0.2)||12.9 (0.4)||0.73||13.7 (0.4)||13.7 (0.2)||14.2 (0.3)||0.36|
| Carbohydrate (%)||54.8 (1.0)||55.0 (0.5)||55.6 (0.9)||0.55||55.2 (1.0)||54.2 (0.5)||53.1 (0.9)||0.13|
| Fat (%)||32.1 (0.9)||32.2 (0.4)||31.5 (0.7)||0.61||31.2 (0.7)||32.2 (0.4)||32.8 (0.6)||0.11|
| Restaurant meals/wk||3.4 (0.3)||2.8 (0.1)||3.0 (0.2)||0.24||2.9 (0.3)||3.1 (0.1)||2.9 (0.2)||0.90|
|PA|| || || || || || || || |
| Vigorous activity times/wk||5.9 (0.7)||3.8 (0.3)||3.1 (0.5)||0.002||7.1 (0.8)||6.8 (0.4)||5.9 (0.7)||0.26|
| Hours of vigorous activity/wk||4.9 (0.6)||3.5 (0.3)||2.3 (0.5)||0.001||8.4 (0.8)||7.7 (0.4)||6.5 (0.7)||0.09|
| Moderate activity times/wk||3.6 (0.7)||3.3 (0.3)||2.9 (0.5)||0.35||2.8 (0.7)||3.7 (0.3)||2.5 (0.6)||0.75|
| Hours of moderate activity/wk||2.6 (0.5)||2.0 (0.2)||2.1 (0.4)||0.42||2.9 (0.6)||2.9 (0.3)||2.3 (0.5)||0.43|
| Total hours of PA/wk||7.6 (0.9)||5.6 (0.4)||4.4 (0.7)||0.006||11.7 (1.1)||10.9 (0.6)||9.0 (1.0)||0.08|
| Television hours/wk||2.9 (0.2)||3.1 (0.1)||3.5 (0.1)||0.011||3.2 (0.2)||3.3 (0.1)||3.6 (0.1)||0.09|
| Any strength activity (yes/no; %)||52.3||44.1||37.1||0.016||64.6||65.8||64.1||0.93|
| Strength activity times/wk||1.9 (0.3)||1.7 (0.1)||1.5 (0.2)||0.27||2.4 (0.4)||3.0 (0.2)||3.0 (0.3)||0.27|
| Meets PA recommendations (yes/no; %)||51.1||48.8||40.4||0.09||69.3||64.5||55.1||0.01|
Compared with the M and G groups, female participants in the L group participated in significantly more hours of vigorous activity, more bouts of vigorous physical activities in the past month, more total hours of PA, and watched fewer hours of television per day. Similar trends were seen in male adolescents, but data for hours of vigorous PA, total hours of PA, and hours of television per day reached marginal significance (p = 0.09, p = 0.08, p = 0.09, respectively), whereas the data for bouts of physical activities did not reach significance. A greater percentage of the female weight losers reported engaging in strength training, although the number of strength training activities did not differ across weight change groups. For male adolescents, engagement in strength training was higher than in female adolescents, but engagement and the number of activities differed little across weight change groups. Male adolescents in the weight L group were more likely to meet the Healthy People 2010 PA requirements, with a similar trend in female adolescents where the data reached marginal significance (p = 0.09).
Adolescents Who Are Overweight and At-Risk-for-Overweight: L, M, and G Groups
Results for adolescents considered overweight and at-risk-for-overweight (≥85th percentile for weight previous year) are presented in Table 4. The results suggested that fewer female and male adolescents in the L group considered themselves overweight. Fewer of the male adolescents in the L group indicated that they would like to weigh less and reported fewer weight loss attempts. Relatively more female adolescents in the L group tried not to gain weight. There were no significant differences for overweight female or male adolescents on diet variables or frequency of restaurant use each week.
Table 4. Overweight and at-risk-for-overweight (previous year) adolescents and weight-related attitudes and behaviors, diet, and PA [means (SE)] and tests of linear trend and contrasts, adjusted for weight 1 year ago, race, and family income
| ||Female adolescents||Male adolescents|
| ||L (n = 36)||M (n = 144)||G (n = 27)||Test of L vs. M + G†||L (n = 77)||M (n = 185)||G (n = 21)||Test of L vs. M + G†|
|Weight|| || || || || || || || |
| Like to weigh more (%)||*||*||*||*||*||*||*||*|
| Like to weigh less (%)||*||*||*||*||46.9||63.3||71.0||0.02|
| Consider self overweight (%)||59.3||87.7||96.6||<0.001||31.3||56.0||74.9||<0.001|
| Consider self underweight (%)||*||*||*||*||*||*||*||*|
| Tried to lose weight (%)||81.2||62.8||58.0||0.26||18.8||30.5||57.7||0.02|
| Tried not to gain (%)||88.7||52.2||63.5||0.04||*||*||*||*|
|Diet|| || || || || || || || |
| Energy (kcal)||2010 (147)||1920 (73.9)||2065 (188)||0.92||2566 (138)||2477 (88)||2170 (268)||0.22|
| Protein (%)||12.9 (0.8)||13.9 (0.4)||13.0 (1.0)||0.54||13.6 (0.6)||13.8 (0.4)||14.5 (1.1)||0.49|
| Carbohydrate (%)||55.7 (1.9)||53.3 (0.9)||53.1 (2.4)||0.28||55.6 (1.5)||53.2 (1.0)||52.5 (2.9)||0.21|
| Fat (%)||32.3 (1.6)||33.7 (0.8)||31.0 (2.0)||0.97||31.1 (1.1)||32.3 (0.7)||34.2 (2.1)||0.19|
| Restaurant meals/wk||3.0 (0.4)||2.3 (0.2)||3.8 (0.5)||0.83||2.9 (0.3)||2.9 (0.2)||2.8 (0.6)||0.98|
|PA|| || || || || || || || |
| Vigorous activity times/wk||7.1 (1.2)||3.9 (0.6)||2.1 (1.4)||0.006||7.4 (1.2)||7.1 (0.8)||4.4 (2.4)||0.33|
| Hours of vigorous activity/wk||5.4 (1.1)||3.3 (0.5)||1.3 (1.3)||0.02||7.9 (1.1)||7.5 (0.7)||5.4 (2.1)||0.36|
| Moderate activity times/wk||4.9 (1.6)||4.4 (0.8)||3.3 (1.8)||0.59||2.6 (0.8)||3.1 (0.5)||1.4 (1.5)||0.74|
| Hours of moderate activity/wk||2.5 (1.0)||3.2 (0.5)||1.6 (1.2)||0.91||3.2 (0.8)||2.5 (0.5)||1.2 (1.5)||0.21|
| Total hours of PA/wk||7.6 (1.9)||6.7 (0.9)||3.1 (2.2)||0.23||11.5 (1.5)||10.2 (1.0)||6.9 (2.9)||0.17|
| TV hours/wk||3.0 (0.3)||3.1 (0.2)||3.8 (0.4)||0.18||3.3 (0.2)||3.6 (0.1)||4.1 (0.4)||0.14|
| Any strength activity (yes/no; %)||53.3||43.9||35.1||0.19||76.0||63.6||43.4||0.01|
|Strength activity times/wk||2.1 (0.6)||1.8 (0.3)||1.7 (0.7)||0.61||2.6 (0.5)||2.9 (0.3)||4.2 (1.0)||0.20|
|Meets PA recommendations (yes/no; %)||57.4||49.4||46.0||0.36||71.3||61.6||37.9||0.01|
Similar patterns to the total sample were seen in PA for the overweight and at-risk-for-overweight adolescents across the three weight change groups. Female adolescents in the L group participated in more bouts of vigorous activity and more hours of vigorous activity per week compared with the M and G groups. Male adolescents in the L group were more likely to engage in strength training, and relatively more met the recommended PA requirements. There are a number of trends that were not statistically significant but are worth noting; male and female adolescents in the L group participated in more hours of PA per week and watched less television per week than the M or G groups.
The aims of this report were to present a methodology for determining meaningful weight change groups in adolescents and to apply this method to a national sample of 1726 adolescents to identify behaviors associated with successful weight loss. To the best of our knowledge, this is the first research that describes a theoretically and statistically based definition for adolescent weight loss and applies it to a large national sample of adolescents. This study has a number of important findings regarding weight loss in adolescents. In general, the results suggest that the L group participated in relatively more PA and relatively less television viewing.
The method we described, using 1SD of ΔWT_Z to determine the L, M, and G groups appears to be valid. The method has both theoretical and statistical validity and when applied to the sample, showed the expected direction of changes in weight. The female adolescents in the L group reduced their weight by ∼11% of their body weight, whereas the male adolescents in the L group reduced their weight by ∼8% of their body weight. The female adolescents initially overweight and at-risk-for-overweight who were in the L group lost 14% of their body weight, and the corresponding male adolescents reduced their body weight by 12%. These weight losses represent significant changes in body weight even for adults and are more remarkable for adolescents who are growing and would, on average, gain weight during this period. These numbers are also similar to the 10% weight loss, which is recommended for adults to experience medical benefits of weight loss (17), although the change in risk factors associated with the weight loss shown in adolescents still needs to be evaluated.
In interpreting these results, we are guided by theory on negative energy balance and consistency in the results between the total sample and overweight sample. Overall, the results suggested that more vigorous exercise, more overall PA, and less sedentary activity are associated with being in the L group. These results are similar to findings in adults from the NWCR. Almost all of the NWCR participants engaged in PA and 72% of the registry exceeded American College of Sports Medicine recommendations for a minimum PA level, and over 50% of the registry exceeded the energy expenditures for optimum activity levels (33).
The negative energy balance expected for dietary intake was not demonstrated in this study. In both genders, there were no significant differences among the L, M, and G groups for any of the dietary variables or on restaurant usage. It is possible, but not probable, that dietary intake is not associated with weight loss in adolescents. More likely, the null results are an artifact of the error associated with the use of a single 24-hour dietary recall as a measure of diet. There is significant intra-individual variability of food intake from day to day, and one 24-hour recall does not capture that variability (34, 35, 36, 37). In this study, the female adolescents who lost weight reported a mean intake of 81 calories less per day than those who gained weight (Table 3). Although not statistically different, over a year, this small caloric deficit could account for some of the changes in weight. Measures that account for daily variability (such as multiple recalls) should be utilized before conclusions are drawn regarding the importance of dietary quality in adolescent weight loss. In addition, this study evaluated current dietary intake, and it is possible that previous year's dietary intake might have been more closely associated with the L, M, and G groups. Finally, bias in self-reporting diet is known to occur, especially in the direction of under-reporting intakes with obesity (38).
It is noteworthy that among male adolescents, previous year's weight differed across the three weight change groups, but current weight was similar across the groups (see Table 2). In female adolescents, there were significant differences in current weight and previous year's weight by the weight change groups. The lack of differences in current weight for male adolescents presents a quandary. If we could manipulate the data, the perfect scenario would be one in which the groups have no differences at the previous year's weight (i.e., they start at the same weight) and show differences at the current weight, which leads them to be classified in the L, M, and G groups. This was not the case using these methods as described in this paper. We controlled for previous year's weight in an attempt to manage the differences seen among groups in the total sample. We also performed the same analyses for adolescents who were overweight and controlled for previous year's weight. It is possible that the lack of differences in the current weight in male adolescents could have tempered the final results for the male adolescents.
It is also worth noting that there were counterintuitive differences by L, M, and G groups in the proportion trying to lose weight. These differences could reflect a perception of those in the L group that they do not need to lose weight any longer. Alternatively, it may be that efforts aimed at losing weight are just not very successful and that those successfully managing their weight do so through living a healthy lifestyle that includes increased PA and decreased sedentary activity.
There are a number of strengths in this study that enhance our ability to draw conclusions from these findings. This study improves on the design of the NWCR because it is a large population-based national sample that includes a racially diverse population. Furthermore, in the current study, we were able to include a comparison group (gainers). In addition, this study incorporates a theoretically based method to compare the weight change groups by gender. As in all studies, weaknesses also need to be taken into account. All of the measures, including current weight and previous year's weight, are based on self-reported measures. The high correlations found between self-reported current weight and measured current weight lessen our concerns about using self-reported weights, but we cannot account for changes in bias when self-reporting weight 1 year ago and current weight. The methods described in this study are applied to a population; thus, we cannot differentiate intentional and unintentional weight loss. This study is different from the NWCR in that it includes adolescents who may have unintentionally changed their body weight. In addition, one administration of a 24-hour dietary recall is not designed to capture the individual variability in dietary intake. Finally, the measures of PA were brief and based on self-report. Future studies can improve on this study by including multiple dietary recalls and objective measures of PA in a longitudinal design.
Overall, this study contributes to the literature by defining adolescent weight loss and by demonstrating that moderate and vigorous PA are associated with weight loss in the general population of adolescents and in the adolescents who are overweight or at-risk-for-overweight. This study provides a method to determine successful adolescent weight loss for researchers and provides useful information for clinicians and others who work with adolescents.
There was no funding/outside support for this study.
Nonstandard abbreviations: NWCR, National Weight Control Registry; PA, physical activity; NHANES, National Health and Nutrition Survey; WT, weight; L, loss; M, maintain; G, gain; SD, standard deviation.
The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.