Predictors of Weight Loss in Young Adults

Abstract

Background:

Increased understanding of weight loss among healthy young people in naturalistic settings could inform the development of effective weight control programs. The aim of this study was to describe loss in BMI over 7 years in a population-based sample of healthy young adolescents (mean age 17 years at beginning of follow-up) and identify determinants of BMI loss.

Design and Method:

Data were available for 681 participants in the Nicotine Dependence in Teens Study (1999–2012), a longitudinal investigation of adolescents in Montreal (Canada). Loss in BMI was assessed between age 17 and 24 years. Potential predictors of BMI loss including age, sex, mother's education, worry about weight, physical activity, screen time, and cigarette smoking were studied in multivariable logistic regression.

Results:

Males and females gained 2.0 and 1.4 BMI units, respectively, on average, between age 17 and 24 years. However, 9% of males and 14% of females experienced a loss in BMI ≥1.0 unit. Female sex and a higher BMI at age 17 were associated with a higher probability of BMI loss, but none of age, mother's education, physical activity, screen time, or cigarette smoking were associated with BMI loss between ages 17 and 24.

Conclusions:

Whereas BMI increased on average between age 17 and 24 years in a population-based sample of healthy young people, 12% of participants experienced a loss in BMI ≥1 unit. Weight loss was highest among the heaviest persons and did not affect the prevalence of underweight. No single behavior at age 17 stands out as associated with predicting BMI loss.

Introduction

The prevalence of overweight and obesity among children and adolescents in developed countries has increased substantially over the past 30 years. Worldwide, 24% of boys and 23% of girls were overweight or obese in 2013.¹ In Canada, 26% of boys and 22% of girls were overweight or obese.¹ These prevalences are worrisome because overweight children are more likely to become overweight adults² and because overweight and obesity increase the risk of chronic diseases, including diabetes, coronary heart disease, and certain types of cancer.³ Numerous risk factors for overweight and obesity among youth have been identified including among others, lack of physical activity (PA), poor diet, parental overweight, and low family socioeconomic status (SES).^4–11

An age range of particular interest in terms of weight change is the transition from adolescence to young adulthood as offspring move away from parental control and independence increases. Studies of the so-called “Freshman 15” (i.e., the terminology used to describe the weight gain experienced by some young people when they move away from home to university) shows mean weight gains of 0.7–3.4 kg (which is, in fact, not close to the 15 pounds implied by the “Freshman 15” label).¹² Weight increases are apparently most pronounced in the first year of university.¹³ In the Cardia study, weight gains from ages 18 to 24 years were greater than weight gains between ages 25 and 30 in both men and women.¹⁴

In contrast to weight gain, there are few studies on weight loss among young adults outside of clinical interventions or studies on anorexia and bulimia.¹⁵ In a British birth cohort examined in young adulthood for predictors of weight change, Viner and Cole¹⁶ found that persons who lost weight between ages 18 and 30 years were more likely to be female, of higher SES, and to have played sports more frequently. Boutelle and colleagues¹⁷ reported that more overall PA, more vigorous exercise, and less sedentary activity were associated with weight loss in 1726 youth ages 16–18 years who participated in the 1999–2000 or the 2001–2002 National Health and Nutrition Survey study. Importantly, fewer teens in the weight loss groups endorsed efforts at trying to lose weight, compared with the maintenance and gain weight groups. There is also some evidence that smoking may be used as a weight loss strategy. The prevalence of smoking is twice as high among normal-weight adolescent girls who report trying to lose weight compared to same-age girls not trying to lose weight.¹⁸

Although many studies have evaluated obesity prevention and weight loss interventions for young people,^19–21 the results in terms of sustained weight loss are generally disappointing.^22,23 This may relate, in part, to a lack of understanding about the natural course and determinants of weight loss in population-based samples of healthy young people, such that increased understanding of weight loss in naturalistic settings could inform the development of effective weight control programs. The aim of this current study was therefore to describe loss in BMI over 7 years in a population-based sample of healthy young adolescents (mean age 17 years at beginning of follow-up) and identify determinants of BMI loss between ages 17 and 24. The determinants investigated were selected based on factors known to affect weight in young people as well as on the data available and included age, sex, mother's education, worry about weight, PA, screen time, and cigarette smoking.

Methods

Data for this study were available in the Nicotine Dependence in Teens (NDIT) Study.²⁴ NDIT is a longitudinal investigation of 1294 students recruited in 1999–2000 from all grade 7 classes in a sample of 10 high schools in Montreal, Canada, which were selected to include French and English schools, schools located in disadvantaged and advantaged neighbourhoods, and schools in rural, suburban, and urban districts. Its primary objectives were to study the natural course and determinants of cigarette smoking and nicotine dependence in novice smokers. The main source of data was self-report questionnaires administered in class at school every 3 months from grade 7 to 11 (1999–2005), for a total of 20 survey cycles during high school. Questionnaires were also completed 3 and 6 years after graduation from high school (i.e., in 2007–2008 and 2011–2012 in survey cycles 21 and 22, respectively) when participants were age 20 and 24 years, on average, respectively. In addition to its primary objectives, NDIT also embedded studies on obesity (i.e., height and weight were measured by trained technicians in survey cycles 1, 12, 19, and 22; in survey cycle 21, data on height and weight were collected in self-reports) as well as on PA and sedentary behavior, which were measured in all survey cycles.

Participants and parents provided written informed consent. The study was approved by the Direction de santé publique de Montréal, McGill University Institutional Review Board and the Centre de Recherche du Centre Hospitalier de l'Université de Montréal Ethics Review Board.

Study Variables

Loss in body mass index

Height (in meters) and weight (in kilograms) were measured at age 17 and 24 (in survey cycles 19 and 22) by trained and certified technicians. If there was a discrepancy between the two measures (i.e., >0.5 cm for height or >0.5 lbs for weight), a third measure was taken. The mean was calculated (if there were three measures, the two closest measures were used). BMI was calculated as weight (kg) / height² (m), and participants were categorized into weight status subgroups (normal weight, overweight, or obese) based on cutoffs from Cole and colleagues.²⁵ Because measured height and weight at ages 17 and 24 correlated highly with self-reported height and weight assessed at age 20 in survey cycle 21 (r = 0.82 between ages 17 and 20 years; r = 0.82 between ages 20 and 24 years), self-report data at age 20 were imputed for 239 participants missing measured height and/or weight data at age 17. Similarly, self-report data at age 20 were imputed for 150 participants missing measured height and/or weight at age 24. Loss in BMI, ascertained by subtracting BMI at age 17 from BMI at age 24, was categorized as yes (i.e., BMI was lower at age 24 than at age 17) or no.

Potential predictor variables

Data on potential predictor variables, including PA, screen time, cigarette smoking, and worry about weight were drawn from survey cycle 19 in grade 11 so that the value of the variable preceded the reference period for the loss in BMI outcome (which was measured in survey cycles 19, 21, and 22). We therefore made use of the longitudinal study design, strengthening the basis for causal inference. If data on a particular variable were missing in survey cycle 19, we drew data from survey cycle 18, and if data were also missing in survey cycle 18, we drew data from survey cycle 17 (given that survey cycles 17, 18, and 19 were all in grade 11).

Number of bouts of moderate and vigorous physical activity (MVPA) was measured in a 7-day recall (which included a list of 28 moderate and vigorous activities)²⁶ by: “Think about the physical activities that you did last week from Monday to Sunday outside your regular school gym class. For each activity that you did for 5 minutes or more at one time, mark an ‘X’ to show the day(s) on which you did that activity.” The number of bouts of moderate and vigorous activities were summed and used as a continuous variable in the analyses.

Screen time was assessed by: “How many hours of television (including video movies) do you usually watch in a single day on…….weekdays? weekends?”; and “How many hours do you usually play video or computer games, or use the Internet in a single day on…. weekdays? weekends?” To obtain the number of hours of screen time per week for each item, the number of hours on weekdays and weekends were multiplied by 5 and 2, respectively, and then hours of television and hours of video or computer games were summed to obtain the total number of hours of screen time per week.

Worry about weight was assessed by: “During the past 3 months, have you been worried or stressed by your weight?” Response options included not at all, a little bit, quite a bit, and a whole lot. For analysis, responses were categorized as not at all versus all other response options.

Cigarette smoking was assessed in a past 3-month recall,²⁴ which measured number of days on which participants had smoked during each month and number of cigarettes smoked per day on average. The total number of cigarettes smoked in each of the past 3 months was calculated by multiplying number of days by average number of cigarettes smoked per day. For analysis, participants were categorized as smoked at least one cigarette in the past 3 months (yes, no).

Sociodemographic indicators included age, sex, and mother university-educated (yes, no, or unknown). Mother's education was measured by: “How much education has your mother had?” Response options included did not finish high school, high school graduate, vocational or technical school, CEGEP, university, don't know, not applicable, and other, which were categorized for analysis as any university, no university, or unknown.

Statistical Analysis

After descriptive analyses, we used one-way analysis of variance (ANOVA) supplemented by Tukey's test to study differences in change in BMI across weight status subgroups in males and females separately. Tukey's method is used in ANOVA to create confidence intervals (CIs) for all pair-wise differences between factor level means while controlling the error rate to a level specified. It is important to consider the error rate when making multiple comparisons because the probability of a type I error for a series of comparisons is greater than the error rate for any one comparison alone. To counter the higher error rate, Tukey's method adjusts the confidence level for each individual interval so that the resulting confidence level is equal to the value specified.

The association between each potential predictor variable and loss in BMI was ascertained in multivariable logistic regression. Clustering induced by children attending the same elementary school was accounted for using generalized estimating equations, with an exchangeable correlation matrix and robust (sandwich) estimators of standard errors. In order to differentiate between trivial and more substantial weight loss, we examined any loss in BMI (yes, no), loss ≥0.5 BMI units (yes, no), and loss ≥1 BMI unit (yes, no) in three separate models. Among adults of average height, a 1-unit change in BMI is approximately equivalent to a 6-pound change in weight. A 1-unit change is closer to 7 pounds among tall adults and 5 pounds among shorter adults. Because cigarette smoking was not statistically significant in any model, it was excluded from the analyses presented herein. Also, we tested interaction terms for sex and MVPA as well as for sex and screen time at age 17. Neither interaction term was statistically significant in any model and therefore was not considered further. All analyses were conducted using SAS software (version 9.1; SAS Institute Inc., Cary, NC).

Results

A total of 1294 students were recruited into the NDIT cohort. Data on BMI were available for 911 participants (70% of 1294) at both ages 17 and 24. Of these 911, 54 participants who were underweight²⁰ at either age 17 and/or 24 and who lost weight during follow-up were excluded given that weight loss in these individuals might indicate a health issue. Among the 54, only 2 males and 11 females with a “normal” BMI at age 17 were underweight at age 24. Therefore, of the 911 participants with BMI data at ages 17 and 24, only 13 (1%) who were normal weight became underweight during the 7-year interval. In addition to the 54 excluded because of an underweight status, 176 participants missing data on one or more potential predictor variables were excluded. The final analytic sample therefore included 681 participants (Fig. 1).

Figure 1.

Derivation of the analytical sample.

Table 1 compares selected characteristics at cohort inception of participants retained for the current analysis with those of participants not retained (from among all 1294 students recruited to participate in NDIT). Participants who were excluded from the analytical sample becasue of missing data were less likely to have a university-educated mother, reported more screen time, and were more likely to have ever smoked (and among those who smoked, smoked fewer cigarettes per month, on average) compared to participants included in the analysis. Though age was statistically significantly lower among participants retained, the difference was trivial. Finally, a higher proportion of males in the retained group were worried about their weight.

Table 1.

Comparison of Selected Characteristics at Cohort Inception of Participants Retained for Analysis (n = 681) with Those Not Retained (n = 613): NDIT Study, Montreal 1999–2012

	Retained (n = 681)	Not retained (n = 613)	p value
Age, mean (SD)	12.6 (0.4)	12.7 (0.7)	0.000
Male, %	47.9	48.5	0.813
Mother university educated, %	46.8	39.8	0.039
French speaking, %	27.8	32.7	0.054
BMI, mean (SD)	20.2 (3.8)	19.9 (3.9)	0.209
Weight status^a, %			0.502
Normal	72.7	75.6
Overweight	16.6	15.4
Obese	10.7	9.0
Worried about weight, %
Total	38.0	35.7	0.426
Males	33.8	25.8	0.036
Females	41.6	45.0	0.405
Screen time, hours/week, mean (SD)	28.1 (19.1)	30.4 (21.5)	0.049
Ever smoked, %	23.6	41.6	0.000
No. of cigarettes per month in past 3 months, mean (SD)	1.8 (15.4)	18.0 (81.1)	0.000

According to BMI-for-age percentile.

NDIT, Nicotine Dependence in Teens Study; SD, standard deviation.

Among participants retained for analysis, at age 17 years 25% smoked cigarettes, and the median (interquartile range) number of cigarettes smoked per month among those who smoked was 20 (1–200). The mean (SD) BMI was 22.9 (3.8), which is “normal” weight according to the CDC.²⁷ Fifteen percent of participants were overweight and 6% were obese. Twenty-four percent of males worried about their weight, compared to 63% of females. Participants reported 10.6 bouts of PA per week. On average, participants spent 25.8 hours per week (approximately 3.7 hours per day) watching television, playing video games, or using a computer.

Change in Body Mass Index

Mean BMI increased from ages 17 to 24 in both males and females in all weight status subgroups (Table 2). There were minimal increases in height across weight status subgroups, so that changes in BMI were owing primarily to changes in weight. Among participants who were normal weight at age 17 years, 22% gained weight to the extent that they were categorized as overweight at follow-up; 2% (n = 8) of normal weight participants at age 17 were obese at follow-up.

Table 2.

Change in BMI, Height, and Weight from Age 17–24 Years (Survey Cycle 19–22) According to Sex and Weight Status Subgroup at Survey Cycle 19 (n = 681): NDIT Study, Montreal 1999–2012

		Females				Males
	Total n = 355	Normal weight n = 279	Overweight n = 54	Obese n = 22	Total n = 326	Normal weight n = 239	Overweight n = 64	Obese n = 23
Median BMI	21.8	21.1	26.6	31.9	22.1	21.3	25.7	32.2
Loss in BMI, %
Total	28.5	26.5	33.3	40.9	19.3	13.0	34.4	43.5
≥0.5 units	17.8	12.9	33.3	40.9	13.2	6.3	28.1	43.5
≥1 unit	14.1	9.0	31.5	36.4	9.2	3.4	20.3	39.1
Loss in BMI among participants with BMI >median, %
Total	36.5	33.8	37.0	54.6	31.9	20.8	43.8	41.7
≥0.5 units	26.4	18.0	37.0	54.6	23.3	10.8	40.6	41.7
≥1 unit	21.9	12.2	33.3	45.5	17.8	6.7	28.1	33.3
BMI, mean (SD)
survey cycle 19	22.8 (3.8)	21.3 (1.7)^a,b	26.7 (1.4)^a,c	33.2 (4.5)^b,c	23.1 (3.8)	21.3 (1.7)^a,b	26.2 (1.5)^a,c	33.0 (2.7)^b,c
survey cycle 22	24.2 (4.5)	22.6 (2.6)^a,b	28.5 (4.2)^a,c	34.1 (5.0)^b,c	25.0 (4.2)	23.5 (2.7)^a,b	27.7 (3.8)^a,c	33.1 (5.5)^b,c
change	1.4 (2.6)	1.3 (2.1)	1.7 (4.0)	1.0 (4.6)	2.0 (2.8)	2.3 (2.2)^b	1.5 (3.5)	0.1 (4.8)^b
Height (m), mean (SD)
survey cycle 19	1.63 (0.06)	1.63 (0.06)	1.62 (0.06)	1.65 (0.06)	1.76 (0.07)	1.76 (0.06)	1.76 (0.07)^c	1.72 (0.08)^c
survey cycle 22	1.64 (0.06)	1.64 (0.06)	1.64 (0.06)	1.67 (0.05)	1.78 (0.07)	1.78 (0.07)	1.78 (0.07)	1.76 (0.07)
change	0.01 (0.01)	0.01 (0.01)^b	0.01 (0.01)^c	0.02 (0.04)^b,c	0.02 (0.02)	0.02 (0.02)^b	0.02 (0.02)	0.03 (0.05)^b
Weight (kg), mean (SD)
survey cycle 19	60.8 (11.1)	56.8 (6.1)^a,b	70.6 (6.9)^a,c	89.8 (13.7)^b,c	70.8 (12.2)	65.6 (6.9)^a,b	81.7 (7.6)^a,c	97.9 (12.4)^b,c
survey cycle 22	65.4 (13.4)	61.2 (8.5)^a,b	77.9 (12.5)^a,c	94.7 (15.1)^b,c	79.5 (14.2)	74.9 (10.0)^a,b	88.4 (14.2)^a,c	104.0 (17.1)^b,c
change	4.8 (7.0)	4.5 (5.7)	6.6 (11.1)	5.2 (10.9)	8.4 (8.9)	9.3 (7.2)^b	6.8 (11.7)	3.4 (13.5)^b

Difference between normal weight and overweight was statistically significant according to Tukey's test.

Difference between normal weight and obese was statistically significant according to Tukey's test.

Difference between overweight and obese was statistically significant according to Tukey's test.

Nicotine Dependence in Teens (NDIT) Study; SD, standard deviation.

In females, weight gain over the 7 years was highest in the overweight subgroup (6.6 kg compared to 4.5 and 5.2 kg in the normal weight and obese groups, respectively). The lowest weight gain (4.5 kg) was in the normal weight subgroup. According to Tukey's tests, the mean changes in both BMI and weight were similar in all weight status subgroups.

In males, the normal weight subgroup gained the most weight (9.3 vs. 6.8 kg and 3.4 kg in the overweight and obese subgroups, respectively). The lowest weight gain (3.4 kg) was in the obese subgroup. There were statistically significant differences between normal weight and obese males in the mean BMI change and the mean weight change (i.e., normal weight males experienced greater changes compared to obese males).

Loss in Body Mass Index

Overall, 24% of participants lost BMI (16% lost ≥0.5 BMI units and 12% lost ≥1.0 BMI unit; Table 2). Among participants who were overweight or obese at age 17, 12% experienced weight loss over the 7-year follow-up period that resulted in their being categorized as normal weight at age 24.

Nineteen percent of males experienced a loss in BMI (13% lost ≥0.5 and 9% lost ≥1.0 BMI unit) and 29% of females experienced a loss in BMI (18% lost ≥0.5 and 14% lost ≥1.0 BMI unit). Compared to overweight and obese participants in both sexes, a lower percentage of normal weight participants experienced a loss in BMI ≥1 unit. The proportions of participants who experienced a loss in BMI ≥1 unit was higher among normal and overweight females than among normal and overweight males, respectively. However, among obese participants, the proportions were similar across the sexes (36% in females and 39% in males).

In normal and overweight males and females as well as in obese females, the proportion of participants who experienced a loss in BMI ≥1 unit was greater among those above the median sex-specific BMI (i.e., higher proportions of heavier participants within weight status subgroups lost weight).

Predictors of Body Mass Index Loss

The results across the three multivariable models tested in this analysis were generally similar. Only two variables were statistically significantly associated with loss in BMI in one or more model. Female sex was associated with an increase in the probability of a loss in BMI, although the variable attained statistical significant only in the model that tested any loss in BMI (Table 3). In addition, a higher BMI at age 17 was statistically significantly associated with loss in BMI between ages 17 and 24, and the probability was higher in the models that tested loss ≥0.5 and ≥1 BMI unit, compared to the model that tested any BMI loss.

Table 3.

Crude and Adjusted Odds Ratios (OR) and 95% Confidence Intervals (CI) for Loss in BMI According to Potential Predictor Variables (n = 654)^a: NDIT Study, Montreal 1999–2012

	OR_cr (95% CI)	OR_adj (95% CI)
Any loss in BMI
Age	0.74 (0.47–1.15)	0.64 (0.39–1.04)
Sex
Male	ref.	ref.
Female	1.66 (1.16–2.38)	1.69 (1.14–2.50)
Mother university educated
Yes	ref.	ref.
No	1.43 (0.99–2.06)	1.40 (0.95–2.05)
MVPA	0.99 (0.97–1.01)	0.99 (0.97–1.01)
Screen time	1.00 (0.99–1.01)	1.00 (0.98–1.01)
Baseline BMI	1.15 (1.10–1.20)	1.16 (1.11–1.22)
Loss in BMI ≥0.5 units
Age	0.80 (0.46–1.36)	0.58 (0.31–1.09)
Sex
Male	ref.	ref.
Female	1.42 (0.93–2.16)	1.43 (0.88–2.30)
Mother university educated
Yes	ref.	ref.
No	1.48 (0.96–2.29)	1.45 (0.90–2.33)
MVPA	0.98 (0.96–1.01)	0.98 (0.96–1.01)
Screen time	1.00 (0.99–1.01)	0.99 (0.97–1.01)
Baseline BMI	1.24 (1.18–1.31)	1.26 (1.19–1.33)
Loss in BMI ≥1 unit
Age	0.76 (0.41–1.40)	0.55 (0.27–1.13)
Sex
Male	ref.	ref.
Female	1.62 (1.00–2.61)	1.56 (0.91–2.70)
Mother university educated
Yes	ref.	ref.
No	1.39 (0.86–2.24)	1.29 (0.76–2.19)
MVPA	0.99 (0.97–1.02)	0.99 (0.96–1.02)
Screen time	1.00 (0.99–1.02)	0.99 (0.97–1.01)
Baseline BMI	1.26 (1.19–1.33)	1.28 (1.21–1.36)

n = 164 participants experienced a loss in BMI.

n = 106 participants experienced a loss ≥0.5 BMI units.

n = 82 participants experienced a loss ≥1 BMI unit.

NDIT, Nicotine Dependence in Teens Study; MVPA, moderate-to-vigorous physical activity.

Bolded OR (95% CI) indicate that they are significant (p < 0.05).

Neither MVPA nor screen time at age 17 approached statistical significance in any model. In secondary analyses (data not shown), we tested “worried about weight” rather than BMI at age 17 given that worried about weight was likely on the causal pathway between BMI and loss in BMI. Worried about weight was statistically significantly associated with a loss in BMI ≥0.5 (adjusted odds ratio [OR_adj] = 0.61; 95% CI, 0.38–0.97) and with a loss in BMI ≥1 unit (OR_adj = 0.57; 95% CI, 0.34–0.96).

Discussion

The aim of this study was to estimate the frequency and identify the determinants of weight loss in a population-based sample of healthy young adults. Despite the general increase in BMI, 12% of participants lost ≥1.0 BMI unit. Further 12% of participants who were overweight or obese at age 17 lost enough weight to be categorized in the normal weight group at age 24. In both males and females, heavier persons were more likely to lose BMI. Further within weight status groups, the proportion that lost BMI was higher among those above the group-specific BMI median. These data support the notion that weight loss in this population-based sample of young people occurred primarily among those who were heavier and therefore can be viewed as “weight loss that is not likely to be detrimental to health.” Only 11 participants (1.6% of the original sample) were excluded because they were normal weight at age 17 and lost weight over the 6-year follow-up to the extent that they became underweight. Therefore, overall, in this population-based sample of healthy young people, the primary issue with respect to adiposity is weight gain. Public health messaging and social norms that encourage weight loss to achieve healthy body weights do not seem to result in widespread weight loss that is detrimental to health. This evidence is important to the debate that encouraging young girls in particular to lose weight results in widespread unhealthy underweight.

Thirteen percent of normal weight females lost at least 0.5 BMI units, compared to 6% of normal weight males, and the multivariate analysis supported the notion that females were more likely than males to lose weight. Worry about weight and dieting are much more common in females than males, as are misperceptions about weight status.^28–30 Several reports suggest that many females believe that they are overweight when, in fact, they are within the normal weight range.^29,31 The healthy weight range is wide and young women within this group may lose weight, possibly accounting for the higher proportion that experienced a loss in BMI over the 7-year follow-up period.

Although no data were collected in this study on methods used to lose weight, neither level of PA nor screen time at age 17 were associated with loss in BMI between ages 17 and 24. This finding, in addition to the observation that there were few notable increases in MVPA and that there were no marked decreases in screen time, suggests that, in general, participants did not change their activity level in order to lose weight.

This study did not provide evidence of weight loss that is detrimental to health (only 1.6 % of the sample with BMI data available at both ages 17 and 24 became underweight). However, it is worrisome that males and females gained 8.4 and 4.8 kg, on average, between ages 17 and 24 years. Whereas some of the weight gain in males may relate to later pubertal development, most females have completed puberty by age 17 years.³² These findings underscore that weight gain during young adulthood is likely a far more critical public health issue than weight loss. In addition, the data underscore that because level of PA and screen time do not seem to be associated with weight loss in a naturalistic setting, these behaviors (which are fundamental to healthy weight control) should be central to public health interventions that target overweight and obesity among young adults.

Limitations

All data on potential predictor variables were based on self-report at age 17, which may have resulted in misclassification that attenuated the estimates of the associations between potential predictor variables and the outcomes. Lack of data on diet and alcohol use may have resulted in residual confounding. The data were subject to selection bias owing to loss to follow-up. Finally, use of a selected sample may limit the generalizability of the findings. However, at least at cohort inception, the NDIT sample resembled a representative population-based sample of similar-aged adolescents.¹⁹

Conclusions

In the population-based sample of healthy young people, 12% experienced a loss in BMI ≥1 unit between ages 17 and 24 years. Weight loss was highest among the heaviest persons and did not affect the prevalence of underweight. Therefore, it is unlikely that the observed weight loss was detrimental to health. These data provide evidence that may allay concerns that public health weight loss messaging is associated with weight loss that is harmful to health.

Footnotes

Acknowledgments

The NDIT Study was funded by the Canadian Cancer Society (grants 010271, 017435). The funders were not involved in the design or conduct of the study; collection, management, analysis, or interpretation of the data; or preparation, review, or approval of the manuscript. J.O.L. is a Canada Research Chair in the Early Determinants of Adult Chronic Disease.

Author Disclosure Statement

No competing financial interests exist.

References

, Fleming

, Robinson

, et al. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–2013: A systematic analysis for the Global Burden of Disease Study 2013. Lancet, 2014; 384:766–781.

Guo

, Chumlea

. Tracking of body mass index in children in relation to overweight in adulthood. Am J Clin Nutr, 1999; 70:145S–148S.

World Health Organization. Obesity and overweight. Fact sheet No. 311. Available at www.who.int/mediacentre/factsheets/fs311/en Last accessed August 11, 2014.

Steinbeck

. The importance of physical activity in the prevention of overweight and obesity in childhood: A review and an opinion. Obes Rev, 2001; 2:117–130.

McMurray

, Harrell

, Deng

, et al. The influence of physical activity, socioeconomic status, and ethnicity on the weight status of adolescents. Obes Res, 2000; 8:130–139.

Giammattei

, Blix

, Marshak

, et al. Television watching and soft drink consumption: Associations with obesity in 11- to 13-year-old schoolchildren. Arch Pediatr Adolesc Med, 2003; 157:882–886.

Berkey

, Rockett

, Field

, et al. Activity, dietary intake, and weight changes in a longitudinal study of preadolescent and adolescent boys and girls. Pediatrics, 2000; 105:E56.

Maffeis

. Aetiology of overweight and obesity in children and adolescents. Eur J Pediatr, 2000; 159(Suppl 1):S35–S44.

Parsons

, Power

, Logan

, et al. Childhood predictors of adult obesity: a systematic review. Int J Obes Relat Metab Disord, 1999; 23(Suppl 8):S1–S107.

10.

Must

, Tybor

. Physical activity and sedentary behavior: A review of longitudinal studies of weight and adiposity in youth. Int J Obes (Lond), 2005; 29(Suppl 2):S84–S96.

11.

Fulton

, Dai

, Steffen

, et al. Physical activity, energy intake, sedentary behavior, and adiposity in youth. Am J Prev Med, 2009; 37(1 Suppl):S40–S49.

12.

Crombie

, Ilich

, Dutton

, et al. The freshman weight gain phenomenon revisited. Nutr Rev, 2009; 67:83–94.

13.

Gropper

, Simmons

, Connell

, et al. Weight and body composition changes during the first three years of college. J Obes, 2012;(2012):634048.

14.

Burke

, Bild

, Hilner

, et al. Differences in weight gain in relation to race, gender, age and education in young adults: the CARDIA Study. Coronary Artery Risk Development in Young Adults. Ethn Health, 1996; 1:327–335.

15.

Neumark-Sztainer

, Wall

, Story

, et al. Dieting and unhealthy weight control behaviors during adolescence: Associations with 10-year changes in body mass index. J Adolesc Health, 2012; 50:80–86.

16.

Viner

, Cole

. Who changes body mass between adolescence and adulthood? Factors predicting change in BMI between16 year and 30 years in the 1970 British Birth Cohort. Int J Obes (Lond), 2006; 30:1368–1374.

17.

Boutelle

, Hannan

, Neumark-Sztainer

, et al. Identification and correlates of weight loss in adolescents in a national sample. Obesity (Silver Spring), 2007; 15:473–482.

18.

Strauss

, Mir

. Smoking and weight loss attempts in overweight and normal-weight adolescents. Int J Obes Relat Metab Disord, 2001; 25:1381–1385.

19.

Kitzmann

, Dalton

3rd , Stanley

, et al. Lifestyle interventions for youth who are overweight: A meta-analytic review. Health Psychol, 2010; 29:91–101.

20.

Stice

, Shaw

, Nathan

. A meta-analytic review of obesity prevention programs for children and adolescents: The skinny on interventions that work. Psychol Bull, 2006; 132:667–691.

21.

Flodmark

, Marcus

, Britton

. Interventions to prevent obesity in children and adolescents: A systematic literature review. Int J Obes, 2006; 30:579–589.

22.

Showell

, Fawole

, Segal

, et al. A systematic review of home-based childhood obesity prevention studies. Pediatrics, 2013; 132:e193–e200.

23.

Wang

, Wu

, Wilson

, et al. Childhood obesity prevention programs: Comparative effectiveness review and meta-analysis. Report no.: 13-EHC081-EF. AHRQ Comparative Effectiveness Reviews. Agency for Healthcare Research and Quality: Rockville, MD, 2013.

24.

O'Loughlin

, Dugas

, Brunet

, et al. Cohort profile: The Nicotine Dependence in Teens (NDIT) Study. Int J Epidemiol, 2014 Jul 13. pii: dyu135. [Epub ahead of print]

25.

Cole

, Bellizzi

, Flegal

, et al. Establishing a standard definition for child overweight and obesity worldwide: International survey. BMJ, 2000; 320:1240–1243.

26.

Ainsworth

, Haskell

, Whitt

, et al. Compendium of physical activities: An update of activity codes and MET intensities. Med Sci Sports Exerc, 2000; 32(9 Suppl):S498–S504.

27.

Centers for Disease Control and Prevention. Healthy Weight: It's not a diet, it's a lifestyle! Available at www.cdc.gov/healthyweight/assessing/bmi/adult_bmi/index.html Last accessed September 15, 2014.

28.

Mendes

, Araújo

, Lopes

, et al. Determinants of weight loss dieting among adolescents: A longitudinal analysis. J Adolesc Health, 2014; 54:360–363.

29.

Wardle

, Haase

, Steptoe

. Body image and weight control in young adults: International comparisons in university students from 22 countries. Int J Obes (Lond), 2006; 30:644–651.

30.

Field

, Aneja

, Austin

, et al. Race and gender differences in the association of dieting and gains in BMI among young adults. Obesity (Silver Spring), 2007; 15:456–464.

31.

Strauss

. Self-reported weight status and dieting in a cross-sectional sample of young adolescents: National Health and Nutrition Examination Survey III. Arch Pediatr Adolesc Med, 1999; 153:741–747.

32.

Herman-Giddens

. Recent data on pubertal milestones in United States children: The secular trend toward earlier development. Int J Androl, 2006; 29:241–246; discussion, 286–290.