PAPER Association of weight change, weight control ... - Nature

International Journal of Obesity (2004) 28, 1134–1142 & 2004 Nature Publishing Group All rights reserved 0307-0565/04 $30.00 www.nature.com/ijo

PAPER Association of weight change, weight control practices, and weight cycling among women in the Nurses’ Health Study II AE Field1,2*, JE Manson2,3, CB Taylor4, WC Willett2,5,6 and GA Colditz2,5 1 Division of Adolescent Medicine, Department of Medicine and Department of Psychiatry, Children’s Hospital Boston and Harvard Medical School, USA; 2Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, USA; 3Division of Preventive Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, USA; 4Department of Psychiatry and Behavioral Medicine, Stanford University School of Medicine, USA; 5Department of Epidemiology, Harvard School of Public Health, USA; and 6Department of Nutrition, Harvard School of Public Health, USA

OBJECTIVE: To assess the association of weight cycling with weight change, weight control practices, and bulimic behaviors. METHODS: A nested study of 2476 young and middle-aged women in the Nurses’ Health Study II who provided information on intentional weight losses between 1989 and 1993. SAMPLE: In total, 224 women who were severe cyclers, 741 women who were mild cyclers, 967 age- and BMI-matched controls (noncyclers), and 544 women who did not weight cycle and maintained their weight between 1989 and 1993 completed a questionnaire in 2000–2001 assessing recent intentional weight losses, weight control practices, and weight concerns. RESULTS: After controlling for age and body mass index (BMI) in 1993, when weight cycling was initially assessed, mild cyclers gained an average of 6.7 pounds (lbs) more and severe cyclers gained approximately 10.3 lbs more than noncyclers between 1993 and 2001. Weight cyclers preferred to change their diet rather than to exercise to control their weight. Severe weight cyclers were less likely than noncyclers to use frequent exercise as a weight control strategy (odds ratio [OR] ¼ 0.8, 95% confidence interval (CI) 0.6–1.1). Cyclers were also more likely than noncyclers to engage in binge eating (mild cyclers: OR ¼ 1.8, 95% CI 1.4–2.4; and severe cyclers: OR ¼ 2.5, 95% CI 1.7–3.5). Independent of weight cycling status, age, and BMI, women who engaged in binge eating gained approximately 5 lbs more than their peers (Po0.001). CONCLUSIONS: Weight cycling was associated with greater weight gain, less physical activity, and a higher prevalence of binge eating. Low levels of activity and binge eating may be partially responsible for the large amount of weight regained by weight cyclers. International Journal of Obesity (2004) 28, 1134–1142. doi:10.1038/sj.ijo.0802728 Published online 20 July 2004 Keywords: women; weight gain; weight control; weight cycling; bingeing

Introduction In the United States obesity has become a serious public health problem. According to the third National Health and Nutrition Examination Survey (NHANES), 34% of adults in the United States are overweight and an additional 30.5% are obese.1 The rapid increase in the prevalence of obesity is

*Correspondence: Dr AE Field, Division of Adolescent Medicine, 300 Longwood Ave, Boston 02115, MA, USA. E-mail: [email protected] Received 19 November 2003; revised 4 April 2004; accepted 9 May 2004; published online 20 July 2004

thought to be largely due to lifestyle factors, such as diet and physical inactivity. Although the prevalence of overweight and obesity are increasing, the desire to be thin is widespread. Dieting to lose weight is common among adolescent females,2–4 as well as adult women.5–7 In cross-sectional studies, researchers have observed a strong association between dieting and being overweight or obese.7,8 In large prospective studies, weight loss attempts9,10 and weight cycling11,12 have been observed to be predictive of major weight gain among young and middle-aged women. The association of weight cycling to weight gain is of interest because during the past decade,

Weight change and weight cycling among women AE Field et al

1135 weight loss and weight cycling have been investigated as predictors of morbidity and mortality. Although some of the earlier studies observed that weight loss or cycling was associated with increased health risks,13–16 more recent studies have not observed an association12,17,18 However, the latter studies have observed that weight and weight gain, correlates of weight cycling, were predictive of developing morbidities. Unfortunately, little is known about why women who intentionally lose weight are paradoxically more likely than their peers to gain weight. Weight gain is an important independent risk factor for the development of type II diabetes,19 hypertension,20 cardiovascular disease,21 and some cancers,22–24 thus understanding the reasons for the excessive weight gain associated with cycling could be of public health importance. To assess the correlates of weight cycling, we conducted a study among a sample of young and middle-aged women participating in the Nurses’ Health Study II.

Methods The Nurses’ Health Study II (NHS II) was established in 1989 and consists of 116 671 female nurses, 25–43 y at baseline, who responded to a mailed questionnaire about their medical history and lifestyle and health behaviors. Followup questionnaires have been sent to participants biennially since 1989. Additional details have been reported previously.25 Height, weight, and recalled weight at age 18 y were ascertained in 1989 and current weight was assessed in 1989 and on each follow-up questionnaire. Body mass index (BMI, kg/m2) was calculated from self-reported information on weight and height. BMIs less than 16 or greater than 60 were assumed to be data errors, so their values were set to missing and not used in the analysis. The BMI classification scheme in the US Dietary Guidelines was used to classify women into underweight (16 BMI o18.5), healthy weight (18.5rBMIr24.9), overweight (25rBMIr29.9), and obese (BMIZ30).26 The 1993 NHS II survey included questions on weight losses that were specifically designed to address the longterm health consequences of intentional weight loss. They were developed after extensive discussion among investigators from the Nurses’ Health Study, Centers for Disease Control and Prevention, and University of Minnesota. These questions included the number of intentional weight loss episodes of varying magnitude (5–9, 10–19, 20–49, and 50 þ lbs) over the past 4 y and between 18 and 30 y of age. The information on intentional weight losses between 1989 and 1993 was used to classify women as non, mild, or severe weight cyclers. The question, ‘Between 1989 and 1993, how many times did you lose each of the following amounts of weight on purpose?’ The responses were 0, 1–2, 3–4, 5–6, and 7 or more times for each of the magnitudes of weight loss. To be consistent with the magnitude of the weight loss required by Field et al,12 French et al,27 and Williamson et al28 in their studies of the relation between intentional weight loss and

disease, to be classified as a severe weight cycler, we required that a woman report intentionally losing 20 or more lbs. To ensure that the cyclers were women who had repeatedly lost weight, we required that women intentionally lost weight three or more times between 1989 and 1993 to be classified as severe weight cyclers. Women who had intentionally lost 10 or more lbs three or more times, but did not meet the criteria for severe weight cycling, were classified as mild weight cyclers. Women who did not meet the criteria described above for mild or severe weight cycling were classified as nonweight cyclers. The information on body weight, reported on each of the biennial questionnaires, was not used to define weight cycler status.

Sample Based on their weight and responses to the 1993 intentional weight loss questions, a sample comprising all of the severe cyclers, a subset of the mild and noncyclers, one control for each of the weight cyclers matched on BMI and age, and a sample of women whose 1989, 1991, and 1993 weights were within 5 lbs of one another were invited to participate in a substudy. Two groups of noncyclers were included in the sample because women whose 1989, 1991, and 1993 weights were within 5 lbs of one another were believed to represent the women with the most effective weight control practices, whereas, noncyclers who were not weight stable were likely to represent the general group of women who did not make repeated large changes in weight (ie, weight cycling), but gained some weight over time. The 3931 women selected for participation were sent an invitation letter, which explained that participation was voluntary, and a supplementary questionnaire. Supplementary questionnaire The questionnaire assessed current weight, highest and lowest weight in the past 4 y, weight concerns, weight control behaviors, binge eating, intentional and unintentional weight loss (and regain) patterns during the past 4 y, dietary restraint, disinhibition, and attitudes about exercise. Women were asked for their highest and lowest weights during the past 4 y, as well as their current weight. The validity of selfreported weight and height were examined among 140 women, 40–65 y of age, participating in the Nurses’ Health Study;29 self-reported weight was highly correlated with the average of two standardized measurements taken approximately 6 months apart by trained interviewers (r ¼ 0.97). Dietary restraint was measured with questions from the restraint scales from two instruments: the Three Factor Eating Questionnaire (TFEQ)30 and the Herman and Polivy Restraint Scale.31 Two instruments were used because there is controversy over whether dietary restraint should refer to people who are successful at limiting their intake (ie, being restrained vis-a`-vis their intake) or, the majority of dieters, who try to limit their intake but break down and splurge from time to time. This latter group is frequently International Journal of Obesity


1136 characterized as chronic (unsuccessful) dieters. Laessle et al32 found that the Herman and Polivy Restraint Scale is best suited for identifying weight fluctuators and chronic and unsuccessful dieters; Whereas, Stunkard & Messick’s TFEQ is more appropriate for identifying successful caloric restriction. We augmented the TFEQ restraint assessment with several questions that assessed attitudes about and practices of exercise to control weight. One new questions asked ‘to control my weight I would rather change my diet than exercise regularly.’ The other new items were similar to existing questions on the restraint scale, but substituted the word ‘exercise’ for ‘diet.’ The new items were: ‘life is too short to worry about exercising to control my weight,’ ‘while on a diet to lose weight, if I eat a food that is not allowed, I exercise afterwards to make up for it,’ ‘exercising does not help me to lose weight or keep from gaining weight,’ ‘I do not like to exercise, so I rarely stick to an exercise program,’ ‘I feel guilty if I do not exercise,’ ‘I make myself exercise in order not to gain weight,’ and ‘if I eat too much, I exercise to make up for it.’ Weight control behaviors were measured in several ways. Women were asked to indicate the methods they use to control weight in general, as well asked to report all of the methods they used for their most recent 10–19, 20–49, and 50 lb weight losses. The options they were given were: ‘eat a low calorie diet,’ ‘eat a low fat diet,’ ‘eat a high protein diet,’ ‘count calories,’ ‘not eat dessert,’ ‘skip meals/fast,’ ‘commercial weight loss program,’ ‘gastric surgery,’ ‘limit portion size,’ ‘not eat between meals,’ ‘use Slimfast or other shakes,’ ‘use products containing Olestra,’ ‘weigh/measure myself frequently,’ ‘decrease alcohol intake,’ ‘exercise frequently,’ ‘use over the counter diet pills,’ ‘use prescription diet pills (eg, Phentermine, Orlistat, or other drugs),’ ‘smoking,’ or ‘other.’ In addition, women were asked how often they had used the following methods to control their weight: exercise, diet pills (over the counter), laxatives, and vomiting. Women who reported at least monthly use of diet pills, laxatives, or vomiting to control weight were labeled as engaging in unhealthy weight control practices and those who reported at least monthly use of laxatives, or vomiting to control weight were classified as purgers. Women were also asked how often during the past year they have eaten in a discrete amount of time (eg, within any 2-h period), an amount of food that is definitely larger than most people would eat during a similar amount of time and under similar circumstances. Women who reported having any of these large eating episodes were then asked whether they felt out of control and as if they could not stop during the episode. Women who reported large eating episodes at least monthly and reported feeling out of control during the episodes were classified as binge eaters.

Sample for analysis After three mailings approximately 70% (n ¼ 2751) of the women returned completed questionnaires. The sample was International Journal of Obesity

comprised of 821 mild weight cyclers, and 248 severe weight cyclers, 1070 age- and BMI-matched controls (noncyclers), and 612 women who did not weight cycle and maintained their weight between 1989 and 1993. Age was not associated with willingness to participate, however, BMI was inversely associated with participation among the non (Po0.05) and mild (Po0.001) weight cyclers. Women were excluded from the analysis if they reported a pregnancy in the past 4 y (n ¼ 146), did not provide information on recent weight losses (n ¼ 96) or were missing information on weight at the follow-up (n ¼ 33); thus 2476 women remained for analysis.

Statistical methods Univariate differences were assessed with w2-tests and Wilcoxon rank-sum tests. Multivariate linear regression was used to assess cross-sectional differences in BMI, as well as changes in weight between 1993 and 2001. All models included age, BMI in 1993, and weight cycling status in 1993. To assess differences in weight control beliefs and behaviors, we used logistic regression models that included age, BMI in 1993, and weight cycling status in 1993. Stratified analyses and interaction terms were used to assess whether the associations with weight cycling differed by weight status (ie, healthy weight vs overweight women). Nonweight cyclers who were not weight stable were the reference group.

Results Although only 14% of the women who had been mild cyclers and 27% of the severe cyclers between 1989 and 1993 continued to weight cycle between 1997 and 2001, many women reported at least one intentional weight loss during the follow-up. Approximately 41.5% of the women reported that they had intentionally lost 10–19 lbs at least once in the past 4 y, 29.3% reported that they had intentionally lost 20– 49 lbs, and 5.1% reported having intentionally lost 50 lbs at least once in the past 4 y. Unintentional weight loss was less common. The percentage of women who had unintentionally lost 10–19, 20–49, and 50 or more lbs were 16.1, 8.2, and 1.4%, respectively. Despite the high prevalence of weight loss, most of the women gained weight during the follow-up. The mean weight change between 1993, when weight cycling status was first assessed, and the end of follow-up (2000–2001) was a gain of 11 lbs; however, 10% of the women gained more than 30 lbs and 10% lost at least 6 lbs. Weight status in 1993 was associated with weight gain. Overweight women gained an average of 2.2 lbs more than women in the healthy weight range and obese women gained an average of 3.4 lbs more than women in the healthy weight range. Weight cyclers were heavier than noncyclers in 1993 (Table 1) and weight cycling status between 1989 and 1993 was strongly related to making at least one intentional


1137 Table 1

Weight control and weight change patterns among a sample of young and middle-aged US women Weight cycling status (1989–1993) Weight stable noncyclers (n ¼ 544)

Noncyclers, not weight stable (n ¼ 967)

Mild cyclers (n ¼ 741)

Severe cycler (n ¼ 224)

Age (y) in 1993 BMI (kg/m2) in 1993 Weight change 1993–2001a At least one intentional 10–19 lb weight loss (1998–2001) At least one intentional 20–49 lb weight loss (1998–2001) At least one intentional weight loss ( 50 lbs (1998–2001)

38.7 (4.4)w 21.6 (3.4)w 5.9 (10.0)w 20.9%w 5.3%w 0.9%z

39.9 (4.2) 26.1 (2.7) 8.7 (15.2) 42.2% 25.4% 2.6%

39.8 (4.1) 25.9 (2.7) 15.4 (17.7)w 51.6%w 42.8%w 7.4%w

40.2 (4.1) 27.3 (2.5)wz 18.8 (23.1) w 55.8%y 58.8%w 18.9%w

Weight cycling status in 1998–2001 Noncycler Mild cycler Severe cycler

97.5% 2.0% 0.6%

95.1% 3.3% 1.6%

85.5% 9.6% 4.9%

72.4% 13.6% 14.0%

55.9%w 30.7%z 60.5% 68.8%z

46.7% 36.0% 62.1% 74.2%

47.6% 47.2%w 66.4% 80.2%y

38.8%zz 42.4% 60.7% 76.8%

12.9% 5.5% 0.6%w 1.0% 5.7%w

16.2% 6.2% 3.1% 1.0% 11.2%

18.6 % 12.1%w 5.4%y 3.0%y 18.2%w

21.9%z 15.2%w 11.2%w 5.7%w 26.2%wz

Exercises frequently to control weight Eat a low calorie diet to control weight Eat a low-fat diet to control weight Lowered caloric intake or changed dietary composition to control weight Skipped meals/fasted to control weight Used over the counter diet pills to control weight Used prescription diet pills to control weight Used vomiting or laxatives to control weight Engaged in binge eating at least monthly

z Pr0.05 compared to noncyclers who were not weight stable from 1989 to 1993. yPr0.01 compared to noncyclers who were not weight stable from 1989 to 1993. wPr0.001 compared to noncyclers who were not weight stable from 1989 to 1993. zPr0.02 compared to mild weight cyclers. aDifference between weight (lbs) reported on 1993 questionnaire and 2000/2001 supplementary questionnaires.

weight loss of 10 or more lbs between 1997 and 2001 (Table 1). Independent of age and BMI in 1993, mild weight cyclers were two times more likely (odds ratio (OR) ¼ 2.4, 95% confidence interval (CI) 1.9–2.9) and severe weight cyclers were four times more likely (OR ¼ 3.7, 95% CI 2.7– 5.1) than noncyclers to report at least one intentional 20– 49 lbs loss between 1997 and 2001. The associations were even stronger when comparing noncyclers to cyclers for having made at least one intentional weight loss of 50 or more lbs (OR ¼ 3.0 and 7.2 for mild and severe cyclers, respectively). Women who had made intentional weight losses were more likely than their peers to have had unintentional weight losses, but after controlling for intentional weight loss, weight cycling status was not independently associated with unintentional weight losses (data not shown).

Dietary restraint, physical activity, and weight cycling Weight cycling was associated with many potential behavioral predictors of weight change, including dietary restraint, exercise patterns, and dietary weight control strategies. Dietary restraint, as measured by the TFEQ, had a modest, but significant (Po0.001), positive association with weight cycling status (Figure 1). Dietary restraint as measured by Herman and Polivy’s scale had a much stronger positive association with weight cycling status (Figure 1).

Moreover, both the weight fluctuation and the concern for dieting subscales of the Herman and Polivy’s Restraint scale had strong positive associations with weight cycling status. Severe cyclers had the highest restraint scores, followed by mild cyclers, and weight stable women had the lowest scores on the Herman and Polivy’s Restraint scale. Overall, the exercise factor, which we developed to mirror the TFEQ restraint scale, but assessing exercise behaviors and beliefs instead of dietary ones, had a modest inverse association with weight cycling status. Although weight cyclers were more likely than noncyclers to report that if they were on a diet and ate a food that was not allowed they would exercise to make up for it, severe weight cyclers were approximately 60% more likely than noncyclers (odds ratio [OR] 1.6, 95% confidence interval [CI] 1.2–2.1) to report that they rarely were able to stick to an exercise program (Table 2). Weight stable women engaged in significantly more activity than other women. The median hours of vigorous activity was 1 h per week for the weight-stable women vs 0.75 h per week for non, mild, and severe cyclers (Figure 1). Moreover, weightstable women engaged in a total of 3.5 h per week of leisuretime activity vs 2.5 for noncyclers and mild cyclers and 2.1 h per week for severe cyclers (Figure 1). The lower activity level reflects that mild (OR ¼ 1.2, 95% CI 1.0–1.5) and severe cyclers (OR ¼ 1.8, 95% CI 1.3–2.4) were more inclined than noncyclers to change their dietary habits rather than use exercise as a weight control method (Table 2). International Journal of Obesity


1138

Figure 1 Dietary restraint scoresy and hours per week of activityz by weight cycling status among a sample of young and middle-aged female nurses. The white bars represent the women who were weight stable between 1989 and 1993. The bars with black diagonal stripes on a white background represent women who were not weight cyclers or weight stable between 1989 and 1993 (ie, the controls). The black bars represent women who were mild weight cyclers between 1989 and 1993. The bars with white diagonal strips on a black background represent women who were severe weight cyclers between 1989 and 1993. yMean scores on restraint scales. zMedian hours per week of activity 2000–2001.

Table 2

Associationa (odds ratios [OR] and 95% confidence intervals) of dietary restraint variables and weight cycling status among women in the NHS II substudy

Rather change diet than exercise (n) Age- and BMI-adjusted OR Life is too short to worry about dieting (n) Age- and BMI-adjusted OR Exercises after eating foods not allowed on diet (n) Age- and BMI-adjusted OR Eat less after eating foods not allowed on diet (n) Age- and BMI-adjusted OR Rarely stick with an exercise program (n) Age- and BMI-adjusted OR Make self-exercise to prevent weight gain (n) Age- and BMI-adjusted OR

Weight stable noncyclers (n ¼ 544)

Noncyclers, not wt stable (n ¼ 967)


Severe cyclers (n ¼ 224)

178 0.9 (0.7–1.1) 155 0.9 (0.7–1.3) 113 1.3 (0.9–1.8) 234 0.8 (0.6–1.0) 164 0.8 (0.6–1.0) 231 1.0 (0.8–1.3)

411 1.0 (ref) 260 1.0 (ref) 133 1.0 (ref) 475 1.0 (ref) 425 1.0 (ref) 357 1.0 (ref)

351 1.2 (1.0–1.5) 121 0.5 (0.4–0.7) 136 1.4 (1.1–1.8) 424 1.4 (1.1–1.7) 344 1.1 (0.9–1.4) 270 1.0 (0.8–1.2)

133 1.8 (1.3–2.4) 32 0.5 (0.3–0.7) 44 1.6 (1.1–2.4) 123 1.3 (0.9–1.7) 129 1.6 (1.2–2.1) 61 0.7 (0.5–1.0)

a

Adjusted for age and BMI in 1993.

Weight control strategies, bulimic behaviors, and weight cycling Most women in the sample reported using a variety of methods to lose or maintain weight. A low-fat diet was the most common approach to controlling weight, but the prevalence did not vary significantly by weight status (63% of women with BMI o25, 62.9% of overweight women, and 56% of obese women) or weight cycling status (61% of weight stable women, 62% of noncyclers, 66% of mild cyclers, and 61% of severe cyclers). However, mild and severe cyclers were significantly more likely than noncyclers to report using a low calorie diet (Table 1) or a high-protein diet to control their weight. Weight cyclers differed from noncyclers in terms of several behavioral strategies for controlling weight. Severe cyclers were more likely (OR ¼ 1.5, 95% CI 1.1–2.2) than noncyclers to report skipping meals or fasting as a weight control strategy, but International Journal of Obesity

they were significantly less likely to report using frequent exercise as a weight control strategy (OR ¼ 0.8) (Table 3). In addition, mild (OR ¼ 2.1) and severe (OR ¼ 3.1) cyclers were significantly more likely to have used a commercial weight loss program to help control their weight. Overall, a relatively small proportion of the women engaged in bulimic behaviors or other potentially unhealthy weight control strategies. Approximately 8.6% of the women used over the counter diet pills, 3.9% used smoking, and 2.1% using vomiting or laxatives (purging) to control weight. Weight cyclers were significantly more likely than noncyclers to use diet pills (OR ¼ 2.0–3.1) and purging (OR ¼ 3.0–6.8) to control their weight (Table 3), but weight cycling status was not independently associated with using smoking as a weight control strategy (data not shown). In addition to being more likely to use unhealthy weight control strategies, weight cyclers were more likely to engage


1139 Table 3 Associationsa (odds ratios [OR] and 95% confidence intervals) of binge eating and weight control practices with weight cycling among women in the NHS II substudy Weight stable noncyclers (n ¼ 544)

Noncyclers, not wt stable (n ¼ 967)


Severe cyclers (n ¼ 224)

Episodes of overeating (n) Age- and BMI-adjusted OR

56 0.7 (0.5–1.0)

195 1.0 (ref)

209 1.6 (1.3–2.0)

86 2.2 (1.6–3.0)

Binge eating episodes (n) Age- and BMI-adjusted OR

31 0.8 (0.5–1.3)

108 1.0 (ref)

134 1.8 (1.4–2.4)

58 2.5 (1.7–3.5)

Exercises frequently (n) Age- and BMI-adjusted OR

304 1.1 (0.8–1.3)

452 1.0 (ref)

353 1.0 (0.8–1.2)

87 0.8 (0.6–1.1)

Prescription or OTC diet pills (n) Age- and BMI-adjusted OR

33 0.7 (0.4–1.2)

84 1.0 (ref)

116 2.0 (1.5–2.7)

52 3.1 (2.1–4.6)

Laxatives or vomiting (n) Age- and BMI-adjusted OR

5 0.7 (0.2–2.2)

9 1.0 (ref)

21 3.0 (1.4–6.7)

12 6.8 (2.8–16.6)

OTC diet pills, laxatives, or vomiting (n) Age- and BMI-adjusted OR

6 0.3 (0.1–0.7)

29 1.0 (ref)

37 1.8 (1.1–2.9)

20 3.8 (2.1–6.9)

a

Adjusted for age and BMI in 1993. OTC ¼ over the counter.

Table 4

Correlates of weight change between 1993 and 2001 among young and middle-aged women in the United States B (s.e.)a Adjusted for age, BMI, and other covariates listed in the column

Adjusted for age and BMI Cycling status Weight stable noncycler Mild cycler Severe cycler Dietary restraintc Concern for dieting (HP) Weight fluctuation (HP) Factor 1 (TFFQ) Exercise Physical activity (h/week)c Binge eating Use of vomiting or laxatives z

7

yy

Fully adjustedb

3.4 (1.0)yy 6.7 (0.8)* 10.3 (1.2)*

3.5 (1.0)yy 6.9 (0.8)* 9.9 (1.2)*

2.6 (1.0)z 6.6 (0.8)* 9.1 (1.3)*

3.6 (1.0)yy 6.7 (0.8)* 9.7 (1.2)*

3.6 (1.0)yy 7.4 (0.8)* 9.8 (1.2)*

2.7 (1.1)z 6.9 (0.9)* 9.2 (1.3)*

0.1 (0.1) 0.4 (0.1)yy 0.6 (0.1)* 1.8 (0.2)* 0.5 (0.1)*

1.2 0.4 0.9 1.3

1.2 0.4 0.9 1.1 0.3

1.0 0.5 0.8 1.3

1.2 0.5 0.9 1.3

1.1 0.5 0.8 1.2 0.2 3.5 0.3

(0.1)* (0.1)yy (0.1)* (0.2)*

(0.1)* (0.1)7 (0.1)* (0.3)* (0.1)yy

(0.1)* (0.1)* (0.1)* (0.2)*

(0.1)* (0.1)c (0.1)* (0.2)*

4.7 (1.0)* 1.3 (2.3)

(0.1)* (0.1)* (1.0)* (0.3)* (0.1)7 (1.1)7 (2.5)

a

Po0.05. Po0.01. Po0.001. *Po0.000. Beta ¼ pounds gained between 1993 and 2001 per unit increase in the exposure variable. bAdjusted for age, BMI and all of the factors listed in the table (weight cycling status, restraint scales, activity, binge eating, and purging). cAdjusted for age, BMI, and weight cycling status.

in overeating episodes and binge eating. Overall, 22% of the women reported overeating (eating an amount of food that is definitely larger than most people would eat in similar circumstances) and 61% of those women (13% of the sample) reported a lack of control during the overeating episodes (ie, binge eating). Weight cyclers were significantly more likely than noncyclers to engage in both overeating and binge eating (Table 3). For example, weight cyclers were more than twice as likely as noncyclers to report binge eating at least monthly (OR ¼ 1.8 for mild cycling, OR ¼ 2.5 for severe cycling) (Table 3).

Weight cycling and weight gain Despite the increased prevalence of intentional weight losses, weight cyclers gained more weight than noncyclers between 1993 and 2001. Independent of age and BMI in

1993, mild cyclers gained an average of 7 lbs more than noncyclers and severe cyclers gained approximately 10 lbs more than noncyclers who were not weight stable between 1989 and 1991 (Table 4). Although weight control practices, activity patterns, bulimic behaviors, and level of dietary restraint differed by weight cycling status, only activity patterns, binge eating, and dietary restraint were independently associated with weight changes from 1993 to 2001. After adjusting for age, BMI, and weight cycling status in 1993, the concern for dieting subscale of Herman and Polivy’s restraint scale had a significant positive association with change, but the modified version of the weight fluctuation subscale was inversely associated with weight change. The dietary restraint scale adapted from the TFEQ Restraint scale was also inversely associated with weight change (b ¼ 0.6, Po0.001). Each one unit increase above the mean score on the scale was associated with gaining International Journal of Obesity


1140 approximately 0.6 lbs less between 1993 and 2001 than noncyclers who scored at the mean. However, of the psychological measures, the exercise factor that we developed had the strongest association with weight change (b ¼ 1.8, Po0.001). When the four restraint subscales were entered into one model, all of the factors remained significant; however, the exercise factor was somewhat attenuated and the concern for dieting subscale of Herman and Polivy’s restraint scale became much stronger (Table 4). Hours per week of leisure-time activity between 2000 and 2001 had a significant inverse association with weight gain from 1993 to 2001. Unhealthy weight control methods, such as the use of diet pills, vomiting, and/or laxatives was unrelated to changes in weight. However, independent of dietary restraint, age, BMI, and weight cycling status in 1993, women who engaged in binge eating at least monthly gained an average of 5 lbs more from 1993 to 2001 than women who did not binge eat (b ¼ 4.8 lbs, Po0.001). After adjusting for dietary restraint, age, BMI, activity, and binge eating, weight cycling status in 1993 remained strongly associated with weight change between 1993 and 2001. Mild weight cyclers gain approximately 7 lbs and severe cyclers gained 9 lbs more than noncyclers; whereas, the women who had been weight stable between 1989 and 1991 gained approximately 3 lbs less than noncyclers (Table 4).

Discussion We observed that weight gain was common and that women who had been weight cyclers gained significantly more weight than their peers during the follow-up. Most of the women had used dietary strategies to control their weight, but severe weight cyclers were significantly less likely than their peers to use exercise as a weight control strategy and were significantly more likely to use unhealthy weight control strategies. It is possible that the lack of activity is at least partially responsible for the large weight gains observed among the severe cyclers during the follow-up. Although there is debate about whether activity aids in weight loss, there is a growing body of literature that suggests that activity is helpful for weight maintenance and prevention of weight regain.33–36 In a previous analysis in the Nurses’ Health Study II, we observed that the vigorous activity was associated with less weight regain and that women who engaged in 5 or more hours per week of vigorous activity gained less weight than their inactive peers.35 Schoeller et al33 estimated that an average of 35 min a day of vigorous activity would result in weight maintenance among 32 middle-aged women who had recently lost weight. In our study, few women were engaged in the activity level recommended by Schoeller and the percentage was highest among weight-stable women and lowest among weight cyclers. Excessive weight and weight gain are major public health concerns in the United States. Obesity is a risk factor for International Journal of Obesity

developing hypertension,37,38 type II diabetes,38 certain cancers,22 cardiovascular disease,39 and total mortality,40 thus to design appropriate interventions, there is a need to identify factors that predict weight gain. Moreover, given the high prevalence of obesity, there is a need to identify factors that facilitate successful maintenance of weight loss. Although the majority of the weight-stable women, noncyclers, and cyclers ate a low-fat diet to control weight, we did not observe an association between eating a low-fat diet as a weight control strategy and weight change. Cyclers differed from noncyclers in terms of unhealthy weight control practices, such as skipping meals and purging. Owing to the study design, we do not know whether these maladaptive weight control practices were a cause or consequence of the weight gain. Prospective studies are needed to better understand these relationships; however, some prospective data support for our finding of a strong relationship between binge eating and weight gain. In the Growing Up Today Study, comprised of 16 882 of the NHS II offspring, boys who engaged in binge eating gained significantly more weight that their peers.41 Moreover, in several studies binge eating severity lessened with weight loss,42 thus it is more likely that the binge eating predated as opposed to resulted from the weight gain. Our results build on those of Pasman et al43 and French et al,44 who observed that women who diet to lose weight gained, or regained, more weight than their nondieting peers. Despite high prevalence rates of intentionally losing weight between 1998 and 2001 and a high prevalence of dieting, most of the women in the sample gained weight and women who had been mild or severe weight cyclers between 1989 and 1993 gained the most weight. Thus, our results are similar to those of Kroke et al,11 who observed that among a cohort of adults, weight cycling was the strongest predictor of weight change over a two period. Our study builds on the previous work on weight loss efforts and weight gain by including assessment of weight control strategies, dietary restraint, binge eating, and beliefs about weight control. We were therefore able to assess why weight cycling may be associated with weight gain, rather than just stating that weight cyclers gain more weight than noncyclers. Our findings suggest that the difference between women who weight cycle and those who do not weight cycle may be related to their outlook on physical activity. Severe weight cyclers were less likely than noncyclers to exercise to prevent weight gain and were more likely to prefer changing their diet rather than their activity patterns to control weight. Although a reduction in energy intake can lead to weight loss, our results suggest that more needs to be done to encourage physical activity to prevent weight gain or regain. A necessary first step will be to better assess the perceived barriers to activity, including lack of time and space and dislike of being active, among women who have a history of weight losses and regains. There are several limitations to the present study. First, we collected information on dietary restraint at the same time as


1141 we collected information on weight change. Therefore, it is possible that the dietary restraint scores changed in response to weight gain. However, given that few women gained weight for the first time in their adult life, this problem is true for most studies of adults. Another limitation is that we were unable to assess the specific weight control strategy that was used for weight loss or weight maintenance. It is therefore possible that some of the strategies were successful for weight loss, but the loss was not maintained and therefore the strategy did not appear to have an association with weight change in our analysis. Third, the sample was not a random sample of women in the United States. The women were sampled from the Nurses’ Health Study II; as a result there are relatively few women of color in the sample. Moreover, by design the sample was enriched with weight cyclers, so the overall estimate of weight gain should not be generalized to the general population, among whom weight cycling is less common than in the subgroup of women included in the analysis. In conclusion, our findings suggest that women who weight cycle gain more weight and place a lower priority on physical activity as a weight control strategy than their noncycling peers. Activity has numerous health benefits, including helping to prevent weight regain, thus clinicians should be encouraged to motivate their patients with a history of weight cycling and unsuccessful dieting to find a form of activity that they can incorporate into their daily lives. In addition, clinicians should discourage their patients from using unhealthy means of weight control, which may increase the risk of binge eating and do not appear to help to control weight. Future studies are needed which prospectively assess the relationship between weight cycling, binge eating, and weight change. To tease apart the order of association it will be essential for the study to include a sufficiently large sample size to be able to assess whether weight cycling predicts the onset of binge eating or binge eating predicts the onset of weight cycling and whether these two factors have independent associations with subsequent weight changes. Given the relatively young average age of onset of binge eating, it would be best to study these associations in a cohort of adolescents and young adults who will be followed into mid-adult life. The prospective association between physical activity and weight cycling should also be investigated. Moreover, future studies should consider assessing whether other family members have a history of obesity, major weight gain, and disordered eating in order to tease apart a familial predisposition to weight gain or disordered eating from weight gain due to weight control practices influenced by other personal and social factors.

Acknowledgements We thank Dr Nan Laird for her thoughtful comments and suggestions on the manuscript and Dr Meir Stampfer for his

suggestions on items to include on the questionnaire. Dr Field was partially supported by the Boston Obesity Nutrition Research Center (DK 42600). Additional funding was provided by a research Grant (CA50385-09) from the National Institutes of Health, and a First Independent Research Support and Training Award (R29) Grant (HL57871-01) from the National Institutes of Health.

References 1 Flegal KM, Carroll MD, Ogden CL, Johnson CL. Prevalence and trends in obesity among US adults, 1999–2000. JAMA 2002; 288 (14): 1723–1727. 2 Field AE, Camargo Jr CA, Taylor CB, Berkey CS, Frazier AL, Gillman MW, Colditz GA. Overweight, weight concerns, and bulimic behaviors among girls and boys. J Am Acad Child Adolesc Psychiatry 1999; 38 (6): 754–760. 3 Field AE, Wolf AM, Herzog DB, Cheung L, Colditz GA. The relationship of caloric intake to frequency of dieting among preadolescent and adolescent girls. J Am Acad Child Adolesc Psychiatry 1993; 32 (6): 1246–1252. 4 Casper RC, Offer D. Weight and dieting concerns in adolescents, fashion or symptom? Pediatrics 1990; 86 (3): 384–390. 5 Wilfley DE, Schreiber GB, Pike KM, Striegel-Moore RH, Wright DJ, Rodin J. Eating disturbance and body image: a comparison of a community sample of adult black and white women. Int J Eat Disord 1996; 20 (4): 377–387. 6 Jeffery RW, French SA. Socioeconomic status and weight control practices among 20- to 45-year-old women. Am J Public Health 1996; 86 (7): 1005–1010. 7 Williamson DF, Serdula MK, Anda RF, Levy A, Byers T. Weight loss attempts in adults: goals, duration, and rate of weight loss. Am J Public Health 1992; 82 (9): 1251–1257. 8 Serdula MK, Collins ME, Williamson DF, Anda RF, Pamuk E, Byers TE. Weight control practices of U.S. adolescents and adults. Ann Intern Med 1993; 119 (7 Part 2): 667–671. 9 Korkeila M, Rissanen A, Kaprio J, Sorensen TI, Koskenvuo M. Weight-loss attempts and risk of major weight gain: a prospective study in Finnish adults. Am J Clin Nutr 1999; 70 (6): 965–975. 10 Juhaeri, Steven J, Chambless LE, Tyroler HA, Harp J, Jones D, Arnett D. Weight change among self-reported dieters and nondieters in white and African American men and women. Eur J Epidemiol 2001; 17 (10): 917–923. 11 Kroke A, Liese AD, Schulz M, Bergmann MM, Klipstein-Grobusch K, Hoffmann K, Boeing H. Recent weight changes and weight cycling as predictors of subsequent two year weight change in a middle-aged cohort. Int J Obes Relat Metab Disord 2002; 26 (3): 403–409. 12 Field AE, Byers T, Hunter DJ, Laird NM, Manson JE, Williamson DF, Willett WC, Colditz GA. Weight cycling, weight gain, and risk of hypertension in women. Am J Epidemiol 1999; 150 (6): 573–579. 13 Hamm P, Shekelle RB, Stamler J. Large fluctuations in body weight during young adulthood and twenty-five-year risk of coronary death in men. Am J Epidemiol 1989; 129 (2): 312–318. 14 Lissner L, Odell PM, D’Agostino RB, Stokes III J, Kreger BE, Belanger AJ, Brownell KD. Variability of body weight and health outcomes in the Framingham population. N Engl J Med 1991; 324 (26): 1839–1844. 15 Lee IM, Paffenbarger Jr RS. Change in body weight and longevity. JAMA 1992; 268 (15): 2045–2049. 16 Blair SN, Shaten J, Brownell K, Collins G, Lissner L. Body weight change, all-cause mortality, and cause-specific mortality in the multiple risk factor intervention trial. Ann Intern Med 1993; 119 (7 Part 2): 749–757.

International Journal of Obesity


1142 17 Will JC, Williamson DF, Ford ES, Calle EE, Thun MJ. Intentional weight loss and 13-year diabetes incidence in overweight adults. Am J Public Health 2002; 92 (8): 1245–1248. 18 Lissner L, Andres R, Muller DC, Shimokata H. Body weight variability in men: metabolic rate, health and longevity. Int J Obes 1990; 14 (4): 373–383. 19 Colditz GA, Willett WC, Rotnitzky A, Manson JE. Weight gain as a risk factor for clinical diabetes mellitus in women. Ann Intern Med 1995; 122 (7): 481–486. 20 Huang Z, Willett WC, Manson JE, Rosner B, Stampfer MJ, Speizer FE, Colditz GA. Body weight, weight change, and risk for hypertension in women. Ann Intern Med 1998; 128 (2): 81–88. 21 Willett WC, Manson JE, Stampfer MJ, Colditz GA, Rosner B, Speizer FE, Hennekens CH. Weight, weight change, and coronary heart disease in women. Risk within the ‘normal’ weight range. JAMA 1995; 273 (6): 461–465. 22 Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003; 348 (17): 1625–1638. 23 Huang Z, Hankinson SE, Colditz GA, Stampfer MJ, Hunter DJ, Manson JE, Hennekens CH, Rosner B, Speizer FE, Willett WC. Dual effects of weight and weight gain on breast cancer risk. JAMA 1997; 278 (17): 1407–1411. 24 Trentham-Dietz A, Newcomb PA, Egan KM, Titus-Ernstoff L, Baron JA, Storer BE, Stampfer M, Willett WC. Weight change and risk of postmenopausal breast cancer (United States). Cancer Causes Control 2000; 11 (6): 533–542. 25 Solomon CG, Willett WC, Carey VJ, Rich-Edwards J, Hunter DJ, Colditz GA, Stampfer MJ, Speizer FE, Spiegelman D, Manson JE. A prospective study of pregravid determinants of gestational diabetes mellitus. JAMA 1997; 278 (13): 1078–1083. 26 Dietary, Guideline, Advisory, Committee. Report of the Dietary Advisory Committee on the Dietary Guidelines for Americans: Secretary of Health and Human Services and the Secretary of Agriculture; 1995. 27 French SA, Jeffery RW, Folsom AR, Williamson DF, Byers T. Relation of weight variability and intentionality of weight loss to disease history and health-related variables in a population-based sample of women aged 55–69 years. Am J Epidemiol 1995; 142 (12): 1306–1314. 28 Williamson DF, Pamuk E, Thun M, Flanders D, Byers T, Heath C. Prospective study of intentional weight loss and mortality in never-smoking overweight US white women aged 40–64 years. Am J Epidemiol 1995; 141 (12): 1128–1141. 29 Rimm EB, Stampfer MJ, Colditz GA, Chute CG, Litin LB, Willett WC. Validity of self-reported waist and hip circumferences in men and women. Epidemiology 1990; 1 (6): 466–473.

International Journal of Obesity

30 Stunkard AJ, Messick S. The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res 1985; 29 (1): 71–83. 31 Herman C, Polivy J. Restrained eating. In: Stunkard A (ed). Obesity. WB Saunders Co.: Philadelphia, PA; 1980. pp 208–225. 32 Laessle RG, Tuschl RJ, Kotthaus BC, Pirke KM. A comparison of the validity of three scales for the assessment of dietary restraint. J Abnorm Psychol 1989; 98 (4): 504–507. 33 Schoeller DA, Shay K, Kushner RF. How much physical activity is needed to minimize weight gain in previously obese women? Am J Clin Nutr 1997; 66 (3): 551–556. 34 Ewbank PP, Darga LL, Lucas CP. Physical activity as a predictor of weight maintenance in previously obese subjects. Obes Res 1995; 3 (3): 257–263. 35 Field AE, Wing RR, Manson JE, Spiegelman DL, Willett WC. Relationship of a large weight loss to long-term weight change among young and middle-aged US women. Int J Obes Relat Metab Disord 2001; 25 (8): 1113–1121. 36 Pronk N, Wing R. Physical activity and long-term weight maintenance. Obes Res 1994; 2: 587–599. 37 Yong LC, Kuller LH, Rutan G, Bunker C. Longitudinal study of blood pressure: changes and determinants from adolescence to middle age, 1957–1963 to 1989–1990. The Dormont high school follow-up study. Am J Epidemiol 1993; 138 (11): 973–983. 38 Field AE, Coakley EH, Must A, Spadano JL, Laird N, Dietz WH, Rimm E, Colditz GA. Impact of overweight on the risk of developing common chronic diseases during a 10-year period. Arch Intern Med 2001; 161 (13): 1581–1586. 39 Manson JE, Colditz GA, Stampfer MJ, Willett WC, Rosner B, Monson RR, Speizer FE, Hennekens CH. A prospective study of obesity and risk of coronary heart disease in women. N Engl J Med 1990; 322 (13): 882–889. 40 Manson JE, Willett WC, Stampfer MJ, Colditz GA, Hunter DJ, Hankinson SE, Hennekens CH, Speizer FE. Body weight and mortality among women. N Engl J Med 1995; 333 (11): 677–685. 41 Field A, Austin S, Taylor C, Malspeis S, Rosner B, Rockett H, Gillman MW, Colditz GA. The relation between dieting and weight change among preadolescents and adolescents. Pediatrics 2003; 112: 900–906. 42 Stunkard AJ, Allison KC. Two forms of disordered eating in obesity: binge eating and night eating. Int J Obes Relat Metab Disord 2003; 27 (1): 1–12. 43 Pasman WJ, Saris WH, Westerterp-Plantenga MS. Predictors of weight maintenance. Obes Res 1999; 7 (1): 43–50. 44 French SA, Jeffery RW, Forster JL, McGovern PG, Kelder SH, Baxter JE. Predictors of weight change over two years among a population of working adults: the Healthy Worker Project. Int J Obes Relat Metab Disord 1994; 18 (3): 145–154.