International Journal of Obesity (2002) 26, 969–972 ß 2002 Nature Publishing Group All rights reserved 0307–0565/02 $25.00 www.nature.com/ijo
PAPER Prevalence and correlates of large weight gains and losses RW Jeffery1*, MT McGuire1 and SA French1 1
Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis, USA
OBJECTIVE: To examine the prevalence and correlates of large weight gains and losses over a 3 y period in a heterogeneous population of participants in a study of weight gain prevention. DESIGN AND MEASURES: Analyses based on a cohort of 823 participants in a weight gain prevention study whose weight was measured on at least two of four annual examinations. RESULTS: Weight gains and losses of 5% body weight over 1 y were observed in 9.3 and 15% of the population, respectively. Weight gains among those initially losing weight were significantly greater (3.9 kg) than among those experiencing stable weight (0.8 kg) or a large weight gain (1.5 kg) over the following 2 y. Cumulative weight changes over 3 y were 7 2.6, 1.0 and 7.6 kg among large loss, weight stable and large gain groups, respectively. Large weight loss was more common in smokers, large gains were more common in younger people and in those with a more extensive weight loss history, and stable weight was observed more often in individuals with less extensive histories of weight loss. CONCLUSION: The high prevalence of large short-term weight gains and losses in this heterogeneous population, their apparent resistance to short-term reversal, and the strength of their relationship to longer-term weight trends suggest that rapid weight change over relatively short time intervals is a phenomenon that deserves more research attention. Short periods of rapid weight gain may contribute importantly to rapidly rising obesity rates. International Journal of Obesity (2002) 26, 969 – 972. doi:10.1038/sj.ijo.0802015 Keywords: obesity; weight gain; pattern
Introduction Dramatic increases in the prevalence of obesity over the last two decades have heightened interest in the natural history of weight gain.1 It is well documented that, in US adults, average weight gains of about 0.5 – 1.0 kg per year are seen between the ages of 20 and 50.2 However, the patterning of these weight changes over time have not been carefully studied. It is possible that weight gain with age is a slow process, characterized in most people by small annual increases due to a small chronic excess of energy intake over energy expenditure. On the other hand, it is also possible that weight gain with age often occurs in larger discrete steps (ie short periods of rapid weight gain interspersed with longer periods of stable weight). Because differ-
*Correspondence: RW Jeffery, Division of Epidemiology, University of Minnesota, School of Public Health, 1300 So. Second St, Suite 300, Minneapolis, MN 55454-1015, USA. E-mail:
[email protected]. Received 15 May 2001; revised 18 January 2002; accepted 4 February 2002
ent patterns of weight gain imply different behavioral etiologies, it is believed that a closer examination of them could inform strategies for addressing weight gain with age as a public health issue. We were able to locate only four prior studies that have specifically attempted to examine weight gain patterns over time. Two of these focused on patterns of gain leading to obesity. Haartz and Rimm3 used retrospective reports of maximum weight in each decade of life to describe the patterns of weight changes over time in women between the ages of 50 and 59 y who were members of the weight loss program TOPS. Heitman and Garby4 examined weight changes across successive 5 and 6 y intervals in a representative cohort of Danes who were not obese at age 25 but became so by the end of the 11 y observation period. Although both of these studies were limited by the number of time points measured and the distance between them, they came to similar conclusions. Patterns of weight gain leading to obesity are variable, but the most common pattern includes large weight gains in multiple time periods. The third study of weight gain patterns over time examined
Large weight gains and losses RW Jeffery et al
970 weight changes over four successive 2 y periods in a large cohort of nonsmoking women between the ages of 30 and 50 y.5 An inverse association was found between weight changes in adjacent time periods. That is, weight losses tended to be followed by weight gains and weight gains by weight losses. This tendency was much more striking for losses than gains, however. Women losing 5 kg or more over the first 4 y gained 5.8 kg on average over the next 4 y. Those whose weights were within 3 kg of baseline in the first 4 y gained 1.5 kg in the next 4 y. Those gaining 5 or more kg in the first 4 y gained 0.7 kg. The fourth study was a study of ‘holiday’ weight gain which examined weight gains in 165 adults over a 6 week Christmas holiday period in relation to total weight gain over 12 months.6 Results showed that 51% of a total weight gain of 0.62 kg over the year of observation occurred during the Christmas holidays, suggesting that a substantial portion of adult weight gain may occur in spurts. The present study examined patterns of weight change over a 3 y period in a heterogeneous sample of men and women participating in a study of weight gain prevention. The study assessed the prevalence of large annual gains and losses, the extent to which large weight changes in one year are typically followed by compensating weight changes in the opposite direction in subsequent years, and the extent to which large weight changes in a single year contribute to cumulative weight change over the entire 3 y. Overall, it was thought that a more detailed examination of weight gain and loss patterns over time might contribute to an improved understanding of the process by which people gain weight with age, and thus inform the conceptualization of what might be done to prevent weight gain in adults.
Table 1
Predictor variables Baseline BMI Year 1 weight change (kg) Age Sex: Male Female Education: High school or less Some college College or more Cigarette smoking: Yes No Restraint Regular dieting at baseline: Yes No History of intentional weight loss (total lb) History of unintentional weight loss (total lb) Weight history slope
Methods Data for the present study are from the Pound of Prevention Study (POP), a 3 y trial designed to evaluate whether low-cost educational methods would be successful in helping people to reduce weight gain with age. The design, methods, and overall results of the trial have been presented elsewhere.7 Briefly, about 1200 men and women between the ages of 20 and 45 were recruited to participate in a 3 y study of methods for preventing weight gain (Table 1). All were assessed annually with direct measurement of body weight and self-reported measurement of a variety of other weight-related variables. Half the participants were randomized to an educational intervention delivered primarily through monthly newsletters about healthy eating and exercise behavior. The remainder received no education. The effectiveness of the interventions was assessed through annual evaluation over 3 y. The present study examined data collected at annual examinations to calculate the prevalence and correlates of large weight gains and losses. Large weight gain was defined as an increase of 5% of body weight between baseline and the 1 y follow-up. The following baseline measures were also used in the analyses: International Journal of Obesity
Descriptive statistics for participants
n
Mean (%)
Standard deviation
823 789 824
26.69 0.50 38.52
5.48 3.76 6.54
176 647
21.0 79.0
— —
89 282 452
11.0 34.0 55.0
— — —
109 714 823
13.0 87.0 8.05
— — 4.35
130 693 823 823 823
16.0 84.0 42.33 15.01 0.31
— — 54.48 31.20 0.38
age in years; body mass index (BMI) — measured weight (kg) divided by measured height (m2); sex — male or female; education — high school or less, more than high school but less than college degree, or college degree or more; current cigarette smoking — yes or no; dietary restraint: the Cognitive Restraint Subscale of the 3-Factor Eating Inventory developed by Stunkard and Messick;8 regular dieting at baseline — yes or no; history of intentional weight loss — the total number of times participants reported having lost more than 20 lb (9.0 kg) intentionally in their lifetime; history of unintentional weight loss — the total number of times participants reported having lost more than 20 lb (9.0 kg) unintentionally in their lifetime; weight history slope — the slope of the linear regression line relating time to reported weight at 5 y intervals from age 15 to the present (calculated only for individuals 25 y old).
Analyses Analyses included all participants who did not become pregnant during the study, whose smoking status remained stable throughout the study period, and on whom measured weight was obtained on at least the year 1 or year 3 follow-up visit. The available sample size for baselines year 1 and year 3 was 823, 789 and 761, respectively. The primary questions posed in this paper were: (1) how common are annual weight gains and losses 5% of body weight; (2) how likely are large weight gains and losses experienced in one year to be reversed in the following 2 y;
Large weight gains and losses RW Jeffery et al
971 (3) how strongly are large weight changes in one year related to cumulative weight changes over 3 y; and (4) what are the correlates of large annual weight changes. Four sets of data analyses were conducted to address these questions. The first set calculated the prevalence of gains and losses of 5% of body weight between baseline and the first year of follow-up. The second analysis examined the stability of large weight changes by examining the mean weight change over the subsequent 2 y of individuals who gained 5%, lost 5%, or remained within 5% of their body weight in year 1. The third analysis examined cumulative weight losses over 3 y among individuals in the same three weight loss groups. The final analysis examined correlates of large weight changes. Logistic regression analyses were used to examine the relationships between the set of predictor variables described above and the likelihood of experiencing a large gain or loss in any of the 3 y of the study. Although weight changes observed in this study did not differ by treatment group, treatment group was included as a covariate in all analyses. Both bivariate and multivariate analyses were conducted.
gain of 7.6 kg on average. These means were significantly different from each other as well as from the mean cumulative change seen in the weight stable group of 1.03 kg. The final set of analyses examined factors correlated with having experienced large weight gains and losses in study year 1. Predictor variables considered in both bivariate and multivariate analyses were baseline BMI, sex, education, smoking, dietary restraint, regular dieting, history of intentional and unintentional weight loss, and age. Treatment group was included as a covariate. Three variables were associated with the likelihood of large weight gain. Older age reduced the likelihood of experiencing such a gain (P < 0.05). Regular dieting at baseline (P < 0.01 in bivariate analysis only) and having lost more weight intentionally in the past were associated with a higher likelihood of a large weight gain (P < 0.001 multivariate). Smoking was the only variable predictive of large weight losses. In multivariate analyses, current smokers were 2.25 times as likely as nonsmokers to lose 5% of their body weight in year 1 (P < 0.05). Three variables were significantly predictive of weight stability. Having lost more weight in the past intentionally, higher restraint scores (P < 0.03 in bivariate analysis only), and regular dieting to control weight (P < 0.01 in bivariate analyses only) were all associated with reduced likelihood of having a stable body weight over the next year.
Results Table 2 shows weight changes from baseline to 1 y, from 1 to 3 y, and from baseline to 3 y among individuals who lost 5%, gained 5%, or remained within 5% of their baseline body weight in the first year of observation. Overall, 9.3% of this population experienced a large weight loss, 15% experienced a large weight gain, and 76% stayed within 5% of their baseline weight. In the 2 y span between the first and third follow-up years, all three groups gained weight. Those who had lost 5% of their body weight in year 1 gained significantly more than those in the other two groups. Large weight gains were not reversed during this time span, however. Large gainers actually gained slightly more on average than the weight stable group. Cumulatively over the entire 3 y time span, the impact of large weight gains and losses in year 1 were still clearly evident. Although they did show significant weight regain after year 1, large weight losers remained 2.6 kg below their baseline weight on average. Large gainers on the other hand had a cumulative
Discussion The purpose of this analysis was to describe the prevalence of large weight gains and losses over relatively short periods of time in a nonclinical population to better understand their potential importance as contributors to cumulative weight changes with age. The results indicated that weight gains and losses 5% of body weight over 1 y are quite common. About 25% of this population experienced a change of this magnitude in a single year. Large weight losses in 1 y were predictive of larger than average gains in the next 2 y, but large weight gains in 1 y were not compensated for at all over the next 2 y. Over the entire time span covered by this study, large weight changes in either direction in the short term had significant long-term consequences. Despite weight
Table 2 Mean change in body weight (kg) during follow-up period as a function of change in body weight during year 1 Weight changea Percent body weight change in year 1 Weight loss group ( 7 5%)
n 69
Weight stable group ( 7 5 to þ 5%)
575
Weight gain group ( þ 5%)
117
a
Baseline to year 1 (s.d.) a
7 6.48 (4.19) b 0.22 (1.82) 6.07c (2.48)
Year 1 to year 3 (s.d.) a
3.90 (5.90) b 0.81 (5.30) 1.51b (6.55)
Baseline to year 3 (s.d.) 7 2.58a (4.97) b 1.03 (5.52) 7.58c (6.84)
Within each column means with unshared superscripts differ by P < 0.05.
International Journal of Obesity
Large weight gains and losses RW Jeffery et al
972 regain, large weight losers showed significant cumulative weight loss 2 y later compared to those with stable weight. On the other end of the continuum, large weight gainers showed cumulative weight gain of a magnitude likely to be associated with adverse health outcomes. These results are consistent with those previously reported by Colditz et al5 in indicating that there appears to be stronger defenses against weight loss than gain. They are also consistent with recent data published by Field et al9 indicating that there often is long-term benefits associated with large weight losses, despite the fact that regain is common. Considering the high prevalence of large short-term changes in weight in both directions, and their persistence over time, the present results suggest quite strongly that large short-term weight changes may be an important contributor to weight gain with age. Strengths of the present study include a prospective design, 3 y of observation, a nonclinical sample, and inclusion of both men and women. Limitations include a volunteer sample limiting generalizability, a shorter total time period and fewer weight measurements than would be desirable, a limited age range, and a limited number of predictor variables. Nevertheless it is believed that the results of this investigation underscore important points about short-term weight variability and suggest that more study of its predictors and consequences would be worthwhile.
Acknowledgements This research was supported by the National Institutes of Diabetes and Digestive and Kidney Diseases grant DK 45361,
International Journal of Obesity
with additional funding from the Centers for Disease Control and Prevention. The authors would also like to thank Ms Emily Finch for her assistance with data analyses. References 1 Flegal KM, Kuczmarski RJ, Johnson CL. Overweight and obesity in the United States: prevalence and trends, 1960 – 1994. Int J Obes Relat Metab Disord 1998; 22: 39 – 47. 2 Burke GL, Bild DE, Hilner JE, Folsom AR, Wagenknecht LE, Sidney S. Differences in weight gain in relation to race, gender, age and education in young adults: the CARDIA study. Ethne Health 1996; 1: 327 – 335. 3 Hartz AJ, Rimm AA. Natural history of obesity in 6,946 women between 50 and 59 y of age. Am J Public Health 1980; 70: 385 – 388. 4 Heitmann BL, Garby L. Patterns of long-term weight changes in overweight developing Danish men and women aged between 30 and 60 y. Int J Obes Relat Metab Disord 1999; 23: 1074 – 1078. 5 Colditz GA, Willett WC, Stampfer MJ, London SJ, Segal MR, Speizer FE. Patterns of weight change and their relation to diet in a cohort of healthy women. Am J Clin Nutr 1990; 51: 1100 – 1105. 6 Yanovski JA, Yanovski SZ, Sovik KN, Nguyen TT, O’Neil PM, Sebring NG. A prospective study of holiday weight gain. New Engl J Med 2000; 342: 861 – 867. 7 Jeffery RW, French SA. Preventing weight gain in adults: the Pound of Prevention study. Am J Public Health 1999; 89: 747 – 751. 8 Stunkard AJ, Messick S. The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res 1985; 29: 71 – 81. 9 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: 1113 – 1121.
