Although the rise in overweight and obesity in the United States is well documented, long-term weight loss maintenance (LTWLM) has been minimally explored.
The aim of this study is to estimate the prevalence and correlates of LTWLM among US adults.
Design, setting and participants:
We examined weight data from 14 306 participants (age 20–84 years) in the 1999–2006 National Health and Nutrition Examination Survey (NHANES). We defined LTWLM as weight loss maintained for at least 1 year. We excluded individuals who were not overweight or obese at their maximum weight.
Among US adults who had ever been overweight or obese, 36.6, 17.3, 8.5 and 4.4% reported LTWLM of at least 5, 10, 15 and 20%, respectively. Among the 17.3% of individuals who reported an LTWLM of at least 10%, the average and median weight loss maintained was 19.1 kg (42.1 pounds) and 15.5 kg (34.1 pounds), respectively. LTWLM of at least 10% was higher among adults of ages 75–84 years (vs ages 20–34, adjusted odds ratio (OR): 1.5; 95% confidence interval (CI): 1.2, 1.8), among those who were non-Hispanic white (vs Hispanic, adjusted OR: 1.6; 95% CI: 1.3, 2.0) and among those who were female (vs male, adjusted OR: 1.2; 95% CI: 1.1, 1.3).
More than one out of every six US adults who has ever been overweight or obese has accomplished LTWLM of at least 10%. This rate is significantly higher than those reported in clinical trials and many other observational studies, suggesting that US adults may be more successful at sustaining weight loss than previously thought.
The obesity epidemic in the United States is a significant public health concern. Two-thirds of the US adult population is overweight or obese with a body mass index (BMI) ⩾25, and the prevalence of obesity (BMI ⩾30) has doubled between 1980 and 2004.1, 2, 3 This continued rise in overweight and obesity increases the population's risks of several obesity-related illnesses, such as type 2 diabetes and hypertension.4, 5, 6 Modest weight loss treatment targets of 5–10% of initial body weight are recommended based on the evidence that morbidity and mortality are significantly lower at this level, although any intentional weight loss improves health in patients with comorbidities.7, 8
Helping patients lose weight and keep it off, however, remains a vexing public health problem. On average, current state-of-the-art behavioral weight loss programs produce weight losses of 10% of initial body weight over 30 weeks.9, 10 Although many patients have weight regain, some individuals are successful at both losing weight and maintaining this weight loss over the long term. The identification of individuals who are able to achieve long-term weight loss maintenance (LTWLM) may allow for further development of interventions to aid others in maintaining weight loss.
Despite the wealth of data on obesity rates and efficacy of clinical trials of weight loss treatments, little information exists about the epidemiology of individuals achieving LTWLM. Much of the data to address this issue come from two sources: published clinical trials and the National Weight-Control Registry (NWCR).11, 12 The NWCR is a database of individuals who were recruited initially based on their being able to lose at least 13.6 kg of their body weight (average, 30 kg) and keep it off for at least 1 year (average, 5 years).11 The NWCR, however, is not population based and is therefore not useful for providing estimates of LTWLM in the United States.11
The goals of this study were to provide nationally representative estimates and correlates of LTWLM among US adults. On the basis of the evidence that loss of at least 10% consistently produce beneficial changes to health-risk factors, we defined LTWLM as losing 10% of maximum body weight and keeping it off for at least 1 year.7, 12
Subjects and methods
We analyzed data from the 1999–2006 National Health and Nutrition Examination Survey (NHANES), an annual survey evaluating the health and nutrition of a representative portion of the US population. We included surveys from years 1999 to 2006 because 1999 was the first year when the weight history questionnaire was added to NHANES. The data were collected through in-home interviews using a standard questionnaire, which has been described in previous studies.13
Weight status and weight history were evaluated using several self-reported measures. Self-reported measures were used because measured weight was only available for individuals' current weight, not their maximum weight or weight 1 year ago. Maximum weight was assessed by the question, ‘Up to the present time, what is the most you have ever weighed?’; weight 1 year ago by, ‘How much did you weigh a year ago?’; current weight by, ‘How much do you weigh without clothes or shoes?’; and current height by, ‘How tall are you without shoes?’ The question that assessed maximum weight asked women to exclude maximum weights during pregnancy. Individuals who lost at least 10 pounds in the previous year were asked, ‘Was the change between your current weight and your weight a year ago intentional?’ BMI was calculated by dividing the participant's self-reported weight in kilograms by the self-reported height in meters squared. These values were then classified into six standard BMI categories, <18.5, 18.5 to <25, 25 to <30, 30 to <35, 35 to <40 and ⩾40, using standard National Institutes of Health cut-points.
For this study, we calculated whether or not each individual had achieved 5, 10, 15 or 20% LTWLM, based on the self-reported maximum weight, weight 1 year ago and current weight. For example, for an individual to meet the criterion of ⩾5% LTWLM, an individual's weight 1 year ago, as well as their current weight, would have to be ⩾5% less than their maximum weight.
The demographic variables analyzed included gender, age, race/ethnicity, education level and marital status, according to standard NHANES categories. Weight-related variables included current BMI categories, years since the maximum weight and current weight-control status (attempting to lose and maintain weight, attempting to lose weight only, attempting to maintain weight only and neither attempting to lose nor maintain weight). We also included several health questions taken from NHANES, including the diabetes questionnaire, blood pressure and cholesterol questionnaire, and current health status questionnaire including high cholesterol (yes, no), hypertension (yes, no), diabetes (yes, no) and overall health (poor, fair, good, very good, excellent).
Participants were excluded if their maximum BMI was <25, if they were under the age of 20 or over 84 years. The final sample size included 14 306 individuals. Data were weighted according to the NHANES analytical guidelines to account for the complex sampling design and were analyzed using weighted methods, primarily bivariate and multivariate logistic regression, with SAS version 9.2 (SAS Institute Inc, Cary, NC, USA; 2002–2003). All data presented in the tables represent weighted analyses and reflect nationally representative estimates.
Demographic and weight-control characteristics of 14 306 US adults who were ever overweight or obese, representing 74.6% of the total 1996–2006 NHANES population, are shown in Table 1. The population included almost equal numbers of men (52.3%) and women (47.7%), and included 71.7% non-Hispanic white, 12.5% Hispanic and 11.8% non-Hispanic black participants. Approximately half (53.7%) of the adults completed more than a high school education and nearly two-third (66.6%) were married or living with a partner. In terms of overall health, the majority of individuals reported ‘good’ (32.1%), ‘very good’ (29.9%) or ‘excellent’ health (19.5%). However, losing weight was a current goal for only one-third of individuals, despite the fact that 82.6% of individuals were either overweight or obese.
Overall, among US adults who have ever been overweight or obese, 36.5, 17.3, 8.5 and 4.4%, respectively, have maintained a weight loss of at least 5, 10, 15 and 20% (Table 2). Further description of the individuals who make up these groups can be found in Table 2. Table 3 shows variables associated with LTWLM of at least 10% in the bivariate and multivariate analyses. Participant characteristics that were found to be statistically significant in bivariate analyses (age, gender, race, education, marital status, current BMI and diabetes) were used as covariates in the adjusted analyses.
After adjusting for these demographics, the results of this population demonstrated that (1) women, (2) older adults (age 75–84 years), (3) non-Hispanic whites and (4) those with less than a high school education had a higher prevalence of LTWLM, whereas (5) married or partnered individuals had a lower prevalence of LTWLM. Women had a higher prevalence of LTWLM of at least 10% compared with men (19.2 vs 15.6%; odds ratio (OR)=1.2; 95% confidence interval (CI)=1.1–1.3). Adults with the age of 75–84 years also had greater LTWLM compared with those of 20–34 years (29.3 vs 15.2%; OR =1.5; 95% CI 1.2–1.8). In addition, non-Hispanic whites had a higher prevalence of LTWLM when compared with Hispanic individuals (18.6 vs 12.9%; OR=1.6; 95% CI 1.3–2.0). Those individuals with less than a high school education also had a higher proportion of LTWLM of at least 10% when compared with individuals with more than a high school education (19.3 vs 16.1%; OR=1.3; 95% CI 1.1–1.5). Finally, participants who were married or living with a partner had a lower prevalence of LTWLM when compared with those who were never married (15.4 vs 17.8; OR=0.7; 95% CI 0.6–0.8) or widowed or divorced or separated (15.4 vs 24.5; OR=0.6; 95% CI 0.5–0.8).
Weight was also associated with achieving LTWLM of at least 10%. Adjusted analyses found a higher prevalence of weight maintenance in individuals with a BMI <25 or 25–30 (42.2 vs 13.3%; OR=9.4 and 2.0, respectively; P<0.001) compared with those with a BMI ⩾40. However, individuals with a maximum BMI of ⩾40 had a higher prevalence of LTWLM than those with a BMI in the overweight range (OR=2.1; P<0.001). A greater number of years since maximum weight was also strongly associated with current LTWLM of at least 10%, with individuals who had reached their maximum weight >10 years ago having 30.9 greater odds than those who weighed their heaviest less than 2 years ago (P<0.001). Current weight-control goals were no longer associated with LTWLM in the adjusted analyses.
Health status was also associated with achieving LTWLM of at least 10%. One-fourth of individuals who reported having diabetes experienced LTWLM compared with 16.5% of individuals without diabetes (OR=2.1; P<0.001). Individuals who reported poor overall health had higher rates of LTWLM than those who reported their health as excellent (34.1 vs 17.5%; OR=2.3; P<0.001). Older participants and those with diabetes were significantly more likely to rate their overall health as poor (P<0.001).
Among those who reported losing at least 10 pounds in the previous year, 69.0% reported that the loss was intentional (Table 4). Weight loss was more likely to be intentional in younger individuals, females, non-Hispanic whites, those with greater than a high school education, a history of diabetes or better overall health. Intentional weight loss was less prevalent in the 75–84 year olds as compared with all the other age groups (OR=0.2; P<0.001). In addition, individuals who rated their health as ‘poor’ were less likely to have intentional weight loss when compared with those whose health was rated as ‘excellent’ (OR=3.1; P<0.001).
The average and median amounts of weight lost by those individuals with LTWLM of at least 10% was ∼19.1 and 15.5 kg, respectively. Figure 1 shows the distribution of weight lost by these individuals. The average current weight of individuals with LTWLM of at least 10% was 78.4 kg and their mean age was 49.5 years (Table 5). Individuals reported an average of 14.8 years since they had weighed their maximum weight. Despite losing at least 10% of their maximum weight and maintaining this loss for >1 year, individuals with LTWLM remained overweight with an average BMI of 27.
Table 5 additionally presents a comparison of NHANES participants from our study who have achieved LTWLM of at least 10% with participants from the NWCR.11 Overall, the NHANES participants were older (49.5 vs 45.3 years), less educated and were more likely to be male (47.1 vs 18.8%). In addition, there were significantly more minority members in NHANES as compared with NWCR (10.3 vs 1.4% non-Hispanic African Americans and 9.3 vs 1.1% Hispanics). Participants in the NWCR lost more weight than the NHANES participants (30.0 vs 19.1 kg), although both groups were similar in maximum weights (99.9 vs 97.5 kg).
Our results demonstrate that a significant percentage of US adults are able to lose weight and keep it off. One in three (36.5%) ever-overweight or obese US adults has had LTWLM of at least 5% of their body weight, a population which includes 74.6% of the total NHANES sample from 1999–2006. On average, these individuals achieved a weight loss of 13.2 kg. This finding is in contrast to clinical trials that show less weight loss maintenance success. Though few intervention studies follow individuals for longer than 1–2 years, ∼10–20% of subjects in clinical trials are able to maintain a loss of at least 5% for ⩾5 years.12 The Diabetes Prevention Program showed that ∼37% of participants maintained a weight loss of at least 7%, or an average of ∼6.0 kg, after 3 years.14 Meta-analyses of weight loss intervention studies found that individuals in the experimental arms of clinical trials are able to lose and maintain an average of 3.0 kg or 3.2% of their original body weight for at least 2 years.15, 16
There are multiple potential explanations for finding a higher proportion of US adults accomplishing LTWLM than those reported in clinical trials and other observational studies. First, individuals who participate in clinical trials may be less likely to have long-term success, possibly because this is a selected population.17 Second, some of the current weight loss experienced in our estimates includes unintentional weight loss. Third, there may be a substantial number of individuals who may experience weight gain during a particular time in their life but manage to return to a normal weight, such as those who gain the ‘freshman 15’ when entering college. Given the brevity of such experiences, these individuals would be less likely to be included in clinical trials, but would be captured in our analysis. Finally, we report on individuals who maintain a weight loss of >5% below their maximum weight, whereas clinical trials will often capture participants who maintain weight below their baseline weight upon entry into the trial. Again, clinical trials will be less likely to capture individuals at their maximum weight.
LTWLM was more common among women, older adults, individuals who reported poor overall health and those with less than a high school education. We observed that 19.1% of women vs 15.6% of men (P<0.0001) were able to achieve LTWLM of at least 10% and that recent weight losses were more likely to be intentional for women. This variation in LTWLM by gender may occur as a result of different uses of weight-control behaviors. Neumark-Sztainer et al.18 observed that women were more likely than men to report dieting and using each of the 23 weight-control practices surveyed, supporting our observation that LTWLM was more common in women. Although we found that LTWLM was more common in older adults (age 75–84 years), this population was more likely to have unintentional weight loss. In addition, patients with diabetes had a higher prevalence of LTWLM. This is encouraging, but not surprising, given the consistent recommendations from the American Diabetes Association and other professional organizations for patients with type 2 diabetes to lose weight.19 In addition, individuals with diabetes were much more likely to report poor health, which was also associated with LTWLM. Individuals with less than a high school education also had a higher prevalence of LTWLM. Further research is necessary to examine the reason why these individuals are more successful with LTWLM.
An interesting finding was that >80% of the US adults who were overweight or obese reported their health as ‘good’, ‘very good’ or ‘excellent’. This finding is despite a significant percentage of individuals with hypertension (82%) and/or hyperlipidemia (42%). Obesity has been shown to have a negative impact on self-rated health among adults, even in the absence of comorbidities.20, 21, 22 However, this population does have a percentage of participants who are currently overweight (45.6%), which has been shown to be associated with a slight reduction of self-rated health.20, 23 In addition, 17.4% of individuals were normal weight, which would not be expected to influence overall reported health.
We also found that only one in three individuals had a current weight-control goal to lose weight. Health-related quality of life, or more generally, health status, have been found to influence weight loss effort.24, 25, 26 This raises concerns for the potential implications of overweight and obese individuals with positively rated health and their resultant motivation to lose weight. However, multiple factors influence weight loss motivation, especially in women, including appearance24, 26 and body image.24 Therefore, it is not entirely surprising that a minority of individuals were attempting to lose weight.
Physician advice also has an important role in the decision to lose weight.24 Therefore, it is important for health professionals to understand the true prevalence of long-term weight loss, as it may help to change the underlying beliefs and influence clinical practice. Foster and colleagues observed that physicians rated the treatment of obesity as significantly less effective than therapies for 9 of 10 other chronic conditions, such as asthma and diabetes.27 Tsui et al.28 observed that only one-quarter (27%) of physicians were confident in their weight loss counseling skills. Sciamanna et al.29 found that fewer than half (47.3%) of the obese patients with a weight-related comorbidity (for example, hypertension) reported having received advice to lose weight in the previous year from a health-care professional. Although these studies did not directly measure the belief that patients could actually lose weight and keep it off, they suggest physicians may not believe offering weight loss advice and counseling is a worthwhile activity in clinical practice. An awareness of our findings may encourage health professionals to pursue weight loss counseling for overweight patients.
In addition to informing health-care providers, identification of a significant percentage of the population that is attempting and succeeding in some weight loss may be an important target population for weight maintenance programs. Although the amounts lost are modest, if a substantial number of individuals achieved such losses, it would have a significant public health effect. Particularly, those individuals who have lost at least 5% and kept it off (one in three ever-overweight Americans) may represent a unique opportunity to reach a target population who has had some success but could benefit from greater weight loss efforts.
There are several potential limitations to this study. First, this is a single time point study and we only have information of individuals' current, maximum and weight 1 year ago. Therefore, we cannot determine the total duration of weight loss maintenance. In addition, we are unable to note whether individuals have experienced weight regain or weight cycling. For example, someone with a maximum weight of 100 kg, who weighed 75 kg 2 years ago and currently weighs 85 kg would be classified as achieving LTWLM of >10%. However, it can be argued that this individual still remains more successful at weight loss than he or she was at their maximum weight.
A second limitation is we were not able to determine the method of weight loss. Therefore, the data may include individuals with unintentional weight lost from illness. One in three individuals with LTWLM of at least 10% reported their health as poor. Given the cross-sectional nature of this study, we are unable to determine when the weight loss occurred with respect to the individual's health. In addition, we cannot determine whether those individuals with large amounts of weight loss had undergone bariatric surgery. Increasingly, surgery is a popular option for weight loss in overweight individuals. The number of people overall who undergo weight loss surgery, however, remains too small to affect our results. In 2002, rates of gastric bypass surgery were only 38.6 per 100 000. Despite an increase in surgical rates, we did not observe an increase in LTWLM over time.30
Another potential limitation is that the data are self-reported and therefore are subjected to recall problems in addition to over- and underestimation. Casey et al., however, reported that recall of past weight is relatively stable over time and was affected by a mean underestimation of 1.98 kg for men and 1.86 kg for women. Stevens et al.32 observed a correlation between reported and measured weights of 0.98 for current weight, 0.94 for weight 4 years ago and 0.82 for weight 28 years ago. We expect some degree of inaccurate data reported but, based on the available data supporting the accuracy of current and past weight reporting, these inaccuracies are unlikely to affect the main observations.
A major strength of this study is the use of a large, nationally representative sampling of LTWLM in US adults. We are aware of only a few studies that have examined the epidemiology of LTWLM in US adults. One study, limited to women in Iowa, examined weight loss that occurred between the ages of 18–30 years and evaluated for maintenance of this weight loss (±5%) between the ages of 30 and 50 years.33 Using this narrow definition, investigators observed that only 1.6% of the total sample was classified as having achieved LTWLM.33 A second study, by McGuire et al.34, examined the rate of LTWLM in a random-digit-dial survey that included only individuals who were ever overweight. One in five (20.6%) of the adults in the study intentionally lost at least 10% of their maximum weight and kept it off for at least 1 year. In comparison to the NWCR participants, a selected group of individuals recruited for their weight loss and maintenance efforts, NHANES offers a population-based sample and insight on estimates of LTWLM in the United States.11 Although Weiss et al.13 also examined the NHANES population, their work focused largely on weight regain and only includes participants from the years 1999–2002.
In conclusion, more than one-sixth of individuals who has ever been overweight and obese are able to lose and maintain long-term 10% weight loss, a success rate significantly better than those reported in weight loss clinical trials. Although we find this information encouraging, it remains clear that the great majority of individuals who are overweight are not able to lose much weight and keep it off for the long term. This suggests that the majority of Americans will be unable to maintain a significant amount of weight loss without a significant change in either the efficacy or availability of weight loss and weight maintenance interventions. Further investigation of the LTWLM methods of individuals who are able to achieve success may reveal important knowledge relevant to the creation of successful weight maintenance interventions.