These conclusions were derived from a review of the literature and from Stunkard and McLaren-Hume's [2] observation of 100 consecutive patients treated by conventional diets providing 800 to 1500 kcal/d. Only 12 patients lost 9.1 kg (20 lb) or more, and only 1 patient lost 18.2 kg (40 lb). Thirty-nine patients failed to return to the nutrition clinic for a second visit. Long-term weight losses were equally discouraging.

This article reviews recent studies of the short- and long-term effectiveness of both moderate and severe caloric restriction, as determined by randomized clinical trials. The effectiveness of moderate caloric restriction was assessed by examining all studies published in four behavioral journals in 1974 and 1978 and from 1984 to 1990. All controlled studies of severe caloric restriction that included follow-up data were reviewed. Where possible, current findings are compared with Stunkard's [1], which are often thought to characterize current treatment.

Moderate Caloric Restriction

Research trials usually prescribe a diet of 1200 kcal/d of conventional foods for obese women and a diet of approximately 1500 kcal/d for obese men. Patients are encouraged to consume approximately 15% of calories from protein, no more than 30% from fat, and the rest from carbohydrate. Some trials allow participants to select their own foods, whereas other trials provide greater structure in the form of meal plans, such as that developed jointly by the American Diabetes and American Dietetic Associations [3].

The common practice of prescribing the same diet for all patients creates energy deficits and markedly different weight losses [4]. Thus, a short, mildly obese woman with an energy expenditure of 1500 kcal/d will lose only 0.5 kg (1 lb) a week on a 1000-kcal/d diet. By contrast, a taller, heavier woman with a daily energy expenditure of 2500 kcal/d will lose approximately 1.4 kg (3 lbs) a week on the same diet. This finding underscores the need to estimate, if not measure, the energy expenditure of overweight patients to select an appropriate reducing diet [5].

Behavior Modification

Most obese persons will lose weight when placed on an appropriately restricted diet [6]. The magnitude of their weight loss, after accounting for basal energy expenditure, depends largely on the adequacy of their adherence to the prescribed caloric intake and on the duration of the diet.

Adherence appears to be partly a function of the intensity of the supervision provided by the weight loss program. Thus, persons who are given diet plans or treatment manuals by their health providers and told to "come back in 3 months" receive minimal supervision and are unlikely to lose significant amounts of weight. Similarly, the average weight loss of persons who participate in large-scale, community interventions [7] or self-help programs [8] is modest. The results of such efforts are likely to resemble those described by Stunkard in the 1950s.

The introduction of behavior modification in the 1960s significantly improved the short-term treatment of obesity [9, 10]. Thus, all current research studies of the dietary management of obesity include elements of behavioral treatment, which provides patients with a set of principles for evaluating and modifying their eating and exercise habits as well as their emotional reactions to their weight. Patients keep daily records of their physical activity, food intake, and any difficulties encountered, which may then be reviewed with a dietitian, behavioral psychologist, or other provider. Brownell [11] has provided an excellent account of this approach in a 200-page manual.

The structure of treatment is perhaps as important as the principles and techniques themselves [12]. Specifically, most behavioral weight-loss programs provide weekly instruction to groups limited to 10 to 20 participants. In addition, the length of treatment is specified in advance and usually lasts from 10 to 25 weeks. Thus, patients receive regular and intensive supervision. When this structure is adopted, more traditional dietary interventions may produce weight losses similar to those associated with the strongest behavioral programs [13].

Short-Term Weight Losses

Table 1 provides a summary of randomized trials in which moderate caloric restriction was combined with behavior modification [14]. The studies were published in 1974 and 1978 and from 1984 to 1990 in the following journals: Addictive Behaviors, Behavior Therapy, Behaviour Research and Therapy, and Journal of Consulting and Clinical Psychology. Results of these studies are similar to those published in medical and nutrition journals.

The table shows that current treatment (that is, studies from 1988 to 1990) produces an average weight loss of 8.5 kg in 21.3 weeks and is associated with attrition of approximately 20%. This weight loss is more than double that achieved in earlier studies. In 1974, for example, patients lost only 3.8 kg in treatment that averaged 8.4 weeks.

The increase in mean weight loss from 1974 to 1990 could be attributable to several factors, including the heavier initial weights of patients in the most recent studies; heavier patients typically lose the most weight. The best explanation, however, for the larger losses is probably the longer duration of treatment. The average weekly rate of weight loss of 0.4 to 0.5 kg has not changed since 1974; however, longer treatment periods have given patients more time to lose weight. This finding has led investigators to extend treatment up to 1 year in hopes of producing average losses as great as 20 to 25 kg. In a 40-week study, Perri and associates [15] found that patients lost 9.9 kg in the first 20 weeks of therapy but only an additional 3.6 kg in the second 20 weeks. In a similar, 52-week study [14], patients lost 11.9 kg during the first 26 weeks but only an additional 2.5 kg at the end of the year. In both studies, patients remained substantially overweight at the end of treatment. Thus, the size of the weight losses that can be produced by conventional reducing diets (as combined with behavior modification) appears to be limited. It is unclear whether these limits result primarily from metabolic influences or from reduced motivation and adherence [16].

Long-Term Weight Losses

The studies from 1984 to 1990 (see Table 1) show that patients treated by moderate caloric restriction regained an average of one third of lost weight in the year after treatment. This regain contrasts with the generally good maintenance of weight loss seen in the studies from 1974 and 1978. Patients in the early studies, however, lost smaller amounts of weight and were followed for briefer periods, thus allowing less time to observe possible weight regain.

Longer-term studies indicate that patients regain increasing amounts of weight with increasing time from therapy [17, 18]. Most persons in research trials treated by moderate caloric restriction are likely to return to their baseline weight within 5 years of completing therapy.

Improving Long-Term Results

Perri and colleagues [13, 19] have shown that weight regain can be prevented by continuing care. They view obesity as a chronic disorder requiring long-term treatment, similar to the management of hypertension or diabetes. In a promising randomized trial [19], patients who lost 13.2 kg during an initial 20 weeks of treatment using moderate caloric restriction subsequently received 52 weeks of biweekly maintenance therapy in which they continued to attend group meetings and discuss weight control skills. At the end of this time, they maintained an average weight loss of 12.9 kg. By contrast, patients who received the initial 20-week weight-reduction program, but no maintenance therapy, regained almost 50% of their weight loss during the year. Additional studies have shown that regular patient-therapist contact facilitates maintenance of weight loss, whether the contact is in person, by telephone, or by mail [11].

Adding an exercise regimen to moderate caloric restriction also improves long-term results [20]. Both correlational studies [21] and randomized trials [22] have shown that persons treated by diet plus exercise are more likely to maintain their weight losses than are those treated by diet alone. In addition to increasing energy expenditure, exercise minimizes the loss of fat-free mass that typically accompanies weight reduction [23]. Regular exercise may also have favorable effects on mood, which facilitates adherence [24].

None of the studies summarized in Table 1 included programs of intensive exercise or weight maintenance therapy (as developed by Perri). Thus, the findings summarized in the table represent the short-and long-term results that can be expected with a typical course of treatment lasting 15 to 25 weeks. Better results can be achieved with more intensive interventions.

Severe Caloric Restriction

The failure in the 1950s of moderate caloric restriction to produce adequate weight losses in the significantly obese led investigators to explore the use of very-low-calorie diets (VLCDs) providing 800 or fewer kcal/d [14]. These diets provide large amounts of dietary protein (>50 g/d) of high biological quality and have been shown to spare fat-free mass while producing average weekly weight losses of 1.5 to 2.0 kg in women and 2.0 to 2.5 kg in men [25]. Protein may be obtained from lean meat, fish, and fowl, or from egg and milk sources. In the latter, the protein is powdered and later mixed with water (by the patient) to yield a liquid diet.

Extensive research has shown that current VLCDs, as contrasted with the liquid protein diets of 1976 and 1977, are generally safe when administered to appropriate patients (30% or more overweight) under medical supervision [25, 26]. Recent studies, however, have shown that these diets are associated with an increased risk for gallstones [27, 28], a problem that requires further investigation. Anecdotal reports [29] have also linked VLCDs to an increased risk for binge eating [30]; however, no controlled studies of this issue have been done.

Short-Term Weight Losses

Several uncontrolled studies have shown that treatment using VLCDs for 12 to 16 weeks produces aver-age weight losses of approximately 20 kg in women and 25 kg in men [29–33]. These mean losses may be some-what inflated because of the failure in many studies to account for attrition. A recent study of 407 women, however, included the weight losses of dropouts in the analysis and showed a mean loss of 19.2 kg for 12 weeks of treatment using a liquid VLCD [12]. (The long-term weight losses of dropouts, however, were not reported.) Thus, VLCDs produce average losses that are two to three times greater than those resulting from conventional reducing diets used for the same period.

Despite these impressive short-term results, the success of VLCDs may not be attributable to severe caloric restriction [34, 35]. Foster and colleagues [34] recently showed that an 800-kcal/d liquid diet produced weight losses in 12 weeks that were only slightly (and insignificantly) smaller than those produced by a 420-kcal/d diet (Figure 1). This finding raises the possibility that the success of VLCDs is attributable more to the form in which these diets are served than to their marked caloric restriction. The use of portion- and calorie-controlled servings, as with liquid diets, is likely to facilitate excellent adherence and, thus, large weight losses. By contrast, numerous studies have shown that obese patients, when asked to consume a conventional diet, underestimate their caloric intake by as much as 40% [36]. Thus, when prescribed a 1200-kcal/d diet of conventional foods, patients may in fact consume closer to 1700 kcal/d. It would be interesting to compare weight losses resulting from a 1000-kcal/d liquid diet with those from a 1000-kcal/d diet of conventional foods.

View larger version (18K): [in this window] [in a new window]   Figure 1. Weight loss among 68 women randomly assigned to three dietary conditions. Patients consumed liquid diets providing either 420 kcal/d, 660 kcal/d, or 800 kcal/d during weeks 2 to 13, after consuming a 1200 kcal/d diet of conventional foods the first week. All consumed a refeeding diet during weeks 14 to 19, and a 1200 to 1500 kcal/d diet for the rest of the treatment. There were no significant differences among conditions in weight losses at weeks 13 or 26. Figure reprinted from reference 34, with permission.).

Long-Term Weight Losses

Few controlled studies have examined the long-term results of VLCDs. Table 2 shows the findings of four randomized trials in which patients were followed for a minimum of 1 year after treatment [37–40]. Studies by Sikand and colleagues [38], Wadden and associates [39], and Wing and coworkers [40] all showed that patients regained large amounts of weight in the first 1 or 2 years after treatment. Andersen and colleagues [41] obtained similar results in a randomized trial that compared the effectiveness of a VLCD with that of gastroplasty. (The results of that study were reported in graphic form only, making determination of the exact weight losses difficult.)

Wadden and coworkers [39] showed that a combination of VLCD and behavior modification significantly improved the maintenance of weight loss at the end of 1 year compared with treatment by VLCD alone. Patients in both dietary conditions, however, as well as those treated by a 1200-kcal/d diet combined with behavior modification, regained to their baseline weights within 5 years (Figure 2). The discrepancy between the favorable findings from the Japanese study led by Miura [37] and the results of the American studies is difficult to explain. Miura and colleagues, however, did confirm the benefits of combining a VLCD with behavior modification compared with treatment by VLCD alone.

View larger version (13K): [in this window] [in a new window]   Figure 2. Changes in weight at the end of treatment and during a 5-year follow-up period. Seventy-six women were randomly assigned to treatment by very-low-calorie diet alone (VLCD), behavior therapy combined with a 1200 kcal/d diet of conventional foods (BMOD), or VLCD combined with behavior therapy (COMBINED). Patients in the VLCD condition received a 1200 kcal/d diet the first month, a VLCD during months 2 to 3, and a refeeding diet month 4. Patients in the COMBINED condition consumed the same diet the first 4 months and a 1200 kcal/d diet during months 5 and 6. Those in the BMOD condition consumed a 1200 kcal/d diet during 6 months of treatment. Patients in the COMBINED condition lost significantly more weight at the end of treatment than did those in the two other conditions and significantly more at the 1-year follow-up than did patients in the VLCD condition. There were no statistically significant differences among the three conditions in changes in weight at 3 or 5 years follow-up. (Figure prepared from data reported in reference 39.).

Improving Long-Term Results

Little research has focused on methods to improve the maintenance of weight loss after VLCD, particularly of the kind conducted by Perri and associates [19] with moderate caloric restriction. The most promising findings have been reported by Pavlou and coworkers [42], who found that patients who received a VLCD in combination with an intensive, supervised program of aerobic exercise achieved excellent maintenance of weight loss 3 years after treatment. Participants in this study, however, were mildly obese male police officers who clearly are not representative of persons treated by VLCD in research trials.

A report by Finer and associates [43] similarly suggested that pharmacotherapy might facilitate maintenance of weight loss after VLCD. After patients lost a mean of 13.9 kg during 8 weeks using a 330-kcal/d diet, they were randomly assigned to either dexfenfluramine (a serotonin reuptake inhibitor) or placebo, both in combination with a conventional reducing diet. Patients receiving dexfenfluramine lost an additional 5.9 kg during the next 18 weeks, whereas those receiving placebo gained 3.0 kg. These favorable findings, however, stand in contrast to those of Andersen and colleagues [44], who treated patients for 1 year with dexfenfluramine combined with a VLCD (for the first 4 months) and later with a conventional reducing diet. Patients began regaining weight after 6 months of treatment, despite continued use of the drug, and their weight losses at 1 year were not significantly greater than those of patients who received placebo. Further research is needed to resolve the discrepancy between these two studies and to assess the effectiveness of other maintenance strategies after VLCDs.

Conclusions

Investigators have made significant progress in inducing weight loss in the 35 years since Stunkard's review [1]. Research trials of moderate caloric restriction combined with behavior modification indicate that 80% of patients will remain in treatment for 20 weeks and that approximately 50% will achieve a weight loss of 9.1 kg (20 lbs) or more, compared with only 12% of Stunkard and McLaren-Hume's patients [2]. Approximately 90% of persons treated using a VLCD and behavior modification will attain a loss of this magnitude, and 50% will attain a loss of 18.2 kg (40 lb) or more [34], compared with only 1% for Stunkard and McLaren-Hume [2].

Consistent with Stunkard's observation in the 1950s, however, most obese patients treated in research trials still regain their lost weight. Thus, the long-term results reported here for both moderate and severe caloric restriction confirm charges that "diets don't work" (if this statement is taken to mean that a 15- to 25-week program of diet and behavior modification is not associated with successful weight control 3 to 5 years later). This conclusion, however, is not surprising in view of the treatment of other chronic disorders. Few practitioners would expect to achieve long-term control of hypertension or type II diabetes with only 15 to 25 weeks of treatment. Similarly, a 6-month program of aerobic activity would not be expected to confer cardiovascular fitness 3 to 5 years later. Continued efforts are required in all cases.

Greater attention must be devoted to research on the maintenance of weight loss and, in time, is likely to yield long-term results that are far superior to those summarized in Table s1 and 2. Simultaneously, carefully controlled studies are needed to determine whether cycles of weight loss and regain pose greater threats to health than maintaining an obese but perhaps more stable weight [45]. Stated differently, investigators need to determine whether it is better to have lost (weight) and regained or to have never lost at all. Only by virtue of such studies will practitioners be able to make truly informed decisions concerning the most appropriate use of both moderate and severe caloric restriction.