Foster GD, Wadden TA, Phelan S, Sarwer DB, Sanderson RS. Obese Patients' Perceptions of Treatment Outcomes and the Factors That Influence Them. Arch Intern Med. 2001;161(17):2133-2139. doi:10.1001/archinte.161.17.2133
Copyright 2001 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2001
Despite considerable professional consensus that modest weight losses of 5% to 10% are successful for reducing the comorbid conditions associated with obesity, obese patients often desire weight losses 2 to 3 times greater than this. Examining ways to reduce the disparities between treatment expectations and subsequent outcomes, this study evaluated the role of physical characteristics, treatment setting, and mood in patients' evaluations of treatment outcomes.
This study was conducted in a university outpatient weight loss clinic with a sample of 397 obese individuals seeking weight loss by a variety of modalities. Before treatment, participants' heights and weights were measured, and the Beck Depression Inventory and the Goals and Relative Weight Questionnaire were administered.
Outcome evaluations ranged from 64.4 ± 11.1 kg (mean ± SD) for dream weight to 90.1 ± 19.1 kg for disappointed weight. Initial body weight was the strongest predictor of disappointed, acceptable, and happy weights (β = .90, .76, and .57, respectively). Sex (β = −.37) and height (β = .37) were the strongest determinants of dream weight. Heavier participants chose higher absolute weights, but the weight loss required to reach each of the outcomes was greater for heavier than for lighter patients.
These data signal a therapeutic dilemma in which the amount of weight loss produced by the best behavioral and/or pharmacologic treatments is viewed as even less than disappointing. Patients with the highest pretreatment weights are likely to have the most unrealistic expectations for success.
OBESE PATIENTS and their physicians are often at odds about what constitutes a successful treatment outcome. While there is considerable professional consensus that modest weight losses of 5% to 10% are successful,1- 3 obese patients seeking treatment view things quite differently. Studies suggest that the desired weight losses of obese patients are 2 to 3 times greater (a 22%-34% reduction in body weight) than those recommended by professionals as feasible and health promoting.4- 6 Moreover, these expectations greatly exceed the average 10% reductions in body weight achieved by the best behavioral and/or pharmacologic treatments.7,8 Population studies have reported similar discrepancies between health provider and lay definitions of desirable weights.9 The discrepancy between desired and actual weight losses can result in unrealistic and negative evaluations of treatment outcomes. For example, obese women seeking treatment characterized a 25% weight loss as "one I would not be happy with" and a 17% weight loss as "one that I could not view as successful in any way."5 Thus, the average 10% weight losses produced by the best available nonsurgical treatments are perceived as even less than unsuccessful.
Based on findings from the broader literature on goal setting, when goals remain out of reach and progress toward them is unsatisfying, people experience negative affect,10,11 aversive self-focus,12,13 and impaired performance,14 which often lead to abandonment of their goals.15 Decreased disparity between actual and expected outcomes appears to lessen the negative affect associated with unmet expectations.16,17 This cycle parallels our clinical experience, in which patients dissatisfied with their end-of-treatment (usually after 6 months) weight loss engage in self-critical statements, set unrealistic behavioral goals that are unattainable, experience "failure," and abandon all weight-control efforts. Among obese patients, the less the discrepancy between posttreatment weights and the weights described (before treatment) as "acceptable," the greater the satisfaction with posttreatment weight.5
One approach to decreasing the discrepancy between expected and actual outcomes is to produce larger weight losses. Unfortunately, efforts to increase the magnitude of weight loss, including very-low-calorie diets,18 structured exercise and food-provision programs,19,20 and longer treatment programs,21,22 produce long-term results that are no different than those of standard behavioral methods. An alternative approach to decreasing the disparity between expected and actual outcomes is to alter patients' pretreatment expectations about what constitutes a successful outcome. Efforts to modify patients' outcome evaluations have been hampered by a paucity of data about patients' views and the factors that influence them. In our initial study of this topic, we reported that outcome evaluations were related to body mass index (BMI), body image, and, to a lesser degree, self-esteem.5 These findings, however, came from a selected sample of 60 women enrolled in a research study. Less is known about outcome evaluations in more heterogeneous samples of obese treatment seekers.
To increase our understanding about outcome expectations and the factors that influence them, we have collected data about outcome evaluations from nearly 400 participants in research and clinical programs. Specifically, we evaluated the role of physical characteristics (weight, height, sex, and race), treatment approach (ie, research participants, patients seeking outpatient treatment, and patients seeking surgical treatment for obesity), and mood in outcome evaluations.
Participants were 397 obese individuals who sought weight loss by a variety of modalities offered at the University of Pennsylvania, Philadelphia. Participants were recruited from 3 different samples. The first consisted of 154 women who were evaluated prior to participating in a clinical trial of behavioral weight loss treatments. The second sample consisted of 193 participants (157 women and 36 men) who sought outpatient weight loss treatment in our fee-for-service clinic. Participants in this sample were evaluated on a consecutive basis from January 1996 to November 1998 and given a $25 reduction in their initial assessment fee for completing the measures described below.
The third sample consisted of 50 participants (43 women and 7 men) who presented for a psychosocial assessment prior to undergoing surgical treatment for obesity. Participants in this surgical sample were evaluated on a consecutive basis from August 1997 to April 1999 after an initial consultation with a surgeon. All prospective surgical patients at our institution receive this psychosocial assessment. All participants completed questionnaires as part of their assessment packet. Participants gave written and informed consent as approved by the University of Pennsylvania's institutional review board.
The 397 participants had a mean ± SD age of 43.1 ± 10.9 years, weight of 109.0 ± 28.9 kg, and BMI (calculated as weight in kilograms divided by the square of height in meters) of 39.3 ± 9.5. Most (88.9%) were women, 77.9% were white, 21.3% were African American, and 0.8% were Hispanic. The majority (56.3%) of participants were married, 29.7% were single, 9.5% were divorced, 2.6% were separated, and 1.6% were widowed.
Participants completed these measures before treatment.
Weight was measured on an electronic digital scale (model 5600; Detecto, Webb City, Mo) and height by a stadiometer. Participants were dressed in indoor clothing without shoes.
Various weight loss outcomes were evaluated using the Goals and Relative Weight Questionnaire (GRWQ).5 The GRWQ asks participants to numerically define 4 different weight loss outcomes, as described in Table 1. Participants assign a numerical equivalent (in pounds) to each of these weights. One-week test-retest reliability is extremely high for definitions of happy, acceptable, and disappointed weights (r>0.96 for all) but less so (r = 0.64; P<.001) for dream weight.5
Mood was assessed by the Beck Depression Inventory (BDI),23 a 21-item scale with a range of possible scores from 0 to 63. Higher scores indicate greater dysphoria.
Data are presented as mean ± SD. Paired t tests, adjusted for multiple comparisons, assessed differences among the 4 outcome evaluations using the SPSS software (version 9.0; SPSS Inc, Chicago, Ill). Simultaneous regression analyses (using the SPSS software) were used to identify the principal determinants of each of the 4 outcome evaluations (ie, dream, happy, acceptable, and disappointed weights). For each outcome, the following variables were entered simultaneously: weight, height, age, race, sex, treatment setting (ie, research, clinic, surgery), and mood (ie, BDI score).
Outcome evaluations of the 397 participants ranged from 64.4 ± 11.1 kg (38.4% ± 12.5%) for dream weight to 90.1 ± 19.1 kg (15.7% ± 9.9%) for disappointed weight (Table 1). Each weight was significantly different from the remaining 3 (P<.001 for all). A weight that, on average, would require a 19-kg (15.7% ± 9.9%) weight loss was considered "disappointing," while a 29-kg (24.9% ± 11.0%) weight loss was considered "acceptable."
Simultaneous regression analyses revealed that initial body weight was the strongest predictor of disappointed, acceptable, and happy weights (β = .90, .76, and .57, respectively [Table 2]). Thus, heavier participants chose higher absolute weights. Nevertheless, the weight loss required to reach each of the outcomes was greater for heavier than for lighter patients. For example, participants in the lowest tertile of mean ± SD body weight (82.6 ± 9.1 kg) required a weight loss of 14.2 ± 5.7 kg (16.8% ± 6.0%) to achieve an acceptable weight (68.5 ± 7.0 kg), while those in the highest tertile of body weight (140.0 ± 26.8 kg) required a weight loss of 48.6 ± 22.3 kg (33.6% ± 11.1%) to achieve an acceptable weight (91.5 ± 16.1 kg; P<.001). Similar results were obtained for disappointed, happy, and dream weights (P<.001 for all).
Table 3 underscores the effect of initial weight on outcome evaluations in more clinical terms. Participants were categorized as grade I, II, or III based on the clinical guidelines of the National Heart, Lung, and Blood Institute1 (see Table 3 for BMI levels for each grade). Across all 4 outcomes, grade III participants chose higher absolute weights. However, the percentage weight loss required to achieve these outcomes was 11% to 18% greater for grade III participants than for grade I participants. Thus, grade III patients are likely to have the most unrealistic expectations for weight loss when presenting for treatment.
Sex (β = −.37) and height (β = .37) were the strongest determinants of dream weight, with initial body weight making a modest (β = .26) but significant contribution (Table 2). Thus, taller participants and men chose higher absolute dream weights. Sex (β = −.27) and height (β = .24) also made modest but significant contributions to the explanation of variance in happy weight. In general, the mean ± SD outcome evaluations for men were higher than those for women (111.7 ± 22.9 vs 87.4 ± 16.8 kg for disappointed, 99.5 ± 17.9 vs 77.1 ± 12.1 kg for acceptable, 94.3 ± 14.1 vs 70.2 ± 9.6 kg for happy, and 85.5 ± 11.7 vs 61.8 ± 7.9 kg for dream). Among the remaining predictors, mood was the only variable that contributed significantly to all 4 outcomes, although its effect was weak (β≤.15). Other predictors (eg, race, age) contributed minimally (β≤.12 for all), if at all, to the explanation of the outcomes (Table 2).
Surgery participants (ie, patients who were screened for surgery, including those who did not undergo surgical procedures) chose significantly higher values than research and clinic participants for disappointed, acceptable, and happy weights (unadjusted values in Table 4). However, as suggested by the regression analysis (Table 2), these differences were largely a result of the demographic differences among the samples (Table 4). In general, surgery patients had higher BMIs and were slightly younger than research and clinic patients. Although there were no differences among the groups in the number of African Americans, the research sample had a lower percentage of white participants. By design, the research sample was exclusively female. Thus, after controlling for weight, height, age, race, and sex, the effects of treatment setting disappeared except for a very modest (β = .12) effect on disappointed weight, with surgery participants choosing significantly lower values than research and clinic participants.
There are several principal findings from this study. The first is that, on average, the outcome evaluations in this large, heterogeneous sample of obese treatment seekers were quite similar to those previously reported among a small sample of female research participants.5 The weight loss percentages for each of the 4 outcome evaluations were within 1.3% of the values reported previously. This suggests that unrealistic expectations are not exclusive to female research participants and are likely to be representative of obese patients seeking a variety of obesity treatments.
These data reflect a therapeutic dilemma in which the weight loss produced by the best behavioral and/or pharmacologic treatments is viewed by patients, prior to treatment, as even less than disappointing. The 10% weight loss associated with both behavioral7 and pharmacologic8 treatments is nearly 6% less than the weight loss (15.6%) that participants described before treatment as "could not be viewed as successful in any way" and less than half of the weight loss (24.5%) that was characterized as one "that I would not be particularly happy with . . . but could accept."
It is interesting to note that, among the subgroup of female research participants in this sample (n = 154), a weight that represented a 12% weight loss was considered disappointing, compared with a 17% weight loss considered disappointing in our previous but very similar sample.5 Similarly, the weight losses associated with acceptable, happy, and dream weights were approximately 4% lower among female research participants in the present sample than in the previous sample. Specifically, values for acceptable, happy, and dream weights were 21.2%, 27.0%, and 34.8%, respectively, in the present study compared with 25.1%, 31.4%, and 38.0%, respectively, in the previous study. The reason for this change is unknown, but it may be because the benefits of modest weight loss are receiving more attention in the lay press or because of our larger (154 vs 60) sample. Despite this shift, it is troubling that patients "could not view as successful in any way" an outcome (12% weight loss) that approximates what can be expected. This significant disparity between expected and actual outcomes is consistent with our clinical experience—patients are reluctant to concentrate on maintaining a weight they perceive as far from even acceptable.
The second principal finding was that initial weight was a strong predictor of happy, acceptable, and disappointed outcomes. Despite choosing higher absolute weights, heavier participants' selections resulted in weight losses (expressed as a percentage of initial weight) that were greater than those of lighter patients. For example, participants classified as having grade III obesity perceived a 33% weight loss as acceptable, while grade I participants viewed an 18% weight loss as acceptable. Grade III participants desired nearly twice the weight loss for each of the 4 outcomes. While it is encouraging that heavier patients are apparently realistic in choosing higher absolute outcomes than lighter patients, it is troubling that heavier patients desire larger weight losses, even after controlling for differences in initial weight. Heavier patients may have a stronger biological predisposition to obesity and are almost certain to be characterized by adipocyte hyperplasia, each of which may place limits on weight loss.24,25 Such patients (ie, grades II and III) should be targeted for efforts to modify outcome evaluations, including education about the biological limits of weight loss, the medical benefits of modest weight loss, and strategies to improve body image26 and enhance weight-independent self-esteem.27,28
The explanatory power of initial weight was lower for dream weight than for the other outcome weights and was mostly influenced by sex and height. The impact of sex on definitions of ideal weight has also been reported in population studies.9 The fact that men and taller participants chose higher outcome evaluations suggests that "ideal" versions of body weight may still be based on the Metropolitan Life height and weight tables, which assigned an ideal weight based on sex and height.29 By contrast, the current clinical assessment of relative weight is the BMI, which is independent of sex. Despite its effect on dream and happy weights, it was surprising that sex had no effect on the selection of disappointed weight and had only a slight effect (β = .09) for acceptable weight. A similar but more modest effect was observed for race in the selection of happy and dream weights, while race had no effect on disappointed and acceptable weights. This suggests that when participants select weights that more closely approximate achievable outcomes (ie, acceptable and disappointed), sex and race are not predictive. However, outcomes that are further removed from reality (ie, dream, happy) appear to be influenced by race and sex. It is possible that any race and sex effects were attenuated in a treatment-seeking sample.
As might be expected, treatment setting influenced outcome evaluations, with surgery patients selecting weights that represented larger percentage reductions in weight for all 4 outcomes, despite choosing higher absolute weights (Table 4). This appears to be realistic since surgical treatment provides 2 to 3 times the weight loss associated with more conservative treatments.30 However, surgical participants "could not view as successful in any way" a typical surgical outcome (27% weight loss). A 38% or 57-kg reduction in body weight, clearly beyond the typical surgical outcome, was considered only acceptable. These data are similar to those reported by Rabner and Greenstein,31 who found that, preoperatively, 70% of gastric surgery patients expected to lose 46 to 48 kg (neither initial weight nor mean expected weight loss was reported). It is important to note that the effects of treatment approach were largely a result of demographic differences among the samples, particularly in weight. Controlling for these differences, disappointment in weight was the only outcome affected by treatment approach and the effect was very modest (β = −.12). Clinically, surgical patients are likely to desire larger percentage reductions in weight than nonsurgical patients, but the difference is principally a result of their increased weight.
Mood as assessed by the BDI was the only variable except initial weight to be significantly related to all 4 outcome evaluations, although the effect was modest (β = −.08 to −.15). Thus, the higher the level of dysphoria, the lower the outcome evaluation. This suggests that efforts to enhance mood before treatment may alter outcome evaluations.
Clinically, the replication of unrealistic and negative outcome evaluations in varied treatment approaches suggests that there continues to be a significant disparity between expected and actual outcomes. Previous data suggested that the discrepancy between actual weight loss and various outcomes (assessed pretreatment) was strongly related (r = −0.52 to −0.75) to posttreatment satisfaction. Thus, patients will likely end treatment dissatisfied with their weight. The relationship between satisfaction with weight loss and maintenance of weight loss has not directly been tested. One study4 found no relationship between reaching goal weight and maintenance of weight loss, but no ratings of satisfaction were obtained, and very few patients (17%) reached their goal weight. The belief that satisfaction (the discrepancy between expected and actual outcomes) affects subsequent success is based on the broader literature on goal setting14,15,32 and our own clinical experience. Thus, this belief needs to be empirically validated in a prospective, longitudinal fashion.
Despite a large, heterogeneous sample, this study has several limitations. First, the findings are limited to obese treatment seekers and provide no information on obese persons who do not seek treatment or who do so in settings that are not university based. It is possible that those who seek treatment at tertiary care centers expect more from treatment than others (ie, they are more unrealistic). Alternatively, those seeking treatment at specialized centers typically have made multiple previous attempts to lose weight,33 perhaps making them more realistic. Second, although our sample included men, the number was less than 50, making any conclusions about sex effects tentative. The disproportionate number of women in this sample was a result of the greater prevalence of women among obese treatment seekers and the exclusion of men from the research sample.
Finally, this study was limited to the examination of determinants that, except for mood, are largely unmodifiable (height, race, sex, etc). Additional research is needed to identify other unmodifiable (eg, family history of obesity, weight history) as well as modifiable (eg, attitudes toward exercise, frequency of self-weighing) variables that are related to patients' outcome evaluations. Qualitative research methods (eg, ethnographic studies) may be particularly useful in this regard. While the data in the present study can help identify patients who are most likely to have unrealistic expectations, more research is needed to determine how these expectations may be modified to ones that are achievable (ie, 10% of initial body weight). This will likely include a 2-step process in which modifiable determinants of outcomes are identified (eg, body image, self-esteem, attributions about previous weight loss history) and treatments are developed to alter them. Future research should also focus on how expectations may vary across different treatment approaches (eg, commercial weight-loss programs, pharmacologic approaches, very-low-calorie diets), particularly those that are not university based. Finally, as noted above, it is important to examine the proposed relationships between outcome evaluations and subsequent weight change, given the frequency with which weight is regained after obesity treatment.24,25
Accepted for publication February 22, 2001.
This work was supported in part by grant DK56114 from the National Institutes of Health, Bethesda, Md.
Corresponding author and reprints: Gary D. Foster, PhD, University of Pennsylvania Department of Psychiatry, 3535 Market St, Suite 3027, Philadelphia, PA 19104 (e-mail: firstname.lastname@example.org).