Consistent with results of a review of randomized controlled trials for weight loss interventions , we observed greater improvements in the weight-related measure of HRQOL than in two generic measures in a one-year weight loss trial. Few weight loss studies have used both obesity-specific and generic measures to assess HRQOL outcomes within the same study. Kaukua and colleagues reported stronger correlations between weight change and HRQOL change for the OP Scale, an obesity-specific measure, (r = 0.635) than for SF-36 physical functioning (r = -0.502) .
Results of the two generic measures included in this trial were inconsistent with each other, and the subscales of the SF-36 were variable across domains. For the EQ-5D, weight loss corresponded more closely to HRQOL changes than did weight gain. Weight change (both loss and gain) seemed to correspond closely for PCS scores of the SF-36. Improvements in PCS were greater for the 10+% weight loss category than the 5–9.9% category; weight gain was associated with small reductions in PCS that were further reduced by additional weight gain. For the SF-36 subscales, weight loss was associated with improvement on some domains, but deterioration or no change on others. Variability with respect to changes in HRQOL, both within and across weight loss studies, has been reported previously . Physical Functioning and PCS were most responsive to weight loss in the current study, a finding also reported in some of the weight loss trials reviewed (e.g. [7, 21]). Social Functioning, Role Emotional, Mental Health, and MCS showed poor correspondence with weight change.
Few studies have explored the effects of weight gain on HRQOL. Engel and colleagues  found that changes in weight-related HRQOL for participants in a weight loss trial were similar in degree, but opposite in direction for weight loss and weight gain. That is, weight loss was associated with improved HRQOL and weight gain was associated with reduced HRQOL, and these changes occurred in a linear fashion. Among the individuals who gained 5% or more of their weight in our study, scores on Mental Health, MCS, and Vitality showed the greatest deterioration. Unlike the Engel et al. study, we found improved weight-related HRQOL for the group that gained 0–4.9% of their weight and only a slight decrement for the group that gained 5+% of their weight. Because no generic measures of HRQOL were used in the Engel et al. study, we cannot compare that part of our results to theirs. A prospective cohort study of 40,098 women participating in the Nurses' Health Study  found that women who had gained 5 pounds or more over the course of four years reported significant impairment in SF-36 Physical Functioning, Vitality, and Bodily Pain. In the present study these three domains of the SF-36, as well as several others, exhibited impairments associated with weight gain. As more of the world's population is gaining weight, a potential fruitful focus of future investigations is the effect of weight gain on HRQOL (the current literature focuses on effects of weight loss and cross-sectional differences among BMI groups).
Results of this study support the potential value of assessing HRQOL changes in weight loss trials with more than one instrument. In the best of all worlds we recommend the use of an assessment battery, the approach taken in the Swedish Obese Subjects studies , since each instrument contributes somewhat different information. However, we recognize the practical limitations of this approach in most clinical trials with respect to cost and respondent burden.
We believe our findings have direct relevance for weight loss patients and clinicians/researchers who work with this population. Especially if replicated in other studies of different weight loss interventions, we can use these results to inform patients and clinicians/researchers of what HRQOL changes they can expect to experience with varying amounts of weight loss or weight gain. For example, based on the current findings as well as previous findings  we can say with some certainty that weight loss of at least 5% is most likely to have a positive effect on weight-related physical function and self-esteem, as well as cardiovascular risk factors . Knowledge of this information may serve to keep patients motivated, which as clinicians and patients are well aware, is frequently a challenge. On the other hand, knowledge of the likely adverse effects on HRQOL with increased weight may serve to reinforce the importance of weight maintenance. We know from previous research  that weight regain is associated with reduced weight-related HRQOL.
One of the strengths of this study is that we were able to compare HRQOL changes on three different measures of HRQOL. Although the SF-36 has been widely used in weight loss studies [6, 23] and the EQ-5D has been studied with respect to BMI [24, 25], this study is unique in its comparison of three measures of HRQOL outcomes in weight loss patients. Other strengths of this study include the large sample size (n = 926) and the longitudinal design with one-year follow-up assessment on all HRQOL measures. However, this study is not without limitations. The current sample was predominantly female (84%) and Caucasian (79%), limiting generalizability to other patient populations. Additionally, only a subset of the sample experienced what might be considered clinically meaningful weight loss. Only 30% of the sample lost at least 5% of their baseline weight and only 12% lost at least 10% of their baseline weight. This limits our ability to make inferences about HRQOL changes as a result of more substantial weight loss. On the other hand, having large variability in weight change, including subjects who gained weight, should increase the study's external validity. Another limitation of this study is that only 56% of the trial participants completed the one-year protocol; it is unknown in what ways attrition may limit the generalizability of our findings. In addition, it is unknown whether the results we observed would generalize to other weight loss methods and other placebo-controlled trials for different pharmacological agents. Finally, it would be an over simplification to suggest that changes in HRQOL depend solely on amount of weight change. Health care providers and clinical researchers who treat obese individuals recognize that changes in HRQOL could be influenced by a variety of variables not explored in the current study, such as initial weight loss expectations, satisfaction with weight loss results and the treatment program, self-esteem and other psychological variables, as well as comorbid health. We lack the data to address the potential role of these other variables.