Labes.fmh.utl.pt

International Journal of Obesity (2004) 28, 1124–1133 & 2004 Nature Publishing Group All rights reserved 0307-0565/04 $30.00 Pretreatment predictors of attrition and successfulweight management in women PJ Teixeira1*, SB Going3, LB Houtkooper3, EC Cussler2, LL Metcalfe2, RM Blew2, LB Sardinha1 andTG Lohman2 1Department of Exercise and Health, Faculty of Human Movement, Technical University of Lisbon, Lisbon, Portugal; 2Department of Physiology, Body Composition Research Laboratory, The University of Arizona, USA; and 3Department ofNutritional Sciences, Body Composition Research Laboratory, The University of Arizona, USA OBJECTIVE: This study analyzed baseline behavioral and psychosocial differences between successful and nonsuccessfulparticipants in a behavioral weight management program. Success was defined by commonly used health-related criteria (5%weight loss). Noncompletion was also used as a marker of a failed attempt at weight control.
SUBJECTS: A total of 158 healthy overweight and obese women (age, 48.074.5 y; BMI, 31.073.8 kg/m2; body fat,44.575.3%).
INTERVENTION: Subjects participated in a 16-week lifestyle weight loss program consisting of group-based behavior therapy toimprove diet and increase physical activity, and were followed for 1 y after treatment.
METHODS: At baseline, all women completed a comprehensive behavioral and psychosocial battery assessing dieting/weighthistory, dietary intake and eating behaviors, exercise, self-efficacy, outcome evaluations, body image, and other variablesconsidered relevant for weight management. Participants who maintained a weight loss of 5% or more at 16 months (or 10% ormore of initial fat mass) were classified as successful. Nonsuccessful participants were those who dropped out and completerswho had not lost weight at follow-up.
RESULTS: Of all participants, 30% (n ¼ 47) did not complete initial treatment and/or missed follow-up assessments(noncompleters). Noncompletion was independently associated with more previous weight loss attempts, poorer quality of life,more stringent weight outcome evaluations, and lower reported carbohydrate intake at baseline. In logistic regression,completion status was predicted correctly in 84% of all cases (w2 ¼ 45.5, Po0.001), using baseline information only. Additionalpredictors of attrition were initial weight, exercise minutes, fiber intake, binge eating, psychological health, and body image. Alarge variation in weight loss/maintenance results was observed (range: 37.2 kg for 16-month weight change). Independentbaseline predictors of success at 16 months were more moderate weight outcome evaluations, lower level of previous dieting,higher exercise self-efficacy, and smaller waist-to-hip ratio. Success status at follow-up was predicted correctly in 74% of allstarting cases (w2 ¼ 33.6, Po0.001).
CONCLUSION: Psychosocial and behavioral variables (eg, dieting history, dietary intake, outcome evaluations, exercise self-efficacy, and quality of life) may be useful as pretreatment predictors of success level and/or attrition in previously overweightand mildly obese women who volunteer for behavioral weight control programs. These factors can be used in developingreadiness profiles for weight management, a potentially important tool to address the issue of low success/completion rates inthe current management of obesity.
International Journal of Obesity (2004) 28, 1124–1133. doi:10.1038/sj.ijo.0802727Published online 20 July 2004 Keywords: overweight; correlates; psychosocial; weight loss IntroductionCurrent behavioral interventions to help overweight andobese individuals reduce and manage their weight are onlymodestly successful. Generally, the longer the follow-up *Correspondence: Dr PJ Teixeira, Faculdade de Motricidade Humana, period after treatment, the greater the number of people who Estrada da Costa, Cruz Quebrada, Lisboa 1495-688, Portugal.
regain weight close to pretreatment levels.1 Moderate to large levels of attrition are also frequent, compromising Received 10 November 2003; revised 21 April 2004; accepted 11 May2004; published online 20 July 2004 internal and external validity of many published studies.2 Predictors of success in weight lossPJ Teixeira et al However, why some people succeed at adopting and management intervention. It consisted of weekly group sustaining behaviors associated with weight control while meetings for 16 weeks followed by random assignment to others, undergoing similar treatment programs, do not, ongoing online contact or no contact, for an additional year.
remains largely unknown.3 Personal factors (biological, No differences in weight change between 4 and 16 months psychological, behavioral), especially if they contribute to (the follow-up period) were observed between the online and resistance to long-lasting change, should play a role in no contact groups and data were pooled for this analysis. All determining higher vs lower success rates, along with participants agreed to refrain from participating in any other treatment characteristics, socioecological environmental weight loss program for the duration of the study.
among the three.
Previous behavioral programs have shown pretreatment variables such as initial body weight,4 a history of repeated Subjects met with the intervention team once a week in diet attempts,5 eating self-efficacy,6 psychopathology,7 and groups of approximately 25 subjects, for 150 min per session.
body image8 to be significantly associated with subsequent They were encouraged to make progressive changes in their weight loss. However, with few exceptions (eg eating self- lifestyle (eating habits and physical activity), leading to a efficacy), the number of studies available to determine the moderate daily energy deficit (less 1260–2090 kJ/day (300– true association of these and other factors with success is 500 kcal/day)). A weight loss of about 0.5 kg a week was limited to only a few, and research methodologies (eg targeted and individualized goals for energy intake and psychosocial assessments) have varied substantially. In expenditure were provided to all subjects. Cognitive and addition, other variables potentially relevant for the process behavioral strategies used to improve compliance included of weight control have been insufficiently unexplored.
regular self-monitoring, self-efficacy enhancement, cogni- Suggestive evidence indicates that general, self-regulatory tive restructuring, relapse prevention and problem-solving efficacy,9 an autonomous orientation,10 weight-specific skills, stress management, preventing emotional eating, and quality of life,11 and exercise-related variables,11 assessed social support. Further details on the design and methods of before treatment, may also predict weight change.
this study are available elsewhere.11 Building on a previous report on predictors of short-term weight outcomes,11 the objective of the present study was toidentify baseline correlates of 16-month weight loss in previously overweight and obese women who participated Pretreatment assessments included weight and body compo- in a 4-month behavioral weight management program.
sition, exercise, dietary intake, and psychosocial variables.
Baseline differences between successful and nonsuccessful Body composition was assessed by dual energy X-ray women were analyzed, based on categories of success derived absorptiometry (DXA, Lunar DPX-IQ, software version 4.6).
a priori from health-related guidelines.12 A large psychosocial Minutes per day and energy expenditure (kJ/day) from battery was examined, including several variables that have leisure-time moderate and vigorous physical activities (ie not been previously researched in this context. As attrition is activities with METs 43.8, examples of which were provided a hallmark of nonsuccessful weight loss attempts, we report to participants during the assessment interview) were additionally on pretreatment differences between comple- estimated with the 7-Day Physical Activity Recall13 and ters and noncompleters.
dietary intake was assessed with 3-day food records averagedfrom 1 weekend and 2 week days.
Dieting/weight history was assessed by a questionnaire developed specifically for this study and weight outcome evaluations were assessed with the Goals and Relative Weights Questionnaire,14 which asks subjects to indicate Participants were recruited from the Tucson, Arizona area their ‘dream' weight, and weight values that they would be through newspaper and TV advertisements. Subjects were ‘happy' with, they would consider ‘acceptable', and that required to be between 40 and 55 y of age, have a BMI they would be ‘disappointed' with, at the end of the between 25.0 and 38.0 kg/m2, be a nonsmoker, and be free program. To account for initial values, these variables are from major illnesses to be eligible. The University of expressed as percent of initial weight (thus, the lower the Arizona's Human Subjects Protection, Institutional Review percent value, the more stringent the attitude about each Board approved the study and all participants gave written weight outcome and ‘dream' weight). To assess quality of informed consent prior to participation. Previously, we life, general (SF-36) and obesity-specific (Impact of Weight reported on predictors of short-term weight loss in a subset on Quality of Life-lite) measures were used,15,16 and general of women participating in this research trial.11 The present social support was assessed by five items originally developed study, exclusively focused on long-term weight outcomes, for the Medical Outcomes Study (MOS).17 Depressive reports on all of the 158 women who participated in the symptoms were measured with the Beck Depression Inven- research study and who started the behavioral weight tory.18 Self-esteem was assessed with the Rosenberg's Self- International Journal of Obesity Predictors of success in weight loss PJ Teixeira et al esteem/Self-concept questionnaire19 and self-motivation was where all dropped subjects were assumed to have returned to measured with the Self-Motivation Inventory.20 The Binge baseline weight by 16 months (BOCF), and (ii) a modified Eating Scale,21 the Eating Self-Efficacy Scale,22 and the Eating version of the last observation carried forward method, Inventory23 were used to measure variables related to eating where the last measured weight was used as the final weight behavior; the dimensions of cognitive restraint, eating with 0.2 kg added per each month passed since the last lab disinhibition, and perceived hunger were derived from the assessment (LOCF þ ). The value of 0.2 kg was the approx- Eating Inventory.
imate average monthly weight regain during follow-up in a Self-efficacy for exercise was assessed with the Self-Efficacy recent review of similar studies.30 Although other imputa- for Exercise Behaviors Scale,24 measuring beliefs that a tion methods are generally preferable,31 sufficient weight person can ‘stick with' an exercise program for at least 6 information over time was not available to derive adequately months under varying circumstances. The questionnaire has missing data points. Also, since psychosocial characteristics two subscales of ‘resisting relapse' and ‘making time' with were being used as predictors in primary analyses, they could five items each. The average of all items was used to produce not be employed in multivariate imputation models to the exercise self-efficacy score (Crombach's a ¼ 0.84). Ex- derive missing weight data. The BOCF model, in particular, is ercise perceived barriers were assessed with items from the conservative and offers acceptable protection against type I Exercise Perceived Barriers scale.25 A two-item ‘obstacles' subscale from the original instrument was not included in Two categories of success were defined, based on 16- analyses due to very low internal consistency (a ¼ 0.05); month weight data. Successful participants (n ¼ 53) were items from the two remaining dimensions of ‘time' (three those who lost 5% or more of their initial body weight,12 an items) and ‘effort' (six items) were used to produce the two outcome that generally represents a lower threshold for subscale scores and averaged to calculate the total exercise important physical health benefits.33,34 Nonsuccessful parti- barriers score (a ¼ 0.73). Exercise social support was measured cipants (n ¼ 71) were defined as those showing no weight with 10 items from a scale developed to assess participation/ loss or weight gain (ie weight change Z0.5 kg) at 16 involvement from family and friends with regard to one's months (see Figure 1). The BOCF procedure was used to exercise, the Exercise Social Support questionnaire.26 From define success categories, thus classifying all dropouts as the original scale, three questions loading on a ‘rewards/ nonsuccessful (no change from baseline). All other partici- punishment' dimension were not included in the final pants (n ¼ 34) were excluded from analyses comparing the analysis due to very low item–scale correlations and low two success categories. This procedure was chosen to identify internal consistency (a ¼ 0.29). The average of the 10 two clearly distinct levels of success and minimize misclassi- remaining items, loading on a single dimension (‘participa- fication caused by using a single cutoff separating the two tion/involvement'), was used to calculate the global exercise success levels.
social support composite score (a ¼ 0.87).
Several subjects with weight losses smaller than 5% of Body image was assessed with the Body Shape Question- initial weight displayed disproportionately larger fat losses, naire, which measures concerns with body shape and which were offset by noticeable increases in fat-free mass. To ‘feeling fat',27 and also with the silhouette-based Body Image adjust for body composition changes, an additional success Assessment Questionnaire (self-ideal difference),28 and the classification was created, where in addition to subjects Body Cathexis Questionnaire, which assesses feelings to- losing 5% of initial body weight, subjects losing 10% or more wards various body parts or characteristics.29 Internal of their initial fat mass were also categorized as successful, consistency for all global scales and subscales varied between regardless of weight change. In the absence of specific 0.72 (exercise self-efficacy, making time subscale) and 0.97 health-related criteria for body fat loss, this was an arbitrary (body cathexis).
cutoff; however, it has face validity considering loss of bodyfat and not lean is considered health beneficial. In all, 10additional subjects were considered as successful (total Attrition and success categories n ¼ 63) using this procedure.
In all, 47 participants (29.7%) did not complete assessmentsat 16 months (noncompleters). Of these, 22 women droppedout during the initial 4-month treatment phase and the Statistical analysis remaining could not be reached for follow-up assessments at Analyses were completed using the Statistical Package for the 16 months. The most prevalent reported reasons for drop- Social Sciences (SPSS, version 11.5). Spearman's, rank-order ping out during treatment were lack of time (35%), correlation was used for psychometric variables since a dissatisfaction with the program/staff (22%), personal life majority displayed a significantly skewed distribution.
issues (17%), and health limitations (17%). Weight changes Independent sample t-tests and analysis of covariance were analyzed for completers only and additionally by two (ANCOVA) were used to compare baseline data for com- procedures employed to include baseline data for all starting pleters vs noncompleters and to compare subjects in subjects, following an intent-to-treat model. These proce- different success categories. Logistic regression with back- dures were: (i) the baseline observation carried forward, ward stepwise selection was used to predict group classifica- International Journal of Obesity Predictors of success in weight lossPJ Teixeira et al No wt loss/dropout Weight change (% of initial) Figure 1 Individual percent weight changes at 16 months for 158 starting women.
tion at 16 months based on pretreatment variables. We less active, and reported consuming fewer calories and lower selected a P-value of 0.157 and 0.05 for removal of amounts of dietary carbohydrates and fiber. In addition, they predictors35 and similar final models were obtained. We had dieted significantly more often in the previous year and chose those models which maximized overall classification their weight had fluctuated more times during adulthood, scores while also resulting in the most parsimonious events compared to completers. Noncompleters also reported high- per variable ratio (an event defined as the number of positive er levels of binge eating at baseline. Completers displayed or negative cases in the outcome variable, whichever is more accepting evaluations with regard to weight loss lower, that is, 43 and 57, respectively, for noncompletion including a higher accepting dream weight, and had more and weight loss success); 10 or more events per independent positive scores for quality of life, psychological health, and variable, corresponding in our case to a maximum of four to body image. When analyses were replicated adjusting for five predictors per model, is considered adequate.35 baseline weight or BMI (ANCOVA), results were virtuallyunchanged, even for variables typically correlated withweight such as obesity-specific quality of life, weight out-come evaluations, and body image.
Few variables predicted weight change when only com- After the 4-month intervention program, average weight loss pleters were used in the analysis. In contrast, when data for was 5.1 kg (6.2% of initial weight) in the 136 completing dropouts were included (intent-to-treat models), several subjects. At 16 months, completing participants (n ¼ 111) pretreatment measures were associated with subsequent lost an average of 4.6 kg (5.5%), while mean weight change weight loss. They included an abdominal fat distribution, for all 158 participants was 3.2 kg (3.9%) using BOCF and more frequent previous dieting, more stringent weight loss 3.0 kg (3.7%) using LOCF þ. As shown in Figure 1, a large evaluations, and exercise perceived barriers (negative pre- variability in individual weight change was observed with a dictors), as well as weight-specific quality of life, self- range for completers at 37.2 kg, thereby providing adequate motivation, eating and exercise self-efficacy, and a better database to study individual differences in predictors of body image (positive predictors).
weight change. Average weight change for the successful Using the success criterion based exclusively on body group (n ¼ 53) was 9.5 kg (11.5%), while nonsuccessful weight (and BOCF), several significant differences were participants (n ¼ 71) gained 0.8 kg ( þ 1.0%) using BOCF and found between groups, most of which are depicted in gained 1.3 kg ( þ 1.5%) using LOCF þ. For subjects in the Figure 2. In addition, at baseline, subjects who went on to body composition-adjusted success category (n ¼ 63), mean lose and sustain 5% or more of their initial body weight also weight change was 8.4 kg (10.2%).
reported more minutes of exercise (P ¼ 0.05), higher total Table 1 shows differences between completers and non- fiber consumption (P ¼ 0.027), and a poorer body image as completers at baseline for body habitus, behavioral, and indicated by a greater self-ideal difference in the Body Image psychosocial characteristics. Noncompleters were heavier, Assessment Questionnaire (P ¼ 0.026). When similar com- International Journal of Obesity Predictors of success in weight loss PJ Teixeira et al Baseline differences between completers and noncompleters and correlation between baseline measures and changes in weight T-test comparison Correlation with weight change Waist-to-hip ratio Exercise (kJ/day) Exercise (min/day) Total energy (kJ/kg/day) Carbohydrates (g/kg/day) Protein (g/kg/day) Weight/diet history Number of diets in previous year Years at current weight Life frequency of weight74.5 kg (10 lb) Age when started dieting Weight outcome evaluations ‘Dream' weight (% of initial) ‘Disappointing' weight loss (%) ‘Acceptable' weight loss (%) ‘Happy' weight loss (%) Quality of life and social support Weight-specific quality of life Eating self-efficacya Cognitive restraint Exercise perceived barriers Exercise self-efficacy Exercise social support Body shape concerns Body size dissatisfaction *Po0.05, **Po0.01, ***Po0.001. LOCF+, last observation carried forward with 0.2 kg/month added to the last measured weight (see text for details); BOCF,baseline observation carried forward (ie zero change from baseline for noncompleters). Higher scores indicate higher value for characteristic tested (eg higher qualityof life, higher perceived hunger, more body concerns, etc). Since weight change was coded as baseline weight subtracted to 16-month weight, weight loss isrepresented by a negative weight change (thus, a negative correlation coefficient indicates a positive correlation with weight loss). aOwing to a procedural error, datafor the Eating Self-efficacy Scale could only be collected from 111 participants (71 completers); for this questionnaire, higher scores indicate lower self-efficacy.
bHigher scores indicate more positive feelings with regard to one's body.
International Journal of Obesity Predictors of success in weight lossPJ Teixeira et al Waist-to-hip ratio Dietary carbohydrate (g/kg/d) Number of diets in previous year Nonsuccessful Successful Nonsuccessful Successful Nonsuccessful Successful "Dream" weight (% of initial) "Happy" weight (% of initial) Weight-specific quality of lifea Nonsuccessful Successful Nonsuccessful Successful Nonsuccessful Successful Exercise perceived barriers Exercise self-efficacy Nonsuccessful Successful Nonsuccessful Successful Nonsuccessful Successful Figure 2 Significant baseline differences between successful and nonsuccessful participants. Successful participants (n ¼ 53) completed 16-month assessments andlost 5% of more of initial weight; nonsuccessful subjects (n ¼ 71) dropped out or completed the study without weight loss. aTotal score from the IWQOL-liteQuestionnaire. bA higher score indicates more positive feelings towards one's body.
parisons were performed using the success category adjusted (B ¼ 0.23, P ¼ 0.012), ‘happy' weight loss evaluations for changes in body composition, the same variables emerged as significant.
P ¼ 0.004). This model was highly sensitive to program Table 2 shows baseline differences for the different completion (of 111 completers, only four were wrongly completion/success groups for various dimensions within predicted to dropout) but 20 (45%) of the 44 eventual the IWQOL-lite (weight-specific quality of life), exercise self- efficacy, and exercise perceived barriers questionnaires (w2 ¼ 45.5, Po0.001). For weight loss success, the final model between completers and noncompleters and successful included the waist-to-hip ratio (B ¼ 7.77, P ¼ 0.024), num- ‘losers' vs nonsuccessful ‘losers'.
ber of previous diets (B ¼ 0.29, P ¼ 0.012), ‘happy' weight Logistic regression was used to predict program comple- loss evaluations (B ¼ 0.12, P ¼ 0.002), and exercise self- tion vs noncompletion (n ¼ 158) and also to predict success efficacy (B ¼ 1.00, P ¼ 0.012). Sensitivity for (correct) classi- at 16 months (n ¼ 124). The best model for program fication into the weight loss successful category was 68%, while specificity (percent of participants wrongly predicted as nonsuccessful) was 11% (w2 ¼ 33.6, Po0.001). Overall International Journal of Obesity Predictors of success in weight loss PJ Teixeira et al Baseline differences between completion and success groups for body image, distinguished between completers and non- quality of life and exercise dimensions A review of pretreatment predictors of completion in previous obesity treatment studies reveals mixed results. For example, attrition was positively36–38 and negatively39associated with the number of previous diet attempts, positively40 and negatively41 associated with binge eating, and positively42 and not associated43,44 with initial weight/ Quality of life (IWQOL-lite) BMI level. Having higher initial weight loss expectations36,38 Physical functioning and reporting higher emotional disturbance39 also predicted noncompliance. Baseline depression was associated with poorer adherence, as measured by the number of sessions attended.45 Using comparable methodologies as in thepresent report, three previous studies evaluated prediction ExerciseEffort (perceived barriers) of eventual completion status using multivariate models; Time (perceived barriers) these studies correctly classified subjects as completers or Making time (self-efficacy) noncompleters in 55,39 62,36 and 68–75% of cases,38 using Resisting relapse (self-efficacy) *Po0.05, **Po0.01, ***Po0.001 on t-test comparison.
Unique to this study is the association of pretreatment quality of life measures with long-term outcomes. Scores forthe obesity-specific IWQOL-lite and to a lesser extent for the percent of subjects correctly classified was 84.2 and 74.0%, more general SF-36 quality of life questionnaires were respectively, for completion and 16-month weight manage- positively associated with study completion. Most dimen- ment success.
sions (subscales) of the IWQOL-lite were also related withweight management success in bivariate analyses. Weight-specific quality of life, binge eating, depression, and body image (all significantly associated with study completion) This study analyzed the association of several baseline were strongly intercorrelated in our sample and it is possible personal factors with weight management outcomes after that quality life may have served as a surrogate measure for 16 months. Success was defined as relative weight/body the other variables, with regard to their impact on program composition changes during the study's period. In addition, adherence. In fact, when quality of life was absent in the study completion was considered as a separate marker of multivariate model for completion, depression and binge successful participation. Strengths of this study include eating entered the model significantly, without much loss in success levels defined a priori from health-related criteria, its overall classification score (results not shown). The assessment of body composition changes, a sufficient follow- IWQOL-lite questionnaire evaluates several dimensions of up time to study weight loss and maintenance of weight lost, functioning with a high specificity to weight-related issues a large number and broad scope of baseline assessments, and and their perceived impact on well-being46 and may become the inclusion of all starting participants in statistical a useful tool in the future to help identify participants more analyses, in addition to presenting completers-only results.
likely to experience difficulties during treatment. We found Main results showed that more frequent previous dieting that self-worth and work-related scores were particularly attempts, more stringent weight loss outcome evaluations, a related to weight loss success and/or completion. In agree- lower weight-specific perceived quality of life, and lower ment with these findings, Bennett and Jones38 have shown exercise self-efficacy were associated with poorer long-term that the pretreatment score in an ‘interference' measure (‘How do you feel your weight affects your daily activities?') In the present study, differences at baseline were analyzed significantly predicted attrition, in a study very similar to the between eventual completers and noncompleters. In study- present report.
ing baseline determinants of weight loss success, completers- Excessively optimistic expectations are the norm in only analyses are particularly inadequate as participants individuals seeking obesity treatment,47 who typically place presenting with more barriers to success at the start of the great value on reaching goal weights.48 In the present study, program are likely to be those who dropout preferentially.
prospective weight outcome evaluations reported at baseline Accordingly, as we have shown before,11 substantially were significantly more stringent in eventual noncompleters different results emerge when analyzing predictors for than in completers, suggesting that more realistic expecta- completers only as opposed to using baseline data for all tions towards weight loss are an important, independent participants (see Table 1). We found that multiple baseline cognitive indicator of readiness to complete treatment. It has variables including initial weight, previous dieting, outcome been frequently argued that false hope and the desire to lose evaluations, quality of life, depression, binge eating, and more weight than what realistically can be expected may International Journal of Obesity Predictors of success in weight lossPJ Teixeira et al increase the likelihood of early disappointment and relapse long-term success is possible even after many previous failed upon smaller than wanted changes.49,50 The present and our attempts,58 it appears that reporting a large number of recent previous report11 are the first studies to have shown dieting attempts (eg 4 þ ) may foretell poor outcomes in empirical data to support this claim. Others recently volunteers for formal behavioral treatment programs.
reported that weight goals and dream weights were not Despite the existence of good motivation-based theoretical associated with weight loss, in a completers-only follow-up models for exercise intentions and behaviors,59 exercise- analysis of a 8-week treatment program (47% attrition at 18 related psychosocial correlates of weight loss have received months).51 It may be that having both positive and realistic little attention thus far. We have analyzed three exercise/ weight loss expectations is the most beneficial trait regarding physical activity variables as prospective predictors of weight long-term outcomes. However, as the present study clearly loss and study completion. Scores in these variables were not shows (see Table 1), completers-only and all sample (intent- associated with study completion. However, initial self- to-treat) analyses can produce substantially different results efficacy and perceived barriers to physical activity correlated and caution must be exercised when comparing studies with with weight loss in a consistent manner across analyses, even different analytical procedures regarding noncompleters.32 in completers-only analysis. The scales used for exercise- We noticed that noncompleters reported lower energy related psychosocial variables are relatively short, simple to intake and lower carbohydrate and fiber intake than interpret, and could be used more frequently in the context completers. Several factors may account for these differ- of obesity treatment. In fact, continued motivation to be ences, including lower energy requirements for the former physically active is perhaps the single most important factor group of women, who had a higher percent body fat (thus in long-term weight management and very recent evidence less relative amounts of more metabolically active fat-free shows that high levels of exercise, even when temporarily mass) and also reported increased levels of recent dieting, interrupted by lapses,60 can be adopted and sustained by a which could have induced temporary reductions in resting large number of participants, and are indeed predictive of metabolism and energy intake.52 More simply, noncompl- larger weight losses.61 More empirical research is needed to eters could also merely be dieting and in negative energy study attitudes and cognitions with regard to exercise and balance at a higher extent than completers, at study entry physical activity in the context of obesity treatment.
(which would agree with their most extensive dieting Results from this study should only be generalized to history). Compared to total energy expenditure estimates healthy, middle-aged, overweight or mildly obese women using an equation for overweight women derived from volunteering for formal weight loss treatment. Also, the large doubly labeled water studies,53 completers' energy intake number of statistical tests may have increased chances of was 3% lower than predicted, while noncompleters reported type I error in some of the more exploratory analyses (eg about 12% less calories ingested than would be estimated.
Table 1). Logistic regression analysis, which resulted in Thus, although the mean difference between the two groups parsimonious predictive models, and the fact that associa- is relatively small (616 kJ [147 kcal]) and was not statistically tions were generally observed in the expected direction significant (P ¼ 0.113, data not shown), the possibility exists provide some evidence that chance alone was not respon- that noncompleters underestimated their energy and carbo- sible for our primary findings. Nevertheless, the validity of hydrate intake slightly more than completers, which could the present prediction models needs to be confirmed in also help explain the results (assuming all participants were separate samples.
in energy balance at baseline).
In conclusion, pretreatment predictors of meeting estab- In the behavioral treatment of obesity, previous participa- lished goals for long-term weight management were identi- tion in weight loss programs and previous dieting attempts fied, including easily accessible variables such as dieting are among the most consistent predictors of smaller weight history, outcome evaluations, self-efficacy, and quality of losses, as shown in previous reviews54,55 and more recent life. These results largely confirm our earlier findings with reports.5,11,56 It is possible that volunteers for formal regard to short-term weight loss11 and contribute to the university-based weight loss programs represent a self- development of a research database of moderators of success selected group among the overweight/obese population, in weight management, such as individual readiness profiles, which may be more resistant to treatment and include which is slowly being established in the field.55,62,63 The many individuals who have repeatedly failed to control their understanding that obese patients respond to any program weight.57 A closer inspection of our data showed that among in a very heterogeneous manner reinforces the importance the 26 women reporting four or more diet attempts in the of continuing to assess, prior to treatment initiation, if previous year, 14 (54%) subjects did not complete the study, weight loss candidates are indeed ready for such task.64 only three subjects lost more than 5%, and only one lost10% or more of their initial weight at 16 months. Further,among 17 women initially reporting five or more recentattempts, only one was successful, that is, completed thestudy losing more than 5% of her initial weight. In sum, despite evidence in highly selected groups showing that This study was supported by NIH Grant DK57453.
International Journal of Obesity Predictors of success in weight loss PJ Teixeira et al 24 Sallis JF, Pinski MA, Grossman RB, Patterson TL, Nader PR. The development of self-efficacy scales for health-related diet and 1 Jeffery RW, Drewnowski A, Epstein LH, Stunkard AJ, Wilson GT, exercise behaviors. Health Educ Res 1988; 3: 283–292.
Wing RR, Hill DR. Long-term maintenance of weight loss: current 25 Steinhardt MA, Dishman RK. Reliability and validity of expected status. Health Psychol 2000; 19: 5–16.
outcomes and barriers for habitual physical activity. J Occup Med 2 Anderson JW, Konz EC, Frederich RC, Wood CL. Long-term 1989; 31: 536–546.
weight-loss maintenance: a meta-analysis of US studies. Am J ClinNutr 2001; 74: 579–584.
26 Sallis JF, Grossman RM, Pinski RB, Patterson TL, Nader PR. The 3 Wing RR. Behavioral interventions for obesity: recognizing development of scales to measure social support for diet and our progress and future challenges. Obes Res 2003; 11 (Suppl 1): exercise behaviors. Prev Med 1987; 16: 825–836.
27 Cooper PJ, Taylor MJ, Cooper Z, Fairburn CG. The development 4 Wing RR, Jeffery RW. Benefits of recruiting participants with and validation of a body shape questionnaire. Int J Eating Disord friends and increasing social support for weight loss and 1987; 6: 485–494.
maintenance. J Consult Clin Psychol 1999; 67: 132–138.
28 Williamson DA, Gleaves DH, Watkins PC, Schlundt DG. Valida- 5 Kiernan M, King AC, Kraemer HC, Stefanick ML, Killen JD.
tion of a self-ideal body size discrepancy as a measure of body size Characteristics of successful and unsuccessful dieters: an applica- dissatisfaction. J Psychol Behav Assess 1993; 15 (Suppl 1): 57–68.
tion of signal detection methodology. Ann Behav Med 1998; 20: 29 Secord S, Jourard P. The appraisal of body cathexis: Body cathexis and the self. J Consult Psych 1953; 17: 343–347.
6 Bernier M, Avard J. Self-efficacy, outcome and attrition in a 30 Wing RR. Behavioral weight control. In: Wadden TA, Stunkard AJ weight-reduction program. Cogn Ther Res 1986; 10: 319–338.
(eds) Handbook of obesity treatment. The Guilford Press: New York; 7 Beliard D, Kirschenbaum DS, Fitzgibbon ML. Evaluation of an 2002. pp. 301–316.
intensive weight control program using a priori criteria to 31 Gadbury GL, Coffey CS, Allison DB. Modern statistical methods determine outcome. Int J Obes Relat Metab Disord 1992; 16: 505– for handling missing repeated measurements in obesity trial data: beyond LOCF. Obes Rev 2003; 4: 175–184.
8 Traverso A, Ravera G, Lagattolla V, Testa S, Adami GF. Weight loss 32 Ware JH. Interpreting incomplete data in studies of diet and after dieting with behavioral modification for obesity: the weight loss. N Engl J Med 2003; 348: 2136–2137.
predicting efficiency of some psychometric data. Eat Weight 33 Goldstein DJ. Beneficial health effects of modest weight loss. Int J Disord 2000; 5: 102–107.
Obes Relat Metab Disord 1992; 16: 397–415.
9 Dennis KE, Goldberg AP. Weight control self-efficacy types and 34 Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin transitions affect weight-loss outcomes in obese women. Addict JM, Walker EA, Nathan DM. Reduction in the incidence of type 2 Behav 1996; 21: 103–116.
diabetes with lifestyle intervention or metformin. N Engl J Med 10 Williams GC, Grow VM, Freedman ZR, Ryan RM, Deci EL.
2002; 346: 393–403.
Motivational predictors of weight loss and weight-loss main- 35 Steyerberg EW, Eijkemans MJ, Harrell Jr FE, Habbema JD.
tenance. J Pers Soc Psychol 1996; 70: 115–126.
Prognostic modelling with logistic regression analysis: a compar- 11 Teixeira PJ, Going SB, Houtkooper LB, Cussler EC, Martin CJ, ison of selection and estimation methods in small data sets. Stat Metcalfe LL, Finkenthal NR, Blew RM, Sardinha LB, Lohman TG.
Med 2000; 19: 1059–1079.
Weight loss readiness in middle-aged women: psychosocial 36 Fowler JL, Follick MJ, Abrams DB, Rickard-Figueroa K. Participant predictors of success for behavioral weight reduction. J Behav characteristics as predictors of attrition in worksite weight loss.
Med 2002; 25: 499–523.
Addict Behav 1985; 10: 445–448.
12 Institute of Medicine. Weighing the Options: Criteria for Evaluating 37 Mitchell C, Stuart RB. Effect of self-efficacy on dropout from Weight-Management Programs. National Academy Press: Washing- obesity treatment. J Consult Clin Psychol 1984; 52: 1100–1101.
ton, DC; 1995.
38 Bennett GA, Jones SE. Dropping out of treatment for obesity.
13 Sallis JF, Haskell WL, Wood PD, Fortmann SP, Rogers T, Blair SN, J Psychosom Res 1986; 30: 567–573.
Paffenbarger Jr RS. Physical activity assessment methodology in 39 Yass-Reed EM, Barry NJ, Dacey CM. Examination of pretreatment the Five-City Project. Am J Epidemiol 1985; 121: 91–106.
predictors of attrition in a VLCD and behavior therapy weight- 14 Foster GD, Wadden TA, Vogt RA, Brewer G. What is a reasonable loss program. Addict Behav 1993; 18: 431–435.
weight loss? Patients' expectations and evaluations of obesity 40 Marcus MD, Wing RR, Hopkins J. Obese binge eaters: affect, treatment outcomes. J Consult Clin Psychol 1997; 65: 79–85.
cognitions, and response to behavioural weight control. J Consult 15 Ware J, Snow K, Kosinski M, Gandek B. SF-36 Health Survey Clin Psychol 1988; 56: 433–439.
Manual and Interpretation Guide. New England Medical Center, 41 Ho KS, Nichaman MZ, Taylor WC, Lee ES, Foreyt JP. Binge eating The Health Institute: Boston, MA; 1993.
disorder, retention, and dropout in an adult obesity program. Int J 16 Kolotkin RL, Crosby RD. Psychometric evaluation of the Impact Eat Disord 1995; 18: 291–294.
of Weight on Quality of Life-Lite Questionnaire (IWQOL-lite) in a 42 Clark MM, Guise BJ, Niaura RS. Obesity level and attrition: community sample. Qual Life Res 2002; 11: 157–171.
support for patient-treatment matching in obesity treatment.
17 Sherbourne CD, Stewart AL. The MOS social support survey. Soc Obes Res 1995; 3: 63–64.
Sci Med 1991; 32: 705–714.
43 Ek A, Andersson I, Barkeling B, Rossner S. Obesity treatment and 18 Beck AT, Steer RA. Manual for the Beck Depression Inventory.
attrition: no relationship to obesity level. Obes Res 1996; 4: 295– Psychological Corporation: New York; 1987.
19 Rosenberg M. Society and the Adolescent Self-image. Princeton 44 Honas JJ, Early JL, Frederickson DD, O'Brien MS. Predictors of University Press: Princeton, NJ; 1965.
attrition in a large clinic-based weight-loss program. Obes Res 20 Dishman RK, Ickes W. Self-motivation and adherence to 2003; 11: 888–894.
therapeutic exercise. J Behav Med 1981; 4: 421–438.
45 Clark MM, Niaura R, King TK, Pera V. Depression, smoking, 21 Gormally J, Black S, Daston S, Rardin D. The assessment of activity level, and health status: pretreatment predictors of binge eating severity among obese persons. Addict Behav 1982; 7: attrition in obesity treatment. Addict Behav 1996; 21: 509–513.
46 Kolotkin RL, Crosby RD, Kosloski KD, Williams GR. Development 22 Glynn SM, Ruderman AJ. The development and validation of an of a brief measure to assess quality of life in obesity. Obes Res eating self-efficacy scale. Cogn Ther Res 1986; 10: 403–420.
2001; 9: 102–111.
23 Stunkard AJ, Messick S. The three-factor eating questionnaire to 47 Foster GD, Wadden TA, Phelan S, Sarwer DB, Sanderson RS. Obese measure dietary restraint, disinhibition and hunger. J Psychosom patients' perceptions of treatment outcomes and the factors that Res 1985; 29: 71–83.
influence them. Arch Intern Med 2001; 161: 2133–2139.
International Journal of Obesity Predictors of success in weight lossPJ Teixeira et al 48 O'Neil PM, Smith CF, Foster GD, Anderson DA. The perceived Wadden TA, VanItallie TB (eds) Treatment of the Seriously Obese relative worth of reaching and maintaining goal weight. Int J Obes Patient. The Guilford Press: New York, NY; 1992. pp. 383–410.
Relat Metab Disord 2000; 24: 1069–1076.
56 Pasman WJ, Saris WH, Westerterp-Plantenga MS. Predictors of 49 Cooper Z, Fairburn CG. A new cognitive behavioural approach to weight maintenance. Obes Res 1999; 7: 43–50.
the treatment of obesity. Behav Res Ther 2001; 39: 499–511.
57 Brownell KD. Whether obesity should be treated. Health Psychol 50 Polivy J. The false hope syndrome: unrealistic expectations 1993; 12: 339–341.
of self-change. Int J Obes Relat Metab Disord 2001; 25 (Suppl 1): 58 McGuire MT, Wing RR, Hill JO. The prevalence of weight loss maintenance among American adults. Int J Obes Relat Metab 51 Linde JA, Jeffery RW, Finch EA, Ng DM, Rothman AJ. Are Disord 1999; 23: 1314–1319.
unrealistic weight loss goals associated with outcomes for over- 59 Vallerand RJ. A hierarchical model of intrinsic and extrinsic weight women? Obes Res 2004; 12: 569–576.
motivation in sport and exercise. In: Roberts GC (ed) Advances in 52 Weinsier RL, Nagy TR, Hunter GR, Darnell BE, Hensrud DD, Weiss Motivation is Sport and Exercise. Human Kinetics: Champaign, IL; HL. Do adaptive changes in metabolic rate favor weight regain in 2001. pp. 263–319.
weight-reduced individuals? An examination of the set-point 60 Wing RR, Jeffery RW. Prescribed ‘breaks' as a means to disrupt theory. Am J Clin Nutr 2000; 72: 1088–1094.
weight control efforts. Obes Res 2003; 11: 287–291.
53 Food and Nutrition Board, Institute of Medicine. Dietary Reference 61 Jeffery RW, Wing RR, Sherwood NE, Tate DF. Physical activity and Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, weight loss: does prescribing higher physical activity goals Protein, and Amino Acids (Macronutrients). National Academy Press: improve outcome? Am J Clin Nutr 2003; 78: 684–689.
Washington, DC; 2002.
62 Weiss AR. Characteristics of successful weight reducers: a brief 54 Foreyt JP, Goodrick GK. Prediction of weight management review of predictor variables. Addict Behav 1977; 2: 193–201.
outcome: Implications for practice. In: Allison DA, Pi-Sunyer FX 63 Klem ML, Wing RR, McGuire MT, Seagle HM, Hill JO. Psycholo- (eds) Obesity Treatment: Establishing Goals, Improving Outcomes, gical symptoms in individuals successful at long-term mainte- and Reviewing the Research Agenda. Plenum Press: New York; 1995.
nance of weight loss. Health Psychol 1998; 17: 336–345.
pp. 199–205.
64 USDHHS. The Practical Guide to the Identification, Evaluation, and 55 Wadden TA, Letizia KA. Predictors of attrition and weight loss in Treatment of Overweight and Obesity in Adults. NIHFNational patients treated by moderate to severe caloric restriction. In: Heart, Lung, and Blood Institute: Bethesda, MD; 2000.
International Journal of Obesity

Source: http://www.labes.fmh.utl.pt/obesity/publications/Teixeira%20predictors%20of%20attrition%20and%20wt%20loss%2016m%20IJO%202004.pdf