Reductions in Internal Disinhibition During Weight Loss Predict Better Weight Loss Maintenance

Abstract

The construct of disinhibition, as measured by the Eating Inventory, was recently found to have two factors: internal disinhibition (eating in response to cognitive and emotional cues) and external disinhibition (eating in response to environmental cues). This study examined whether early changes in disinhibition that occurred during a weight loss program predicted later weight loss maintenance. Participants were adults enrolled in a weight loss treatment study (n = 81, 16% men, BMI = 38.4 ± 6.5 kg/m²). Two-thirds of participants were African Americans. Participants received a uniform, meal-replacement-based weight loss program in months 1–3 and one of four nutritionally focused programs for weight loss maintenance in months 4–12. Disinhibition and weight were assessed at clinic visits. Change in internal disinhibition from months 1–3 (i.e., the weight loss period) significantly predicted change in weight from month 4 to 12 (i.e., the weight maintenance period); this remained significant when treatment group, age, gender, ethnicity, baseline weight, baseline depression, baseline internal disinhibition, and initial weight loss were controlled for (P = 0.03). A comparable analysis examining change in external disinhibition found that it was not a significant predictor of weight maintenance (P = 0.43). Participants who experienced the biggest decreases in internal disinhibition during the initial phase of treatment had the most success maintaining their weight loss in the next phase of treatment. Long-term weight loss outcomes may be improved by spending sufficient treatment time teaching strategies for reducing eating in response to internal cues.

INTRODUCTION

Weight loss maintenance is a challenge for nearly all adults who diet (1). Identifying factors associated with greater long-term success is a high priority for obesity research. One tool used in obesity treatment research is the Eating Inventory (EI), a self-report questionnaire that measures three factors: disinhibition, restraint, and hunger (2). Disinhibition has been of particular interest for predicting treatment outcomes, although findings for its predictive ability are mixed (3–5). Exploratory and confirmatory factor analyses conducted recently by Niemeier et al. (6) found that the construct of disinhibition, as measured by the EI, has two factors: internal disinhibition (i.e., eating in response to cognitive and emotional cues) and external disinhibition (i.e., eating in response to environmental cues). Niemeier et al. examined these factors as predictors of weight change in two samples. In the first sample, internal and external disinhibition were measured at pretreatment for participants enrolled in a weight loss trial. Lower levels of internal disinhibition predicted more weight loss at 6 months (P = 0.03) and marginally predicted more weight loss at 18 months (P = 0.06). In the second sample, internal and external disinhibition were measured at entry to the National Weight Control Registry, a database of individuals who upon enrollment have successfully maintained a weight loss for at least 1 year (7). Lower levels of internal disinhibition at entry to the National Weight Control Registry, predicted less weight regain 1 year after entry (P = 0.001). In both samples, external disinhibition was not predictive of weight change.

The purpose of this study was to extend this emergent line of research and determine if changes in internal or external disinhibition that occur in the initial phase of weight loss treatment are predictive of subsequent weight loss maintenance. If a relationship between these variables is demonstrated, it would provide further support for the use of the internal or external disinhibition subfactors of the EI in this research area and highlight these constructs for consideration when refining obesity treatment.

METHODS AND PROCEDURES

Participants were drawn from a sample of adults enrolled in a weight loss study. In the first phase of the study (months 1–3), the same weight loss program was delivered to all participants through weekly, individual, 15-min telephone calls with a clinician. The clinicians taught participants basic behavioral weight control skills and encouraged them to follow a low-calorie diet and consume two liquid meal replacement products each day. Topics covered included using stimulus control to structure the food environment, responding to food cravings, problem solving barriers to physical activity, and obtaining social support for behavior change. At the beginning of the second phase of the study (months 4–12), participants were randomized to follow one of four weight loss maintenance diets: (i) standard calorie control, (ii) reduced energy density eating, (iii) continued use of one meal replacement per day, or (iv) a combination of reduced energy density eating and continued use of meal replacements. In months 4–12, participants had 20 telephone contacts with a clinician, who emphasized continued use of behavioral weight control skills and encouraged participants to adhere to the prescribed weight loss maintenance diet.

Although 238 participants enrolled in the treatment study, attrition at 1 year was high (66%), likely because all participants were recruited from primary care clinics and enrolled at their physician’s prompting; no run-in period of clinic visits or food monitoring was required. Eighty-one participants completed 1 year of treatment and were included in this study. Participants in this study (i.e., the treatment completers) were compared to participants who dropped out (n = 157). Completers had significantly lower baseline BMI than dropouts (38.4 ± 6.5 kg/m² vs. 40.2 ± 6.5 kg/m², P = 0.02). They did not differ at baseline on sex, ethnicity, age, depression, internal disinhibition, or external disinhibition (all P values >0.05).

Assessments were conducted at baseline, the end of the weight loss phase (i.e., month 3), and the end of the weight loss maintenance phase (i.e., month 12). Participants were paid $25 for completing the month 12 assessment. Weight was measured with a digital Seca scale. Internal and external disinhibition were measured with the EI (2) and scored according to the factors’ structure identified by Niemeier et al. (6), whereby internal and external disinhibition are comprised of eight and six items, respectively. True items are scored 1 and false items are scored 0 (with the exception of one item that is reverse scored) and scores are summed, so that higher scores indicate greater disinhibition. For one item, participants use a 1–4 scale, where 1 = rarely, 2 = sometimes, 3 = usually, and 4 = always, and these ratings are re-coded so that a rating of 1 or 2 is scored as 0 and a rating of 3 or 4 is scored as 1. Depression was measured with the Center for Epidemiological Studies Depression Scale (8).

Bivariate linear regressions and simultaneous multiple regression analyses were used to determine if internal or external disinhibition predicted weight change at month 12. Where noted, the following covariates were entered in the regression equation: treatment group; initial weight loss; and baseline levels of weight, depression, and internal or external disinhibition.

RESULTS

At baseline, participants’ BMI averaged 38.4 ± 6.5 kg/m². Sixteen percent of participants were men and two-thirds of participants were African Americans. Mean scores for internal and external disinhibition were 3.0 ± 2.2 and 2.5 ± 1.8, respectively, at baseline and 2.3 ± 1.9 and 1.9 ± 1.7, respectively, at month 3 (i.e., disinhibition generally decreased during treatment). Weight loss from baseline to month 3 averaged 6.1 ± 3.3 kg. Weight loss from baseline to month 12 averaged 5.7 ± 6.1 kg.

Change in total disinhibition during months 1–3 significantly predicted change in weight during months 4–12 (r = 0.28, P = 0.018), so that decreases in disinhibition were associated with decreases in weight. Change in internal disinhibition from month 1 to 3 significantly predicted change in weight from month 4 to 12 (r = 0.36, P = 0.002), such that those who experienced the greatest decreases in internal disinhibition lost the most weight. This finding remained significant when treatment group, baseline weight, baseline depression, baseline internal disinhibition, and initial weight loss were controlled (β = 0.32, t = 2.17, P = 0.03). Change in external disinhibition was not a significant predictor of subsequent weight change when examined in a bivariate regression (r = 0.04, P = 0.71) or when treatment group, baseline weight, baseline depression, baseline external disinhibition, and initial weight loss were controlled for (β = 0.11, t = 0.80, P = 0.43).

The relationship between change in internal and external disinhibition was examined. Change in these variables in months 1–3 was significantly correlated (r = 0.23, P = 0.01). Change in these variables in months 4–12 was not correlated, nor was change in one variable in months 1–3 predictive of change in the other during months 4–12 (all P values >0.05).

DISCUSSION

This study examined whether changes in two, newly identified subfactors of the EI disinhibition scale were predictive of weight change in participants who completed 1 year of weight loss treatment. Participants who experienced the biggest decreases in internal disinhibition during the initial phase of treatment had the most success maintaining their weight loss in the next phase of treatment. Changes in external disinhibition did not predict weight change. Although the small sample size must be considered in interpreting that null finding, the effect size was minimal, indicating that when changes in total disinhibition are predictive of weight change, those findings may be carried by changes in internal disinhibition. Although initial change in internal and external disinhibition is correlated, only 5% of the variance is shared. These results extend those reported by Niemeier et al., in part because the participants in this study differed in several ways from the Niemeier et al. samples: these participants had higher baseline BMIs (approximately one-third had class 3 obesity), the majority were African Americans, and they received telephone-based, individual weight management.

These findings provide further support for the notion that internal and external disinhibition, as measured by the EI, have differential ability to predict weight change. The high attrition rate from the derivation sample is an important limitation of this study, because it is possible that participants who dropped out of treatment were experiencing differential change in disinhibition. Future research should use experimental designs to examine whether reductions in internal disinhibition during treatment may predict long-term outcome. If susceptibility to internal disinhibition is causally related to weight regain, long-term obesity treatment outcomes might be improved by spending more intervention time teaching strategies for reducing eating in response to internal cues.