Predictors of Obesity
Rena R. Wing, Ph.D.
The Miriam Hospital/Brown Medical School
This presentation will highlight some of the methodological issues in the study of weight gain and obesity and attempt to integrate findings from the papers presented into the proposed model.
This topic is clearly complex. First, there are different ways to conceptualize the "weight outcomes." We can discuss development of obesity per se, weight gain, or weight regain. Much of the literature, and several of the papers presented focus on weight loss. It remains unclear whether findings from studies of weight loss can be extrapolated to the other areas.
Second, based on the Law of Thermodynamics, weight changes must be related to changes in energy balance, and consequently to changes in dietary intake or energy expenditure (physical activity; sedentary behavior). However, in many studies, this has been difficult to show. For example, in the weight loss literature, weight loss and/or weight regain is often not significantly associated with changes in physical activity or intake. The lack of association may reflect a variety of methodological shortcomings, including inaccuracies in self-report, lack of precision in the measures, or the time-window covered by the measures.
In addition, much of the literature does not assess predictors, but rather concurrent changes. Changes in weight over a year are examined in relation to changes in diet/activity over that same time period. In these analyses, changes in activity may lead to weight gain or the reverse. More frequent measures of weight and the behaviors are needed to address the temporal sequence.
It also remains unclear whether associations seen in young adults can be extrapolated to older individuals and whether there will be gender and/or ethnic differences in the associations.
Finally, I note that the behavioral variables studied (overall intake and activity) are composite variables and while the composite is what influences weight, it is going to be important to study the more specific behaviors that influence the composite. For example, what are the specific behaviors that lead to high caloric intake? Are there important differences related to eating patterns (number of meals/day; eating breakfast; or specific food choices (soda consumption or amount of dietary variety).
Having raised these methodological issues, I will attempt to summarize what has been presented by showing how the various research areas (epidemiological studies of weight change; treatment studies of weight loss) are each examining different aspects of the proposed ecological model. Clearly more integration of these approaches and a broader view of the problem is needed.