Grouping participants for tailored work-site weight loss programs: a signal detection approach.

Abstract

Although work-site obesity prevention programs have been widely promoted, they have a high rate of attrition and participants find it very difficult to maintain the decreased weight. It is necessary to develop effective work-site programs that match the type of intervention to the participants and offer the necessary support. To this end, higher-order interaction of the causal factors of obesity needs to be analyzed. The subjects were male, white-collar workers (20-64 years of age), in Osaka, Japan. Since conventional methods, such as regression analysis or analysis of variance, cannot deal with the interaction of many variables, signal detection analysis by Kraemer was used to identify the higher-order interaction of multiple predictors of obesity. Out of 15 independent variables, a higher-order interaction consisting of 8 significant variables was identified. Consequently, the subjects were categorized into nine subgroups. It was revealed that the obesity of two groups of workers, 40 or more years old with a high degree of obesity, had different causes: one was related to working conditions, and one was related to smoking cessation. For the other terminal groups, further factors related to obesity were revealed. Although the applicability of the findings is limited, the methodology using signal detection analysis might be applicable to other weight loss programs as a way of facilitating the matching of the type of intervention and the target group.