Psychosocial bias in the diagnosis of obesity.

This study explores demographic and psychosocial variables associated with the assignment of the diagnosis of obesity in a family medicine residency model practice. Three groups of adult patients seen during 1978 were studied: a random sample of active patients, patients diagnosed as obese during 1978, and those never diagnosed as obese. While the prevalence of true obesity (greater than 20 percent above ideal body weight) was similar for men (58 percent) and women (47 percent), more women were diagnosed (222 women vs 87 men) and were more likely to be diagnosed within a year of entering the practice (42 percent women vs 10 percent men). Diagnosed obese women were older, had more psychological problems, and visited the practice more often than nonobese women. Diagnosed obese men were older, more frequently had psychological problems, visited the practice more often, and were more likely to be married than nonobese men. Undiagnosed obese men, however, had fewer psychological problems than nonobese men. The results suggest that physician education should address problems with diagnostic labeling and that researchers should anticipate subtle selection biases in retrospective studies when sampling methods depend on diagnosis.