Predictors of weight change in middle-aged and old men

Studies on weight change and mortality have yielded inconclusive results. This 10-year prospective study was undertaken to improve understanding of factors affecting weight change.

The subjects were 1,143 men, aged 36 to 88 years (mean, 53.3 years) at entry. A questionnaire was filled in at entry and at the end of the follow-up with queries on weight, height, weight at the age of 20, physician-diagnosed diseases, smoking, alcohol use, dietary habits, leisure physical activity, occupation, present occupational activity, living conditions (living alone or cohabiting), and former athletic status. Further information on morbidity was obtained from selected national registers. Factors predicting weight change during the study were identified by stepwise linear multiple regression analysis.

The mean 10-year weight change was 0.8 (range, -29 to +24) kg. Age at entry (beta-coefficient, -0.17, SE 0.02), weight at entry (beta, -0.03, SE 0.01), diabetes at entry (beta, -3.55, SE 1.02), diabetes diagnosed after entry (beta, -3.94, SE 0.96), malignant cancer (beta, -1.60, SE 0.70), being a smoker (beta, - 1.59, SE 0.48), and increased physical activity (beta, - 1.27, SE 0.54) were significantly (p < 0.05) associated with weight loss in the final model. The model explained 13% of the variance of weight change.

The results emphasize the complexity of weight change. Some factors associated with weight change are apparently negatively, and some positively, associated with health. This could explain the equivocal findings on weight change and mortality in the literature.