A time-series analysis of longitudinal weight changes in two adult women

This study was conducted to examine individual longitudinal serial trends in body weight of two moderately obese adult women. Self-measured daily weights (over 5000 data points for subject 1 and over 3500 for subject 2) were averaged for each month and analyzed using a time-series analysis approach. The best statistical model for both women indicated a significant upward trend (gain 30 and 22 lb, respectively) interrupted by linear decreases (P less than 0.05) during weight loss programs over the 16- and 10-year observation periods, respectively. The women participated in six and eight weight control programs, respectively, and consistently lost weight, averaging 9.17 +/- 1.85 lb and 8.00 +/- 3.57 lb. Both subjects regained weight after dieting, in a period approximately 2.25 times as long as the diet. Numerous weight fluctuations occurred and one subject lost and then regained 10 percent of her body weight during two 5-year periods. Results are examined in terms of a possible weight increase protection benefit, as well as the deleterious medical potential that may occur with numerous weight loss attempts. Suggestions for future research are provided regarding weight cycling, seasonal variations, set point theory, aging effects, and the use of self-directed weight loss initiatives in conjunction with formal weight reduction programs. This study may serve as a prototype for exploring weight fluctuation using time-series analyses and for further investigating the etiology and clinical management of obesity.