A quantitative method for estimating and comparing the duration of human satiety responses: statistical modeling and application to liquid meal replacers.

Current methods of analyzing appetite-related self-report data do not allow for representation or statistical comparison of results in terms of common units or response durations. Using data from 13 previous studies, we assessed the suitability of several alternative approaches (interpolation, linear regression, non-linear models) for quantitatively estimating and comparing time to return to baseline pre-prandial levels (TTRTB, min). Curve modeling using the Weibull distribution gave the best fit and ability to determine mean TTRTB values with 95% confidence intervals. We then applied this in a study comparing liquid meal replacers (MR, 190 kcal) to 3 'meals' of similar weight and equal or greater energy content (yogurt, 190 kcal; bagel with cream cheese and juice, 400 kcal; hamburger with bun and soft drink, 400 kcal). While area under the curve data did not significantly differ amongst these, TTRTB was significantly longer for MR than yogurt. When corrected for energy content, TTRTB (min/kcal) was greater for MR than all other treatments. While further method development and validation are needed, the Weibull modeling procedure appears most suitable for estimating and quantitatively comparing durations of appetite-related responses to foods, providing an absolute response measure that can be expressed in commonly understood units.