In Reply The intent of our comments was not to dismiss projection analyses but rather to point out that it may be difficult to extrapolate accurately from past data to future trends in obesity. Changes in obesity prevalence over time have not followed a readily predictable pattern. Projections can vary according to the starting time point chosen and the statistical model used. Joinpoint analyses can identify past changes in slope but provide little guidance to predict future changes in slope.
Katherine M. Flegal, Cynthia L. Ogden. Use of Projection Analyses and Obesity Trends—Reply. JAMA. 2016;316(12):1317. doi:10.1001/jama.2016.11965