Weight Change in Control Groups in Weight-Loss TrialsFriday, April 1, 2016
Readers may recall that the very large Look AHEAD trial, which attempted to show a cardiovascular benefit of weight loss in individuals with type 2 diabetes, ultimately had to be abandoned due to “futility”.
One of the reasons (among others) that this trial did not show any difference between the intervention (weight loss) and the control group, was that the control group, to everyone’s surprise, lost a substantial amount of weight over the course of the trial.
This unexpected weight loss in the control group essentially diminished any possibility of finding a difference between the groups, as the weight curves of the intervention and control groups converged over the course of the trial.
The weight loss in the control group was certainly not what anyone had expected.
In fact, the general assumption in weight-management trials is that weight in control groups (with minimal intervention) will either not change or even increase over time.
It turns out, as pointed out in a paper by David Jones and colleagues published in the recent “statistics edition” of OBESITY, that this assumption is not exactly true.
Rather, their analysis of weight trajectories in participants in almost 30 weight-loss trials, who were randomised to the “control” group, generally receiving minimal interventions (if any), shows that these individuals may lose on average about 1 kg over 12 months (with minimal variation depending on the intensity of the “control” intervention).
Several factors can account for these findings including selection of the population (people wanting to lose weight – which is why they would join a weight-loss trial in the first place) and the fact that even simply weighing and discussing the aim of the study (as during the consent process) may motivate individuals to change their behaviours.
Thus, the authors conclude that uncontrolled evaluations of weight loss programs should assume that, in the absence of intervention, their population would weigh up to a kilogram on average less than baseline at the end of the first year of follow-up.
This, must be considered in any statistical analyses of such studies.