Data from: The statistical mechanics of human weight change

Over the past 35 years there has been a near doubling in the worldwide
prevalence of obesity. Body Mass Index (BMI) distributions in high-income
societies have increasingly shifted rightwards, corresponding to increases
in average BMI that are due to well-studied changes in the socioeconomic
environment. However, in addition to this shift, BMI distributions have
also shown marked changes in their particular shape over time, exhibiting
an ongoing right-skewed broadening that is not well understood. Here, we
compile and analyze the largest data set so far of year-over-year BMI
changes. The data confirm that, on average, heavy individuals become
lighter while light individuals become heavier year-over-year, and also
show that year-over-year BMI evolution is characterized by fluctuations
with a magnitude that is linearly proportional to BMI. We find that the
distribution of human BMIs is intrinsically dynamic—due to the short-term
variability of human weight—and its shape is determined by a balance
between deterministic drift towards a natural set point and diffusion
resulting from random fluctuations in, e.g., diet and physical activity.
We formulate a stochastic mathematical model for BMI dynamics, deriving a
theoretical shape for the BMI distribution and offering a mechanism that
may explain the right-skewed broadening of BMI distributions over time. An
extension of the base model investigates the hypothesis that peer-to-peer
social influence plays a role in BMI dynamics. While including this effect
improves the fit with the data, indicating that correlations in the
behavior of individuals with similar BMI may be important for BMI
dynamics, testing social transmission against other plausible unmodeled
effects and interpretations remains the subject of future work.
Implications of our findings on the dynamics of BMI distributions for
public health interventions are discussed.