Stochastic modeling of obesity status in United States adults using Markov Chains: A nationally representative analysis of population health data from 2017-2020.
Level 4 - case-series / case-control
Cross-sectional survey analysis with retrospective 10-year weight recall and computational Markov modeling.
PubMed 38090680 · doi:10.1002/osp4.697
What was done
Analyzed data from 6,146 US adults in the 2017–2020 National Health and Nutrition Examination Survey (NHANES) who had complete data for current weight and recalled weight 10 years prior. Researchers constructed a Markov transition state matrix combined with bootstrap simulations to model transitions across body mass index (BMI) categories over a 10-year period.
What was found
The sample (n = 6,146; 49% male, 51% female; 37% White, 37% Black, 20% Hispanic, 12% Asian) had a mean current BMI of 30.16 (SD = 7.15) and mean weight of 83.67 kg (SD = 22.04). Mean 10-year weight change was an increase of 3.27 kg (SD = 14.97); 61% (n = 3,735) gained weight and 39% (n = 2,411) lost weight. At survey time, 1% (n = 87) were underweight, 33% (n = 2,058) normal weight, 22% (n = 1,376) overweight, and 43% (n = 2,625) obese. Specific transition probabilities from the Markov matrix were not reported in the abstract.
Why it matters
Demonstrates the feasibility of applying stochastic Markov chain modeling with bootstrapping to national survey data to project population-level shifts into overweight and obesity categories over decade-long horizons.
Limits
Baseline weight from 10 years prior relies on self-report recall rather than direct measurement, introducing potential recall bias. Evaluating only two time points misses intervening weight fluctuations and non-linear trajectories. Specific Markov transition probabilities and fit metrics are omitted from the abstract.
Cited by
- supports Globally, over 2.5 billion people are overweight, and in the United States, 75% of adults are overweight and 42% are obese.