Oral Presentation Australian and New Zealand Obesity Society Annual Scientific Conference 2026

The BMI fallacy: do obesity definitions and categorisations matter for clinical research? (141599)

Efris EK Kartikasari 1 , Brian BR Robinson 1 , Lisa LW Woods 2 , Caz CH Hales 1
  1. School of Health, Victoria University of Wellington, Wellington, New Zealand
  2. School of Mathematics and Statistics, Victoria University of Wellington, Wellington, New Zealand

Background:  Despite established risks, obesity has been associated with improved survival in sepsis, termed the ‘obesity paradox’.1–3 Obesity can be defined using anthropometric, metabolic, or immune parameters.4–7 This study examined how different definitions and categorisations of obesity influence clinical outcomes in acute inflammatory conditions.

Methods: This multi-method study comprised a scoping review, a systematic review and meta-analysis, and a pilot clinical study. The scoping review summarised obesity definitions in clinical research. The systematic review and meta-analysis examined associations between obesity categorisation and sepsis outcomes. The pilot study, currently underway, will inform sample size estimation for a larger study examining obesity definitions, inflammatory markers, and outcomes.

Results: The scoping review included 68 studies. Most (88.2%) defined obesity using body mass index (BMI), including 90.5% of studies supporting and 88.6% refuting the obesity paradox. Retrospective designs predominated. Studies supporting the paradox described patients with obesity as younger, more female, and with greater comorbidity burden, whereas those refuting included more heterogeneous populations with consistently greater chronic disease prevalence.

The systematic review included 29 studies, with 16 in meta-analysis. Compared with BMI < 30 kg/m², obesity (BMI ≥ 30 kg/m²) showed a borderline reduction in mortality (OR 0.76, 95% CI 0.58–1.00). Excluding underweight, no significant association was observed (OR 1.06, 95% CI 0.53–2.13). By category, obesity classes I, II, and III were each associated with lower mortality relative to healthy weight, whereas BMI 30–39.9 kg/m² showed no significant difference (OR 0.88, 95% CI 0.73–1.06).

Conclusion: Evidence for the obesity paradox in sepsis is conflicting and appears sensitive to BMI categorisation and analytic approach. Differences in BMI cut-offs and exclusion of patients with underweight may affect findings. Prospective studies using high validity measures of adiposity are required to distinguish biological effects from methodological bias and inform clinical assessment.