Pulmonology

Joint Bayesian model shows larger FVC decline and less bias in fibrotic ILD analyses

A Bayesian joint mixed effects disease progression model jointly estimated FVC trajectory and mortality risk in fibrotic ILD cohorts. Compared with a linear mixed model of FVC alone, the joint model yielded a higher estimated FVC decline (6.0%/year vs 4.7%/year). The approach better fit non-linear trajectories and increases information per patient, reducing bias from differential mortality when estimating treatment effects. This model is intended for trial analysis or population characterization rather than bedside prognostication. 1

Multidisciplinary pathway greatly increased delivery of guideline-concordant inpatient pediatric asthma care

A hospital medicine quality improvement intervention raised all‑three-component optimal asthma care from 15% to 69% over 12 months. Interventions combined clinical decision support, education, and enhanced interdisciplinary communication aligned with GINA updates. The bundle was associated with reduced length of stay and fewer 90‑day readmissions. Process measures included all-or-none completion of risk assessment, maintenance plan, and a discharge action plan. 2

Two‑year exacerbation history improves prediction of future COPD exacerbations

In COPDGene and NOVELTY cohorts, any moderate or severe exacerbation in the prior two years gave the highest discrimination for next‑year risk. Observed AUCs were 0.69 (COPDGene) and 0.87 (NOVELTY), with improvement over the past‑year standard (ΔAUC 0.03 and 0.12). Net‑benefit analysis favored rolling two‑year windows across clinically relevant treatment thresholds (5%–30%). At least one moderate or severe exacerbation over two years is a pragmatic rule to flag higher exacerbation risk. 3