B58-03 Abnormality Classification Differs Minimally Across Two Oscillometry Reference Equations in a Veteran Cohort

Rationale Clinical interpretation of respiratory oscillometry involves comparing an individual’s measured values with those predicted from their age, sex, height, and weight. In our clinic, predicted values have been based on Oostveen et al. 2013 normative reference equations. Recently, Valach et al. 2025 introduced new reference equations derived from a larger population tested with a contemporary device. Due to its infancy, evaluation is needed alongside the older equation among Veterans before applicable use. The objective of this study is to evaluate the agreement of two prediction equations and explore abnormality reclassification utilizing measures of respiratory oscillometry in a sample of previously deployed military Veterans with respiratory complaints. Additionally, another objective is to make an online calculation application for clinicians and researchers to help compare the two equations on an individual basis. Methods Deployed and symptomatic Veterans were undergoing research or clinical evaluation at one of six Post-Deployment Clinical Evaluation Network sites for clinical evaluation. Oscillometry (Resmon Pro Full® and Quark i2M) was acquired in accordance with standard guidelines and parameters (R5, X5, AX, Fres) were identified for analyses. Paired t-tests were conducted to compare percent predicted mean differences of each Veteran’s oscillometry parameters between the two equations. Reclassification was calculated to assess Veterans that changed abnormality status according to the newer equation and vice versa. Cohen’s kappa coefficient was calculated to assess levels of agreement and AUC from ROC curves was calculated to establish agreement of the newer equation to the “golden standard” Oostveen prediction equation. Oscillometry parameters were deemed normal using z-scores. Results In our sample of 435 Veterans, the Valach (LEAD) equation indicated a slightly higher percentage of abnormal oscillometry compared to Oostveen (25% vs. 22%). T-tests showed significant mean differences across parameters (R5: 8.43%, X5: 32.6%, AX: 34.4%, Fres: 5.99%). Minimal reclassification (<7%) was observed across all parameters. A Cohen’s Kappa coefficient of 0.82 points to strong levels of agreement between the two equations. AUC values ranged from 0.85-0.95, also indicating robust agreement. Demographics and relevant statistics are included in the Table. An R shiny application was created to calculate an individual’s predicted values across two reference equations, and visually display impedance: https://shobhikc.shinyapps.io/fot_app/. Conclusion Despite significant mean differences in predicted values, both Valach and Oostveen exhibited strong agreement in identifying abnormalities. Mean differences likely reflect variations in equation derivation, such as Valach’s inclusion of BMI. Future analyses should consider exposure factors that could contribute to reclassification.