TY - JOUR T1 - Non-invasive assessment of VQ - should we use a two compartment or three compartment model? JF - European Respiratory Journal JO - Eur Respir J VL - 44 IS - Suppl 58 SP - 190 AU - Gareth Jones AU - Geoff Lockwood AU - Stenson Ben AU - Joan Lasenby AU - Robert Ross Russell Y1 - 2014/09/01 UR - http://erj.ersjournals.com/content/44/Suppl_58/190.abstract N2 - Background - Computerised modelling of gas exchange has recently allowed non-invasive assessment of shunt and ventilation-perfusion (VQ) mismatch. We have shown that measurements of arterial saturation (SaO2) made at different levels of inspired oxygen (PiO2) can be computed to reconstruct an oxygen diffusion graph (J Clin Monitor 2014). This then provides a measure of VQ, shift and shunt.Methods - The original computer program was based on a 2-compartment model of the lungs incorporating a single VQ unit, and a shunt component. However it has been suggested that a 3-compartment model (using a high VQ area, a low VQ area and a shunt fraction) would be more accurate. We therefore reassessed our original data using both a 2C and 3C model.Results - Data from the Olszowlka and Wagner gas exchange model, and then data from 16 infants with 56 datasets were compared between the two models.Data from the O&W model showed close correlation with both the 2C and the 3C model although the 3C model did not break the data back into constituent fractions as well as the 2C model. When applied to the clinical dataset, the 2C and 3C models were very close for 33 of 56 datasets, but in the remaining 23 datasets, the 3C model varied. In these cases, the 3C model suggested greater VQ abnormalities (and smaller shunt fraction) than the 2C model. These patients were clinically sicker with more severe lung disease.Discussion - The 2C and 3C models both give acceptable interpretation of most datasets, but in the sickest neonates the 3C model appears to offer a more clinically representative interpretation. ER -