Correlations between computational fluid dynamics and clinical evaluation of nasal airway obstruction due to septal deviation: An observational study

International audience ; Objectives: The primary objective of this study was to determine how computational fluid dynamics (CFD) could be correlated to clinical evaluation of nasal airway obstruction (NAO) in a population of patients with symptomatic septal deviation (SD). The secondary objective was to determine whether CFD could define which side was the more obstructed.Design: This was an observational study. Settings: Few publications have attempted to correlate CFD with clinical evaluation of NAO. This correlation would permit validation and improved interpretation. This study was performed in a university research laboratory specialised in fluid mechanics. Participants: We included patients referred for septal surgery at our centre. Age range was 19-58 years. Preoperative CT scans were performed. All patients with non-structural causes of NAO such as rhinitis, sinusitis or tumoral/autoimmune processes (ie, not due to anatomic obstruction) were excluded. Main outcome measurement: For each nasal fossa, we compared CFD data (total pressure, heat flux, wall shear stress, temperatures, velocity and nasal resistances) with both patient perception scores and rhinomanometry using the Spearman correlation test (r s). Perception scores were graded from 0/4 to 4/4 on each side, based on the patient interview. We also compared CFD-derived nasal resistances with rhinomanometry-derived nasal resistances.Results: Twenty-two patients complaining of NAO with SD were analysed, and 44 analyses were performed comparing each side with its CFD data. Regarding correlations with patient perception scores, the best values we found were heat flux measures (r s = 0.86). Both rhinomanometry and CFD-calculated nasal resistances had strong correlations with subjective perception scores (r s = 0.75, P < 0.001 and r s = 0.6, P < 0.001, respectively). We found a statistically significant difference between RMM-NR and CFD-NR (P = 0.003). Heat flux analysis allowed us to distinguish the more obstructed side (MOS) and the less ...