Comparison of ROX index (SpO(2)/FIO(2) ratio/respiratory rate) with a modified dynamic index incorporating PaO(2)/FIO(2) ratio and heart rate to predict high flow nasal cannula outcomes among patients with acute respiratory failure: a single centre retrospective study

BACKGROUND: High flow nasal cannula (HFNC) is increasingly being used to support patients with acute respiratory failure (ARF) and to avoid need for intubation. However, almost one third of the patients do not respond and require escalation of respiratory support. Previously, ROX index (SpO(2)/FIO(2) [SF] ratio/respiratory rate) has been validated among pneumonia patients to facilitate early recognition of patients likely to fail HFNC and therefore, benefit from timely interventions. However, it has been postulated that incorporation of PaO(2)/FIO(2) (PF) ratio from arterial blood gas (ABG) analysis may better predict the outcome of HFNC compared to indices that utilizes SF ratio. Similarly, heart rate increase after HFNC therapy initiation has been found to be associated with HFNC failure. Therefore, we aimed to compare ROX index with a new modified index to predict HFNC outcomes among ARF patients. MATERIALS AND METHODS: This single centre 2-year retrospective study included ARF patients of varying etiologies treated with HFNC. The modified index incorporated heart rate and substituted PF ratio for SF ratio in addition to respiratory rate. We named the index POX-HR and calculated Delta POX-HR index as the difference pre- and post-HFNC initiation POX-HR. We also recorded ROX index at the time when post-HFNC initiation ABG was done (‘post-HFNC initiation ROX’) and calculated Delta ROX. HFNC success was defined as no need of escalation of respiratory support or discharged to ward within 48 h of HFNC initiation, or successful wean off HFNC for at least 12 h. Evaluation was performed using area under the receiver operating characteristic curve (AUROC) and cut-offs assessed for prediction of HFNC outcomes. RESULTS: One hundred eleven patients were initiated on HFNC for ARF, of whom 72 patients (64.9%) had HFNC success. Patients with HFNC failure had significantly lower values for all the indices. At median of 3.33 h (IQR 1.48–7.24 h), Delta POX-HR demonstrated the best prediction accuracy (AUROC 0.813, 95% CI 0.726–0.900). A Delta POX-HR > 0.1 was significantly associated with a lower risk of HFNC failure. CONCLUSIONS: Our proposed modified dynamic index (Delta POX-HR) may facilitate early and accurate prediction of HFNC outcomes compared to ROX index among ARF patients of varied etiologies.