Comparison of Rox Index and Hacor Score for Predicting Failure of Non-invasive Respiratory Therapy in Acute Hypoxemic Respiratory Failure

INTRODUCTION: Acute hypoxemic respiratory failure (AHRF) is frequently found in critically ill patients and associated with poor outcomes. For those patients who fail COT, the use of non-invasive respiratory therapies (NIRT) like non-invasive ventilation (NIV), or high frequency nasal cannula (HFNC) may prevent the need for intubation and invasive mechanical ventilation. Although HACOR score and ROX index have been validated for their respective indications they have not been compared.(1–3) OBJECTIVES: Comparison of ROX index and HACOR score for predicting NIRT failure and effectiveness as defined by need of intubation. MATERIALS AND METHODS: This prospective observational study, conducted in a tertiary care hospital, evaluated 204 patients over 18 months. Patients diagnosed with AHRF failing conventional oxygen therapy were enrolled and stratified into HFNC (n=58) and NIV (n=146) groups based on clinician discretion. HACOR scores and ROX indices were recorded at 1 and 6 hours of therapy. Failure was defined by the necessity for ETI. RESULTS: Among NIV patients, HACOR score >4.5 at 1 & 6 hour demonstrated moderate realiability. ROX index at cut-off of 4.7 for 1 hour, had sensitivity of 71% and specificity of 59%. And at 6 hours cut-off of 5.4 sensitivity increased to 73% with specificity at 71%, achieving an AUC of 0.738. In HFNC recipients, the ROX index at 6 hours exhibited superior specificity (100%) but moderate sensitivity (37.5%). Whereas HACOR score showed moderate reliability in predicting the need for intubation in these patients. With cut-off of 5.5 provided a sensitivity (55%) and specificity (67%), achieving an AUC of 0.684 (95% CI 0.546-0.823), at 6 hours a lower cut-off of 4.5 at 6 hours improved sensitivity to 75% with a specificity of 66% (AUC of 0.785, 95% CI 0.647-0.923). The study found that both tools could predict NIRT failure, though their predictive accuracy improved over time. HFNC patients exhibited lower failure rates than NIV users for similar severity levels, highlighting its efficacy in milder cases. CONCLUSION: Both the HACOR score and ROX index had appropriately identified the failure of NIRT. The ROX index had improved performance with a higher cut-off in predicting NIV failure at 6 hours. Hence ROX index can be used to predict NIV failure over time with higher cut-offs in resource-limited settings.