ROX index at 12 hours helps predict success of non-invasive respiratory support in patients with COVID pneumonitis

ABSTRACT

Introduction: The ISARIC4C score1 and APACHE II score are used widely to predict mortality in critically ill patients with COVID-19 pneumonitis. These scores, however, do not predict response to specific treatments. Non-invasive respiratory support (NRS) in the form of CPAP/NIV/HFNC has been extensively used to treat COVID-19 pneumonitis. ROX index2 (SpO2/FiO2/Respiratory Rate) is used to predict failure of HFNC in treatment of Acute Hypoxaemic Respiratory Failure and ARDS. However, there are limited data on its efficacy to predict NRS failure in COVID-19 infection. Objective: Whether ROX index can be used to predict response to NRS in both patients for escalation to mechanical ventilation and those where NRS is ceiling of care. Methods: A retrospective study of individuals, SARSCOV-2 positive by RT-PCR, admitted to the ICU and requiring CPAP/NIV/HFNC, in a single centre between October 2020 to January 2021. Respiratory parameters were obtained at initiation of NRS followed by 2, 6 and 12 hours post initiation. NRS failure was defined as the need for mechanical ventilation in those for escalation of support or death in those where NRS was set as ceiling of care. Results: Data (Table 1) for 104 patients (70 men) were analysed. The mean age and BMI were 58.0 years and 31.4kg/m2, respectively, mean Respiratory Rate was 31 and mean SpO2/FiO2 of 144 on admission. In 10 out of 104 NRS was set as ceiling of care. Most patients were treated with CPAP/NIV, and they often used HFNC for breaks and while eating and drinking. Of the 62 patients that failed NRS, 10 had NRS as ceiling of care and a further 10 died without receiving mechanical ventilation. 42 patients underwent mechanical ventilation. 70 were discharged and 34 (32.7%) died in hospital. Baseline and 12 hour ROX index was not significantly different but the mean change between 0 and 12 hour ROX (2.2, 95% CI 0.99 to 3.46;p=0.0005) was significantly higher in those with NRS success. NRS success was predicted by a ROX index value of >5(OR 2.59, 95% CI 1.15-5.85;p= 0.01) and improvement in ROX score by >1 at 12 hours (OR 3.25, 95% CI 1.43 to 7.4;p=0.025). Conclusion: There was a significantly higher increase in ROX index at 12 hours in those with NRS success. A 12 hour ROX index of > 5 or an improvement by >1 are good predictors of success. Patients where NRS failed were older, had higher APACHE II and slightly higher ISARIC-4C score as expected. Discussion: This was a real life study where patients were treated with a combination of CPAP/NIV and HFNC as opposed to just HFNC or CPAP. The advantage of ROX index over P/F ratio is that arterial blood gases are not needed. It can be scored easily by routinely collected vital observations. A large number of patients are likely to be treated outside critical care in light of results from RECOVERY-RS trial3. In these settings, ROX index could be a useful tool for escalation to critical care or planning for symptom palliation as appropriate.