Retrospective Application of ROX Index to Adult Patients With Acute Respiratory Failure due to Pneumonia,ARDS,and/or COVID-19: A Quality Improvement Project

Abstract

Background:

Evidence on mortality indicates the importance of avoiding intubation too soon or too late. When to switch from noninvasive oxygen delivery systems to mechanical ventilation (MV) can be difficult to assess. The Respiratory rate-OXygen index (ROX) is a validated tool to assist clinicians in this decision process. This study evaluates independent variables for opportunities to improve mortality with the retrospective application of ROX.

Methods:

This is a single center, observational, cross-sectional study with IRB approval from UC Davis Medical Center, IRBNet. From January 2020-December 2020, a consecutive patient population was received from an EMR data analyst if ≥18 yrs old, diagnosed with acute respiratory failure due to pneumonia, ARDS, and/or COVID-19, utilized high-flow nasal cannula (HFNC), and subsequently intubated. Data analyst's population included ROX metrics (SpO₂, FIO₂, RR) when charted into Epic EMR in the same date/time stamp between initiation of HFNC and MV start. ROX results measured with an Excel formula, SpO₂/FIO₂/RR; inclusion criteria of <4.88 and exclusion criteria of >4.88. Patients included when ROX result was <4.88 twice, consecutively. Authors collected date/time stamp of the second failed ROX (SFR) for extrapolation. Univariate analyses compared expired/hospice (E) to survived (S) population with independent variables. Excel t-Test: Two-Sample Assuming Unequal Variances used for statistical analyses. Cognizance of one-tail and skewness of data for P-value interpretation validity.

Results:

Authors included n = 76 patients for variable comparisons with a mortality rate of 56.6% (43). We found statistical significance in (mean, 95% CI): Age, P = .005 (E) 65.56 (70.16-61) vs (S) 55.3 (60.85-49.75); HFNC start to SFR in hours, P = .03 (E) 33.99 (50.35-17.62) vs (S) 15.18 (26.42-3.94); SFR to MV start in hours, P = .013 (E) 62.57 (86.12-38.02) vs (S) 28.57 (47.91-9.23). We found no statistical significance in: duration of MV in hours, P = .769 (E) 185.4 (230.91-139.88) vs (S) 222.44 (313.39-131.49); hospital LOS in in days, P = .957 (E) 18.74 (24.78-12.71) vs (S) 26.55 (33.35-19.74); ROX result, P = .507 (E) 3.83 (4.05-3.61) vs (S) 3.83 (4.04-3.61).

Conclusions:

Opportunities to decrease mortality may exist within the periods from HFNC initiation to observed failed ROX and from observed failed ROX to MV start. UC Davis Medical Center will initiate ROX to auto calculate in Epic EMR to find ROX failure quicker and to potentially decrease the time from failed ROX to MV start.