We thank Laserna et al (1) for their interest in our study (2) and their insights. Although the rate-oxygenation index (ROX) index did perform poorly in predicting progression to invasive mechanical ventilation in our cohort of high-flow nasal cannula (HFNC) patients, a model including a number of demographic, clinical, and laboratory variables in addition to the ROX index performed better at predicting ventilation or death by days 1 and 7 with area-under-the-curves (AUCs) of 0.73 and 0.71, respectively. Among the large group of variables, we collected, least absolute shrinkage and selection operator regularization methods (3) identified the following variables to use in our predictive model in order of most to least significant: ROX less than 3.85, do-not-resuscitate (DNR)/do-not-intubate (DNI) order, pulse, Spo2/Fio2 ratio, White race, glomerular filtration rate, alanine transaminase, temperature, systolic blood pressure, hemoglobin, albumin, and C-reactive protein (Table 1). We suspect that DNR/DNI status may have been a proxy for frailty and overall health status.
TABLE 1.
Variables in Model Predicting Time to Ventilation or Death
| Variables | Adjusted Hazards (p) |
|---|---|
| Rate-oxygenation index < 3.85 | 0.50 (< 0.001) |
| Do-not-resuscitate/do-not-intubate order | 0.52 (< 0.001) |
| Spo2/Fio2 ratio | 0.76 (< 0.001) |
| Alanine transaminase | 0.79 (0.02) |
| Estimated glomerular filtration rate | 0.79 (0.001) |
| Systolic blood pressure | 0.84 (0.03) |
| Hemoglobin | 0.87 (0.07) |
| Albumin | 0.93 (0.32) |
| C-reactive protein | 0.96 (0.52) |
| Temperature | 1.21 (0.007) |
| White race | 1.24 (0.017) |
| Pulse | 1.31 (< 0.001) |
These variables were selected from a larger pool of variables using the least absolute shrinkage and selection operator regularization method. The complete list of variables considered also included age, sex, Charlson comorbidity index, body mass index, diastolic blood pressure, respiratory rate, absolute lymphocyte count, d-dimer, and ferritin.
Due to the rapid creation and closing of new ICU and stepdown units during the time period of the study, it was difficult for us to extract and validate data regarding ICU length of stay (LOS), ICU mortality, neuromuscular blockade, and prone positioning. We suspect, however, that the use of prone positioning or neuromuscular blockade would not have been different according to whether a patient received HFNC prior to intubation. However, we agree that future analyses will be more robust with the inclusion of neuromuscular blockade and both awake and postintubation prone positioning.
There was no difference between the groups in percentage of patients who underwent tracheostomy. There was a nonstatistically significant trend toward reduced duration of mechanical ventilation among HFNC patients (both overall and among survivors) compared with controls (Table 2). As these differences were not statistically significant, we are hesitant to draw conclusions from them. On the other hand, tocilizumab use was modestly higher in the HFNC versus the control group (11.6% vs 7.1%; p = 0.025), which may have been due to a greater opportunity to use this therapy prior to intubation.
TABLE 2.
Additional Differences Between Groups
| Outcomes | Matched High-Flow Nasal Cannula | Matched Control | p |
|---|---|---|---|
| Duration of mechanical ventilation (mean [sd]) | 14.4 d (14.6 d) | 16.6 d (13.9 d) | 0.199 |
| Duration of mechanical ventilation among survivors (mean [sd]) | 15.1 d (14.0 d) | 17.4 d (13.4 d) | 0.223 |
| Number requiring tracheostomy (mean) | 32 (7.6%) | 32 (7.6%) | 0.572 |
| Number treated with tocilizumab (mean) | 49 (11.6%) | 30 (7.1%) | 0.025 |
We agree that secular trends are difficult to completely control for as practice continues to evolve throughout the course of the pandemic. Thus, despite our best efforts, the hazard ratio associated with HFNC use may have not been stable throughout the time period studied. For instance, HFNC may have not been employed to its full effect early in the pandemic in which patients tended to be intubated earlier in their course. We find it reassuring, however, that a recent randomized controlled trial found decreased need for invasive mechanical ventilation as well as a trend toward lower mortality among patients treated with HFNC versus standard supplemental oxygen therapy (4). These findings along with those from our own analysis support the use of HFNC for pneumonia and ARDS in appropriately selected patients.
Footnotes
Dr. Garibaldi received funding from Janssen Research and Development, Gilead Life Sciences, and Atea Pharmaceuticals. The remaining authors have disclosed that they do not have any potential conflicts of interest.
REFERENCES
- 1.Laserna A, Barahona-Correa JE, Papadakos P, et al. : High-Flow Nasal Cannula in COVID-19 Pneumonia: Practical Issues. Crit Care Med 2022; 50:e591–e592 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Burnim MS, Wang K, Checkley W, et al. : The effectiveness of high-flow nasal cannula in COVID-19 pneumonia: A retrospective cohort study. Crit Care Med 2022; 50:e253-e262 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Tibshirani R: The lasso method for variable selection in the Cox model. Stat Med 1997; 16:385–395 [DOI] [PubMed] [Google Scholar]
- 4.Ospina-Tascón GA, Calderón-Tapia LE, García AF, et al. : Effect of high-flow oxygen therapy vs conventional oxygen therapy on invasive mechanical ventilation and clinical recovery in patients with severe COVID-19: A randomized clinical trial. JAMA 2021; 326:2161–2171 [DOI] [PMC free article] [PubMed] [Google Scholar]