- COVID-19
- SARS-CoV-2
- intensive care units
- rapid response team
- mechanical ventilation
- noninvasive ventilation.
Introduction
The COVID-19 pandemic has strained hospital resources.1 Optimal timing of noninvasive respiratory support versus invasive mechanical ventilation has not been elucidated. Beginning September 2020, Mayo Clinic allowed early initiation of high-flow nasal cannula (HFNC) on the general care floor for adult patients with COVID-19 to mitigate strain. We hypothesized that this policy will stratify patients who are safe for continued floor care from those who warrant ICU-level care and intubation.
Methods
Subjects
A retrospective analysis of 22,330 rapid response team activations on adult subjects at Mayo Clinic Hospital in Rochester, Minnesota, from January 1, 2015–February 28, 2021, was performed. Demographic information, intubation rates, and mortality for polymerase chain reaction–confirmed subjects with COVID-19 on the general floor initiated on HFNC between September 1, 2020–February 28, 2021, were analyzed. This study was reviewed and approved by Mayo Clinic institutional review board (16-009990). Respiratory metrics—breathing frequency, SpO2, and FIO2—were collected at 2, 6, 12, and 24 h post-HFNC initiation and used to calculate the ROX index, defined as SpO2/FIO2/breathing frequency. Subjects with COVID-19 with hypoxia on the general floor (SpO2 < 90%) while on 6 L/min nasal cannula were initiated on HFNC at 40 L/min, 37°C, and FIO2 0.4. Flow, temperature, and FIO2 were titrated to sustain SpO2 ≥ 92%. To avoid failure-to-rescue situations, HFNC was capped at maximum 60 L/min of flow and FIO2 0.7. In the case of worsening hypoxia or increasing oxygen requirements, decision to escalate care to the ICU was made jointly by the rapid response and primary floor teams. HFNC failure was defined as escalation to invasive mechanical ventilation or death. In the case of HFNC failure, the decision to initiate invasive mechanical ventilation was made by the ICU clinician. Data were analyzed in IBM SPSS Statistics Version 28 (IBM, Armonk, New York) and reported as mean with SD or median with interquartile range (IQR). Categorical data are presented as value, frequency (percentage), and odds ratio. Rapid response activation was analyzed with the Mann-Whitney test. Continuous data were compared with the 2-tailed independent samples t test. ROX index was analyzed with receiver operating characteristic curve and area under the curve. Optimal cutoff was determined by Youden (J) index.
Results
Rapid Response Team Activation Trend
Rapid response team activations since the start of the COVID-19 pandemic in January 2020 remained unchanged from baseline (P = .31), with a median of 57.75 (IQR 3.44) activations per 1,000 admissions.
Study Population
Between September 2020–February 2021, HFNC was initiated on 126 subjects with COVID-19 on the general floor. Rapid response was activated on 56 (44%) of these subjects. Seven of the 56 subjects (12.5%) were deemed stable to continue treatment on the general floor. In 6 of these cases, noninvasive ventilation was optimized. In the remaining case, the subject elected to change code status. Forty-nine of the 56 rapid response activations (87.5%) required escalation to the ICU. Sixteen subjects had multiple rapid response team activations prior to their ultimate ICU transfer. Three additional subjects on the general care floor were directly transferred to the ICU for increased care needs.
Demographics of the 126 subjects are presented in Table 1. Seventy-four (59%) subjects started on HFNC on the general care floor remained stable and required no further escalation of care. Most common comorbidities included obesity (35/74 with body mass index [BMI] ≥ 30 kg/m2, 47%), hypertension (25/74, 34%), type II diabetes mellitus (20/74, 27%), and preexisting respiratory illnesses defined as obstructive sleep apnea, COPD, or pulmonary embolism (36/74, 49%).
Characteristics of Subjects Escalated to the ICU Versus Those Who Remained on the Floor
Fifty-two subjects (41%) started on HFNC required escalation of care and were admitted to the ICU either through rapid response team activation (49 subjects) or direct transfer (3 subjects). Those requiring escalation to ICU were more likely to be male (37/52, 71%) and obese (28/52 with BMI ≥ 30, 54%). In contrast to the group not requiring ICU transfer, this group had a higher incidence of hypertension (33/52, 63%; OR 3.40 [95% CI 1.62–7.15]) and diabetes (27/52, 52%; OR 2.92 [95% CI 1.38–6.16]). Age, obesity and comorbid respiratory illnesses (24/52, 46%) were not significantly different for subjects requiring ICU escalation.
Among the 52 subjects transferred to the ICU, 13 (25%) remained on HFNC up to 60 L/min and FIO2 0.7 without need for alternative management prior to transferring back to general care. Eighteen subjects (35%) required either an increase in HFNC settings beyond the general ward threshold or initiation of positive airway pressure therapy. Twenty-one subjects (40%) required intubation and mechanical ventilation. Overall mortality for those transferred to the ICU was 52% (27/52).
In contrast, of the 74 subjects with COVID-19 who were started on HFNC and remained on the general care floor, only one mortality (1.4%, 1/74) was noted. This subject elected to stay on the general floor following a goals-of-care discussion.
ROX Index
The ROX index in subjects who failed HFNC and required intubation was compared to subjects who remained on HFNC with respiratory recovery. The ROX index was significantly lower for the intubation group at 6 h post-HFNC initiation (7.69 vs 9.08; t test, P = .03), 12 h (7.23 vs 9.36, P = .01), and 24 h (7.13 vs 9.01, P = .04). The ROX index was not significantly different at 2 h after HFNC initiation (8.87 vs 9.21, P = .57). Furthermore, receiver operating characteristic analysis (Fig. 1) identified the ROX index at 12 h as the best indicator of HFNC failure and intubation with area under the curve of 0.71 (95% CI 0.58–0.84). Optimal threshold of ROX index was > 6.64 (sensitivity of 0.82, specificity of 0.50).
Receiver operating characteristic curve for ROX index 12 hours after HFNC initiation.
Discussion
The stable number of rapid response activations throughout the COVID-19 pandemic despite global surges may be explained by Mayo Clinic’s unique policy to initiate HFNC in hypoxic patients instead of bridging with other devices like a pendant cannula or non rebreathing mask. Other institutions have also suggested the use of HFNC in COVID-19 treatment.2-4
There was 1.4% in-hospital mortality rate for the 74 subjects remaining on the general floor versus 52% in-hospital mortality rate for the 52 subjects transferred to the ICU. This retrospective study highlights the difference in outcomes between these 2 groups. Our criteria to escalate to the ICU were hypoxia (SpO2 < 90%) on HFNC 60 L/min of flow and FIO2 0.7. HFNC use on the general floor allows for stabilization of patients with mild disease and allows for allocation of resources to patients at elevated risk of decompensation. Further exploration is needed to delineate the optimal cutoff for ICU escalation for patient safety and resource efficiency.
Several mechanisms explain how HFNC improves symptoms and prevents escalation of mild and moderate COVID-19-related respiratory distress.5 HFNC provides PEEP by recruiting alveoli and improving gas exchange. It decreases anatomic nasopharyngeal dead space, allowing for improved work of breathing. It also supplies warm and humidified oxygen, decreasing the metabolic demand of breathing. Warm humid air facilitates secretion clearance, preventing atelectasis and ventilation-perfusion mismatch. Lastly, HFNC delivers a titratable inspired oxygen concentration at higher flows than conventional nasal cannula.
Furthermore, our data suggest that patients with certain risk factors are at higher likelihood of needing ICU escalation. Our results align with recent research that has shown poorer clinical outcomes associated with hypertension and diabetes.6,7 Our data did not demonstrate a significantly increased risk of ICU admissions with obesity, sex, or age, which contrasts with existing reports.8,9 This may be due to our sample size. Understanding the comorbid conditions that may portend poor outcomes may allow for early identification of patients with COVID-19 at high risk and allocation of appropriate rapid response team and ICU resources.
The ROX index was first developed by Roca et al10 to identify patients at high risk for HFNC failure in the setting of acute hypoxemic respiratory failure associated with pneumonia. Since then, several studies have proposed its use in patients with COVID-19 with hypoxemic respiratory failure.2-4 Our study demonstrates a significant ROX index difference that was seen as early as 6 h in subjects with COVID-19. Moreover, we report a predictive model with acceptable discriminatory ability to identify patients at risk for HFNC failure, using a ROX index > 6.64 at 12 h post-HFNC initiation. Interestingly, this work—along with several recent studies—has identified a more aggressive ROX cutoff compared to Roca et al cutoff of 4.88 identified in pneumonia-related respiratory failure.3,4 This may be attributed to the lower breathing frequency in patients with COVID-19–related respiratory failure (leading to the phenomenon of “happy” or silent hypoxemia) compared to patients with non–COVID-19-related respiratory failure.11,12 Our study contributes to the existing pool of research showing that serial ROX indices may serve as noninvasive, early markers to identify patients with COVID-19 at high risk and may help clinicians appropriately time ICU escalations and invasive mechanical ventilation.
In conclusion, early initiation of HFNC, in conjunction with identifying comorbidities and serially monitoring ROX indices, allowed health care teams to safely and preemptively allocate resources to improve COVID-19 outcomes.
Footnotes
- Correspondence: Qiaonan Zhong MD, Baldwin Building 4th Floor, 221 Fourth Avenue SW, Rochester, Minnesota 55905. E-mail: zhong.qiaonan{at}mayo.edu
The authors have disclosed no conflicts of interest.
- Copyright © by Daedalus Enterprises