Effects of non-invasive respiratory support on gas exchange and outcomes in COVID-19 outside the ICU

Non-invasive respiratory support (NRS) outside of the ICU has played an important role in the management of COVID-19 pneumonia. There is little data to guide selection of NRS modality. We present outcomes of NRS outside the ICU and discuss the effects of NRS on gas exchange with implications for management.


Introduction
The SARS-CoV-2 pandemic has resulted in unprecedented demand on intensive care units (ICUs) [1]. Patients with COVID-19 related acute hypoxemic respiratory failure (AHRF) are frequently managed outside ICU [2], which data suggests is safe and may be necessary to conserve critical resources [3]. Use of non-invasive respiratory support (NRS) including high flow nasal cannulae (HFNC) and positive end-expiratory pressure (PEEP) may support oxygenation and avoid invasive mechanical ventilation (IMV) [4], allowing time for recovery. Selection of optimal modality remains unclear, with a need for reliable non-invasive bedside monitoring to recognise treatment failure and avoid delayed intubation.
In mechanically ventilated patients disease-course and effects on respiratory dynamics and gas exchange have been well described [5]. Comparatively little data is available for self-ventilating patients outside the ICU. Anecdotal observation of highly variable change in ratio of partial pressure of arterial oxygen (PaO 2 ) to fraction of inspired oxygen (FiO 2 ) (P:F) following initiation of continuous positive airway pressure (CPAP), led us to suspect that a subgroup may respond better via known physiological effects of PEEP.
We hypothesised that analysis of gas exchange abnormalities prior to-and post-initiation of NRS (5 and 6 on the WHO Clinical Progression Scale respectively) could characterise a phenotype more likely to respond to greater PEEP afforded by CPAP. Our aims were to guide selection of optimal respiratory support modality and parameters for monitoring treatment outside the ICU.

Methods
We performed a longitudinal observational cohort study of consecutive patients admitted to 6 centres in Ireland in March and April 2020 with COVID-19 (confirmed by nasopharyngeal polymerase chain reaction for SARS-CoV-2) and a radiographic diagnosis of pneumonia causing AHRF (>4L/min oxygen to maintain peripheral O 2 saturation (SpO 2 ) above 92%). The study was approved by the National Research Ethics Board (ref 20-NREC-COV-001) and consent waived. CPAP was administered via face mask with minimum PEEP of 10cmH 2 O. Flow rate for HFNC was capped at 30L/min, limiting PEEP to <3cmH 2 0. Data from first presentation to time of death, intubation or recovery was collected. We calculated P:F and SpO 2 :FiO 2 (S:F) prior to initiation of CPAP or HFNC and within 24hrs.
Change in P:F ratio (ΔP:F) after transitioning between respiratory support modalities are demonstrated in Fig. 1. Transitioning from COT to CPAP was associated with ΔP:F of +79.4 mmHg ( ±15.7, P < 0.0001, Wilcoxon matched-pairs signed rank test). Changes in PaCO 2 were minimal (mean 36.8 ± 1.3 mmHg on COT/HFNC and 38 ± 1 mmHg after commencing CPAP). 17 of 49 patients initially treated with HFNC received 'rescue' CPAP, but this was not associated with significant change in P:F (mean + 29.4 ± 30.5 mmHg, P = 0.36). 10 (59%) of these individuals were subsequently intubated and two died.
For the non-DNR group ΔP:F within 24hrs of commencing CPAP or HFNC was compared according to subsequent requirement for IMV (Fig. 1B), and did not differ significantly. Greater improvement in P:F after initiation of CPAP did not correlate significantly with BMI, age, duration since symptom onset or biomarkers. Nadir P:F was correlated with IMV, with 59.3% of patients with P:F ≤ 150 mmHg subsequently intubated, compared to 15.3% with P:F nadir >150 mmHg (unadjusted risk ratio 8.06, 95% CI 3.88-16.78). The relationship of S:F to P:F was described by the equation SF = 1.167*P:F + 47.99 (p < 0.0001, R 2 0.76, Fig. 2A). An S:F cut-off of 225 had 88% sensitivity and specificity of 92% to determine P:F < 150 mmHg (area under the ROC curve 0.9476, Fig. 2B).

Discussion
Effects of CPAP include lung recruitment, preservation of pulmonary surfactant and redistribution of perfusion [6], whereas loss of surfactant, atelectasis and worsening V:Q mismatch/shunt have been implicated in COVID-19 AHRF [7]. WHO COVID-19 guidelines recommend CPAP in 'selected patients', however selection criteria are not defined [8]. RCT data suggest that non-invasive PEEP may reduce risk of IMV [9]. Though unlikely to modify the disease course, this may allow time for recovery to occur and aid resource allocation.
Increased BMI is associated with worse outcomes in COVID-19 and with increased lung atelectasis [10]. Contrary to our original hypothesis, elevated BMI was not associated with augmented response to PEEP/C-PAP in our study. We did not identify a distinct 'PEEP/CPAP responder' phenotype and initial improvement in gas exchange was not associated with overall improved outcome. S:F correlated well with P:F ratio and offers a non-invasive method of assessing treatment response. S:F ratio <225 predicted P:F < 150 mmHg, which was associated with increased risk of intubation and may serve as a useful benchmark. Rates of IMV were similar across oxygen delivery groups but comparison is precluded by differences in disease severity. We did not observe increased duration of mechanical ventilation in patients intubated following treatment with CPAP.
Limitations of this study include its observational nature and lack of data on pharmacological treatment. Our study illustrates the effect of NRS on indices of gas exchange in COVID-19 AHRF. We believe that a trial of CPAP in those with more severe degree of shunt with careful bedside assessment of response is reasonable, albeit recognising that improvement in oxygenation within the first 24 h is not necessarily reassuring. S:F can be used in lieu of P:F, with a threshold of 225 prompting more intensive monitoring or escalation. Logistical and patient factors should be considered when starting NRS outside ICU, with bedside assessment of respiratory rate, pulse oximetry and work of breathing prioritised over routine arterial blood gas sampling.

Author contributions
CG, MC, AF, RWC and TMcE designed the study, co-ordinated data collection, analysed data and wrote the manuscript. TMcC, GG and RO'C analysed data and provided statistical input. All other authors coordinated data collection, provided editorial input and approved the final manuscript.

Declaration of competing interest
No authors declare any competing interests relevant to the submitted work. No external funding was provided for this study.
RWC reports grants and personal fees from Aerogen, grants and personal fees from GSK, personal fees from Novartis, personal fees from TEVA, grants from Vitalograph, outside the submitted work. TMcE reports personal fees from Pfizer, outside the submitted work.

Fig. 1.
A P:F ratios within 24hrs of transitioning between conventional oxygen and CPAP or HFNC (score of 5/moderate and 6/severe on the WHO Clinical Progression Scale respectively). Data normally distributed, mean change ± SEM, Wilcoxon matched-pairs signed rank test. Figure 1B ΔP:F within 24hrs of commencing CPAP or HFNC stratified according to subsequent intubation or recovery without IMV.