key: cord-349973-7441gjda authors: Ahmed, Y. N.; Gokaraju, S.; Powrie, D.; Amran, D.; El Sayed, I.; Roshdy, A. title: Predictors of CPAP outcome in hospitalised COVID-19 patients date: 2020-06-16 journal: nan DOI: 10.1101/2020.06.14.20130880 sha: doc_id: 349973 cord_uid: 7441gjda Introduction: Throughout March - April 2020, many patients with COVID-19 presented to Southend University Hospital with Acute Hypoxaemic Respiratory Failure (AHRF). Patients were managed in a Specialist Respiratory High Dependency Unit. We present our experience on the usage of continuous positive airway pressure (CPAP) therapy and possible indicators of its success in this patient group. Methods: Data from patients (n=89) requiring mechanical ventilation during the months of March to April 2020, were retrospectively collected and analysed. 37 patients received IMV (Invasive Mechanical Ventilation) without a CPAP trial beforehand. 52 patients underwent a CPAP trial, of which 21 patients successfully avoided intubation and ITU admission. Results: The 52 patients, prior to receiving CPAP had significant respiratory failure as evidenced by a low PaO2: FiO2 (PFR) (mean +/- SD 123 +/- 60 mmHg) and mean SpO2:FiO2 (SFR) (mean +/- SD: 140+/- 50). The main indicators of CPAP success were: higher SFR before and after CPAP, lower respiratory rate (RR) , lower Neutrophil to Lymphocyte ratio (NLR) and higher PFR prior to CPAP. Discussion: CPAP proved successful in 40% of COVID-19 patients presenting with AHRF. SFR, PFR, RR and NLR are predictors of such success. SFR can be used for effective real time monitoring of patients before and after CPAP to identify likelihood of success. Based on our results, we have suggested a modified CPAP management protocol in COVID-19. These findings can guide future studies and will allow improved triage of patients to either CPAP or IMV, in the event of a future COVID peak. As of May 2020, more than 5.9 million people have been diagnosed with including more than 367,000 fatalities. (1) Whilst most patients are asymptomatic or mildly affected, in the United Kingdom, 17% of COVID-19 hospitalised patients were admitted to Intermediate or Intensive Care units. (2) In COVID-19, acute hypoxaemic respiratory failure (AHRF) may be caused by acute respiratory distress syndrome (ARDS) or pneumonia (not meeting the ARDS criteria). The UK National Health Service (NHS) issued specific guidance on the use of NIPPV in COVID-19 in March 2020. Following this guidance, we created an algorithm for the use of CPAP in COVID-19 patients admitted to our hospital ( Figure 1 ). The decision of whether to give the patient a trial of CPAP or to intubate directly was guided by our algorithm. However, it is important to note that the final decision lay with the on call respiratory consultant and the ITU team. We conducted a retrospective cohort study to investigate the predictors of CPAP outcome in these patients. Adult COVID-19 patients (≥ 18 years old) admitted to Southend University Hospital (UK) between the 1st of March 2020 and the 30th of April 2020 that underwent CPAP treatment in the Acute Respiratory Care Unit (ARCU) were included in the study. All patients had been tested SARS-CoV2 RNA positive using Nose and Throat swabs sent for real time RT-PCR (RdRp gene assay). COVID-19 patients who had any form of domiciliary NIPPV for obstructive sleep apnoea, or for any other reason, prior to hospital admission, were excluded from the study. Patients who underwent hospital BiPAP or received CPAP for the treatment of cardiogenic pulmonary oedema were also excluded from the study. The ARCU is a specialised unit staffed by experienced respiratory consultants and nurses. Patients on CPAP were nursed with 1:2 nurse to patient ratio. Emphasis was placed on a secure fitting interface of the CPAP system with a double filter (expiratory filter for the face mask, and exhaust system filter) in order to reduce widespread dispersion of exhaled air in addition to strictly following guidance on use of personal protective equipment (PPE), thus minimising the risk of airborne generating procedures (AGPs) to healthcare workers. (4) Data was collected retrospectively using various hospital databases and patients' medical notes. The ARCU reports to the Intensive Care National Audit and Research Centre (ICNARC). We screened the ICNARC database for all eligible COVID-19 patients and collected their demographic data. Patients escalation status was collected using 'treatment escalation plan' prior to CPAP trial and patients were classed as 'full escalation' or 'NIPPV' ceiling of treatment. The patients' oxygen saturations (SpO2) on room air as recorded by pulse oximetry was recorded on admission. Patients oxygen saturations (SpO2) and supplementary oxygen (FiO2) prior to CPAP were recorded. Respiratory rate (RR), heart rate (HR) and alertness on AVPU scale were also recorded. SpO2 to FiO2 ratio (SFR) was calculated prior to receiving CPAP. SFR post CPAP trial was also recorded using an average of the first three pulse oximeter recordings divided by the average of the first three FiO2 recordings within 30-120 minutes of the start of CPAP. All measurements were taken while patient was in the supine position, and when on CPAP, its pressure was 10 cm H2O. The PFR (mmHg), also known as the Horowitz index was calculated as the ratio between PaO2 (in mmHg) and FiO2. In order to standardise FiO2 across various oxygen delivery systems, the following standardisations were used: The following biochemical markers were recorded upon admission and prior to CPAP: CRP (mg/L), Neutrophils (109/L), Lymphocytes (109/L) and D-dimer (ng/ml). Neutrophil to lymphocyte ratio (NLR) was calculated. In addition, admission chest x-ray findings were reviewed retrospectively by a radiologist to determine if they showed bilateral pulmonary infiltrates, compatible with the diagnosis of ARDS. A successful CPAP trial was defined as the successful weaning of CPAP to other oxygen delivery devices with no re-institution of any form of MV for 72 hours. A 'fail' on the CPAP trial occurred when the responsible respiratory consultant decided to end the CPAP trial, . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint either proceeding to intubation or if their ceiling of treatment was NIPPV they were stepped down to a ward-based ceiling of treatment. The work has been approved ethically and registered in the hospital R&D under registration number 20007. Quantitative data were described by mean (standard deviation) and median (minimummaximum). While categorical variables were summarized by frequency and percent. Receiver Operating Characteristic (ROC) curve analysis by DeLong method was performed to detect the predictive performance of different indices for CPAP success. (8) An optimal cut-off point for each index test was determined by Youden index. (9) We conducted bivariate analysis using Independent sample t-test or Mann-Whitney test based on quantitative variables distribution as well as Pearson's Chi-square test to compare different demographic and clinical parameters between CPAP success and failure groups. Statistically significant, and clinically relevant categorized predictors based on ROC analysis results were fitted in multivariate stepwise backward Wald logistic regression analysis. Pearson's correlation test was also used to study linear relation between PFR and SFR. (10) Decision trees by classification and regression tree (CART) analysis were performed for depicting the prognostic factors associated with success before and after CPAP. The main variables associated with CPAP success were ordered by their relative importance: SFR before and after CPAP, respiratory rate prior to CPAP, NLR and PFR. Performance of each classification tree was determined by calculating accuracy of model's predicted probability with corresponding area under the curve. (11) We did not calculate sample size, however, power analysis for each of AUC for SFR before and after CPAP as well as NLR was at least more than 80%. Statistical analysis was done using IBM SPSS statistics program (12) , and R software with the following packages: "rpart", "rpart.plot" , "pROC" and "ROCR". (13) All statistical tests were two-sided and judged at 0.05 significance level. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . 89 patients with AHRF were identified. 37 patients were intubated directly and 52 patients underwent a CPAP trial in the ARCU. 40.3% of the patients passed the CPAP trial (n=21). All patients survived after passing the CPAP trial while among those intubated 38% died (p<.001). There were no significant differences in demographic data or co-morbidities between those groups who failed or succeeded on CPAP (Table 1) . Diabetes Mellitus was more prevalent in the CPAP failure group (32% vs 14%) but that difference did not reach statistical significance. The CPAP cohort suffered significant respiratory failure as evidenced by a mean PFR (123.15 ± 59.56) and mean SFR (140 ± 50). 48 patients (92%) met the Berlin definition of ARDS confirming the severity of the disease in this group of patients. Both pre and post CPAP median SFR values were significantly higher in those patients that underwent a successful CPAP trial (p <.05) and median NLR was significantly lower. However, ∆SFR did not differ significantly between both groups (p >.05). For all patients, the mean SFR improved significantly post-CPAP (140±50, p<.001) whether patients failed (p.001) or passed the trial (p .016). We detected also significant positive linear relation between PFR and SFR before (r=.417, p.002) and after CPAP (r =.579, p<.001). NLR was a fair predictor for CPAP success with corresponding sensitivity of 76.2%, specificity of 71% and accuracy of 79% (AUC=0.79, p <.001) ( Table 2) . A cut-off value less than 8.21 best predicted CPAP success ( Figure 2 CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint We report a 40% success of CPAP in COVID-19 patients (n=21/52). We were able to identify 5 significant predictors for CPAP success: RR, PFR, NLR, SFR pre-CPAP; and SFR 30-120 minutes after CPAP (Table 2 ). An SFR of 180 post-CPAP was the most accurate predictor of CPAP success (AUC = 0.8) ( Table 2 & is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint may increase the risk of transmission of COVID-19 to healthcare workers. (19) The avoidance of endotracheal intubation carries many benefits not least of which is the reduced burden on scarce ITU resources during a pandemic. Our data suggest that monitoring should focus mainly on 2 parameters: (1) Work of breathing and (2) Oxygenation. CPAP helps recruiting more functioning lung units. (17) Patients with good lung compliance can generate high tidal and minute volume before feeling short of breath and the RR started to increase significantly. In our study, we showed a significant difference in the RR existed between the failure and success groups before CPAP (Mean±SD 29.8±7.4 versus 25.5± 6.4 cycles per minute respectively, p=0.037) ( Table 1) . We have demonstrated the value of SFR. Monitored before and 30-120 minutes after the CPAP trial, it showed high sensitivity and specificity in detecting CPAP outcome. In our opinion, such an approach can significantly improve the safety of the CPAP trial out of the ITU settings. SFR offers many advantages over PFR. It is based on a non-invasive, continuous and real time pulse oximeter monitoring. By using CPAP and monitoring SFR, there is reduced need of ABG, cost and complications related to this procedure. Moreover, SFR is non-invasive measurement, easily monitored remotely thus reducing exposure to staff from risk of infection related to aerosol generating procedures (AGPs). However, we appreciate many limits of the SFR secondary to the potential inaccuracy of the Based on our results, we suggested a modified CPAP management protocol in COVID-19 patients. (Fig 4) The modified algorithm needs further confirmation in a larger study. Before a CPAP trial, the best predictor of success is SFR. with RR < 25/ min are predicted to fail on CPAP trial. (Figure 4) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint Our study is not without limitations. First, it is a retrospective observational study conducted on a small group of patients. Our results need confirmation in a larger prospective work. Second, many mechanisms can contribute to AHRF even in a single disease (e.g. Pneumonia, ARDS, but also pulmonary embolism which is not rare in COVID-19). Third, it remains to be determined if the CPAP trial affected the outcome of intubated patients in comparison to those promptly intubated from the start. Furthermore, we did not use High Flow Nasal Cannula (HFNC) which may have a role. A randomised controlled trial may be needed for full comparison between the 3 treatment modalities of HFNC, CPAP and early IMV. CPAP has a significant role in the management of COVID-19 patients presenting with AHRF. We have demonstrated that an experienced medical team following protocol driven management can successfully use CPAP to manage patients with COVID-19 and AHRF. In our group of patients, its success rate was 40%. SFR, PFR, NLR and RR are predictors of such success. SFR is a non-invasive, real-time measurement which can be used effectively to monitor these patients before and after CPAP to identify likelihood of success and reduce the need for frequent ABGs. The proposed modified CPAP management algorithm provides an initial step for better identifying those patients who will likely succeed on CPAP therapy. Thus, reducing the need for unnecessary IMV, and its associated complications, which can relieve pressure on scarce ITU resources during a pandemic. This algorithm will strengthen decision making should we encounter a second wave of COVID- . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. Non-parametric analysis was applied to quantitative variables summarized by median(range) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint *ABG needed only ONCE prior to start of CPAP trial, with further follow up by O2 saturation ** CPAP interface: full-face non-vented mask with expiratory viral filter (or whole face mask) Consider awake proning to identify patients who will benefit from proning post-intubation. In anxious/distressed patients, consider small doses of opioids or benzodiazepines. STAFF TO USE STRICT PPE. Contact CPAP Respiratory consultant on Rochford ward (ext 5474, 5475) if further advice needed . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint *Target SpO2 88-92% if risk of T2RF ** If no ceiling of care ǂ First Decision tree for prediction of CPAP success before CPAP procedure, accuracy = 78.8% and corresponding AUC= 0.870 ǂ ǂ Second Decision tree for prediction of CPAP success after CPAP procedure, accuracy = 87.7% and corresponding AUC= 0.867 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.14.20130880 doi: medRxiv preprint Situation report -128 Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study Distinct phenotypes require distinct respiratory management strategies in severe COVID-19. Respiratory Physiology & Neurobiology available online 11 Use Of Non-Invasive Ventilation For Patients With COVID-19: A Cause For Concern COVID-BioB Study Group. Characteristics, treatment, outcomes and cause of death of invasively ventilated patients with COVID-19 ARDS in Noninvasive ventilation in acute respiratory failure: which recipe for success? The American-European Consensus Conference on ARDS. Definitions, mechanisms, relevant outcomes, and clinical trial coordination Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach OptimalCutpoints: An R Package for Selecting Optimal Cutpoints in Diagnostic Tests Discovering statistics using IBM SPSS statistics. Sage An introduction to recursive partitioning: Rationale, application, and characteristics of classification and regression trees, bagging, and random forests Released 2015. IBM SPSS Statistics for Windows R: A language and environment for statistical computing. R Foundation for Statistical Computing Noninvasive Ventilation in Acute Hypoxemic Nonhypercapnic Respiratory Failure: A Systematic Review and Meta-Analysis Clinical characteristics of 113 deceased patients with coronavirus disease 2019: retrospective study Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study Recruitability and effect of PEEP in SARS-Cov-2-associated acute respiratory distress syndrome British Thoracic Society Standards of Care Committee. Non-invasive ventilation in acute respiratory failure Ventilation Techniques and Risk for Transmission of Coronavirus Disease, Including COVID-19 Clinical use of pulse oximetry: Official guidelines from the Thoracic Society of Australia and New Zealand Correlation between the levels of SpO2and PaO2. Lung India Specialty guides for patient management during the coronavirus (COVID-19) pandemic. (2020) Evaluation of the SpO2/FiO2 ratio as a predictor of intensive care unit transfers in respiratory ward patients for whom the rapid response system has been activated National Institutes of Health, National Heart, Lung, and Blood Institute ARDS Network. Comparison of the SpO2/FIO2 ratio and the PaO2/FIO2 ratio in patients with acute lung injury or ARDS Comparison of the spo2/fio2 ratio and the pao2/fio2 ratio in patients with acute lung injury or acute respiratory distress syndrome Development and Validation of a Clinical Risk Score to Predict the Occurrence of Critical Illness in Hospitalized Patients With COVID-19 Neutrophil-to-lymphocyte ratio as an independent risk factor for mortality in hospitalized patients with COVID-19 The diagnostic and predictive role of NLR, d-NLR and PLR in COVID-19 patients Neutrophil-to-lymphocyte ratio and lymphocyte-to-C-reactive protein ratio in patients with severe coronavirus disease 2019 (COVID-19): A metaanalysis Neutrophil-to-lymphocyte ratio as an independent risk factor for mortality in hospitalized patients with COVID-19