1. IntroductionNovel coronavirus disease SARS-CoV-2 emerged in December 2019, rapidly became pandemic and it was the cause of the so-called severe acute respiratory coronavirus disease (Covid-19). So far, more than 127 million cases were confirmed worldwide, with more than 2.7 million deaths [[1]

World health organization (WHO) Coronavirus Disease (COVID-19); https://covid19.who.int. Date last updated: March 31 2021. Date last accessed: March 31 2021.

]. Covid-19 typically affects respiratory tracts, leading to pneumonia and acute respiratory failure (ARF). [[2]Stawicki SP Jeanmonod R Miller AC Paladino L Gaieski DF Yaffee AQ et al.The 2019-2020 Novel Coronavirus (Severe Acute Respiratory Syndrome Coronavirus 2) Pandemic: A Joint American College of Academic International Medicine-World Academic Council of Emergency Medicine Multidisciplinary COVID-19 Working Group Consensus Paper.] Morbidity and mortality related to Covid-19 are due to complications, especially acute respiratory distress syndrome (ARDS), which occur in up to 15% of cases. [[3]Grasselli G Tonetti T Protti A Langer T Girardis M Bellani G et al.Pathophysiology of COVID-19-associated acute respiratory distress syndrome: a multicentre prospective observational study.,[4]Li LQ Huang T Wang YQ Wang ZP Liang Y Huang TB et al.COVID-19 patients' clinical characteristics, discharge rate, and fatality rate of meta-analysis.].The ratio of arterial (PaO2) to inspired (FiO2) partial pressure of oxygen (P/F ratio), is currently utilized to assess the severity of respiratory failure in patients with ARDS [[5]Definition Task Force ARDS Ranieri VM Rubenfeld GD Thompson BT Ferguson ND Caldwell E Fan E et al.Acute respiratory distress syndrome: the Berlin Definition.] and correlates to mortality rate [6Goligher EC Kavanagh BP Rubenfeld GD Adhikari NK Pinto R Fan E et al.Oxygenation response to positive end-expiratory pressure predicts mortality in acute respiratory distress syndrome. A secondary analysis of the LOVS and ExPress trials., 7Sakr Y François B Solé-Violan J Kotfis K Jaschinski U Estella A et al.SOAP and ICON Investigators. Temporal changes in the epidemiology, management, and outcome from acute respiratory distress syndrome in European intensive care units: a comparison of two large cohorts., 8Villar J Blanco J del Campo R Andaluz-Ojeda D Díaz-Domínguez FJ Muriel A et al.Spanish Initiative for Epidemiology, Stratification & Therapies for ARDS (SIESTA) Network. Assessment of PaO2/FiO2 for stratification of patients with moderate and severe acute respiratory distress syndrome.]. P/F ratio has been recently proposed as a prognostic marker in Covid-19 [[9]

World Health Organization (2021). COVID-19 clinical management: living guidance. 25 January 2021. World Health Organization. https://apps.who.int/iris/handle/10665/338882. License: CC BY-NC-SA 3.0 IGO.

,[10]Cortinovis M Perico N Remuzzi G. Long-term follow-up of recovered patients with COVID-19.]. However, P/F ratio may be poorly representative of the severity of hypoxemia in patients with ARDS [[11]Aboab J Louis B Jonson B Brochard L. Relation between PaO2/FIO2 ratio and FIO2: a mathematical description.,[12]Variability of indices of hypoxemia in adult respiratory distress syndrome.] and does not consider the level of respiratory muscles effort and hyperventilation of hypoxemic patients and do not discriminate patients according to their degree of hypocapnia [[13]Non-invasive respiratory support paths in hospitalized patients with COVID-19: proposal of an algorithm.]. In addition, considerable evidence supports that alteration of ventilation perfusion rate assessed as pulmonary dead space fraction [[14]Nuckton TJ Alonso JA Kallet RH Daniel BM Pittet JF Eisner MD et al.Pulmonary dead-space fraction as a risk factor for death in the acute respiratory distress syndrome.] or ventilatory ratio [[3]Grasselli G Tonetti T Protti A Langer T Girardis M Bellani G et al.Pathophysiology of COVID-19-associated acute respiratory distress syndrome: a multicentre prospective observational study.] are associated with mortality in ARDS [[15]Lucangelo U Bernabè F Vatua S Degrassi G Villagrà A Fernandez R et al.Prognostic value of different dead space indices in mechanically ventilated patients with acute lung injury and ARDS.] and severity of COVID-induced ARDS [[3]Grasselli G Tonetti T Protti A Langer T Girardis M Bellani G et al.Pathophysiology of COVID-19-associated acute respiratory distress syndrome: a multicentre prospective observational study.].In a seminal paper, Mays emphasized the axiom that PaO2 and arterial carbon dioxide tensions (PaCO2) are inversely related [[16]A graphical analysis of the respiratory gas exchange: The.] and suggested that estimation of the severity of ventilation/perfusion mismatch may be optimized standardizing PaO2 for PaCO2 by using the formula: standardized PaO2 (STPaO2) = 1.66*PaCO2 + PaO2 - 66.4 [[17]An arterial blood gas diagram for clinical use.]. In the current pilot observational study we evaluated if substituting PaO2 with STPaO2 in calculating P/F ratio may better stratify patients according to outcome failure, defined as needs of invasive mechanical ventilation (IMV) and/or death in patients with COVID-19.2. Material and MethodsThe Institutional Ethical Committee approved the study protocol and patients had to sign written informed consent before enrollment. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline [[18]Von Elm E Altman DG Egger M Pocock SJ Gøtzsche PC Vandenbroucke JP Initiative STROBE The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies.].Patients were prospectively recruited in the period October 31th 2020-January 31th 2021 after admission to pulmonology wards of the following hospitals: IRCCS S.Orsola-Malpighi (Alma Mater University of Bologna); Policlinico Umberto I, (Sapienza University of Rome) and Central Hospital of Bolzano. Inclusion criteria were laboratory-confirmed SARS-CoV-2 infection (positive result of real-time reverse transcriptase-polymerase chain reaction assay from either nasal or pharyngeal swabs, or lower respiratory tract aspirates); presence of consolidation and/or ground glass opacities at Chest X-ray and/or at computed tomography of lungs [[19]Martínez Chamorro E Díez Tascón A Ibáñez Sanz L Ossaba Vélez S Borruel Nacenta S Radiologic diagnosis of patients with COVID-19.] and presence of acute respiratory failure. Acute respiratory failure was identified when pO2 was [20]

British medical journal (BMJ) Best Practice - Acute respiratory failure. https://bestpractice.bmj.com/topics/en-us/853. Date last reviewed: 21 Apr 2021. Date last updated: 13 May 2020. Date last accessed: 22 May 2021.

] Exclusion criteria were needs of endotracheal intubation and invasive mechanical ventilation before Pulmonology wards admission and history of chronic respiratory failure. For each study subject we collected clinical history, arterial blood gas analysis (ABGs) data (PaO2, PaCO2, pH, HCO3, FiO2) at hospital admission and at the time of admission to the Pulmonology Unit, respiratory supports applied throughout hospital stay and date of death or recovery from respiratory failure. PaO2 was standardized for PaCO2 by using the formula: standardized PaO2 (STPaO2) = 1.66*PaCO2 + PaO2 - 66.4 [[17]An arterial blood gas diagram for clinical use.]. P/F, STPaO2 and STP/F were calculated for each subject. For STP/F and P/F, we use data from ABG collected on the first day of admission in Pulmonology Unit with the study subject that had inspired oxygen at a fixed FiO2 for at least 10 minutes [21Cakar N Tuŏrul M Demirarslan A Nahum A Adams A Akýncý O et al.Time required for partial pressure of arterial oxygen equilibration during mechanical ventilation after a step change in fractional inspired oxygen concentration., 22Weinreich UM Thomsen LP Hansen A Kjærgaard S Wagner PD Rees SE. Time to steady state after changes in FIO(2) in patients with COPD., 23Sasse SA Jaffe MB Chen PA Voelker KG Mahutte CK. Arterial oxygenation time after an FIO2 increase in mechanically ventilated patients., 24Utada K Matayoshi Y Fujita F Nakamura K Matsuda N et al.Equilibration period for PaO2 following alteration of FIO2 in mechanically ventilated patients., 25Chiumello D Coppola S Froio S Mietto C Brazzi L et al.Time to reach a new steady state after changes of positive end expiratory pressure.]. Occurrence of ARF, was identified when PaO2 was 2 = 0.21. Outcome failure was defined as needs of invasive mechanical ventilation (IMV) and/or death. We also evaluated the relationship between duration of ARF and P/F and STP/F. Recovery from ARF occurred before pulmonology ward discharge of the study subjects. Duration of ARF was expressed in days from emergency room (ER) admission to the first day of recovery from ARF (subjects died during hospital stay were censored). End of follow-up for each study subject was fixed hospital stay discharge (for survivors) or date of death.Continuous variables are presented as mean value and standard deviation (±SD), median, minimum and maximum values. Categorical ones are expressed by frequencies and percentages. To define accuracy of PF and STPF to predict study outcomes we used the receiver-operating characteristic (ROC) curve and compared the area under curve (AUROC) deriving from the use of conventional P/F vs. STP/F ratio. Comparisons between AUROC of PF and STPF for the study outcomes were made by De Long's test [[26]DeLong ER DeLong DM Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach.] and Best threshold for the ROC analysis was calculated using the Youden index point [[27]Perkins NJ Schisterman EF. The inconsistency of "optimal" cutpoints obtained using two criteria based on the receiver operating characteristic curve.]. Categorical variables were analyzed using one-way analysis of variance (ANOVA) or χ2-square test, when appropriate. Associations between parameters were calculated using Spearman correlation test. P ≤  0.05 was considered statistically significant. Analysis was performed using IBM SPSS Statistics version 21. Using Buderer's formula, we empirically calculated a minimum sample size of 284 subjects to reach 70% of sensitivity and 70% of specificity [[28]Statistical methodology:I. Incorporating the prevalence of the disease into the sample size calculation for sensitivity and specificity.]. In previous studies the average prevalence of outcome failure and mortality, respectively, were 43,7% and 19,6%. [29Bonnet N Martin O Boubaya M Levy V Ebstein N Karoubi P et al.High flow nasal oxygen therapy to avoid invasive mechanical ventilation in SARS-CoV-2 pneumonia: a retrospective study., 30Chalmers JD Crichton ML Goeminne PC Cao B Humbert M Shteinberg M et al.Management of hospitalised adults with coronavirus disease-19 (COVID-19): A European Respiratory Society living guideline., 31Gupta A Nayan N Nair R Kumar K Joshi A Sharma S et al.Diabetes Mellitus and Hypertension Increase Risk of Death in Novel Corona Virus Patients Irrespective of Age: a Prospective Observational Study of Co-morbidities and COVID-19 from India., [33]Garufi G Carbognin L Orlandi A Tortora G Bria E. Smoking habit and hospitalization for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-related pneumonia: The unsolved paradox behind the evidence.].3. ResultsWe enrolled 349 consecutive patients. Characteristics of the study population and outcomes are described in Tables 1A and 1B. Outcome failure was observed in 113 patients (32,4%) and 58 patients died (16.6%). Median survival was 18.5 days (range 4-65, mean 21.0±13.4) and 13.0 days (range 0-65, mean 16.6±13.3) calculated, respectively, from symptoms start to date of death and from ER admission to date of death. All deaths were caused by acute respiratory failure due to Covid-19. Median duration of ARF was 23 days (range 2-58, mean 21.6±9.9).

Table 1A Characteristics of the study population and arterial blood gas analysis data at the time of Emergency Room (ER) admission.

Considering outcome failure, AUROC was 0.747; (95% CI 0.693-0.801) for STP/F and 0.742 for P/F (95% CI 0.687-0.797), with an advantage for STP/F comparing to P/F, but not statistically significant (p=0.59), as shown on Figure 1A. Analyzing only in-hospital mortality as outcome (Figure 1B), only AUROC of STP/F showed enough accuracy comparing to AUROC of PF (0.710; 95% CI 0.638-0.782 vs. 0.688; 95% CI 0.650-0.846); this difference was statistically significant, p=0.0189.

Figure 1Predictive receiver-operating characteristic (ROC) curve of the study population by outcome failure (A) and deaths (B) for STP/F and P/F.

PaO2, STPaO2 and PaCO2 showed not enough accuracy according to the ROC curve both by outcome failure and deaths (AUC<0.7). By outcome failure, PaO2, STPaO2 and PaCO2 showed AUC of, respectively, 0.533 (95% CI 0.491-0.616), 0.582 (95% CI 0.520-0.645), and 0.574 (95% CI 0.509-0.639). By deaths, AUROC of PaO2 was 0.599 (95% CI 0.520-0.677), AUC for STPaO2was 0.623 (95%CI 0.544-0.701) and AUC of PaCO2 was 0.617 (95% CI 0.532-0.702).

According to ROC analysis, the best cut-off for STP/F was, respectively, 170 for outcome failure and 125 for deaths. The best cut-off for P/F was, respectively, 180 for outcome failure and, even if AUROC was not enough accurate, 150 for deaths. Sensibility, specificity, positive predictive value and negative predictive value for STP/F and P/F are shown on Table 2. There were 146 subjects with STP/F ≤170 and 203 with STPF >170. Among subjects with sP/F≤170, outcome failure rate was 46,8% and morality rate 23.2%. In comparison outcome failure rate and mortality rate were, respectively, 12.3% and 7.5% for the subgroup with STP/F>170. These differences were statistically significant (p=0.000). There were 115 subjects with STP/F ≤125 and 234 with P/F>125. Among those with STP/F ≤125, outcome failure was 59.1% and mortality rate 33%. In comparison outcome failure rate and mortality rate were, respectively, 19,2% and 8,8% for the subgroup with STP/F >125. These differences were statistically significant (p=0.000).

Table 2Predictive power of standard PF and PF in the study population according to outcomes (SE = sensibility; SP = specificity; PPV=positive predictive value; NPV=negative predictive value).

Stratifying by deaths (Table 3) mean value of both P/F and STP/F showed statistical significant differences between groups: patient died due to Covid-19 ARF had a mean P/F 149.9±67.5, mean STP/F 129.8±58.4 versus, respectively, 197.3±83.3 and 181.0±81.7 of survivors subgroup (p=0.000 and p=0.000). No differences between groups were observed according to PaO2, while the subgroup of patients who died showed lower values of PaCO2 and STPaO2 compared to the survivor subgroup (pTable 3). Similar results were observed by outcome failure (need of IMV and/or deaths) with the exception of PaCO2 (Table 3). Duration of ARF (Table 4) was inversely associated with P/F and STP/F (p=0.000 and p= 0.000), but not with paO2, STPaO2 and paCO2. Age and Respiratory Rate at the time of admission were positively related with duration of ARF (p

Table 3Differences in terms of PaO2, PaCO2, standard PaO2 (STPaO2), P/F and STP/F by outcome failure and deaths in the study population (ANOVA).

Table 4Association between age, respiratory rate, PaO2, PaCO2, standard PaO2 (STPaO2), P/F, STP/F and ARF duration in days (Spearman's correlation)

4. DiscussionThe main finding of the current investigation is that accuracy of STP/F to predict death was higher than conventional P/F (0.710; 95% CI 0.638-0.782 vs. 0.688; 95% CI 0.650-0.846, p= 0.012), Figure 1 and Table 2. Interestingly, STP/F is accurate and superior to P/F in predicting in-hospital mortality, but not outcome failure (defined as deaths or need of IMV), as if the need of IMV is not affected by STP/F or P/F values. STP/F can predict mortality in all patients of our study population.Mean PaCO2 of the study population was inferior to 40 mmHg, and mean respiratory rate was 22±5 breaths per minute (Table 1B). This confirms the hypocapnic compensation of hypoxemia in Covid-19, mainly obtained by increase of tidal volume. Moreover, mean value of PaCO2 of the subgroup died for ARF due to Covid-19 was inferior to the one of the survivors subgroup (Table 3), p=0.037, as if low PaCO2 might suggest risk of further ARF worsening, even if AUROC curve of paCO2 is not enough accurate to predict outcome failure.Prevalence of never smokers in our study population was 85%; this reflects data emerged from literature about Covid-19 [[34]Lippi G Sanchis-Gomar F Henry BM. Active smoking and COVID-19: a double-edged sword., [35]Gattinoni L Chiumello D Caironi P Busana M Romitti F Brazzi L et al.COVID-19 pneumonia: different respiratory treatments for different phenotypes?.]. Smoking can modulate immunity reducing its effectiveness. Thus could result in a less reactive inflammatory response during Covid-19, preventing the cytokine storm responsible of the progression of the disease in ARF due to Covid-19 and explain the lower prevalence of current or former smokers in Covid-19 reported studies. [[34]Lippi G Sanchis-Gomar F Henry BM. Active smoking and COVID-19: a double-edged sword., [35]Gattinoni L Chiumello D Caironi P Busana M Romitti F Brazzi L et al.COVID-19 pneumonia: different respiratory treatments for different phenotypes?.]. We can speculate, in addition, that usually a fraction of current smokers or former smokers may be affected by chronic obstructive pulmonary disease (COPD). Having a respiratory chronic disease, COPD patient might have a more preventive social behavior strategy and respect strictly rules such as wearing masks and to respect physical distances. Moreover COPD patients could probably early recognize Covid-19 related respiratory symptoms and signs, leading them to have an early access to medical consultation and/or ER.

According to our findings, STP/F better describes ARF due to Covid-19 in its hypocapnic nature. Using STPaO2 instead of PaO2 (standard P/F versus P/F) better describes this phenomenon and could better relate to prognosis, in particular in-hospital mortality.

Defining all mechanisms responsible for ARF during SARS-CoV-2 pneumonia with one parameter is not simple, since pathophysiology of lung injury due to Covid-19 is multifactorial and the impact of every single compensatory mechanism varies between subjects and through the course of the disease. [36Simonson TS Baker TL Banzett RB Bishop T Dempsey JA Feldman JL et al.Silent hypoxaemia in COVID-19 patients., 37Goh KJ Choong MC Cheong EH Kalimuddin S Duu Wen S Phua GC et al.Rapid Progression to Acute Respiratory Distress Syndrome: Review of Current Understanding of Critical Illness from COVID-19 Infection., 38Treating hypoxemic patients with SARS-COV-2 pneumonia: Back to applied physiology.] In Covid-19, inflammation and oedema in alveoli are the main responsible of hypoxemia in the early phases of disease, so that P/F reasonably relate to severity of diffusing impairment here. With the progression of disease (consolidation phase), V/Q mismatch and shunt mechanism become prevalent, so that hypoxemic ARF becomes less responder to implementation of FiO2 due to incapacity to improve PaO2 in non-ventilated alveoli. In lung regions where shunt is prevalent, P/F could be not so representative of severity of the disease. [[39]Covelli HD Nessan VJ Tuttle WK Oxygen derived variables in acute respiratory failure., [40]Tobin MJ Laghi F Jubran A. Why COVID-19 Silent Hypoxemia Is Baffling to Physicians.] Reducing partial pressure of carbon dioxide (PaCO2) represents a protective mechanism: low values of PaO2 increase minute ventilation in response to chemoreceptor stimulation. Hyperventilation is a feedback mechanism to correct hypoxia at the expense of PaCO2 reduction and left shift of the HbO2 dissociation curve. In this way, tachypnea and hyperpnoea, generated by a rise of minute ventilation through increasing respiratory rate and tidal volume, compensate both hypoxemia and prevent blood acidosis [[41]Copin MC Parmentier E Duburcq T Poissy J Mathieu D Lille COVID-19 ICU Group Anatomopathology Time to consider histologic pattern of lung injury to treat critically ill patients with COVID-19 infection.].Notably, the presence of microvascular thrombosis in subjects with Covid-19 ARF, highlighted by the increase of D-dimer and alveolar dead space; may contribute to the severity and progression of hypoxia observed in Covid-19 [[42]Khan M Adil SF Alkhathlan HZ Tahir MN Saif S Khan M et al.COVID-19: A Global Challenge with Old History, Epidemiology and Progress So Far.]. Several data showed that outcome is related to dead space through measurement of ventilatory ratio in typical ARDS and in Covid-19 [[3]Grasselli G Tonetti T Protti A Langer T Girardis M Bellani G et al.Pathophysiology of COVID-19-associated acute respiratory distress syndrome: a multicentre prospective observational study.]. These measurements in subjects in spontaneous breathing are not obtainable, so that STPF could represent a surrogate of the ventilatory ratio.There is an urgent need to identify patients at higher risk of intubation and death, since de novo ARF plays a central role in Covid-19, being responsible for morbidity and mortality [[4]Li LQ Huang T Wang YQ Wang ZP Liang Y Huang TB et al.COVID-19 patients' clinical characteristics, discharge rate, and fatality rate of meta-analysis.,[43]Lai X Liu J Zhang T Feng L Jiang P Kang L et al.Clinical, laboratory and imaging predictors for critical illness and mortality in patients with COVID-19: protocol for a systematic review and meta-analysis.-43]. In addition, defining the best setting where to allocate patients affected by SARS-CoV-2 pneumonia could play a central role in this emergency era for health care resources worldwide. Finding a parameter which could help clinicians to detect early which patient will need more resources, in particular the need of respiratory support and so Pulmonology Unit hospitalization, may optimize Covid-19 outcomes and improve costs-benefits ratio.This is the first study assessing the role of standard paO2 in relationship to prognosis in acute respiratory failure: this pilot study identifies STPF as a better predictor of mortality than PF in Covid-19 ARF. We propose the use of STP/F because, from a pathophysiological point of view, it better describes the compensatory mechanism present in hypoxemic ARF typical of Covid-19 and our study showed that is more accurate in discriminating prognosis. STPaO2 is a parameter obtainable simply in standard practice using a formula validated since years [[17]An arterial blood gas diagram for clinical use.].Limits of this study are its observational nature and the short enrollment phase due to its pilot nature. Moreover it does not take into account patients with ARF due to Covid-19 admitted directly from ER to ICU; this could explain the relatively low outcome failure and mortality ratio seen in our study (respectively 32.4% and 16.6%). However, outcome failure, as defined by need for invasive mechanical ventilation and/or death, was online with previous literature describing patients outside ICU setting [29Bonnet N Martin O