key: cord-0270835-xyzb6lnv authors: Kotok, D.; Rivera Robles, J.; Girard, C.; Shettigar, S.; Lavina, A.; Gillenwater, S.; Kim, A.; Hadeh, A. title: Chest X-ray Severity and its Association with Outcomes in Patients with COVID-19 Presenting to the Emergency Department date: 2021-11-01 journal: nan DOI: 10.1101/2021.10.31.21265672 sha: bcd432b1b95ce5d3e7c09da37e48fe4e7c04530f doc_id: 270835 cord_uid: xyzb6lnv Background: Severity of radiographic abnormalities on chest X-ray (CXR) in patients with COVID-19 has been shown to be associated with worse outcomes, but studies are limited by different scoring systems, sample size, patient age and study duration. Data regarding the longitudinal evolution of radiographic abnormalities and its association with outcomes is scarce. We sought to evaluate these questions using a well-validated scoring system (the Radiographic Assessment of Lung Edema [RALE] score) using data over 6 months from a large, multi-hospital healthcare system. Methods: We collected clinical and demographic data and quantified radiographic edema on CXRs obtained in the emergency department (ED) as well as on days 1-2 and 3-5 (in those admitted) in patients with a nasopharyngeal swab positive for SARS-CoV-2 PCR visiting the ED for COVID-19-related complaints between March and September 2020. We examined the association of baseline and longitudinal evolution of radiographic edema with severity of hypoxemia and clinical outcomes. Results: 870 patients were included (median age 53.6, 50.8% female). Inter-rate agreement for RALE scores was excellent (ICC = 0.84, 95% CI 0.82 - 0.87, p < 0.0001). RALE scores correlated with hypoxemia as quantified by SpO2-FiO2 ratio (r = -0.42, p < 0.001). Admitted patients had higher RALE scores than those discharged (6 [2, 11] vs 0 [0, 3], p < 0.001). An increase of RALE score of 4 or more was associated with worse 30-day survival (p < 0.01). Larger increases in the RALE score were associated with worse survival. Conclusions: The RALE score is reproducible and easily implementable in adult patients presenting to the ED with COVID-19. Its association with physiologic parameters and outcomes at baseline and longitudinally makes it a readily available tool for prognostication and early ICU triage, particularly in patients with worsening radiographic edema. The significant morbidity and mortality associated with the COVID-19 pandemic has resulted in expedited research and development of preventive, diagnostic and therapeutic strategies to decrease the incidence and severity of disease associated with SARS-CoV-2 (1, 2) . The ability to provide accurate prognosis at time of diagnosis remains limited and relies mostly on baseline patient characteristics such as age and known comorbidities (3) (4) (5) . Combined radiographic, biomarker and artificial intelligence methods are emerging to assist with more accurate prognostication and identification of biologic phenotypes (6, 7) . Of the available diagnostic tools, plain chest radiography remains a simple and readily available tool that is accessible even in low-resource areas, some of which have been severely affected with extreme rates of infection and mortality (8) . The Radiographic Assessment of Lung Edema has been shown to be accurate and reliable between observers with baseline assessment of CXR in patients with COVID-19 and non-COVID-19 acute respiratory distress syndrome (ARDS) (9) (10) (11) . Although baseline radiographic edema has been shown to be associated with worse outcomes, whether the longitudinal evolution of radiographic edema as quantified by the RALE score is predicative of need for intensive care unit (ICU) admission, mechanical ventilation and 30day survival remains largely unexplored. We sought to independently evaluate the utility of the RALE score in a multi-center cohort of adult patients diagnosed with COVID-19 early in the pandemic and assess its association with severity of hypoxemia at baseline, to assess its ability to predict clinical outcomes at time of ED evaluation and in the early posthospitalization period in those patients requiring hospital admission. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. From March 2020 -October 2020, we collected data from the electronic medical records within the Cleveland Clinic system from 18 different clinical sites. We included all patients 18 years or older diagnosed with COVID-19 using nasal swab PCR. Patients with prior tracheostomy or those for whom the duration of symptoms associated with COVID-19 were unclear were excluded. Cleveland Clinic Institutional Review Board (IRB) exempted the study from IRB approval (FLA 20-038). We collected baseline clinical data obtained at the time of emergency department (ED) visit (including demographics, comorbidities, physiologic and laboratory variables). We collected chest X-rays at time of ED evaluation. For patients admitted to the hospital, we also collected chest X-rays at two additional time periods: early (1-2 days) and late (3-5 days) after admission to the hospital. Fraction of inspired oxygen (FiO2) was calculated using 1 liter per minute (lpm) supplemental O2 to 3% FiO2 conversion and capped at 15 lpm O2. We examined the association of radiographic edema with hypoxemia, need for hospital admission, ICU admission, need for mechanical ventilation within 7 days of admission and 30-day mortality. For those admitted, we examined the association of worsening radiographic edema with 30-day survival and need for mechanical ventilation. We quantified the radiographic edema by calculating the RALE score as originally described by Warren and colleagues (9) Briefly, the RALE score assesses two aspects of pulmonary edema in four quadrants divided horizontally by a line through the first branch of the left main bronchus and vertically by the spinal column: consolidation extent (quantified as percent of CXR quadrant involved: 0% = 0; <25% = 1; 25-50% = 2; 50-75% = 3; >75% = 4) and density (1 = hazy; 2 = moderate; 3 = dense). Edema extent and density are multiplied for each quadrant and the total RALE score is obtained by addition of all quadrant scores (0-48) ( Figure 1 ). Two independent reviewers (DK and SG) scored all CXRs. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. We assessed interrater agreement (DK, SG) using the interclass correlation coefficient (ICC) with a two-way mixed agreement model (12, 13) . We compared baseline characteristics using Wilcoxon signed-rank and Fisher's exact tests for continuous and categorical variables, respectively. We applied Pearson's correlation test to examine for baseline association between the RALE score and hypoxemia, i.e., S:F ratio (ratio of pulse oximetric saturation [SpO2] and fraction of inspired oxygen [F]). To derive the fraction of inspired oxygen, for every liter per minute (lpm) of inspired supplemental oxygen the fraction was increased by 0.03 (3%), capping at 15 lpm after which F was set to 1. We divided the distribution of RALE score observations into four quartiles (Q1: 0 -0, Q2: 1 -2, Q3: 3 -7, Q4: 8 -32), and then examined associations of RALE quartiles with categorical clinical outcomes. We used Cox proportional hazard models adjusting for age, severity of hypoxemia and history of diabetes to examine the effects of baseline RALE scores on 30-day survival and need for intubation within 7 days of admission. To assess for the impact of radiographic edema worsening on outcomes, we considered different thresholds of RALE score increase from baseline CXR and then compared outcomes for patients with vs. without worsening edema. All statistical analyses were performed with the R statistical software, version 4.1 (14). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) 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 November 1, 2021. ; https://doi.org/10.1101/2021.10.31.21265672 doi: medRxiv preprint We reviewed the electronic medical records of 949 patients based on eligibility criteria. 21 patients did not have CXRs done on ED evaluation, while 58 patients had incomplete baseline and/or outcome data ( Figure 2 ). A total of 870 patients were subsequently included in the study (Table 1) . 376 (43%) patients had a normal CXR (RALE = 0). We further examined baseline characteristics of patients based on RALE quartiles (Table S1 ). (which was not certified by peer review) 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 November 1, 2021. ; https://doi.org/10.1101/2021.10.31.21265672 doi: medRxiv preprint A total of 1185 of CXRs were reviewed from different study periods. Inter-rate agreement between the two reviewing clinicians (DK, SG) for the RALE score was good (interclass correlation coefficient = 0.84, 95% CI 0.82 -0.87, p < 0.001). Patients in the higher RALE quartiles were more likely to be male, older, have history of obstructive sleep apnea, heart failure, hypertension, hyperlipidemia, chronic kidney disease, lung disease, diabetes and immunosuppression (p < 0.05, Table S1) but not tobacco use (p = 0.28). Ethanol consumption was less common in the higher RALE quartiles (p = 0.002). They were also more likely to be tachypneic, have lower hemoglobin levels and higher neutrophil-to-lymphocyte ratios. At baseline, there was a significant association between severity of hypoxemia and RALE score (Pearson's r = -0.42, p < 0.001, Figure S1 ). The RALE score was significantly higher in patients requiring hospitalization (2 [6] [7] [8] [9] [10] [11] ) compared to those who were discharged home (0 [0-3], p < 0.001) ( Figure 3A ). In those patients requiring admission, a higher RALE scored was observed in those requiring ICU admission (4 [8] [9] [10] [11] [12] [13] [14] ) compared to those that did not (5 [0-9], p < 0.001) ( Figure 3B ). A higher baseline RALE score was associated with unadjusted 30-day mortality (5 [8] [9] [10] [11] [12] [13] [14] [15] vs 0 [1] [2] [3] [4] [5] [6] , p < 0.001) and need for mechanical ventilation (8 [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] .56], p = 0.003 and p = 0.0003, respectively) ( Figure S2 ). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Longitudinal evolution of radiographic edema in hospitalized patient and its association with outcomes Among hospitalized patients, 408 had baseline, 91 had early-period and 123 had late-period CXRs. There were significant differences between baseline and early period (6 [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] vs 14 [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] , p < 0.001) as well as baseline and late (6 [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] vs 11 [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] , p < 0.001) period RALE scores. There was no difference between the early and late period (p = 0.51) ( Figure S3) . A 4-point increase in RALE score or more from baseline to early period was associated with worse 30-day survival (HR 2.1 [1-2 -3.5], p = 0.006) (Figure 4 ). This was also true for increase from baseline to late (HR 2.7 [1.7 -4.2], p = 0.002. Further increase in the RALE score was associated with worse survival ( Table 2 ). Although we did not observe identical trends in need for mechanical ventilation, a larger increase in the RALE score from baseline to early (>=6) or a later increase in the RALE score of >=4 (baseline to late) was associated with higher likelihood of requiring mechanical ventilation (HR 2. Table 2 ). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Table 2 . Cox regression analysis for progressive increases in RALE score, 30-day survival, and 7-day probability of not requiring mechanical ventilation. A RALE score increases of at least 4 points from baseline to either early or late periods was associated with worse 30-day survival. A larger early increase (baseline to early by 6 or more) or a later (baseline to late by 4 or more) increase in the RALE score was associated with higher probability of requiring mechanical ventilation. All analyses were done after adjustment for baseline hypoxemia, age and history of diabetes. In this multicenter, retrospective, broadly inclusive study of 870 patients presenting to the emergency department with symptoms concerning for and diagnosed with COVID-19 using nasal-swab PCR between March and September 2020, we examined the association of radiographic edema on CXRs using the RALE score with physiologic parameters and clinical outcomes. Patients with worse radiographic edema as quantified by the RALE score on admission were more likely to be hypoxemic, require hospital admission, ICU admission, mechanical ventilation, and higher 30-day mortality. In addition to these baseline associations, we were able to demonstrate that progressive worsening of radiographic edema over the course of hospitalization for those requiring admission conferred a higher risk of death and need for intubation. This risk increased with worsening of radiographic edema even after adjustment for variables commonly associated with worse outcomes in patients with COVID-19 such as history of diabetes, age, and baseline hypoxemia. The association between baseline radiographic edema and worse outcomes has been consistently demonstrated globally in a multitude of smaller studies using several scoring scores demonstrating an association between CXR severity and clinical outcomes (15) (16) (17) (18) (19) (20) (21) . Our study is the largest to date utilizing the RALE score, which has been previously shown to have high rates of inter-observer agreement across different diseases including pneumonia, ARDS and heart failure (10, 11, 21) . In addition to re-demonstrating previously reported data (i.e. worse radiographic edema being associated with worse outcomes) we were also able to demonstrate a very low-mortality in those with minimal radiographic edema (Q1 and Q2 with RALE score of 2 or less). This findings further reinforces the notion that screening CXRs in patients with asymptomatic or mild COVID-19 disease is not required (22) . However, if performed, a normal CXR at time of ED visit is an indicator of a favorable prognosis. The ability of CXR to identify patients who are at low risk for progression to severe disease while also providing prognostic value for those with moderate-severe (Q3 and Q4 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) 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 November 1, 2021. ; https://doi.org/10.1101/2021.10.31.21265672 doi: medRxiv preprint RALE score of 3 or more) radiographic edema provides a readily available tool at time of ED evaluation. Combined with deep-learning, radiographic evaluation using CXRs in COVID-19 has recently been shown to have an even higher predictive value compared to CXR or clinical data alone (7) . Although baseline radiographic edema has been repeatedly shown to be of predictive value, the value of longitudinal evaluation using CXR shortly after admission remains largely unclear. Small studies examining the evolution of CT abnormalities showed mixed results. This was likely due to a combination of heterogeneity in disease severity and timing of CT imaging (16, 23) . Studies examining the association of baseline and longitudinal evolution of radiographic edema on CXR in patients with ARDS showed mixed results. These outcomes are presumably due to non-pulmonary ARDS affecting outcomes and the variability in agreement between clinicians with regards to what is labeled as ARDS (11, 24, 25) . In the case of COVID-19 (with or without ARDS), where initial pathology requiring hospital admission is often confined to the lungs, we hypothesized that tracking radiographic edema may provide prognostic information outside of vital and clinical monitoring. The significant association between worsening radiographic edema with worse outcomes even after adjustments for confounding factors suggests that progression of COVID-19 pulmonary disease is adequately captured on CXR and periodic radiography may assist in prognostication in hospitalized patients with COVID-19. There are several limitations associated with our study. The time-frame during which the study was performed did not allow for adjustment for anti-viral treatments (such as dexamethasone and remdesivir) which have been shown to decrease severity of radiographic edema (26) as these were introduced later during the study period within the healthcare system (late August-September). We All rights reserved. No reuse allowed without permission. (which was not certified by peer review) 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 November 1, 2021. ; https://doi.org/10.1101/2021.10.31.21265672 doi: medRxiv preprint were also not able to adjust for any variables other than those obtained at baseline when performing the longitudinal analyses, raising the question of whether other organ failure outside of the pulmonary system (renal) and fluid status have contributed to radiographic deterioration that is not directly associated with COVID-19 pulmonary disease. Lastly, while the study is a multi-center study across multiple hospitals and states in the United States, practices and procedures early in the pandemic were often institution-based and not necessarily guided by the Centers for Disease Control and Prevention and/or the World Health Organization. As such, institutional standards of treatment may not have been reflective of other healthcare systems around the same time period. Despite these limitations, we excluded patients for which there was incomplete data that would otherwise require imputation and potentially affect our analyses. Instead, the study protocol structured the CXR sampling periods to provide early, standardized meaningful clinical data that may assist clinicians in prognostication and decision making in patients with COVID-19. In conclusion, we were able to demonstrate the pragmastim and reproducibility of the RALE score in patients with COVID-19 which showed significant associations between severity of radiographic edema and clinical outcomes both at baseline and longitudinally in patients hospitalized with COVID-19. The degree of progression as dictated by the RALE score was linked to worse outcomes. Furthermore, our study demonstrates that longitudinal assessment of radiographic edema may provide prognostic value and assist clinicians in prognostication in patients with COVID-19. The effect of antiviral therapy on these findings is unclear and requires further investigation. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Incomplete baseline or outcome data N = 58 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) 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 November 1, 2021. ; https://doi.org/10.1101/2021.10.31.21265672 doi: medRxiv preprint Figure 3 . Comparison of median RALE score by need for hospital and ICU admission. Patients req hospital admission had a higher median RALE score compared to those discharged home (p < 0.001) (P A). Similarly, in patients admitted to the hospital -those who required ICU admission had a higher media RALE score compared to those who did not (Panel B) (p < 0.001). 1 7 requiring (Panel dian All rights reserved. No reuse allowed without permission. (which was not certified by peer review) 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 November 1, 2021. ; https://doi.org/10.1101/2021.10.31.21265672 doi: medRxiv preprint Figure 4 . Kaplan-Meier estimates of 30-day survival comparing RALE score quartiles. We observe significantly lower likelihood of survival at 30 days when Q3 and Q4 were compared to Q1 (** p = 0.002 **** p < 0.001, respectively). There was no significant difference in survival when Q1 and Q2 were comp 1 8 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) 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 November 1, 2021. ; https://doi.org/10.1101/2021.10.31.21265672 doi: medRxiv preprint All rights reserved. No reuse allowed without permission. (which was not certified by peer review) 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 November 1, 2021. ; https://doi.org/10.1101/2021.10.31.21265672 doi: medRxiv preprint Critical care crisis and some recommendations during the COVID-19 epidemic in China Outpatient Management of Patients With COVID-19: Multicenter Prospective Validation of the Hospitalization or Outpatient Management of Patients With SARS-CoV-2 Infection Rule to Discharge Patients Safely Clinical characteristics of family-clustered onset of coronavirus disease 2019 in Jilin Province Association Between Hypoxemia and Mortality in Patients With COVID-19 Clinical characteristics of refractory COVID-19 pneumonia in Wuhan, China Biomarker-Based Classification of Patients With Acute Respiratory Failure Into Inflammatory Subphenotypes: A Single-Center Exploratory Study All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder Prognostication of patients with COVID-19 using artificial intelligence based on chest x-rays and clinical data: a retro1. Jiao Z et al. Prognostication of patients with COVID-19 using artificial intelligence based on chest x-rays and clinical data: a ret India's COVID-19 emergency Severity scoring of lung oedema on the chest radiograph is associated with clinical outcomes in ARDS The evolution of radiographic edema in ARDS and its association with clinical outcomes: A prospective cohort study in adult patients Early Changes Over Time in the Radiographic Assessment of Lung Edema Score Are Associated With Survival in ARDS Forming inferences about some intraclass correlation coefficients The Intraclass Correlation Coefficient as a Measure of Reliability Population risk factors for severe disease and mortality in COVID-19: A global systematic review and meta-analysis Clinical and radiological changes of hospitalised patients with COVID-19 pneumonia from disease onset to acute All rights reserved. No reuse allowed without permission. (which was not certified by peer review) 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 Chest X-ray severity index as a predictor of in-hospital mortality in coronavirus disease 2019: A study of 302 patients from Italy Chest x-ray severity score in COVID-19 patients on emergency department admission: a two-centre study Chest X-ray for predicting mortality and the need for ventilatory support in COVID-19 patients presenting to the emergency department Baseline chest X-ray in coronavirus disease 19 (COVID-19) patients: association with clinical and laboratory data COVID-19) infection: findings and correlation with clinical outcome Utility of Screening Chest Radiographs in Patients with Asymptomatic or Minimally Symptomatic COVID-19 in The value of repeated CT in monitoring the disease progression in moderate COVID-19 pneumonia: A single-center, retrospective study All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder The evolution of radiographic edema in ARDS and its association with clinical outcomes: A prospective cohort study in adult patients Accuracy of the chest radiograph to identify bilateral pulmonary infiltrates consistent with the diagnosis of acute respiratory distress syndrome using computed tomography as reference standard Dexamethasone may improve severe COVID-19 via ameliorating endothelial injury and inflammation: A preliminary pilot study