E1760. Outcome Predictive Performance of Admission Chest Radiographs in Hospitalized COVID-19 Patients: A Retrospective Cohort of 240 Patients
  1. Russell Reeves; Thomas Jefferson University
  2. Corbin Pomeranz; Thomas Jefferson University
  3. Andrew Gomella; Thomas Jefferson University
  4. Aishwarya Gulati; Thomas Jefferson University
  5. Brandon Metra; Thomas Jefferson University
  6. Anthony Hage; Thomas Jefferson University
  7. Baskaran Sundaram; Thomas Jefferson University
Defining the utility of CXR in evaluating patients with COVID-19 is still in its nascent stage, particularly for admitted patients with COVID-19 pneumonia. The purpose of this study was to determine the prognostic value of a CXR severity scoring system in patients admitted with COVID-19 pneumonia. Outcomes of interest included intubation, need for chronic renal replacement therapy (CRRT), extracorporeal membrane oxygenation (ECMO), and death.

Materials and Methods:
Patients admitted to an urban multicenter health system from March 16 to April 13, 2020, with COVID-19 confirmation on real-time reverse transcriptase-polymerase chain reaction (RT-PCR) were retrospectively identified. Admission CXRs were independently graded by three cardiothoracic radiologists and three diagnostic radiology residents on a 0 to 24-point scale based on the extent and severity of COVID-19 pneumonia. Demographic variables, clinical characteristics, and admission laboratory values were collected. Inter-rater reliability among attending, resident, and the combined group of graders was assessed with Kendall's W statistic. Multivariate logistic regression was used to evaluate the relationship between patient characteristics, laboratory values, CXR severity score, and clinical outcomes.

A total of 240 patients met inclusion criteria: 142 males (59.2%), median age 65 [50-80]. Inter-rater reliability for CXR scoring was high among the attending, residents and combination of graders (Mean 0.686 ± 0.051). There was no difference in outcomes across gender, race, ethnicity, or those with either a history of lung cancer or COPD. Average CXR severity was an independent predictor of death (OR: 1.17, 95% CI 1.10-1.24, p<0.001). CXR severity was a non-independent predictor of CRRT (OR: 1.15, 95% CI 1.04-1.27, p=0.007) and intubation (OR: 1.14, 95% CI 1.08-1.21, p < 0.001) but not ECMO.

In this study, we developed an accurate and reliable tool for classifying COVID-19 severity, which can be used both at the attending chest radiologist and junior resident level. This study identifies the laboratory, clinical and radiographic data that predict important patient outcomes such as death, intubation, and the need for CRRT. These findings are supported by prior work, validating the utility of CXR for patient prognostication, while adding new insight into COVID-19 infection during the initial presentation. In the setting of a high pretest probability of COVID-19 infection or with a quick turnaround of the rapid RT-PCR COVID-19 test, this chest x-ray scoring system may be used prospectively to predict patient outcomes, although future prospective validation is warranted.