CoVA: An Acuity Score for Outpatient Screening that Predicts Coronavirus Disease 2019 Prognosis

Abstract

Background We sought to develop an automatable score to predict hospitalization, critical illness, or death for patients at risk for COVID-19 presenting for urgent care. Methods We developed the COVID-19 Acuity Score (CoVA) based on a single-center study of adult outpatients seen in respiratory illness clinics (RICs) or the emergency department (ED). Data was extracted from the Partners Enterprise Data Warehouse, and split into development (n = 9381, March 7-May 2) and prospective (n = 2205, May 3-14) cohorts. Outcomes were hospitalization, critical illness (ICU or ventilation), or death within 7 days. Calibration was assessed using the expected-to-observed event ratio (E/O). Discrimination was assessed by area under the receiver operating curve (AUC). Results In the prospective cohort, 26.1%, 6.3%, and 0.5% of patients experienced hospitalization, critical illness, or death, respectively. CoVA showed excellent performance in prospective validation for hospitalization (expected-to-observed ratio (E/O): 1.01, AUC: 0.76); for critical illness (E/O 1.03, AUC: 0.79); and for death (E/O: 1.63, AUC=0.93). Among 30 predictors, the top five were age, diastolic blood pressure, blood oxygen saturation, COVID-19 testing status, and respiratory rate. Conclusions CoVA is a prospectively validated automatable score for the outpatient setting to predict adverse events related to COVID-19 infection.

The Clinical Course of COVID-19 in the Outpatient Setting: A Prospective Cohort Study

“…We found no major difference in the diagnostic yield between resting and ambulatory SaO 2 in this setting. The predictive value of outpatient oxygen saturation values has been previously described [ 5 , 24 ]. One study provided pulse oximeters to participants with COVID-19 who presented to an ED or outpatient testing center; 29% required subsequent hospitalization.…”

“…In addition, no single oxygen saturation reading alone predicted outcome, as there was overlap in both the resting and ambulatory oxygen saturations of those who remained at home and those who were hospitalized. Therefore, pulse oximetry may be most useful as an adjunct to clinical monitoring of populations over 40 years of age or at-risk populations [ 5 , 25 ].…”

“…In persons who are hospitalized, the clinical features of COVID-19 and disease course are well described [ 2–4 ]. However, most SARS-CoV-2-infected persons are not hospitalized, and relatively little is known about the progression of symptoms, clinical outcomes, and severity predictors among outpatients [ 5–7 ]. The prevalence and time course of unique clinical features of COVID-19 such as the occurrence of low oxygen saturations with a delayed patient sense of dyspnea, or “silent hypoxemia,” have yet to be fully characterized [ 8–10 ].…”

Development and validation of a prognostic COVID-19 severity assessment (COSA) score and machine learning models for patient triage at a tertiary hospital

“…Given the heterogeneity in symptoms on presentation and the potential outcomes, [7,8] it is not surprising that many of the tools proposed so far have relied on complex sets of parameters or specialized laboratory markers. The recently published COVID-19 Acuity Score (CoVA) developed from electronic health record (EHR) data of out-patients in the Boston area (n = 9381), for instance, contains 30 items, including presence of intracranial hemorrhage and hematological malignancy [9]. While it has been shown to predict hospitalization, critical illness, and death with accuracies of 0.76-0.93 for the area under the receiver operating characteristic (AUROC, ranging from 0 to 1 with a value of 0.5 indicating no class separation above randomness), only 15% (n = 1404) of the cases in the development cohort had a confirmed positive SARS-CoV-2 test.…”

“…Del Valle et al described correlation between prognosis and serum interleukin (IL)-6, IL-8, tumor necrosis factor (TNF)-α and IL-1β, which, though predictive, are expensive non-routine tests [10]. Several tools also rely on chest x-ray findings, some of which use machine learning (ML) to automatically classify digital images [9,11,12]. The value of chest x-rays, however, has been called into question, as no lesions specific for COVID- 19 have so far been identified, and images may appear normal despite pulmonary symptoms [13].…”