Risk score predicts death, respiratory decompensation in suspected COVID-19
Click Here to Manage Email Alerts
A risk score based on patient comorbidities, obesity, vital signs, age and sex accurately predicted death or respiratory decompensation within 7 days among patients with pneumonia, suspected COVID-19 or confirmed COVID-19, data show.
The COVAS score could be used to help physicians “evaluate the millions of patients with acute symptoms” that resemble COVID-19, Adam Sharp, MD, MSc, an emergency medicine physician at the Kaiser Permanente Los Angeles Medical Center, and colleagues wrote in the American Journal of Emergency Medicine.
“Our score is designed to inform hospitalization decisions for patients evaluated in the ED or urgent care, or to triage those who may need labs drawn when resources are constrained, particularly during surges of infected patients,” Sharp told Healio Primary Care.
The researchers retrospectively analyzed 26,600 visits (derivation cohort = 21,280; validation cohort = 5,320) that adults made to 15 southern California EDs between March 1 and April 30. They used logistic regression models to analyze 1,079 adverse events that occurred among the participants. Based on this analysis, Sharp and colleagues identified 15 variables for inclusion in the risk score that were significantly linked to the primary outcome, a composite measure of death or respiratory decompensation within 7 days. These variables were then assigned a score point value.
Zero points were given for:
- BMI of 39.9 kg/m2 or lower; and
- systolic BP of 106 mm/Hg to 115 mm/Hg earn 0 points.
One point each was given for:
- diagnosis of congestive heart failure within the past 12 months;
- diagnosis of coagulopathy in the past 12 months;
- diagnosis of diabetes within the past 12 months; and
- 110 heartbeats or more per minute.
Two points each were given for:
- being aged between 50 and 59 years;
- diagnosis of dementia, seizures and/or dysphagia within the past 12 months;
- diagnosis of malnutrition or a protein deficiency in the past 12 months;
- BMI of 40 kg/m2 or higher;
- systolic BP mm/Hg of 105 or higher;
- oxygen saturation rate of between 93 SpO2 and 94 SpO2;
- fever; and
- breaths totaling 20 to 24 per minute.
Three points each were given for:
- being aged 60 years and older; and
- diagnosis of a fluid and electrolyte disorder in the past 12 months.
In addition, five points were given for breaths totaling 25 or more per minute and seven points were given for having an oxygen saturation rate of 92 SpO2 or higher.
COVAS scores can range from zero to 34, according to the researchers. Patients in the derivation cohort with a COVAS score of 5 or less had a 1.5% or less risk for an adverse event, while patients with a score of 12 or higher had a 15% or more risk for an adverse event. Similar rates were observed for the validation cohort.
Sharp and colleagues wrote that COVAS yielded an area under the curve (AUC) of 0.891 (95% CI, 0.880-0.901) in the derivation cohort and 0.895 (95% CI, 0.874-0.916) in the validation cohort. The sensitivity of COVAS ranged from 100% (score = 0) to 41.7% (score = 15 or higher), while specificity varied from 13.9% (score = 0) to 96.8% (score = 15 or higher).
Also, among a subgroup of 3,252 patients with pneumonia, the AUC for COVAS was 0.780 (95% CI, 0.759-0.801) in the derivation cohort and 0.832 (95% CI, 0.794-0.870) in the validation cohort. Among another subgroup of 2,059 patients with COVID-19, the AUC in the derivation cohort was 0.867 (95% CI 0.843-0.892) and in the validation cohort it was 0.837 (95% CI 0.774-0.899).
Sharp said using ED patients instead of those who are hospitalized to calculate the risk score “highlighted the unique contribution of COVAS to other COVID-19 related scores.”
“The level of care and the type of care available for patients admitted to the hospital is obviously very different from those who are sent home from the ED,” he continued. “Most patients are sent home after an ED evaluation, even if they are sick with COVID-19. A score designed to inform physician-patient decisions about hospitalization must include all patients, not exclude the majority of patients who are sent home.”
Perspective
Back to Top
It is always interesting to see which variables researchers use to develop risk scores and review how easily translatable the findings are in different settings. Looking at Sharp and colleagues’ study, methodology of the study is largely sound and valid.
I am interested in which vital signs and comorbidities were included in the final model. The model they developed is similar to other prediction models published by other research groups, including here at the Cleveland Clinic. The findings of this study are intriguing but are not really a game changer in terms of identifying which patients health systems should or should not admit.
Associate program director of the critical care fellowship, Cleveland Clinic
Assistant professor of medicine, Lerner School of Medicine of Case Western Reserve University
Collapse
Read more about
Click Here to Manage Email Alerts