Each of these scores were designed to predict mortality and used to determine who can safely be sent home. A low risk CURB-65 score (0 or 1), confers a 0.6-2.7% risk of mortality.7 A low risk PSI/PORT score (<90) confers a 0.1-2.8% risk of mortality.8 Comparing the utility of the two, CURB-65 may not identify patients requiring ICU admission as well as the PSI. In addition, CURB-65 does not consider patients’ comorbidities (e.g. COPD), which may have a major impact on outcomes in COVID-19 patients. While CURB-65 is considerably faster to compute, with less inputs, this advantage matters less in the age of electronic records and resources. The PSI/PORT places a larger emphasis on age than CURB-65, assigning points by absolute age (i.e., a 70-year-old gets 70 points), which seems more consistent with what we know about COVID-19’s high mortality in the elderly.
In both of these cohorts, community acquired pneumonia was generally defined as a combination of clinical (e.g. fever, cough, dyspnea, rales) and radiographic (e.g. infiltrate on CXR) findings in the absence of risk factors for healthcare. Neither the CURB-65 or PSI/PORT studies differentiated between viral and bacterial pathogens as a cause for the pneumonia, although the incidence of viral-associated CAP may be up to 29%, with rhinoviruses and influenza being the most common.9-10
Recently, a clinical prediction tool was developed to risk-stratify patients specifically diagnosed with viral pneumonia - the MuLBSTA score.10 The aim of this was to predict clinical characteristics that affect mortality in patients with viral pneumonia. Interestingly, there are predictors of adverse outcomes that correlate with the clinical characteristics that are reported in COVID-19 patients. The presence of a multilobar infiltrate, low lymphocyte count, smoking history, and advanced age all were independent risk factors for mortality in this population - all relatively consistent with risk factors from our COVID cohort. But… before we get too excited, we must take note that this single-center, retrospective, not-externally-validated design may lead to bias and unknown applicability and generalizability.