A More Accurate Measurement of COVID-19 Hospitalizations

While preventing infection was the initial focus of the COVID-19 pandemic response, with increasing population immunity and variant transmissibility, the current focus has shifted to reducing hospitalization and deaths, particularly in vulnerable communities (1). During the recent surge in disease activity driven by the Omicron variant, an increased proportion of “COVID-19 hospitalizations” were incidentally discovered infections in patients newly hospitalized for other reasons (2-6), resulting in decreased measurements of in-hospital disease severity and mortality compared to prior disease surges (6-9). However, estimates of the proportion of total COVID-19 hospitalizations accounted for by these incidental infections range widely from 15% to 68% (2-6), due to heterogeneity in case definitions for these incidental infections and variability across populations with respect to vaccination status and other risk factors for severe COVID-19.

We propose utilizing the CDC criteria for severe COVID-19, based on need for supplemental oxygen or oxygen saturation measured below 92%, to define incidental COVID-19 hospitalization (10). To study the impact of this case definition, we reviewed medical records of SARS-CoV-2 PCR positive patients admitted to LAC+USC Medical Center, a safety net hospital serving predominantly Latino and low-income patients in Los Angeles, California, during the local Omicron variant surge between 12/10/21 to 1/19/22. We abstracted data on age, vaccination and prior infection history, disease severity assessed by oxygen requirement, hospital length of stay, and mortality via retrospective chart review.

Using this case definition based on the CDC criteria for severe disease, 67.5% of SARS-CoV-2 PCR positive hospitalized patients would not have met criteria for a COVID-19 hospitalization. These patients had significantly lower median age (44 years vs. 57 years), median hospital length of stay (2 days vs. 3 days), and in-hospital mortality (3.5% vs. 14%) (see Table). While unadjusted analysis did not show significant association between exposure to vaccine or prior infection and incidental infection (OR=0.79 [0.53-1.17], p=0.24), exposure to vaccine or prior infection was associated with incidental infection upon adjustment for age using logistic regression (OR=0.58 [0.38-0.89], p=0.01).

The high frequency of incidental COVID-19 infection among hospitalized patients detected using the case definition based on lack of oxygen requirement exceeds the rates reported in previous studies that used more stringent case definition based on complete absence of COVID-19 symptoms (2) or were performed during periods of the pandemic prior to the Omicron variant surge (3). However, the high frequency of incidental COVID-19 is very similar to measurements based on case definition of severe COVID-19 (6) or correlates such as administration of steroid treatment (5) during the Omicron surge. Given that non-severe COVID-19 infections not requiring supplementary oxygen can generally be treated on an outpatient basis, we propose that incidental infections in hospitalized patients should not be considered as COVID-19 hospitalizations in public health statistics used to inform the public or make policy decisions. One caveat is that patients with non-severe COVID-19 are hospitalized at a higher rate than patients without COVID-19 (4), which may reflect non-respiratory complications of COVID-19 including thrombosis or multisystem inflammation or exacerbation of underlying chronic diseases, although these complications are often difficult to attribute directly to COVID-19 in individual patients. An updated case definition resulting in more accurate measurement of COVID-19 hospitalizations will facilitate more effective health policy and trust with the public.