Even though omicron appears to produce less-severe illness, on average, than earlier variants of the coronavirus, the staggering spike in cases is leading to surging hospitalizations, stressing the entire U.S. health-care system to the breaking point. But cases are far outstripping hospitalizations. In fact, the growing divergence between mostly mild cases and hospitalizations is causing many to focus on the latter as the soundest metric for the state of the pandemic.

But a new development is complicating the effort to track that figure. There are increasing reports in hospitals of what’s being called “incidental COVID”: instances in which people are admitted for other reasons — gastrointestinal bleeding, say, or cancer surgery — but then test positive for the virus, usually as part of a routine screening.

In late December, for example, Centers for Disease Control and Prevention Director Rochelle Walensky raised eyebrows when she suggested that “many children are hospitalized with COVID as opposed to because of COVID.” She was accused in some quarters of downplaying pediatric hospitalizations, but Anthony Fauci, President Joe Biden’s chief medical adviser, made almost identical comments regarding children. A recent New York Times article cited several New York hospitals that reported that 50 to 65% of their hospitalized COVID patients actually had incidental COVID.

It’s clearly important to distinguish people in the hospital “with” COVID from those in the hospital “for” COVID. If, in fact, 50 to 65% of hospitalized COVID patients are now there “with” COVID — including very mild or even asymptomatic cases — then hospitalization figures will vastly overstate the virus’s toll and over-predict the number of deaths. New York Gov. Kathy Hochul, a Democrat, recently announced plans to ask hospitals to specify how many patients were admitted with COVID. “I just want to always be honest with New Yorkers about how bad this is,” she said.

Unfortunately, there are no nationally accepted methods for disaggregating the “with” and “for” groups — and doing so, as we found when we looked at our own patients, is more difficult than it may first seem.

There are clear political dimensions to this issue: From the start of the pandemic, some people have downplayed COVID deaths, particularly those in patients with medical co-morbidities — and now see reports of incidental COVID as justification. The “it’s just a bad cold” crowd, meanwhile, is saying it’s another sign that it’s time to stop fretting about the virus and get on with our lives.

Neither argument cuts ice. For one thing, this issue has emerged because the pandemic has changed. Until recently, the vast majority of hospitalized patients testing positive for the coronavirus were there unambiguously because of COVID complications, mostly involving the lungs. Plenty of hospitalized COVID patients still fit that description, as we found when we took a close look this past week at the 50 or so hospitalized COVID patients in our large teaching hospital in San Francisco.

Using fairly strict criteria that included admission diagnoses for maladies like COVID pneumonia or respiratory failure, we determined that at least half our COVID patients were admitted “for” complications of their coronavirus infections. But beyond that, we learned that the distinction between “with” and “for” is surprisingly nuanced, and the fraction you get depends on the methods you use.

The complexity of the issue is easily seen when looking at specific patients. One man, in his 60s with a history of cardiovascular disease, clearly had been suffering from COVID at home. His primary symptom at first was a cough, but then he passed out and was brought to the hospital by ambulance. He was admitted with a diagnosis of cardiac arrest and in the hospital soon showed evidence of severe COVID pneumonia, which had probably precipitated his cardiac arrhythmia. While clinical notes all discuss his COVID infection, a cursory search of the diagnostic codes in the medical record might have led to a characterization of the patient as having incidental COVID (because cardiac arrest, not COVID, was the reason for his admission).

We have seen some patients in whom COVID was truly incidental. For example, our hospital has cared for several vaccinated teenagers who were admitted for non-COVID-related problems like appendicitis; they tested positive but were asymptomatic. Among adults, a man in his 60s with prostate cancer was admitted for bowel obstruction and kidney failure, and then tested positive for the coronavirus — another straightforward case of incidental COVID.

But even there, while characterizing his case as incidental would be clinically accurate, it might inadvertently minimize the impact of that patient — and others like him — on the health-care system. First, we don’t know how his COVID will progress. And every COVID patient, incidental or not, needs to be treated as infectious, which translates into time-consuming isolation procedures for hospital personnel.

In some cases, the question of whether a hospitalization is “for” COVID seems unanswerable. We have seen patients with sickle-cell disease admitted because they were in intense pain — yet they also tested positive for the virus. While diffuse body aches are a hallmark of a phenomenon called sickle-cell crisis, COVID can also cause severe body pain. And so, while we might be inclined to put these patients in the incidental COVID category, a relatively mild case of COVID may have triggered their sickle pain crisis. Similarly, another patient “with” COVID developed a blood clot in her brain. While she was not technically hospitalized “for” COVID, we know that COVID  increases the risk of blood clots.

Another new challenge with this highly contagious variant is patients who acquire the disease during their hospitalization. Such cases were very rare with prior, less transmissible, variants. But since omicron came onto the scene, hospitals are seeing more of them, despite aggressive infection control, testing and vaccination programs. A measure that hinges on admission diagnoses for COVID-related illness would characterize such patients as having incidental COVID, even though COVID will undoubtedly increase the length of their stay, and, for some, their risk of a bad outcome

Given this complexity, we at the University of California at San Francisco have struggled for a method to determine the fraction of hospitalized COVID patients who might correctly be characterized as having incidental COVID. When we initially filtered patients by COVID-related admission diagnosis, we came up with an estimate that 50% of our patients were there with incidental COVID. But, for some of the reasons described above, we concluded that this was probably an overestimate.

To try to get a more accurate picture, we added a search term to our analysis: whether the patient was given the antiviral medication remdesivir. Remdesivir is currently recommended for patients who have COVID pneumonia. (Based on a study published two weeks ago in the New England Journal of Medicine, we recently began giving a shorter course of remdesivir — three days instead of five — to patients with lesser symptoms who were at risk for severe progression. We did not add this group to our tally of people hospitalized “for” COVID.) Adding the therapeutic use of this drug to our search criteria, while excluding the patients who received three-day treatments, led to an estimate that approximately one-third of our COVID patients are truly incidental — significantly lower than the 50% estimate from our original search. Applying these same search criteria to patients in our hospital at the peak of last summer’s delta surge, we characterized about half as many (10 to 15%) as having incidental COVID.

Of course, tying the definition of “for” COVID to a medication creates its own problems, since new medications and guidelines will undoubtedly emerge, and physician judgment influences treatment choices. But given the interest in this subject — and its relevance for gauging the severity of any given coronavirus wave — we recommend that the CDC swiftly convene a group to establish a definition for incidental COVID, the first step toward creating a useful metric. Until we have a definition and a universal metric, the concept is likely to confuse people more than enlighten them — and to become partisan fodder. The distortions will occur in both directions. Overestimates of the incidental COVID rate will make the current wave seem milder than it really is. At the same time, undercounting incidental cases will lead to overstatements of COVID’s current severity. Both kinds of errors will cloud our vision at a time when we need a clear-eyed view of the pandemic.

Sara G. Murray is an associate professor in the department of medicine, and serves as associate chief medical information officer, at the University of California, San Francisco. Rhiannon Croci is a clinical informatics specialist on the health informatics data science and innovation team at the University of California, San Francisco Medical Center. Robert M. Wachter is a professor and chair of the department of medicine at the University of California, San Francisco.


Only subscribers are eligible to post comments. Please subscribe or to participate in the conversation. Here’s why.

Use the form below to reset your password. When you've submitted your account email, we will send an email with a reset code.