Early predictions underestimated how long patients with COVID-19 would need to stay in hospitals and how many would require intensive care, according to a study looking at California and Washington state.

Doctors wheeling patient through hospitalShare on Pinterest
Early predictions about COVID-19 and hospital treatment in the U.S. may have been wrong.

All data and statistics are based on publicly available data at the time of publication. Some information may be out of date.

A principal goal of physical distancing and lockdown measures during the COVID-19 pandemic is to prevent healthcare services from becoming overwhelmed.

In the United States, policymakers had to rely almost entirely on data from China, where the pandemic started, to inform their early estimates of how it would impact hospitals.

A study by researchers at the University of California (UC) Berkeley and Kaiser Permanente now suggests that this led to significant underestimates of the average time patients would stay in hospitals, how many would need treatment in an intensive care unit (ICU), and the case fatality risk.

“The hospital resources needed to meet the needs of severely ill patients are substantial,” says Joseph Lewnard, an assistant professor of epidemiology at UC Berkeley and lead author of the paper. “We found that observations from China may not provide a sufficient basis for anticipating the U.S. health care demand.”

The researchers tracked 1,328 confirmed cases of COVID-19 in hospitals run by Kaiser Permanente in Washington state and California up to April 9, 2020.

They monitored their length of stay, admission to ICU, and mortality rate.

Modeling estimates using data from hospitals in China usually assume that about 30% of hospitalized patients will require ICU care.

Of the patients in the U.S. hospitals, however, the probability of ICU admissions was 40.7%, and the probability of death was 18.9% for those with COVID-19 who doctors had admitted by April 9, 2020.

Similarly, in China, the average length of hospital stay among those who died was 7.5 days, whereas, in the U.S., the average stay was 11 days for survivors and 15 days for nonsurvivors.

A widely used modeling study from Imperial College London in the United Kingdom assumes an average hospital stay of 8 days. But the new study found that in the U.S., 25% of patients were in the hospital for 16 days or more.

The mortality rate among all patients in the analysis with completed hospital stays was 18%, with an increased risk of dying for males and older people.

The scientists report their findings in the BMJ.

While the reasons for the differences between the burden on hospitals in China and the U.S. remain unclear, the authors caution against reliance on models that researchers have based on the experience of other countries.

“The spread of COVID-19 and its impact on local healthcare systems show differences across the world,” says Vincent Liu, a research scientist at Kaiser Permanente in Oakland Northern California and an author of the paper.

“Healthcare systems differ, and their capabilities and structure have an effect on the local response and the impact of the surge. So, it’s really important to understand how our own data agree with or, in some cases, differ from the experience we’ve seen in other countries.”

In their paper, the researchers note that the severe disruption in healthcare services that follows surges in patients with severe COVID-19 — and that observers have seen in other regions — has spared the west coast of the U.S., thus far.

But they warn that unchecked, a surge in severe cases of COVID-19 could overwhelm hospitals, as happened in Italy and, more recently, in New York City.

The study also brought some good news. Extrapolations from the hospital data suggest that transmission rates leveled off after the implementation of physical distancing interventions in California and Washington state.

“When people engaged in protecting themselves and their communities through social distancing, their efforts translated into a substantial reduction in the transmissibility of the disease […] Those efforts are going to be critical for this next phase, in which social distancing measures are gradually relaxed. We need our communities to stay really engaged because these data show that even the actions of individuals and small groups can really impact the spread of the virus.”

– Vincent Liu.

The authors acknowledge two important limitations of their study.

First, they note that the outbreak did not overwhelm the capacity of hospitals in the region they studied, so their data are likely to reflect standard clinical practice. This may not be the case in regions where there was a more pronounced surge in admissions.

Second, all the subjects in the study had commercial health insurance, so they were likely to be more socioeconomically advantaged than uninsured patients.

The hospital data and estimates of the rate of transmission, therefore, may not reflect the situation in the wider population.

For live updates on the latest developments regarding the novel coronavirus and COVID-19, click here.