This model projects a fairly flat hospital census for COVID-19 patients in Erie County, despite some fluctuations, until social distancing is reduced.

UB Data Show Local COVID-19 Hospitalizations at Plateau

Published May 8, 2020

Executive director Peter Winkelstein, MD, and his team at UB’s Institute for Healthcare Informatics, are analyzing COVID-19 hospitalization data every day and sharing their conclusions weekly with the Erie County Department of Health (DOH) and the local hospital systems’ medical leadership.

“We are sitting on a powder keg. If we lift the lid too quickly, it blows up. But if we are careful and watch closely, we can safely open. ”
Executive director of UB’s Institute for Healthcare Informatics
Print

Peter Winkelstein, MD

The numbers indicate that Erie County reached a peak in COVID-19 hospitalizations when they hit 258 in late April.

In the context of the global pandemic and how it is playing out in Western New York, that statistic isn’t as grim as it first appeared, according to Winkelstein, a clinical professor of pediatrics and chief medical informatics officer of UBMD Physicians’ Group and Kaleida Health.

“It’s true that on April 27, our COVID-19 hospitalizations spiked to 258,” he says. “But overall, the level of Erie County hospitalizations from COVID-19 since then has been more or less flat.”

Collaborating With Local Health Care Leaders

The UB team includes faculty members, postdoctoral fellows and graduate students from the Department of Biomedical Informatics in the Jacobs School of Medicine and Biomedical Sciences, and from the School of Public Health and Health Professions.

It works in close collaboration with the DOH and other organizations in Erie County, including Kaleida Health, Erie County Medical Center, Catholic Health, UBMD, Independent Health Association and HEALTHeLINK.

“Based on data, modeling and science, our local health care systems have stepped up to the challenges created by the COVID-19 pandemic,” says Erie County Commissioner of Health Gale R. Burstein, MD, a clinical professor of pediatrics. “They have worked together to expand acute care and subacute care capacities to meet our community’s hospital needs.”

Models Provide Timely, Crucial Information

The key goal of the models that Winkelstein and his team develop is to forecast when and how quickly there might be a surge in COVID-19 hospitalizations so that hospitals and public health officials can be better prepared.

The models use an application developed by researchers Gabriel Anaya, MD, a trainee in the clinical informatics fellowship program, and Sarah G. Mullin, a student in the doctoral program in biomedical informatics, that epidemiologically models COVID-19 hospitalizations, intensive care cases, ventilator use and other factors.

“The challenges posed by this pandemic require collaboration and innovation from every unit in UB’s academic health center,” says Michael E. Cain, MD, vice president for health sciences and dean of the Jacobs School.

“While faculty physicians care for COVID-19 patients in partner hospitals, UB’s informatics experts are providing our region’s policymakers and hospital administrators with the information they need to make critical decisions in real time,” he says.

Best to Think in Terms of ‘Plateau,’ Not ‘Peak’

The UB models are based on premises drawn from data in the scientific literature that has so far been generated by the experiences of other areas with COVID-19.

The team looks closely at doubling time, which is the amount of time it takes for the number of COVID-19 hospitalized patients to double. Winkelstein says that after a fairly rapid increase in cases in March, doubling times have slowly decreased.

“We look at an average of the number of cases in the hospitals for the past five to seven days,” Winkelstein says. “There are daily fluctuations. But if you smooth it out, we are about flat. It might be trending up, but if the hospital census grows at all, it will do so much more slowly than we thought it would just a few weeks ago.”

That change in the data is driving a change in the terminology used by Winkelstein and his team. They talk of the “plateau” of cases in Western New York, rather than the “peak” of cases.

“We suggest that people stop talking about the peak,” Winkelstein says. “What’s happening here is that we have hit a plateau, but COVID-19 is not going away. We will be here for a while.”

Winkelstein says this model shows “the bullets we dodged by implementing strong social distancing.” The orange peaks signify the number of hospitalizations that would have been seen otherwise.

Maintaining Social Distancing Remains Key

This plateau is being seen in the number of COVID-19 hospitalizations and the number of COVID-19 patients in the intensive care units.

And while Winkelstein emphasizes that this is extremely good news, he notes that all of it is contingent on the continuation of social distancing.

“These data give us confidence that we are not following exponential curves, where we would see a surge in hospitalizations, which could blow up the health care system,” he says. “The major explanation is that social distancing in Erie County is working better than anyone expected.”

He notes that with a plateau, the hospitalizations will continue at a fairly constant pace so long as social distancing is maintained at the current rate.

“Our model shows we are very sensitive to social distancing,” he explains. “It shows that even a small change in social distancing may result in a large change to the hospital census, so it has to be watched very closely.” 

A model prepared by the UB team illustrated what might happen if people in the community reduced their social distancing by just 10 percent by either May 15 or May 31 without other precautions in place, such as mask wearing and widespread testing. 

“In both cases, we see a large increase in the hospital census,” says Winkelstein, with inpatient numbers potentially exceeding capacity in Erie County.

This graphic shows that while the number of people hospitalized with COVID-19 has drifted upward in the past few days, the general trend over the past two weeks has been toward a flattening.

Hospital Census Numbers Provide Best Window

“We know that the actions that we take today will be reflected in future hospitalization data,” Burstein says. “Continued social distancing is a vital non-pharmaceutical strategy that will slow community spread of SARS-CoV2 (the virus that causes COVID-19) until a safe and effective vaccine is available.”

So hospital census numbers are the critical piece to watch in the coming weeks, Winkelstein adds. “This is why we concentrate on hospital census data. The numbers of patients in the hospitals are the best window on what’s happening in the community and on revealing what proportion of people in the community with the virus end up in the hospital.”

Recent data show that there are about 1,500 hospital beds occupied in Erie County out of a total of approximately 2,800. The total of intensive care unit beds in the county is about 500 and currently, about 60 percent of those are occupied.

Winkelstein notes that the number of people on ventilators is still high, with about 75 percent of people in the ICU on ventilators, but there is still capacity in the ICUs.

The critical variable, he adds, is still social distancing.

“We are sitting on a powder keg. If we lift the lid too quickly, it blows up. But if we are careful and watch closely, we can safely open.”