Analytics help hospitals in states with surging COVID-19 volumes

Houston Methodist is in the midst of a massive COVID-19 surge, seeing 50% more patients than at the height of its previous surge in the spring and on track to reach more than double that peak.

“We’re probably heading up to at least two to three times what our initial surge was,” said Roberta Schwartz, Houston Methodist’s executive vice president and chief innovation officer. “We are definitely surging.”

The health system is based in Texas, which is one of more than half of states that are experiencing an increase in COVID-19 cases. The state’s governor on Thursday issued an executive order suspending surgeries at hospitals in four counties in an effort to ensure there’s capacity for COVID-19 hospitalizations.

Hospitals are increasingly using predictive models to forecast regional COVID-19 surges and demand for healthcare services, said Gurpreet Singh, health services leader at consulting firm PwC. While many of the predictive models that hospitals used in early stages of the pandemic focused on COVID-19, they’re now adjusting them to forecast demand for other services as they re-open for non-emergency care.

Singh estimated that, before the pandemic, about a quarter of hospitals used analytics for predicting demand. Now, he said he’s seen it at least half of hospitals he’s worked with recently.

Predictive analytics have helped Houston Methodist leadership better allocate resources across its eight hospitals as it prepares to care for even more COVID-19 patients.

By looking at its emergency department, admission volumes, and what proportion of those patients test positive for COVID-19, Houston Methodist has been able to identify which hospitals are in communities with a growing number of COVID-19 cases and forecast how high that next surge might be.

The forecasts stem from crunching numbers and creating dashboards on data visualization software, Schwartz said.

“I have a whole new set of COVID dashboards,” she said. “Things that didn’t exist six months ago.”

Renown Health in Reno, Nev., is leveraging its existing data efforts to help predict capacity and patient outcomes.

The health system has experienced an uptick in hospitalizations in recent weeks, but still has “considerable capacity available,” said Dr. Anthony Slonim, Renown Health’s president and CEO. That’s in part because the health system converted two floors of its parking garage into an alternative care site with about 800 beds, doubling its number of inpatient beds in the process.

The Renown Institute for Health Innovation, a collaboration between Renown Health and the Reno-based Desert Research Institute, launched the Healthy Nevada Project in 2016. The population health study is designed to better understand how clinical, genetic, social and environmental factors can predict patients’ risks for various conditions.

More than 50,000 people in Nevada have volunteered to share DNA samples with the project since it opened.

By linking COVID-19 test results to data collected from the project, Renown researchers are working to build models that predict COVID-19 length of stay to better inform hospital capacity. They’re also studying genetic factors that put patients with COVID-19 at risk for severe illness and long-term complications, Slonim said.

“Data are fundamental to running a healthcare organization,” Slonim said.

Leaders at Northwell Health credit a predictive model that the health system built with helping to inform where to increase capacity and redistribute equipment and staff as they worked through New York’s COVID-19 surge. Now that cases are decreasing in the state, Northwell Health is building on the model so it will be able to determine when a resurgence might be taking place.

The New Hyde Park, N.Y.-based health system is down to roughly 270 patients with COVID-19, significantly less than 3,600 inpatients at its peak in April.

“We’re starting to get back towards normal, (but) with the sense that we still have to be very, very careful,” said Dr. Mark Jarrett, Northwell Health’s chief quality officer. That’s where data and analytics comes in, he said. “You need (data for prediction), because otherwise, you’re working in the blind.”

Predicting COVID-19 surges tends to rely on tracking data on patients who test positive for COVID-19, as well as publicly available data about COVID-19 cases in the region.

To more precisely predict hospitalizations, hospitals could consider integrating data that point to people’s behavior in their region—such as using aggregated location data from cell phones to illustrate trends on which areas are abiding by social distancing recommendations, said Ash Shehata, national sector leader for healthcare and life science at consulting firm KPMG.

“There’s going to be a bigger and bigger portion of third-party data that sits outside of a health system … that organizations are going to have to bring into their calculations,” Shehata said.

He added that it’s important to model demand for during and after a surge as a hospital re-opens for non-emergency care and elective procedures. Large health systems with facilities in multiple states may face particular challenges predicting patient behavior, as it may vary significantly between markets.

“A lot of the decision points here are not clinical, but they’re consumer,” Shehata said. “It’s about how comfortable are the consumers in resuming their healthcare.”


Tags: covid-19, pandemic

Thanks! You've already liked this