ITHACA, N.Y.–(BUSINESS WIRE)–#hospitalriskmanagement–COVID-19 pandemic models designed to anticipate the need for medical resources including ICU hospital beds have overestimated the needs across the nation. A Louisiana Department of Health and Hospitals program manager uses Palisade’s @RISK decision support software to guide policy decisions with data probabilities that more accurately predict and align with what hospitals are experiencing.
Overestimates are rooted in uncertainty – unknown variables. Henry Yennie serves as a program manager for the Louisiana Department of Health and Hospitals. He supports the public health department for emergency planning and response. Yennie has relied on @RISK’s Monte Carlo simulations since Hurricane Katrina in 2005.
Yennie said that deterministic models based on “uncertain” variables have produced similar overestimates. Probabilistic, or stochastic, models from @RISK examine a range of possible outcomes and how likely each is to occur. He said deterministic “give me a number” models have been wildly inaccurate during the pandemic, sometimes by a factor of 10 or more.
“Some of these same models forecasted tens of thousands or more than 100,000 fatalities. @RISK estimated 90 percent probability range of 2,700 to 3,800 fatalities, which has proven to be aligned with what we are experiencing,” Yennie said. “I just quit paying attention to those other models.” Yennie worked with Palisade’s Custom Development team to build a model platform using @RISK and other products within the DecisionTools Suite following Hurricane Katrina. “The models have become key pieces of data our team provides to department and state leadership,” Yennie said.
Danger of “Give Me the Number”
While helping to support a statewide hospital evacuation plan in the aftermath of Katrina to prepare for the next crisis, Yennie witnessed how a “give me a number” approach from deterministic modeling can stray from actual results.
Chasing “give me the number” produced an estimate that became embedded in the institution and all evacuation plans. The methodology in the underlying assumptions, including overestimates, were forgotten but the final number remained. He said “the number” is dangerous because it leads to complacency in its acceptance that shapes important decisions.
“That was a defining moment for me in emergency management. I am not a statistician. One of things that attracted me to @RISK is that it is powerful tool for any level user,” Yennie said. “It gives us the capability to create models that can lend credibility to what we are asked to do.”
Following Hurricane Katrina, more than 13,000 missing persons were reported. “We were able to use Neural Tools to forecast and focus recovery teams on missing people who were likely still alive. Overall, we were over 92 percent successful in those estimates using these tools,” Yennie said.
Healthcare Risk Analysis
The healthcare industry is providing better management of resources and improving clinical performance. Prior to the COVID pandemic, a team from the Johns Hopkins Center for Health Security and representatives of the Mayo Clinic used @RISK to calculate pandemic supply needs based on operational considerations in hospitals.
Three years after Katrina, Yennie and his team used the @RISK model to estimate fuel needs for hospital generators – a problem that became apparent after Hurricane Gustav, which left hospitals without electricity, in some cases for weeks. “We were getting calls 24 hours a day, including one hospital that said they needed fuel in the next hour or children will die,” he said. The @RISK model allows for managing resources and prioritizing fuel needs to keep hospitals operational.
“In world of emergency preparedness, we seriously have to avoid the danger of locking in to ‘give me a number’ approaches,” Yennie said. “I need a range of probabilities that I can start with and plan downstream from there. The true scope of the challenge is difficult to look at if we are not using probabilistic methods.”
Click here to read the full article on Henry Yennie’s pandemic modeling in Emergency Management magazine. In addition, an on-demand version of his webinar “Emergency Preparedness Planning Before and During a Crisis” can be viewed here. Due to popular demand, Yennie will also be doing an additional pandemic planning webinar for the Florida Society of Healthcare Risk Management and Patient Safety on October 16th at 12:00PM EDT. For more details or to register, please click here.
With more than 150,000 users globally, Palisade Company provides the world’s largest comprehensive risk and decision analysis software platform. Combining over 35 years of expertise and partnerships with over 90% of the Global Fortune 100 across a multitude of verticals, Palisade Company is uniquely positioned to offer a full suite of analytics tools that includes:
- @RISK for Monte Carlo simulation of data
- PrecisionTree for decision trees
- TopRank for “what if” sensitivity analysis
- NeuralTools for predictive neural networks
- StatTools for statistical analysis & forecasting
- Evolver for sophisticated optimization
- RISKOptimizer for Monte Carlo simulation & optimization
Headquartered in Ithaca, NY, Palisade Company helps companies optimize their decision making in many industries that range from finance to oil and mineral exploration, real estate to heavy manufacturing and pharmaceuticals to aerospace, as well as top business schools around the world. For more information on Palisade Company, call 800-432-7475 or visit www.palisade.com.
Scott Yates, Onpoint Communications (for Palisade)
Global Director of Marketing, Palisade Company
Tel: +1 252 813 7277