Federal Models Showed Up To 4,000 Kentucky Deaths If Social Distancing Were Abandoned
More than 4,000 people in Kentucky could die over the course of the coronavirus pandemic if social distancing and other mitigation tools were abandoned, according to a “best guess” model created by federal health officials in early April.
The virus has so far claimed 185 Kentuckians, and the state is on track for far fewer deaths than scenarios outlined in the federal models.
The model was included in documents prepared by the federal Department of Health and Human Services that were obtained by the Center for Public Integrity and shared with the Kentucky Center for Investigative Reporting. The documents also show models for neighboring states including Ohio (more than 11,000 deaths), West Virginia (more than 2,000 deaths) and Indiana (more than 6,000 deaths).
The models map out several scenarios the potential outbreak could take from the best case to worst case scenarios and then create a “best guess” or most-likely scenario. The documents say they aren’t meant to be read as predictions, but instead could be used to help public officials plan for different scenarios.
After the federal agency released state-level models to other agencies in early April, it doubled the fatality rate it used in its parameters. The impact on individual states using this new, higher death rate was not part of the documents provided to KyCIR.
Along with models of deaths and hospitalizations, the documents use data from symptoms reported by emergency rooms to predict potential coronavirus hotspots. Not all hospitals in the state participated in the data sharing behind the analysis, but three Kentucky counties made the list of potential future hot spots, as of early April: Campbell, Rowan and Fayette.
The number of cases in those three counties has risen since the models were created, although Fayette and Campbell already had a significant number of confirmed positives.
Gov. Andy Beshear has previously cited projections used by the White House which show that, with interventions such as social distancing and business closures, deaths could reach between 1,300 and 3,200 in Kentucky.
Beshear said at the time that those numbers were “optimistic.”
A separate prediction model by the Institute for Health Metrics and Evaluation at the University of Washington currently projects 407 deaths in Kentucky by August 4. This is significantly lower than the 815 deaths the institute had projected for Kentucky as of April 1, about two weeks into social distancing.
Beshear said during his daily briefing on Monday that the models have changed, thanks in part to the sacrifices Kentuckians have made to maintain social distancing.
“We are seeing a reduction, a flattening, a plateauing that we haven’t seen in the past because of what you are doing,” Beshear said. “What that tells me is that we’ve changed the model entirely.”
Kentucky’s Commissioner of Public Health, Dr. Steven Stack, said that the state uses all the available coronavirus models. “We use them all, we look at all the publicly available ones we can find, we use the ones that we can create internally, and the ranges are very wide, depending on the methodology.”
The governor’s office declined to answer questions about the models obtained by the Center for Public Integrity. It is unclear whether these models have been shared with states such as Kentucky. These documents have previously been unreported and the federal government has not yet made its own models public.
Teresa Waters, a professor and chair of Department of Health Management and Policy at University of Kentucky’s College of Public Health, says models like these shouldn’t be read as hard-and-fast predictions, but can add to our understanding of the virus and the paths it might take.
“The early models had really horrific projections of cases, deaths and overrunning hospital capacity and overrunning hospital capacity, but they also mapped out potentially different trajectories with social distancing,” Waters said. “As we get more experience with social distancing and we can see the impact in particular areas, we are sort of pruning off the really awful trajectories and… getting more data points, so we can see what path we seem to be on.”
Waters said the models can also help us understand the impact of mitigation actions such as closing businesses and issuing guidelines for social distancing. These models show the danger of re-opening the state too early as protesters in Kentucky and elsewhere across the country have begun pushing for governors to lift restrictions.
According to the available data, less than one percent of Kentucky’s 4.5 million people have tested positive for the coronavirus, Waters said. That means most people in the state are still at risk of contracting the virus and a hasty re-opening could change the projections yet again — this time, trending towards more deaths. “We’re all still susceptible, and re-opening could put us back on one of those paths that we pruned off of our model,” Waters said.
ER Visits Signal Hotspots
In addition to case scenarios, the federal HHS documents include potential future hotspots based on emergency-room data.
More than two weeks before areas see a sharp increase in COVID-19 cases, the documents explain, the disease’s presence can be signaled by emergency room visits by people presenting symptoms of the disease including fever, shortness of breath and coughing.
Kentucky’s statewide alert based on a rise in the number of ER visits with such symptoms was signaled on March 9, just days after the governor declared a state of emergency.
Waters of UK says it’s difficult to know how accurate a prediction model based on these criteria would be, because of disparities in who goes to the emergency room and who gets tested for the virus.
Health data show that Kentucky also has a high rate of chronic respiratory diseases, many with symptoms parallel to those of COVID-19.
“Symptoms are a very rough measure,” Waters said. “A lot of those people will not be COVID-19 positive.”
At the time the federal models were issued, Kentucky already had more than 1,300 confirmed coronavirus cases. Major urban areas in Jefferson and Fayette counties saw larger clusters, while Daviess, Hopkins and Kenton counties had more than 60 cases each. Fayette and two other counties made the HHS list of potential hot spots to come.
At the time, Fayette County had 188 cases. As of Wednesday it has 221 cases, the second-highest number of cases in the state. New cases appear to be leveling off in recent days.
Public health officials in both Rowan and Campbell counties say the number of positive cases are still relatively low, but they are prepared for any potential surge in cases.
“When you are looking at a model, you want to look at the worst-case scenario and make sure you plan for that. And if that doesn’t happen, that’s a good thing,” said Laura Brinson, the Public Health Impacts Administrator for the Northern Kentucky Health Department, which includes Campbell County. “You were prepared for the worst thing that could happen and if that doesn’t happen, you feel better for it.”
Campbell County currently has 59 positive cases and 7 deaths from the coronavirus. It had already recorded 41 cases when the models were made.
Greg Brewer, the public health director of the Gateway District Health Department that includes Rowan County, said that Rowan County seems to be the most likely county in his five-county district for a COVID-19 outbreak, because it is more populous and the Morehead State University campus usually puts people in close proximity. But with the university closed and with social distancing in place, the county has so far only seen two positive cases. Both of those cases came after the federal models were made. Brewer says the district is prepared for a sudden influx in cases, and he wouldn’t be surprised to see the number of cases rise as testing becomes more widespread.
Brewer said testing capacity has outpaced the demand under the current tiered system where tests are reserved for front line workers and those with severe COVID-19 symptoms.
“People are hunkering down and lying low whenever they get symptomatic,” Brewer said. Others are told by health officials they have symptoms matching COVID-19, but they aren’t serious enough to warrant testing for the virus. “Those are the things that kind of skew it a little bit,” Brewer said.
Contact Jared Bennett at firstname.lastname@example.org.