SIMULATION: How the rural healthcare crisis could affect East Tennessee's readiness for coronavirus

East Tennessee is already suffering from a shortage of hospital beds and - depending on a number of variables - the coronavirus could test the limits of that system. Click here if you want to go straight to the simulation. IMPORTANT: The simulation displays much better on desktop than mobile.

Background

Disclaimer: I am not a medical doctor and have no background in epidemiology or virology. Additionally, I have absolutely no clue how COVID-19 will spread across East Tennessee. Below there is a simulation and the results of numerous simulation runs. It is not meant to be predictive. It is only meant to show the user how different potential outcomes could affect our healthcare system. I have no clue which outcome is most the likely, but many of the possible outcomes demonstrate problematic numbers for our healthcare system.

Since 2010, 13 hospitals have closed across the state of Tennessee, three of which were in East Tennessee. The loss of these hospitals have had many complex negative effects on the individuals they served, but, more simply, it is has also decreased the inventory of staffed hospital beds in the region.

In rural East Tennessee, there is around 1.3 staffed hospital beds for each 1,000 people. This rate is well below the national average of 2.8 beds per 1,000 people. Even if you add Knox County, which comes in at 3.3 beds per 1,000, the regional average is still more than a half point below the national average or 2.2 beds per 1000 people.

Most of the time these numbers - though low - are fine from an available inventory standpoint. In the graph below, you can see the average number of available (or not used) hospital beds for each county in 2018.

First, it is important to note that the residents of Grainger, Morgan and Union counties do not have a single hospital in their counties, which is why the available beds is at zero. Beyond those three counties, most counties generally have between 20 and 50 beds available on most days. Knox county is the outlier, averaging 376 open beds. [Note: If you want to explore this data yourself and see more information about the hospitals in your community, you can final all the above information and a lot more in the Joint Annual Report produced by the Tennessee Department of Health.]

The amount of open beds in our region might serve the need on a daily basis, but it does not provide for much wiggle room. A surge from a major outbreak of COVID-19 in our region could overwhelm our system.

The COVID-19 virus is a type of coronavirus, which is the same type of virus that causes the common cold. The problem is it is believed to be much more severe - and more deadly - than the common cold. Specifically, it causes lower respiratory tract infections - as opposed to the upper respiratory issues of the cold - and can quickly turn into pneumonia.

Though East Tennessee has not had a confirmed case of COVID-19 yet, it is only a matter of time. It is pretty clear that we, as a nation, have moved from a posture of containment into a posture of mitigation and that coronavirus will traverse the country.

The Simulation

Second disclaimer: There is not a lot of data available on COVID-19 yet. It is a new virus, so we do not yet know how fast it spreads, how long it lives on surfaces, and other information that would allow for better forecasting of the spread of the disease and the severity of the disease.

The impact of COVID-19 on East Tennessee is an open question. To examine some possible outcomes and their impact on our healthcare system, I have developed a simulation to model how a surge could affect the availability of hospital beds in East Tennessee. The simulation is modeled using four variables:

1. Infection Rate - This represents the percent of people in East Tennessee infected by the Coronavirus. As a baseline, between 9 million (2.75%) and 45 million (13.75%) Americans get the flu each year (CDC). As an example, if you choose a 5% infection rate that would mean 5% of the population of say Anderson County would be infected by the virus or about 3,824 of the 76,482 population (via U.S. Census data) of Anderson county would be infected.

2. Hospitalization Rate - There is not enough data to know the hospitalization rate for COVID-19. Hospitalization rates for the flu are about 1.5% (CDC), whereas, Italy is experiencing a 10% hospitalization rate for COVID-19.

3. Average Nights in Hospital - Again we don't know, but the average hospital stay for flu is 6.3 days (H-CUP). Since the beginning of the COVID-19 outbreak, a 14-day quarantine period has been standard, though hospital stays might be longer or shorter.

4. Speed of Spread - There is not enough data to know how fast COVID-19 will spread. Additionally, if we engage in preventative measures we can slow the spread. This variable represents the number of weeks it take to reach the full infection rate (i.e., 10% of pop). The more weeks, the slower the spread. [Note: This variable is based on the spread of the disease being nearly evenly distributed. The model does accept that the first handful of the weeks after the virus is introduced it will spread slowly. After the initial period (9 weeks), the rest of the cases are evenly distribute. It is likely that the spread of the virus will not follow an even distribution, but will likely be bimodal with an increase at introduction, a decrease during summer, and an increase in the fall. But since we have no real understanding of the distribution of the spread, the even distribution is the most conservative choice.]

If COVID-19 in East Tennessee ends up being similar to the flu, it will more than likely not overwhelm our system.

In simulation 1, the four variables in the model are set to similar values as a low frequency flu season. About 2.5% get the virus and 1.5% of those are severe enough to require hospitalization. The ones that go to the hospital stay an average of 6 days, and the spread of the virus is set to 30 weeks. By setting the speed of spread to 30 weeks, this model assumes that there will be a number of weeks where the spread of the disease is greatly reduced and that most of the spread will be within the 30-week window. It is also still fairly conservative, because the nature of the even distribution does not allow for unpredictable spikes that are common in outbreaks.

Regardless, simulation 1 clearly slows that our healthcare system can take a small influx of coronavirus patients. Problematically, early reporting and research has found COVID-19 to be worse in many ways.

First, the Italian government has found the hospitalization rate of confirmed COVID-19 cases to be around 10%. This is probably overestimated, because many cases are never confirmed (e.g., some people are asymptomatic or have mild enough symptoms that they never go to the doctor). The hospitalization lengths have also been longer than the flu due to both 1) the 14-day quarantine period being used and 2) the ease in which COVID-19 develops into pneumonia.

In simulation 2, I updated the model to include a higher hospitalization rate (10%) than seasonal flu and a longer average hospital stay length (14 days) than the flu. In this model, we start to see some problems. While the three counties without hospitals would always be in the red, two additional counties - Loundon and Jefferson - went red, meaning that these counties do not have enough staffed hospital beds to accommodate all the sick people in their county. The current model results in 24 people needing to leave their home county to get medical treatment.

Simulation 2 is still based on the infection rate of a mild flu season. Simulation 3 assumes that the infection rate of the coronavirus will be closer to an aggressive flu season (13.5%). This is where we start seeing problems.

In simulation 3, based on the average number of staffed hospital beds in East Tennessee during 2018, our system would be short by more than 730 beds. Additionally, every county, expect Hamblen County, is in the red. In this scenario, the COVID-19 virus will substantially overload our current healthcare system and will necessitate rapid large-scale changes.

Again, I am in no way saying that this is what is going to happen. I am just saying that our healthcare system in East Tennessee is already stressed and the coronavirus could present a substantial strain on the system. This strain will not only affect coronavirus patients, but all patients that need to access healthcare.

One thing that is important to note: we can make changes to our lifestyles and our society in order to slow the transmission of the disease.

As the graphic above from the CDC shows, if we slow the transmission of the disease by doing all the things the CDC tells us to do (i.e., wash hands regularly, don't touch your face, avoid contact with sick people, engage in other social distancing practices, and stay home if you are sick), we can flatten the curve and reduce the spike to our healthcare system.

Simulation 4 shows what would happen to available hospital beds if we can manage to slow the transmission rate of the coronavirus. Overall, in this run, we are still in the red, but we are only 97 beds short, which is much more manageable than the 730 beds in simulation 3.

Try the simulation for yourself

The four models that I chose to run were based on some of the very limited data available and a healthy amount of guessing. Please try the simulation out for yourself and feel free to reach out to me with thoughts or ideas on how I can make it better or add to it.

Reminder! This is a simulation. It is not a forecast. It has zero predictive ability. It is only meant to understand how our healthcare will fair under various conditions.

Use the controls on the right to change the infection rate and other variables.

Summary

As I have said, I am not a medical doctor and have no background in epidemiology or virology, but I know that our healthcare system in East Tennessee has been negatively impacted by the closure of numerous rural hospitals during the last five years. We need systemic change to fix our healthcare system and make it more accessible to all.

Beyond those systemic changes, in the short term regional hospitals, especially the bigger ones, can quickly ramp up the number of available beds. But to do this, they also need staff and supplies for those beds. While that can happen, it does mean substantial adaptation.

Finally and most importantly, we can all engage in positive personal decisions to try to slow the spread of this disease and flatten the curve of COVID-19's spread, resulting on less strain on our healthcare system.