COVID-19 Forecast for El Paso County – May 28
Behind the models with Phoebe Lostroh
Before we begin our COVID-19 coverage, we first want to acknowledge another current public health issue: racialized violence and police brutality nationwide. After videos surfaced of a white Minneapolis police officer pinning his knee on the neck of George Floyd, a 46-year-old Black man who died after crying out that he couldn’t breathe, protests erupted across the country, including in Colorado Springs. The officer has been charged with murder and manslaughter. Floyd’s death is no isolated incident, and news coverage has largely shifted to these important national conversations.
On Friday, Colorado College administrators sent an email titled, “Our Commitment to Speak Out Against Racism.” An excerpt:
“Sadly, such behavior is not new, but with amplified visibility because of social media and already heightened anxieties around COVID-19, the murders of Breonna Taylor and George Floyd (and Trayvon Martin, Michael Brown, Eric Garner, Freddie Gray Jr., Walter Scott, Oscar Grant III, Philando Castile, Sandra Bland, Tamir Rice, Botham Jean, Atatiana Jefferson, Ahmaud Arbery, and more) are hitting home right now.”
We at the CC COVID-19 Reporting Project are critically reflecting at this time about our role as journalists, and we hope this is part of a conversation that leads to lasting change. Our inbox is always open.
Phoebe’s Forecasts
Whether colleges are able to hold in-person classes in the fall could depend on local factors and public health decisions made about the spread of the virus in specific localities.
So we’re going to kick off each week with projections from CC’s Phoebe Lostroh, an associate professor of molecular biology.
While on scholarly leave, she is serving as Program Director in Genetic Mechanisms, Molecular and Cellular Biosciences at the National Science Foundation. Lostroh has been monitoring the spread of COVID-19 in Colorado Springs and El Paso County since the beginning of the outbreak and is producing a series of models to track and project the spread of the disease.
Note: These forecasts represent Dr. Lostroh’s own opinion and not necessarily those of her employer.
Predicted cumulative cases
Predicted new cases each week
Says Lostroh: The Colorado Department of Public Health has granted a “variance” to Colorado Springs allowing restaurants to open with some precautions in place. Officials will revoke the variance if Colorado Springs has more than 715 new cases in any two-week period. In the best-guess scenario, Colorado Springs may cross this threshold sometime between June 25 and July 2. In a two-week period, 715 new cases corresponds to 100 cases per 100,000 residents in Colorado Springs.
Predicted new hospitalizations each week
Says Lostroh: Our local hospitalization rate for new cases has now dropped to 3%, but the average percent of hospitalizations over the past month remains at 6%. I do not have any information on local availability of ventilators or information regarding antivirals or other medicines to treat symptoms of COVID-19, nor about medications needed to keep someone on a ventilator.
May 28 is the start of the 12th week since the first case was detected in El Paso County, and 88 El Paso Country residents have died of COVID-19 since March 5. The most likely outcome is 260 new cases by next week (June 4), 291 the week after that (June 11), and 334 the week after that (June 18). The curve predicts 3,906 total cases by July 9, but the prediction is less accurate as time passes. Last week, my best guess was 1,700 total cases on May 28, and instead, we have 1,686. That means the true outcome was a little better than my best guess. These predictions are made assuming that more than 10% of our population has been exposed to the virus so almost everyone is susceptible, and a simple projection of exponential increase is thus appropriate.
Cumulative COVID-19 cases, El Paso County, May 28
Says Lostroh: Between days 1-18, there was an explosive exponential increase. (An exponential curve looks like a straight line in a semi-log plot). Then we see the curve flatten out from days 19-35, most likely as a consequence of actions taken prior to March 12. Notably, the Bridge Club that was the site of the first outbreak closed on March 12; on or around March 12, contact tracing began. Between days 35-63 or so, we see a pretty slow rise, most likely as a consequence of the stay-at-home orders and contact tracing followed by self-isolation as necessary. After day 63, we start to see the curve increasing a bit faster, especially as we leave the stay-at-home orders. Memorial Day was May 25, which may have been an occasion for people to gather. While May 25 was beautiful and perfect for gathering outdoors, May 24 was rainy and cold. Data from May 22-May 28 have the most uncertainty because of lags in reporting. Restaurants in Colorado Springs were allowed to open with limited capacity and some precautions starting on May 24, while the county has requested a variance for churches and the zoo.
Forecasting stages 1, 2, and 3
Says Lostroh: Stage 1 is the last eight days of exponential increase before the contact-tracing for the earliest cases and the stay-at-home orders start to have an effect. Stage 2 captures the very end of stay-at-home and the first few days after we went to “safer at home,” and it is the flattest part of the curve of cumulative cases. Stage 3 is my best guess because it captures 14 data points while omitting today’s value, which is most likely to change.
Forecast for the next four weeks of cumulative cases in El Paso County
Behind the Models: Q-&-A with Lostroh about her work
After sending us the models, Lostroh sat down — well, actually it was a Zoom call — with us to talk about her models, how she comes up from them, and what they’re telling us. The following has been edited for length and clarity.
CC COVID-19 Reporting Project: So how do you come up with these models?
Phoebe Lostroh: They come from me making graphs and fitting lines to the graph that describe the increase of cases in our El Paso County. I literally use an Excel spreadsheet and put in the cases reported by the county every day, then track them in that spreadsheet.
CCRP: You then extrapolate from that?
Lostroh: I look at it like a microbiologist would look at any case of microbial growth. We put time on the X-axis — the horizontal axis — and then we put the total amount of cases on the Y-axis. I then look at whether the increase is linear or exponential, and also use knowledge of things like when we started the stay-at-home [order] and when we ended it, and try to figure out how different kinds of equations describe those lines most accurately.
CCRP: Last week the “best guess” outcome you projected was 1,700 cumulative cases by this week.
Lostroh: That’s right. I get that by looking at the last 14 days for which we have pretty good data. The case reporting bounces up and down a little bit for the most recent three or four days, but after that, it stabilizes. When I make a graph, I’m looking at the total number of cases that we’ve had for that 14-day window, fit an exponential curve to that graph, and then just solve that equation to look into the future. That’s how I arrive at these numbers. As I say above, last week, my best guess was 1,700 total cases on May 28 and instead we have 1,686. That means the true outcome was a little better than my best guess.
CCRP: You plot out three scenarios, looking into the next three weeks. The first is “worst case,” the second is “optimistic,” and the third is “best guess.” Where do those categories come from?
Lostroh: I look at the picture of total cases over time and how it changes. What I see is that just at the very beginning of the stay at home order — between March 26 and the week that followed — cases were still rising pretty quickly. I call that my “worst case” scenario. I use the line that fits those points to project how many new cases there would be starting from the number of cases we have now. That’s what I call my worst case — and actually it’s not the worst case. It’s a lot milder than what the infection was for the first 21 days in Colorado Springs. But I think it’s the most reasonable worst case, so that’s what I used. Similarly, for my “optimistic” projection I looked at the very end of the stay-at-home orders where the number of cases is increasing very slowly and probably because we were all staying at home. I figured the most optimistic we can be is that the cases will continue to increase at that rate. And then the “best guess” is using the most recent seven to 14 days to fit a line and see, is the situation changing either compared with the “worst case” or the “optimistic.”
CCRP: You then predict new hospitalizations. That’s based on the first numbers, to some extent, but on more than that?
Lostroh: We know that in El Paso County the rate of people who are testing positive entering the hospital at all is 6% or more, depending on the week. And then the estimates I used for the hospital beds needed are based on what Colorado and CDC recommend, which is that about 75% of people who go into the hospital need a regular bed, versus 25% who will go into the ICU.
CCRP: Based on these projections, unless we are hit by a worst-case scenario, we will have enough beds through the next six weeks?
Lostroh: Yes, and that’s based on looking at newspaper articles that document how many hospital beds of different kinds are available in El Paso County. I’m extrapolating from there that about a third of them are available at any given time. So I know approximately what our capacity is.
CCRP: So, if your “optimistic” or “best guess” scenarios occur, we'll be OK, and if “worst case” happens then it’s a serious problem? Is that the takeaway from these models?
Lostroh: The takeaway is that if we continue to behave as we were behaving about seven days ago, during the safer-at-home order, we’ll probably be fine.
But if we decrease non-pharmaceutical intervention measures — wearing face coverings and washing our hands a lot — we could go back to a curve more like the worst-case scenario. The best thing we can do is to increase our social mixing and get to see each other, but be so consistent using face coverings and social distancing and washing our hands that we can keep that curve just as flat as it was during the “optimistic” scenario.
I just don’t think that's very likely to happen because I don’t think that’s how people behave. The way that microbial growth works is that it’s generally exponential, and so is the spread of infectious disease. The problem is that when it gets bad enough for us to notice, it might already be too late to intervene. The idea is that you want to intervene before you get to that point based on exponential growth of the infection in our community.
About the CC COVID-19 Reporting Project
The CC COVID-19 Reporting Project is a student-faculty collaboration by Colorado College student journalists Miriam Brown and Arielle Gordon, Journalism Institute Director Steven Hayward, Visiting Assistant Professor of Journalism Corey Hutchins, and Assistant Professor of English Najnin Islam. Work by Phoebe Lostroh, Associate Professor of Molecular Biology at CC and National Science Foundation Program Director in Genetic Mechanisms, Molecular and Cellular Biosciences, will appear from time to time.
The project seeks to provide frequent updates about CC and other higher education institutions during the pandemic by providing original reporting, analysis, interviews with campus leaders, and context about what state and national headlines mean for the CC community.
📬 Enter your email address to subscribe and get the newsletter in your inbox each time it comes out. You can reach us with questions, feedback, or news tips by emailing ccreportingproject@gmail.com.