How NOT To Look at a Pandemic

In 2018 my wife and I were in the middle of a major outbreak. The origin of the outbreak was within a few miles of us. It started slowly, but by the third day it was obvious that a lot of people would be affected. The outbreak spread like wildfire, slowly at first and then with increasing rate of spread in exponential fashion. As it raged people began to panic. Local authorities quickly realized they would be unable to contain the spread and they called for federal assistance in the form of a Crisis Management team. The Feds locked down the local community, restricted travel and took charge of available resources. Daily public briefings were held, reporting on the extent of the outbreak, detailing the daily spread, the hotspots, the allocation of resources and explaining strategies for the short term future. In spite of this, the outbreak raged out of control, threatening more and more people. Efforts shifted away from trying to contain the outbreak to trying to mitigate individual losses. Over the course of about two weeks, the darned thing ran out of new places to spread, and gradually petered out. Some folks paid a high cost, others lost nothing. At the end of it all, the Crisis Management Team had not extinguished the outbreak. They had some small scale successes around the edges, but ultimately it didn’t stop spreading until it had run its own natural course. This outbreak was not an infectious disease, it was the Spring Creek Wildfire in Colorado, second largest in state history in terms of acres consumed. One hundred and thirty five of our neighbors lost their mountain homes.

I tell you that story to tell you this one.

Respiratory virus outbreaks and forest fires have a lot of similarities. They require a spark, an initial point of entry. They require a susceptible population with environmental or cultural conditions that allow for it to start to spread. They both thrive on unaffected regions and can expand slowly at first, then rapidly. They both are capricious and nondiscriminatory in their destructiveness. They both are largely impervious to the puny efforts of man to contain them once they are in full swing. And neither of them goes on forever. Eventually they run out of susceptible hosts (or unburned trees) and the factors that sustain the contagion peter out.

Here are some things that I’ve tried to keep in mind in regard to the COVID-19 pandemic. I don’t really have an overarching theme or any grand predictions, but here goes:

Projections/Predictions: Epidemiologists live for moments like these. Early on, once the disease is identified and shown to have some predilection for a substantial mortality rate, the epidemiologic sleuths begin their investigation. Early reports of how many people will be infected by a single index case (patient “Zero”) are calculated, giving the so-named R-0 value. That early R-0 value allows the predictions to begin. These things start slow before they grow rapidly before they slow down again. The pros run models based on projected rates of rise (among other variables) and come up with a range of possible outcomes. This range may or may not contain the actual final numbers, but that’s okay, it’s just an educated guess. What gets reported to the public however, is generally the high end of that range. We’ve all heard the reports of “Experts say this virus could infect XXX million people and result in the deaths of XX million!” As a general rule, however, it is likely that the eventual real-life numbers will be closer to the middle of the range than to either of the extremes (high or low). If you can find it reported, look at the overall range, not just the sensationalist “worst-case scenarios”. Remember the old adage, if you ask a barber, he’s going to tell you that you need a haircut.

Rate of Increase: Be it SARS, MERS, COVID-19 or seasonal influenza, viral outbreaks start slow, then grow at steadily increasing rates until they reach a maximum and then start to decline. Consider the rate of growth to be the proportional day-to-day increase in number of confirmed cases. For example, if you have 100 cases today and tomorrow the total has climbed to 130, the daily rate of increase is 30%. Understanding that this daily rate of increase is not constant is key to understanding the trajectory of an outbreak. Any predictions based on “at the current rate of growth XX people will be infected” will frankly be wrong because in the real world the rate DOES NOT remain constant. In general the rate is slow at first, steadily increases to a maximum, then starts to steadily decline. There may be some small fluctuations up and down, but the overall trend follows that same pattern. By my calculations some representative rates of increase for the most recently reported data are: U.S. 32%, Italy 8%, Spain 13%, France 15%, Germany 19%.

The art of modeling epidemics hinges on trying to gauge the steepness and the duration of the increase. These two factors determine the trajectory. This can be a little like driving on a twisty mountain road by looking in your rearview mirror, as you have to rely on yesterday’s data. In the last week, Italy’s daily rate of increase has continued a trend of decreasing and is now below 10%. And, lo and behold, they have reported a decrease in the absolute number of new daily cases over the last three days. I’ve noticed that once the rate of increase drops below10% a decrease in the absolute number of new cases comes soon after.

Be leery of graphs: Make sure you understand what is being graphed, what scale is being used and how stretched or compressed the depiction is. They can be visually deceptive and can add to your own uncertainty as you try to make sense of it all.

Distortion of time: When an outbreak has been covered in the news media, blogs, social media and whatnot to the extent the COVID-19 outbreak has, where it is virtually the only thing being discussed, it is easy to get caught up in a distortion of the time scale. It seems like it has been going on forever, right? At the beginning of March, Italy had barely 1000 confirmed cases. Now, a mere three weeks later it is showing signs of peaking. I find it useful to step back and reconsider the overall timeframe.

Although I do not have a crystal ball or a time machine, I am confident (based on past similar outbreaks such as SARS, personal experience in other outbreaks such as H1N1 influenza during my years in medical practice and my own observations of the real-time data available to the public) that this outbreak will run its course and burn out, just as the Spring Creek Wildfire did in 2018.

Disclaimers: I am a retired physician. I am NOT an epidemiologist. I am NOT an infectious disease specialist, and whatever expertise I may have in infectious disease is in the clinical realm, not the statistical realm. I do NOT claim to know when this epidemic will peak in the U.S. and make NO predictions regarding death rates, impact on the medical personnel and infrastructure of the country, stock market or political repercussions. I recognize that data is imperfect, and temper my analysis with a willingness to reassess as new information comes to light.

One comment to “How NOT To Look at a Pandemic”
  1. Thank you Muldoon for clarity and perspective that is sadly not going to quell the power-grab and waste of trillions, aside from the psychological health of the nation and the world. I hope Trump decides to stop this madness certainly no later than Easter if not sooner.

    Also, can you make this into one of your world-famous limericks? 🙂

Comments are closed.