This was originally published on rarelyread.org on 3/20/2020, before I moved my blog to medium.com.
After seeing John Burn-Murdoch’s excellent charts plotting COVID-19 case counts and death rates by country, I wanted to see the same information for each state. In particular, I wanted to see visually what the trend is for Massachusetts and where we might be headed. Are we still seeing exponential growth? How do we compare to Italy?
I couldn’t find a similar chart for the various US states, so my partner and I created some:
Day 1 for each state is defined as the first day that state reported at least 30 COVID-19 cases, the rationale being to compare disease progression in each state once COVID-19 had a demonstrated foothold. Lines depicting the early disease course in Italy as well as an imaginary state where the disease burden increases by a third each day are provided as reference points.
California, New York, Florida, Washington and Illinois are already prominent in the news. But New Jersey, Michigan and Connecticut quickly jump out as additional states of concern. Of course, these trends could change quickly (for better or for worse) especially in states like Connecticut that have fewer data points.
Because US state populations vary wildly, we made a second chart of the infection rates per 100,000 people:
The per capita chart calls out Louisiana, Colorado and Massachusetts as outliers and makes clear that although concern is appropriately high in both California and Washington, proportional effects on the latter state are much higher. Unfortunately, New York has it the worst using either measure.
One obvious, unfortunate conclusion from both charts is that no US state has peaked, nor come close to doing so.
I’m a safety net emergency physician in Massachusetts and work part-time in clinical informatics and data analytics. I won’t try to discuss all things COVID-19 in one post, nor am I qualified to do so, but I do suggest that anyone with an interest in this pandemic should read the Imperial College London report on COVID-19 modeling for the US and UK, which reportedly helped convince the Trump administration to change course and embrace social distancing.
The report is not for the faint of heart. Their baseline model, where no protective measures were taken, predicted 80% of the population being infected, with over 2 million deaths in the US (about twice as many as die each year due to heart disease and cancer combined). Intensive care unit (ICU) capacity would be exceeded by a factor of 30 — a complete wipeout of the healthcare system. Various mitigating and suppressing strategies were modeled; only the most stringent measures (universal social distancing, which we’re now all familiar with) being predicted to “flatten the curve” without overwhelming the healthcare system.
I include the image above (arrow and ICU capacity label added) because the sheer magnitude of the mismatch between resources and need are not conveyed by most “flatten the curve” charts I’ve seen. That pitiful little horizontal line at bottom? That’s the UK’s capacity for treating critical disease. You can see that even a combination of less-stringent measures is predicted to blow capacity out by nearly tenfold.
The most interesting prediction in my opinion relates to the road ahead: where does all this end? The chart below, which predicts disease rates if suppressive measures were turned on and off based on how many COVID-19 cases required ICU treatment in a given week. You can see how with wave after wave of disease, it rapidly escalates after a month or two each time, with little suggestion of meaningful herd immunity.
To me, this suggests that the most desirable among feasible medium-term end points is to maintain suppression until reasonably certain that all extant cases within a region are known and well isolated. At that point, social restrictions could be relaxed for such regions while discouraging contact with uncontrolled regions and aggressively monitoring for recurrent cases.
Given this, it is unwelcome to see that California and New York are decreasing testing to conserve personal protective equipment (PPE) and testing bandwidth/supplies. I’m neither an epidemiologist nor an infectious disease expert, but this strikes me as a short-term, desperation measure.
Whoo! This is going to be a long road. Let’s all hope that a game-changing vaccine or treatment is developed soon.
I’m planning to create a similar chart of deaths by state, but there are currently too few data points (8 states have at least 5 deaths as of 3/21/2020). Unfortunately, it is likely there will be many more in a few days and I will post an update. In the meantime, stay happy & healthy, stick together (at a safe distance!) and let me know what you think!