Cases Numbers Comparison vs 1 Year Prior

Seasons of Covid (in Charts)

Interactive Site Directly Compares Numbers Between Years

I’ve spent most of the past 2 years caring for Covid-19 patients as an emergency physician while helping my hospital monitor the pandemic in my role as a clinical informaticist. As the country struggled with the Delta variant this summer, I was grateful that, through some combination of vaccines, policy, and luck, the impact on my home state of Massachusetts was less than it had been in the Spring of 2020. But I was worried about a winter surge.

Unlike last year, we have vaccines now (for those willing to take them — but don’t get me started) and a wider range of treatments. But the Delta variant, being far more infectious than last winter’s viruses, works against that — and the new Omicron strain is even more contagious. Finally, lockdowns are more or less history. At least in non-healthcare settings, people are out living their lives and being much less careful. How do these protective and exacerbating factors play out now that messy reality is lumping them together?

I found my eye constantly darting between the same dates in 2020 and 2021, back and forth, to control for seasonal factors. By seeing where we were for cases, hospitalizations, and deaths compared to the year before, I hoped to get a clue of where we might be headed. I checked several great Covid sites* looking for a way to interactively compare dates between years. To my surprise, I couldn’t find one with the visualization I wanted**. With the necessary data all publicly available, I decided to build and publish one, with the source code freely available.

Hospitalizations chart.
Hospitalizations by Year for CT, MA & NH

The line chart above is by now familiar to most everyone, with one difference. For each state, the solid lines represent data from the current year, while the dashed lines are the data on the same date from the year before. Where the solid line is higher, that state is doing worse this year than last, and vice versa.

See how the dashed lines extend further right than the solid ones, which end December 29th? That’s because the dashed lines are showing what happened in Jan-Feb 2021 which, as I write this, is less than one year ago. Showing that time period lets us see what last year’s trends were for the unknown months ahead (Jan-Feb, 2022). For instance, we can see that Massachusetts hospitalizations peaked on January 7th last winter. Switching to look at cases, it’s Jan 12th. The timing of the nationwide peaks were very similar. If that pattern were to hold this year, we’d be less than two weeks away from starting back down!

To make the chart more interactive and useful, I added controls to select states, dates, and the data of interest, and a nifty hover window with details on each point. After many bug fixes, I thought it was a pretty helpful tool that might be worth sharing, but I still wasn’t able to see a state’s cases, hospitalizations, and deaths all together on a single chart. Doing that presented two big challenges. First, there are many more cases than deaths, making it difficult to see both extremes well without using multiple y-axis scales, which is confusing. Second, with six lines per state (two each for cases, hospitalizations, and deaths), any multi-state comparison would be one big tangle of spaghetti.

My solution is the chart below, which shows the percent change in each series, rather than the values themselves. Where the plot is above the horizontal Equality line, numbers have gone up compared to the prior year. Where the plot is below Equality, numbers have gone down. In the unlikely event that the numbers for each year were exactly the same for a month, the plot would exactly follow the Equality line for that time period.

Percentage Change by Year for Massachusetts, July-Dec

To improve viewing across a wide range of percent changes, a logarithmic scale is used. Note that no portion of this chart extends into the future as we need both past and current data to compute the percent changed.

Looking at my home state of Massachusetts, we can make several observations at a glance. Although this summer’s Delta wave sent case numbers high above those of the previous year, hospitalizations were generally lower, rising only moderately above equality for about six weeks.

For deaths, this disconnect was even more dramatic, reaching equality in mid-September and sticking there for about a month before retreating well below.

Percentage Change by Year for Massachusetts, Nov-Dec

Looking later in Fall, we can see an almost eerie degree of stability starting in mid-November where cases followed those of the previous year almost exactly, hospitalizations were about 30% lower, and deaths about 50% lower. With last year’s peak occurring in early January, it appeared the vaccines and other health measures were holding the line against the Delta variant. But was this just the calm before the storm as we were about to be clobbered by the Omicron variant?

It certainly was, looking only at cases. In the thirteen days from Dec 16th to Dec 29th, cases rose from the equality line up to double those of the previous year (which were also rising, just not as fast). However, two weeks in, we can see that the lines for hospitalizations and deaths remain flat. This is very fortunate, as many of our hospitals are already overwhelmed by staff shortages, deferred care needs, and the Delta variant. These may well drift upward in the coming weeks, but every day they remain flat makes a crushing Omicron surge less likely.

Percentage Change by Year for Massachusetts, Jul-Dec

Looking at Florida, a very different picture emerges. The Sunshine State experienced a terrible Delta variant onslaught, with both hospitalizations and deaths above equality from August late into the Fall. Their spike in cases related to Omicron has been steeper than most: from an 80% decrease to a 207% increase, or a fifteen-fold explosion in less than three weeks! And unlike Massachusetts, their hospitalizations are already climbing sharply, though they remain only half of what they were at the same time last year.

Once I had these two chart types working, the final piece that seemed to be missing was a big-picture view that would quickly show which states were currently doing better/worse compared to last year. I added a summary table with sortable column headers to address this.

Summary Table, Sorted by Change in Cases

Sorting by the change in cases (above), we see big increases in northern states, somewhat surprisingly joined by tropical Hawaii and Florida. Looking at hospitalizations (below), most states have seen improvements compared to last year. As expected, states with low vaccination rates are somewhat over-represented among those doing worse. The notable exceptions are Maine and Vermont, although Vermont’s hospitalization rate is well below the national average, so it’s only “bad” compared to its exceedingly low rate last winter.

Summary Table, Sorted by Change in Hospitalizations

So, what’s next with Covid-19? These charts aren’t a model; they don’t make predictions, so I’m trying to figure that out like everyone else. What I’d most like to know is the extent to which Omicron infections are protective against other strains, particularly Delta (as of late December 2021, there’s very preliminary evidence that it is). With Omicron already dominant among new cases, we should have that answer within weeks. Then, if past is prologue, some other twist will come along. Still, I’m optimistic to be past the Winter Solstice, with each day longer than the one before.

Well, that’s it for my charting site. Please, poke around and let me know if you pick out any remarkable trends or have feedback. Next winter, I hope to be busying myself with entirely different things!

* I’ve spent countless hours exploring 91-Divoc, Our World In Data, and the NY Times Covid sites and owe them much gratitude.

** The NY Times comes close, charting each state compared last January’s maximum. However, this is still comparing dates in Nov 2021, Dec 2021, etc to a single day in January 2021, rather than the same date from a year before.

About the author: Tom Seufert is a clinical instructor for Harvard Medical School, board-certified emergency physician, and Director of Emergency Informatics for Cambridge Health Alliance.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Tom Seufert, MD

Tom Seufert, MD

Emergency physician, clinical informaticist, software engineer, author, dad.