There has been discussion lately concerning the rate of new cases of COVID-19 in the U.S. The daily rate of new cases reached a peak of 31,564 on April 10, dropped to 20,658 by May 28, but has leveled off and risen slightly to 21,084 through June 9, according to data compiled by the Johns Hopkins Center for Systems Science and Engineering. However, this doesn’t begin to tell the whole story, as states are in different phases of the COVID-19 pandemic. Comparing state-level data over time shows that there are actually many broad similarities to how the pandemic progresses in each state. The few exceptions to these common characteristics are also interesting.
For the analysis, new-case-per-day data from the Johns Hopkins CSSE database is used, and 7-day moving averages for each state (and D.C.) are taken to smooth out the data, which goesthrough June 9. Most states follow a pattern of a rise to a peak, followed by a less-steep decline. The plots below show the percent decline from the peak versus the number of days since the peak occurred. Note that the absolute rates or rates per capita are not considered, only the progression of new case rates following a peak. (I put that in bold so people won’t complain about it – that would be a different article.) Also, most states have increased their rate of testing over the last month or so, while at the same time the corresponding fraction of tests that are positive rates has gone down; we don’t attempt to correct for these factors in this simple analysis.
The Johns Hopkins data can be divided into geographic regions (Northeast, Midwest, South, and West), which is done here to keep the plots from being too crowded. Interestingly, states within each region for the most part tend to have similar characteristics.
For the northeast region, most states are at least six weeks past their peak, and the current number of cases is a small fraction of what occurred at the peak. The dashed lines show the general path that most states follow, from upper left to lower right; the other plots will have the same dashed lines to guide the eye.
New York, New Jersey, and Pennsylvania reached their peaksnearly two months ago, Rhode Island, Massachusetts, and Connecticut roughly a month and a half ago, while New Hampshire and especially Maine had their peaks more recently. Note, however, that they all do follow the same general path.
Vermont is an outlier here, having reached very reduced levels of new cases somewhat earlier, with a resurgence over the last few weeks (shown by the arrow), although that still puts them on par with Pennsylvania after 60 days. Sometimes media outlets will claim a surge where that is a gross overstatement, but this case may qualify as one.
In the Midwest region, most states are in the middle area, similar to New Hampshire, but they are still are following the same basic path. Minnesota and Wisconsin had their peaks more recently. North Dakota has had a slightly faster drop from the peak.
There are two outliers: Michigan and Missouri. Michigan had been very similar to northeast states, but has had a recent surge in cases, shown by the arrow; average cases has increased nearly five-fold. However, it could be that a large number of older cases were added to the database or there is a spurious data point; I believe the CSSE database counts cases when they are recorded, not when they were actually diagnosed, and the increase in the average was due to a very large, single-day spike.Missouri is unusual in that they had a peak a longer time agocompared to most Midwest states, but the drop-off has been much slower than in most other states.
In the south region, most of the peaks have been within the last three weeks or so, but for the most part as a group they still follow the same path. Florida, North Carolina, and South Carolina are at their all-time highs. Louisiana is more like Pennsylvania, D.C. is more similar to many midwest states, and Delaware is similar to Connecticut. West Virginia has had a much steeper decline than most, while Georgia and Oklahoma are more like Missouri, with a very slow drop-off.
In the west region, we see a wide variety of situations. The states in the upper left have had very recent peaks, with California and Utah, at their peak; these are similar to many southern states.Colorado and Wyoming are slightly more advanced than most Midwest states, but not as much as most northeast states.
There were many more cases of resurgences, as shown by the arrows. Hawaii, Idaho, Montana, and Washington were in various stages of decline before having a recent increase (still below their peaks of a month or more ago). Unlike Michigan, these all appear to be a real trend and not due to a few spurious data points, although the per capita rates are still very low for Hawaii and Montana.
Alaska had reduced case rates as low as 10% of their peak, but then reached a new high a few days ago. Oregon was unusual in that they had peaked and then declined for only a couple of weeks before having their rate increase to new highs. Nevada is the real outlier in the west; they have almost returned to the highs reached nearly two months ago after having dropped by about 50%. Washington’s resurgence over the last three weeks has been similar to Nevada’s, except it started from a lower base.
Obviously these plots don’t capture all of the nuances of the data, including periods of time where there are minor deviations from a smooth rise and fall of the rates. But they do show a lot of the commonality among the progression of states’ new-case averages over time. Some states that have been more open all along or opened up earlier, like Georgia, Missouri, and Oklahoma, have seen a much slower decline in case rates, but nevertheless they have still declined. One wonders what will happen to states that have remained mostly closed until recently, and whether recent increased activity due to protests and riots will see more states with a resurgence in the future.