Hi Everyone,
We have some trends that are looking good, but some that are looking bad. In the US outside of NY, the number of new reported cases since day 50 (40 days ago) is showing a significant downward trend (P < 0.0001) (Fig. 1). Given that the number of new tests continues to increase (Fig. 2), this is a real sign that the actual number of new cases has been declining.
This conclusion must be tempered, however, by a trend in the number of new cases per test (Fig. 3). This had been declining steadily for about 30 days. However, for the past 20 days, this number has been statistically flat (P = 0.3456), suggesting that the actual number of new cases per day has stopped decreasing. Interestingly, the beginning of this period coincides with the beginning of “opening up” and lessening of social-distancing practices.
The data on cumulative new cases per test for the US outside of NY still fits a modified logistic growth curve (Fig. 4), and projections from this curve indicate a passing of most of the first wave of infections around day 95 (Fig. 5). Specifically, for the US outside of NY, on days 93, 121, and 159 (1 June, 29June, and 6 Aug, respectively—these dates are about 1 – 4 days later than projected last time), 95%, 99%, and 99.9% of the total infections during the first wave of the virus will have occurred. On these dates, the minimum number of infections yet to come are 84,166, 18,592, and 1,956. On days 112 and 146, the expected minimum number of new cases will be 1,128 and 27.
Fig 1
Fig 2
Fig 3
Fig 4 Fig 5
The good trend in death rates continues in the US outside of NY (and in NY) (Fig. 6). For the US, this trend is highly statistically significant (P < 0.0001) after day 45 (45 days ago). Extrapolation of this linear trend puts the “zero deaths” day around day 145 (23 June), about two weeks earlier than projected last time. As always, though, there is a substantial error associated with this projection—a zero death day of 266 (21 Nov ) is within the margin of error.
Fig 6
In North Carolina, the number of new reported cases per day continues to climb (Fig. 7). When this statistic is corrected for number of tests, the number of new cases per test has taken an ugly turn upward (Fig. 8). This trend began about 20 days ago, is highly significant (P = 0.0018), and coincides with “opening up” and relaxation of social distancing—a really bad trend.
Not surprisingly, given this trend, the data on cumulative new cases per test continues to deviate systematically from a modified logistic growth curve (Fig. 9)—the actually data is more linear from day 71 than the fit. Consequently, no projections will be made at this time.
Fig 7
Fig 8
Fig 9
Last time, there appeared to be a significant trend in NC for deaths decreasing. Unfortunately, that trend has disappeared with recent spikes in deaths over the past few days (Fig. 10). Death rates appear to be on a steady increase, indicating we have not yet reached the crest of the wave of deaths associated with the virus.
Fig 10
In conclusion, I’m less optimistic than I was last time. The best news is that new deaths in the US continue to decrease, although the opposite seems to be happening in NC. The bad news is that the number of new cases per test seems to have plateaued in the US outside of NY, suggesting that we may be seeing the effects of relaxing social distancing practices. The trend in NC is even worse, with the number of cases per test looking like it’s actually increasing.
One bit of related news that people may be interested in is that I received a message from CVS Pharmacy indicating that individuals can schedule a coronavirus test. As far as I can tell, this offer is open to anyone, it’s free, and results are returned in 2-4 days. So if you have been holding back on getting together with loved ones for fear of contracting or passing on the virus, this might be an opportunity for you to do that with some reassurance. There may be other places doing this as well.