Hi Everyone,
For the most part, trends that we have seen before are continuing. There are some new trends in southern states, however, that I examine here for the first time because they are pretty dark. Please also note that I am presenting the data in a different format. Instead of having two graphs, one for the raw data and one for the moving average, I am now presenting only one graph, in which the points are the raw data and the line is the 5-day moving average. All statistical analyses, though, are based on the raw data.
US Cases
· The number of new cases per day in the US minus NY continues to decline slowly (Fig. 1). This trend is linear after day 50 (P = 0.0142), and projection of that trend indicates there will be no new cases on day 401 (5 April 2021), a long time away. Partly this slow decline is due to increases in new cases in a number of states (see below).
· The number of new tests continues to rise linearly (Fig. 2), a good trend.
· The number of new cases per test also continues to decline in the US outside of NY (and in NY) (Fig. 3; P = 0.0087 after day 78). Projection of this trend yields a 0-new-cases day of 222 (8 October). We will have to see if this trend continues downward. If not, there will be new infections after this date.
· The US data on cumulative new cases per test continues to be fit pretty well by a modified logistic growth curve (Fig. 4), although there still some small systematic deviations, with the actual data being more linear recently than the fitted curve.
· Projections of this curve still indicate that most of the first wave of infections has already passed. Specifically, for the US outside of NY, on days 104, 137, and 177 (12 June, 15 July, and 24 Aug, respectively—these dates are about 1-2 days later than projected last time because of the deviation from the curve; they are also in line with the projected 0-new-cases-per-test day above), 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 90,373, 18,505, and 2,619. On days 135 and 173, the expected minimum number of new cases will be 921 and 132.
Fig 1 Fig 2
Fig 3 Fig 4
Fig 5 Fig 6
US Deaths
· The downward trend in new deaths in the US outside of NY (and in NY) continues unabated (Fig. 6). This trend remains linear, and predicts there will be no new deaths on day 141 (19 June)—4 days from today! If that seems too soon, bear in mind that yesterday there were only 284 deaths in the US outside of NY, down from a peak value of around almost 3,000. Day 207 (23 August) is within the margin of error. These dates remain unchanged from last time, and bode well for the passing of the effects of the first wave of infections.
NC Cases
· The number of reported new cases in NC continues to trend upward (Fig. 7). Moreover, this upward trend appears to be accelerating: the quadratic term in a regression of new cases vs. day is highly significant (P < 0.0001). This is obviously potentially bad news.
· This trend is also seen in number of new cases per test (Fig. 8). Here the trend is increasing linearly after day 73 (12 April) (P = 0.0003). It’s not clear what the cause of this increase is. It started long before NC began relaxing social-distancing practices and started “opening up”.
Fig 7 Fig 8
NC Deaths
· The number of deaths per day in NC continues to plateau, with no statistical evidence for either an increase or decrease (Fig. 9).
Fig 9
Southern Trends
You may have read that a number of states have recently shown an increase in new covid cases. Most of these reports give the number of reported cases, so those trends could be explained largely by an increase in testing. Below, I show data for number of new cases per test, designated “ratio”, for seven southern states (FL, SC, GA, AL, LA, TX and LA; data from
https://covidtracking.com/api ). These are states that have taken aggressive approaches to “opening up” their economies. As the figure shows, in all seven states there is a significant upturn in new cases per test (quadratic regression, with quadratic coefficient significant at least P < 0.05, most with P < 0.001).
In each of these states, the upturn starts around day 80 (19 May), roughly corresponding to when these states opened up. However, we should be cautious about ascribing cause to these upturns, since NC shows a similar uptick (see above), but has been much more cautious about opening up. Moreover, while my analysis shows there has definitely been a reversal of the downward trend in these states, it is too early to tell whether the upward trend will continue or begin to plateau. Nevertheless, the contrast between what’s happening in these states and in the US as a whole, which shows a steady decrease in number of new cases per test, is striking, and should give everyone pause. We have a foreshadowing of a very bad trend that everyone, including politicians, should pay attention to and be proactive about.
Conclusions
As a whole, trends in the US are promising. The number of new cases per test (which is likely a good indicator of the real number of new cases), continues to decline, with a predicted 0-new-cases day of 8 October. Projections from the cumulative new cases per test are even more optimistic, with a projection that all but 0.1% of infections from the first wave will have occurred by 24 August. Deaths also continue to decline steadily, with a projection of 19 June as the day when no new deaths will be recorded.
Of course, this all assumes that the trends continue as they have. However, the data from southern states indicate that an upsurge of infections is beginning. That clearly means these dates will be pushed back to later in the year. Just how far, we’ll have to wait and see. Clearly, though, these new trends are bad news and need new efforts to reverse them.
In North Carolina, similar adverse trends are continuing. The number of new cases per test seems to be increasing at a steady rate, and death rates are showing no sign of a downturn. Even if the “opening up” is not the cause of these trends in NC, we will still need to double down on social-distancing measures to reverse these trends. Let you politicians know you support this (if you do).
Stay safe!
