Important update:
The total number of U.S. Covid-19 cases that is being reported (currently 215,344) includes 3 categories of cases:
1) Currently active cases (201,354 when this is being posted on 4/2/20).
2) Recovered cases (8,878 when posted on 4/2/20).
3) Deaths attributed to Covid-19 (5,112 at the same time on 4/2/20).
https://www.worldometers.info/coronavirus/
-------
What is important about this information?
1) When you see the term "Total number of Covid-19 cases," it refers to the total number of persons who have or have had the disease. Strictly speaking, this is referred to in epidemiology as the "cumulative prevalence" of Covid-19.
2) The cumulative prevalence is higher than the current or "point" prevalence of Covid-19, and the longer the pandemic goes on, the less and less the total cumulative prevalence will tell us about the current prevalence of the disease.
3) The total number of currently active cases is much more important than the number of past cases (including those who have recovered) in estimating the future trajectory of the pandemic.
4) Including "recovered" cases in charts that are intended to be used to help guage effectiveness of Covid-19 prevention efforts obscures, rather than clarifying the changes in current Covid-19 prevalence from day to day.
Q: Does this mean that the upward trajectory in the number of cases that we're seeing in most graphs is being exaggerated, due to the misguided inclusion of recovered cases in the graphs?
A: Yes.
Q: How can we correct for this, to have a better idea of how well our prevention efforts (social distancing, sheltering in place, etc.) are working to date?
A. Until we begin to see graphs that distinguish between current and past cases, the best we can do is use available data on recovered cases from other sources.
Q: Are records being kept as accurately about recovered cases as they're being kept about deaths and current cases?
A: No. Of all the patients diagnosed with Covid-19, only a subset (mostly those with severe cases that resulted in hospitalization) have had their outcomes tracked and recorded.
Q: So the actual number of recovered cases could be substantially higher than the number of recovered cases that have been confirmed to be recovered?
A: Correct, although this won't become fully apparent for another week or two, until it has been at least two weeks since most of the persons with Covid-19 first developed the illness. In the coming weeks, the total number of recovered cases will approach and then surpass the total number of active cases.
Q: The graphs that have been charting the growth curve trajectory of Covid-19 in the U.S. and overseas - - have they been sorting out the recovered cases?
A: Generally, no, they haven't, partly because most nations haven't had a huge number of recovered cases yet.
Q: Isn't the purpose of a "growth curve trajectory" graph to plot the growth in the number of "active" as opposed to "past or present" cases?
A: Yes, insofar as the goal is to use the growth curve trajectories to make predictions about the future impact of the illness on each nation, and to gauge the effectiveness of interventions or need for additional interventions.
Q: Let's say we start now, by only including the active Covid-19 cases in the U.S. growth curve figures. What would that tell us?
A: It would tell us several things:
1) That the slopes in most of the growth curve figures that have been published to date have suggested that the prevalence of current Covid-19 cases has been growing slightly to somewhat more quickly than it actually has.
2) That the total number of active U.S. Covid-19 cases (201,686) has been somewhat lower than most of the published graphs and reports have suggested.
For example, when it is reported that the total number of U.S. Covid-19 cases is over 215,000 in the morning of April 2nd, 2020, this is misleading, because it includes persons who recovered from the disease and those who succumbed to it.
The actual total number of active U.S. Covid-19 cases is 7.5% lower than the overall total number of (past and present) cases, but no one has been making that clear as of yet.
3) It has taken more than 5 days for the number of U.S. Covid-19 cases to double since March 28th, when the total number of cases was reported to be 104,686.
Q: How does this compare with the current charts that show the time it takes the number of U.S. Covid-19 cases to double?
A: The current charts indicate that, at present, it is only taking 3 days for the number of U.S. cases to double.
Q: What is the reason for the discrepancy, if it is clear that it took at least 5 days for the 104,686 cases to double?
A: The main reason is that those charts have been exaggerating the rate of increase in the number of U.S. Covid-19 cases, because they have been plotted based on the day to day "moving averages" in the numbers of cases. Since the doubling times in early March were as low as two days, those numbers have had an undue (and distorting) impact on the growth curve trajectory plots in the graphs.
Q: Is it ok to plot growth curve data using "moving averages" rather than actual numbers of active cases from day to day?
A: Both kinds of charts can be used to plot changes over time in any outcome of interest. Some people are particularly interested in moving averages (such as in charting prices in the stock markets) but they provide different kinds information.
However, it is misleading to present a graph based on moving averages without making it clear to the reader that the growth curve trajectory looks much steeper when it is based on a moving average with a very rapid initial rate of increase.
While a moving average can be useful by emphasizing the central tendency in the data, and smoothing the curve (making it look less jagged and more regular), it has the disadvantage of obscuring potentially informative changes in the data that should not be ignored.
In the case of plotting the growth curve trajectory for U.S. Covid-19 cases, it would have been better to plot the actual increases in the "point prevalences" for each date, rather than combining these numbers into "moving averages."
When the actual increases in prevalence are used, rather than the moving averages, they show that, while Covid-19 has been increasing steadily in prevalence in the U.S. during the past three weeks, there has been much more progress in reducing the rate of increase in Covid-19 cases than the charts displaying moving averages have suggested.
Specifically, while the charts based on the moving averages suggest that the number of U.S. cases is doubling every three days, the actual doubling time that is calculated on the basis of the point prevalence on each specific date shows that - as of the past 5-6 days, it has actually begun to take more than four days for the number of U.S. Covid-19 cases to double. It has taken 5 days for the number of Covid-19 cases on March 28th to double, as of 4/2/20, and, with an estimated average increase of ~28,000 cases per day, based on the current rate of increase in new cases, it is mostly likely that the actual doubling time will soon approach 6 days (i.e., the current rate of increase in new cases supports an estimate of 5.88 days for the 164,620 U.S. cases reported at midnight on March 31st to double).
Q: Why does this matter?
A: It matters because it tells us that our efforts to slow the rate of increase in the number of active Covid-19 cases in the U.S. are starting to have the desired effect. This tells us that we should keep at it, and that if we do, the growth trajectory can be flattened more quickly than many may realize.
Q: Will that be the end of the Covid-19 epidemic?
A: No, but it will be a key step forward. Once the rate of growth is flattened out to the point that it takes more than two weeks for the number of active Covid-19 cases in the U.S. to double, there will be one recovered case for every new case, and the number of active cases will stop growing at that point.
Data from China and South Korea has shown that continuing the prevention efforts (social distancing, washing thoroughly, wearing face masks, self-quarantining upon infection with Covid-19) would most likely cause the number of new cases to decrease each week.
------
So, there are two types of important milestones to work toward:
1) Keep increasing the doubling time until it takes more than two weeks for the number of U.S. Covid-19 cases to double. We've already increased it from 2 days to 5 days, and we've done that since the middle of March. That was the hardest part. Now, the slowing of the growth trajectory is coming along in bigger chunks of time. When it takes more than two weeks to double, it will stop doubling altogether and the growth in the size of the epidemic will cease, because there will then be one newly recovered patient for every new case of Covid-19 (i.e, zero "net" new U.S. cases).
2) Maintain the very same efforts that are gradually slowing the rate of growth, because once the rate of growth is brought to zero, continuing to make the same efforts will cause the number of new Covid-19 cases to decline.
3) There are some promising developments in medical research that are likely to improve patient outcomes and hasten recovery from the disease. In addition, there is some possibility that medical research will help to prevent Americans from developing Covid-19 in the not-too-distant future. For example, a top researcher stated on a major network at 11 pm Wednesday evening that an important research breakthrough with respect to an existing (already FDA approved) anti-viral vaccine is likely to be announced before the end of this week.