I've been thinking about transitioning to a blog for a while, and this week finally did it. I've been frustrated by my inability to add images to Facebook posts, and want to be able to organize these posts a little better (such as tagging them by topic). Also, I have a hunch about 75% of my Facebook friends have snoozed me for the rest of 2020 because they are so tired of seeing coronavirus posts, so I thought it was time to take all that content off Facebook. My goal here is to have the world's simplest blog, but I have added a page of useful links. There are links below to comment and subscribe to receive notifications.
Another part of the makeover has been to redo the graphs to make them simpler and more readable, and add a graph for deaths by 100,000 population. I've expressed my misgivings before about whether we should really be using cases by population to compare countries, as we know that confirmed cases are only a small minority of all infections, and countries that are testing more will have more cases even if they don't have more actual infections than countries that are testing less. I think deaths are a better measure of an epidemic's severity, although we may see over time that countries with similar numbers of infection have quite different fatality rates. (Although one could argue that even in this case, we should assess severity by how many people died, and not on the basis of the number of relatively mild cases.) At any rate, I want to pay more attention to deaths going forward, and deaths per population seems like a more solid measure to me than cases per population.
So here we are, graphs of both cases and deaths shown on a linear scale, log scale, and by 100,000 population.
Notes about graphs: Week 1 in each graph is the week after the country reached 100 confirmed cases. All data points are cumulative through Thursday of that week. The U.S. epidemic is thus modeled as 1 week behind Italy's epidemic. South Africa's epidemic is modeled as 2 weeks behind U.S. epidemic.
South Africa continues to add new cases at a fairly fast rate; the number of cases is doubling every 12 days, and SA has added almost 7,000 cases in the past week (so approximately 1,000 cases a day). Of course, this isn't many compared to the U.S.'s over 20,000 new cases a day, but it's the slope of the curve in South Africa that we should be watching. As you can see, the slope is increasing in the graphs of cases.
At a national level, the U.S. hasn't yet seen an increase in cases as states have opened up in the last few weeks (and the rate of deaths has slowed down, as we've noted before) - but there is a lot of variation in how the epidemic is progressing at the state and county level. I've mentioned the New York Times' coronavirus dashboard before but it's worth mentioning again as it gives an incredible amount of data and detail down to county level.
Another great resource that I just discovered this week is University of Virginia's dashboard which has data for every state and county in the U.S. and virtually every country in the world. I'm going to be pulling my data from this site from now on. It's easier to use (in my view) than the Johns Hopkins dashboard and I have a bit more confidence in the data than the data at worldometers.com. Some data points line up exactly between UVA, JH, and Worldometers, and some vary by 2% or 3%. I'm not exactly sure why, and I'm not all that concerned about that small level of difference, but I have confidence in the UVA site and find it a lot of fun to use. Take a look if you want to produce your own visualizations or find out what's happening in a certain location. It will graph cases (confirmed and active), deaths, and recovered (including in stacked graphs), and will show those indicators plus tests on a map. They even offer 2-minute YouTube video tutorials to show you how to use the site.
There continues to be a lot of debate in the U.S. and South Africa (and I'm guessing all over the world) over whether the cure is worse than the disease (i.e. whether lockdown measures are going to do greater damage than the pandemic itself). I've written before about the three-way tug-of-war to protect public health, economic health, and mental and social health. And let me say that I think it's a completely valid question to ask whether lockdown measures will cost more lives than they save. (Recognizing that no one can give a definitive answer to that question, we can only give it our best educated guess using the information we have now.)
With that said, I think it's worth addressing a recent controversy in South Africa. Professor Glenda Gray, who is a respected physician and scientist, claimed in public remarks that she had been told by medical colleagues that there had been a sharp increase in child malnutrition as a result of the lockdown and called for an end to the lockdown (at least as currently being implemented). The Ministry of Health responded with a statement saying in part: "There has been a reduction in the number of cases of malnutrition that have been seen at Chris Hani Baragwanath Academic Hospital POPD [the hospital Dr. Gray claimed was seeing increases in malnutrition] and the total admissions during the month of March and April 2020, when compared to the previous 4 years. To illustrate this, in April 2019 there were 2885 patients seen and 500 admissions. However, in April 2020 there were 834 patients seen and 146 admissions."
I bring this up just to say that we should absolutely be looking for evidence of unintended consequences of pandemic control measures - but that we should rely on *data* in these discussions, and not just anecdotes, heartbreaking as individual stories may be. This pandemic is revealing a lot of fault lines and injustices that were already present in our societies, from South African schools lacking basic hygiene facilities to U.S. hospitals lacking the reserves of space and supplies to adequately care for a surge in patients. I'm so glad that so many people are paying attention to these injustices, getting angry, and taking action. But if those data about Chris Hani hospital are correct, there was a huge problem of malnourished children before the pandemic struck, too, and at least according to that one data point the problem isn't any worse this year. (I have no reason to doubt those data - but if there's more to the story, please let me know what I'm missing.)
Speaking of data, probably the best news I heard this week was that the CDC is embarking on a huge, year-long study that will give us SARS-CoV-2 antibody data from 25 US cities/metropolitan areas. Without such data, we will continue to be in the dark regarding basic questions such as how much of the population has actually been infected, how close we are to herd immunity, how the virus is spreading (and how fast), and what the actual infection fatality rate is. At present, there are so many questions we simply can't answer. I liked New York Times columnist Nicholas Kristoff's call this week for "epidemiological humility" regarding all we don't know yet. Beside the things I just mentioned, we're still learning about which routes of transmission are most efficient and which probably aren't much of a risk (such as surfaces, per newly-released CDC guidance), how the virus kills people, and long-term health effects for those it doesn't kill (such as children).
Finally, a mea culpa. Remember on May 12 when I said that the U.S. had reached an inflection point and that active cases had started to decline? I was too eager to see good news, and too quick to assume that the U.S. would follow the same path as Italy and other countries I'd looked at, in which active cases would hit a peak and then see a sustained decline. I shouldn't have declared a trend based on just two data points, and in fact on May 13 active cases started climbing again and are still climbing. Lesson learned; I won't do that again.
https://www.worldometers.info/coronavirus/country/us/ (accessed 21 May 2020)