June 11 covid-19 update

I'll start today with some very good news from South Africa. There has been a lot of concern regarding how co-infection with HIV and TB will affect risk of death from covid-19, and South Africa is the first place in the world to have data on this issue. The good news is that co-infection with HIV or TB seems to increase fatality rates from covid-19 only 2-3 times. It's still an open question how much the "benefit" of South Africa's young population will count against factors like high prevalence of HIV and TB and limited healthcare resources, which will increase fatality. But I am really encouraged to hear that TB and HIV don't make a bigger impact on fatality, and I'm going to go on record as saying that I expect fatality rates in South Africa will stay lower than those in older populations, including the U.S. Let's keep watching the data on this.

This article summarizes the research about TB and HIV co-infection, which is fully presented in this hour-long webinar. The webinar also covers a lot of other issues, such as negative impact of the pandemic and lockdown on health services and adherence to medications for people living with HIV and TB. While younger adults have a higher risk of TB and HIV and a much lower for risk of dying from covid-19 than older adults, older adults have a higher risk of high blood pressure and diabetes which increases their (already high) risk of death by covid-19 a lot more than TB or HIV. Uncontrolled diabetes increases risk of death by covid-19 approximately 13 times, and more than half of covid-19 deaths in Western Cape to date have been among individuals with diabetes.

South Africa passed 1 million tests today, which is also good news. The bad news is that South Africa has more than doubled its number of cases and deaths in the past 2 weeks. Cumulative deaths from covid-19 increased from 577 on May 28 to 1284 on June 11, and 3/4 of deaths (966) have been in Western Cape. These numbers are still quite low; Brazil is currently seeing around 1000 deaths per day, and at its peak the U.S. was seeing around 2000 per day. But the trend isn't good. While the Ministry of Health calculated the mortality rate at 2.2%, my estimate of South Africa's case fatality rate (CFR) is 4.7%, although this is still lower than the U.S. and only a third of Italy's. (As a reminder, I divide deaths to date by cumulative infections as of 2 weeks ago to account for lag between infection and death.)

The U.S. is running in place - there's still lots of activity (e.g. increasing testing rates), but we're not really getting anywhere against this virus. We passed 2 million confirmed cases this week. This Atlantic article provides a good summary of the state of the U.S. epidemic, concluding, "As the long, hot summer of 2020 begins, the facts suggest that the U.S. is not going to beat the coronavirus. Collectively, we slowly seem to be giving up... Americans have not fully grasped that we are not doing what countries that have returned to normal have done." Americans stayed home less than Europeans (according to mobility data) during our piecemeal "lockdown", and now are staying home even less as restrictions lift.

Cases and hospitalizations are increasing in a number of states. In the most populous state (California), daily cases are increasing and the daily death count has seen little decrease since mid-April. My brilliant, California-dwelling epidemiologist friend Dr. Rebecca Fielding-Miller was on San Diego news this week to talk about how we're not halting this pandemic, using the analogy that although we've been pumping the breaks, we're still barreling down the hill and have not stopped the car. With a population of 40 million, California is equivalent to a medium-sized country (2/3 the population of Italy and South Africa, or 4 Swedens). Here are some visualizations from the New York Times' website, which offers these graphs for any state (as well as a mountain of other state, county, and city-specific data for the U.S.).

If you want to know whether your state should be re-opening, take a look at the map below, which is from covidexitstrategy.org (lots more info there). States are assessed according to criteria established by the White House. There's a lot of red on this map.

A couple more regionals look at the U.S., from the Atlantic's Covid Tracking Project, which is releasing data-dense blogs every week or two. The first visualization shows that national declines in infections (as a percentage of tests) and deaths have been driven by the Eastern states (including of course the New York City metro area), but that actually these metrics are increasing for the rest of the country. The second image visualization, from the June 11 blog, makes the point of regional variation even more strongly, using data for new cases. This blog also notes that hospitalizations from covid-19 have increased in 11 states, which means climbing cases isn't just due to increased testing (which is increasing, but still below targets).

Tom Frieden, former head of the CDC, wrote an Op-Ed yesterday with the non-subtle title, "Amateur epidemiology is deterring our covid-19 response", which identifies a number of common mistakes in tracking the epidemic as well as data that would more effectively track the epidemic. He warns against relying too much on confirmed cases or number of tests done. As repeatedly discussed in this blog, we know confirmed cases are vastly underestimating true infections, and cases can increase due to increased testing alone. He also makes a good point that recommendations are constantly being updated in response to emerging data, but it's not that experts are making mistakes, but simply that we're making progress in understanding this disease. (Now and then I'm stunned to remember that only 6 months ago, no one on earth had even heard of this virus!)

Dr. Frieden also provides a link to a useful summary which distills in a few pages much of what we currently understand about the epidemiology of this virus. One hot topic this week has been what percentage of infections are transmitted by asymptomatic individuals. You may have seen the furor after a World Health Organization official (Maria Van Kerkhove, WHO's technical lead on the pandemic) suggested that onward infection from asymptomatic individuals is "very rare". Last week, a peer-reviewed article was published which claimed that asymptomatic persons contribute 40-45% of onward infections, and a number of other scientists have written in to say that this estimate is far too high (see Comments at the end of the article). I agree that attributing nearly half of onward infections to asymptomatic cases does not seem credible.

Much of this debate hinges on the difference between asymptomatic and pre-symptomatic. You may remember from an earlier post that cross-sectional analyses tend to over-estimate the proportion of infected individuals who are asymptomatic, and that most individuals do develop symptoms eventually. The research summary shared by Dr. Frieden cites research showing that people are most infectious in the few days around when symptoms emerge (but including the 2 days before symptoms appear), and that viral shedding is highest in the first week of showing symptoms. The document also included data on percent of cases that are asymptomatic, based on a number of studies. I was amazed to see that two studies of cruise ship populations (which we might expect to be similar in terms of age and other factors) gave wildly different estimates of percent asymptomatic - of 18% and 81%! Van Kerkhove is quoted as saying, “Every question we answer, we have 10 more."

In parting, here are this week's graphs (using data from the Johns Hopkins dashboard for Italy and the U.S., as the UVA dashboard was down today, so the numbers may be very slightly different than what the UVA dashboard would have shown - I can update them with UVA figures next week).