Top news, reports and insights for today:
- Breakthrough: First U.S. evidence about community seroprevalence from Santa Clara California
I’m excited to report that today’s top headline is about the announcement of results from a vital study of community seroprevalence of infection with SARS-CoV-2 in Santa Clara county, one of the first hot spots in the U.S. outbreak. The new study being covered by C.N.N. and many other outlets is about the first attempt to document the true extent of the epidemic by testing an entire community, not just those who are sick. Epidemiologist like me have been trumpeting the need for this information since the beginning. There are two main take-away messages from this study. First, as expected, the number of infections is substantially greater than our current testing suggests. In this study of blood samples from 3,300 residents of Santa Clara county recruited through Facebook, between 2.5 and 4.2% were found to have antibodies that suggest prior infection with SARS-CoV-2. Somewhere between 48,000 and 81,000 people in the county have been infected long enough ago to have mounted an antibody response. That’s 50-80 times more cases than current testing can account for. It’s an estimated range rather than 1 hard number to account for the uncertainty in the calculation. This is the first U.S. evidence I am aware of that tells us the seroprevalence (or what proportion of the population has been infected). The second vital clue that this study provides, which is an extension of the first, is that they can now approximate a truer picture of the case fatality rate (CFR). Doing so was not the purpose of this study and there are a bunch of assumptions made to get to this number, but they are approximating a CFR of between 0.12-0.2%. That estimate places COVID-19 within the range of lethality seen for the seasonal flu. It is also orders-of-magnitude smaller than the WHO estimate of 3.4%, which is biased upward by the low ascertainment rate (the proportion of all infections that are captured in testing). Now for some important caveats. First, this study was released as a “preprint” and has not been peer reviewed. There may be big flaws that haven’t been adequately checked. Also, they recruited participants through Facebook, which means its not a representative sample of the community. That’s a big deal, because the strength of a seroprevalence estimate hinges on how well the sample matches the population studied. They tried to make some adjustments, but we will have to wait to see if other studies arrive at similar conclusions. There is a second study in Los Angeles that has a better sampling design and it’s due to release findings on Monday. Additionally, these results are based on samples taken on 2 days in early April. We need to repeat this approach using data from other time frames to make sure our results aren’t biased. Also, the antibody test they used is still being studied; we don’t yet know how much we can trust it. Inaccuracy in that particular assay could bend results up or down. It is crucial to keep in mind that this study, while it uses data on antibodies, does not tell us anything about immunity (see the article in today’s Top Pick of the Day). We still don’t know whether the presence of antibodies tells us who is protected. That will take time.
The bottom line: Taken as a whole, this is an important breakthrough. When combined with an earlier small seroprevalence study in Germany, it tells us that the epidemic is much more widespread than our current testing indicates, and that the overall lethality of the disease is likely to be closer to the seasonal flu. This provides much-needed information for planners, epidemiologists and the public.
- U.S. deaths hold steady, rising 7%
Today, an additional 2,070 deaths were reported in the U.S., a cumulative increase of 7% (see today’s graph). While the absolute numbers on Friday were lower than Thursday, this doesn’t count as a decline because the difference is within the span of the error bars. Despite widespread talk of re-opening the country, we have not yet met the criterion for peak deaths (sustained declines in daily deaths). If anything, the numbers have been flat since April 7, with daily fluctuations in both directions. Because many states did not enact strong social distancing until the end of March, there is a 2-3 week lag in when those measures take effect. The flattening of deaths makes sense in this regard, and suggests that our epidemic control measures are working to some extent. We are helping hospitals keep up with demand, but it means the epidemic is likely to last longer at this pace. The White House coronavirus coordinator, Dr. Deborah Brix has said it won’t be safe to re-start the economy until we have achieved 14 days of consistent declines in new deaths. We certainly are not there now. As usual, the overall numbers mask state and regional differences. Over the last 3 days, hot spots are seen in Minnesota (+40%), North Carolina (+42%), Virginia (+50%), West Virginia (+57%), Connecticut (+54%), Delaware (+41%), Massachussetts (+47%), Maine (+44%), Maryland (+41%), and Rhode Island (+47%).