Day 74: R0 and Rt

Today's numbers are edging ever closer to 2 million cases, hovering around 1,950,000 at the moment. A few numbers have leaked out of North Korea: four doctors and 180 soldiers have died of coronavirus. (These deaths are not included in the world total.)

The UK has joined Italy and New York City in hitting the news for its statistically suspicious death rate. Just by checking death certificates for mention of COVID-19, the Office for National Statistics picked up 15% more deaths than previously counted. A redditor noted even more of a death disparity in the ONS chart summary points:

The provisional number of deaths registered in England and Wales in the week ending 3 April 2020 (Week 14) was 16,387; this represents an increase of 5,246 deaths registered compared with the previous week (Week 13) and 6,082 more than the five-year average.

Of the deaths registered in Week 14, 3,475 mentioned “novel coronavirus (COVID-19)”, which was 21.2% of all deaths; this compares with 539 (4.8% of all deaths) in Week 13.

That's a difference of 2,607 unexpected deaths still not attributed to COVID-19. Unlike Italy and NYC, the health care system in England and Wales is not overwhelmed, so there's not a lot besides coronavirus to explain the extra deaths.

Circumstances have forced PlagueBlog to add the USS Theodore Rooosevelt to the germ boat list at #23, due to a death onboard from coronavirus. The unnamed sailor was found unresponsive last Thursday, two weeks after having tested positive, and died in intensive care on Monday. The case count aboard #23 is 585 positive out of a crew of 4,800. Rumor has it that four other infected sailors are in a hospital ashore; presumably the rest remain aboard.

Likewise, the USNS Mercy hospital ship has earned the title of Germ Boat #24 with its own outbreak. In this case they are combatting the outbreak not in the Diamond Princess way of Germ Boat #23, but by removing the seven infected sailors and their contacts to isolation ashore.

R₀, the basic reproduction number, is a somewhat vaguely defined measure of the number of people one infected person will infect in a naive population taking no precautions. For example, R₀ for measles, an extremely contagious and truly airborne disease, is generally estimated at 15. The CDC has estimated R₀ for coronavirus at 5.7 (revising an initial estimate of 2.45).

R₀ is contrasted with the effective reproduction number R (or R_t, or R_e), which is the average number of people infected by one infected person in the current situation (of precautions, partial herd immunity, etc.). Gabriel Goh at OpenAI has written a fun tool for playing with these values and others to see how they affect "the curve". Kevin Systrom, cofounder of Instagram, has a writeup about R_t as a useful, local metric of how individual US states are doing. He's pretty pessimistic about Massachusetts and Rhode Island:

Which states have the epidemic least “under control?” To answer this, I plotted states where the best case (eg. low end of HDI) is above 1.0, indicating the true value of Rt is almost certainly above 1.0. Surprisingly, Rhode Island, Maryland and Massachusetts sit at the top. Part of this may be that Rhode Island is earlier in its infection curve, but seeing large states like Massachusetts and Texas above 1.0 is worrisome—especially because none of these states have hit the headlines as being trouble spots.

P.S. Massachusetts' numbers are up 5% again today. Hospitalization numbers have dropped out of the report for undocumented reasons. (Speaking of which, the Boston Globe published a hospital-by-hospital breakdown of coronavirus cases, though Steward Health Care hospitals are omitted.) Middlesex holds a commanding lead over Suffolk County, and at least one Middlesex County mayor has decided to confuse matters by issuing hiss own mask guidelines in contradiction of the CDC and state guidelines.