Moderna’s trial so far included 30,000 people: like the Pfizer trial, subdivided into two well-matched halves, half of which got the vaccine and half of which a placebo. From the NYT article:
In Moderna’s study, 95 people contracted the coronavirus: five who were vaccinated, and 90 who received placebo shots of saltwater. Statistically, the difference between the two groups was highly significant. And of the 95 cases, 11 were severe — all in the placebo group.
Moderna hence claims about 95% efficacy. From the BBC:
Moderna’s vaccine appears to be easier to store as it remains stable at minus 20C for up to six months and can be kept in a standard fridge for up to a month.
Pfizer’s vaccine needs ultra-cold storage at around minus 75C, but it can be kept in the fridge for five days.
Important, neither Moderna’s nor Pfizer’s vaccines appear to have severe side effects — though malaise and fatigue for 1-2 days seems to be reasonably
(2) Dr. Campbell discusses the Pfizer vaccine, and mRNA vaccines more generally, how they work, and the shape of the year to come. (Moderna’s data were not out yet at the time of making the video.)
Here is Dr. Seheult, again on the Pfizer vaccine but also mRNA vaccines more generally. (This is a new, promising technology. Traditional vaccines contain either attenuated virus, or viral proteins that by themselves elicit an immune response.)
In this slightly older video, Dr. Campbell hails a “game changer in [medical] research” — a pseudonymized database of NHS patients that can be searched by various (combinations of) criteria to extract data for retrospective studies: https://opensafely.org About 40% of all NHS patients in England are in this database. (Israel’s largest HMO, Clalit Health Services, has been operating a similar database for some time. See for example this preprint.)
Pseudonymized means, in plain English: you can identify individual patients by record ID in the system, but you cannot find out who is the person corresponding to the record.
One example of a question one can “ask” quite quickly is: take all the patients with lupus or rheumatoid arthritis (194, 637 of them), find the subset who are taking hydroxychloroquine as an immunomodulator for these conditions (30, 569 patients), and compare the fatality rate from COVID19 between the groups with and without HOCq. Turns out, it’s the same to within statistical noise (0.23% vs. 0.22%).
(3) You have probably heard that Elon Musk had himself tested 4 times in a row on the same day and tested twice positive, twice negative. How is such a thing possible?
Well, we already know that RT-PCR testing in the field (i.e., with actual patients) has a false negative rate of about 30%. (With laboratory samples, it’s negligible — the problem lies primarily with the sampling.)
So if you have 30% false negatives and you are positive, what are the odds that you would pull two out of four tests positive? This is actually a simple statistical exercise: there are sixteen possible outcome combinations for four tests
- all positive: one combination. Odds: 0.70^4=24.01%
- one negative, three positive: four combinations. Odds: 0.70^3 x 0.30 x 4 = 41.16%
- two negative, two positive: six combinations. Odds: 0.70^2 x 0.30^2 x 6 = 26.46%
- three negative, one positive: four combinations. Odds: 0.70 x 0.30^3 x 4 = 7.56%
- all four negative: one combination. Odds: 0.30^4=0.81%
(In general, the odds are (1-p)^k x p^(n-k) n!/(k! (n-k)!) where p is the fraction of false negatives, n is the number of tests, and k the number that turns up positive.)
In other words: purely based on 30% false negatives, if you have four tests done in a row, you have about one chance in four to get the same outcome as Elon Musk. (BTW, good to know: if you have just two tests done in a row, there is still a 9% chance of a false negative: the odds are 49% both tests turn up positive and 42% that you get one positive and one negative.)