Thread by @DrEricDing

Mulling 🇬🇧 disclosure that new #B117 variant could increase #COVID19 deaths ~30%. 🇬🇧 is expected to prolong & tighten lockdown—Schools may close until Easter, overseas travelers maybe quarantined in hotels for 10 days. I’ve looked at data—it’s worrying. 🧵
nytimes.com/2021/01/24/wor…

2) Notably, 6 out of 10 studies conducted on the mortality risk show significantly increased risk of #COVID19 death, ranging from 28% to 91% higher relative risk. They are circled in red below. A 7th (purple) study also shows a 30% increase, albeit not quite sig, but compatible.

3) Notably, I want people to focus on the two studies from @LSHTM - circled in green. These were prospective cohorts that used full Cox models that detailed followed individual person-time from beginning to end—much better than others. These 2 show 28-35% #COVID19 death increase.

4) The other intriguing study is the Exeter study that reported the 91% increase—much higher. It is also prospective I believe (else it’d say retrospective) and reported Hazard Ratios like the 2 LSHTM Cox Models, which means they likely did detailed risk set sampling (very good).

5) One key thing when reading epidemiology is that not all data is equal. and remember while correlation is not necessarily always causation, every *causation exists as a correlation*. The difference in knowing which is which—is the science of epidemiology. And prospective best.

6) And in epidemiology, not all data is alike. Randomized trials (usually) best—keyword is usually—unless u screw it up. Next best is PROSPECTIVE COHORT studies — that are often analyzed via Cox model that estimate hazard ratios like the two LSHTM studies with 28-35% increase.

7) for this reason, with the 6-7 out of 10 studies show increased risk, and 3 showing no higher risk, I tend to believe the **PROSPECTIVE STUDIES** generally. The 3 null ones were either retro or used odds ratio (usually not prospective data). Follow the best data. Not all equal.

8) And that’s why the UK 🇬🇧 researchers highlighted the two LSHTM studies and the Exeter study — and the Imperial one also not retrospective. All of these showed higher risk. Hence their conclusion (circled red) below.

9) BOTTOMLINE: Though we need more data, I think the increased #B117 mortality risk is somewhat likely real, and should be taken seriously, at least in precautionary principle. And we know B117 is a beast already w/ 60% (40-80%) more infectiousness, it is blowing up our control.

10) Hence, also why Denmark CDC (@SSI_dk) is also incredibly alarmed by the **70% increase per week** of #b117 that is taking over Denmark. They are now genome sequencing *very single case* 🇩🇰 starting this week. That shows they are uber serious and concerned—see 🧵 #COVID19

https://twitter.com/drericding/status/1353297240610189315

11) For this same worrisome reason on #B117 variant’s infectiousness and potential deadlier effects, we definitely need to switch to premium masks... specifically the KN95, KF94, PPF2 varieties... many EU countries mandating it.

https://twitter.com/drericding/status/1353186978519539712

12) To be clear—Before we thought it was just more contagious—which is also bad because more contagious means infecting more people and raising the N dead. But now the new data says it’s deadlier by itself as well by ~30%. That’s what my whole thread is about—#B117 double whammy.

13) Forgot to include the link to the UK 🇬🇧 NERVTAG report

New and Emerging Respiratory Virus Threats Advisory Group (NERVTAG) on new coronavirus (COVID-19) variant B.1.1.7.

From:
🇬🇧 Department of Health and Social Care and Government Office for Science
assets.publishing.service.gov.uk/government/upl…