Tweet

Jon Levy

5 Aug, 12 tweets, 4 min read

1. I've been frustrated with how #COVID19 cases among the vaccinated have been reported, and I feel like it can be done better. I'm not an expert in this space, but I wanted to toss out some thoughts to #epitwitter and see if some collective wisdom could emerge. 🧵

2. First, to be clear, this is about public presentation of data in the media, not optimal study design to determine vaccine efficacy. My premise: Reporting just number of "breakthrough" cases lacks context, and "breakthrough" cases divided by number vaccinated lacks meaning.

3. So what is meaningful? Starting with cases, I'd argue that new cases per 100,000 vaccinated vs. new cases per 100,000 unvaccinated is a good starting point. CA does this - for the past week, it was 7 for vaccinated and 33 for unvaccinated.

cdph.ca.gov/Programs/CID/D…

4. The temptation is to jump from there to vaccine efficacy. Obviously this doesn't control for differences in populations, public health measures, etc. But it is far more useful than saying "99.99% of vaccinated people didn't get COVID", or "there were x breakthrough cases."

5. And you don't need to say "the vaccine gave you an 80% reduction in risk of getting COVID". You can say "COVID cases among the vaccinated were 80% lower than among the unvaccinated this week." Simple, direct, and factual.

@BostonGlobe

6. But the reporting needs to also include hospitalizations and deaths. @BostonGlobe reported today that hospitalizations in MA among the vaccinated increased by 92 from Jul 10 to 31. But that was it. What does 92 mean to the average reader? Is that a big or small number?

7. If I read MA DPH raw data correctly, there were 480 new hospitalizations over this period. If ~ 4.3M were vaccinated and ~2.6M were unvaccinated, that would mean 15 per 100,000 unvaccinated vs. 2 per 100,000 vaccinated hospitalized. Ballpark, of course.

8. Here, differences in population matter more than for cases (especially the fact that many of the unvaccinated are under 12). Better to remove young kids from numerator and denominator, or age adjust. I can't do that from public data, but the state agencies sure can.

9. But even without these refinements, this comparison still seems useful at first order, vs. how things are reported now. Doing this for deaths is trickier, given lag periods and changing denominators. But simple factual statements with caveats could be made.

@BostonGlobe

10. This feels like it would give people much better info. Not to pick on @BostonGlobe, since every media outlet reports the same way, but today's article said the number of breakthrough cases grew 75% during this 3 week period. That could create panic without a comparison group.

11. On the other hand, the same article said that just 0.18% of vaccinated people had breakthrough cases, which could give people a misleading sense that vaccines offer 99.8% protection without a comparison group.

12. So, #epitwitter and those more well versed in the world of ID epi than me - how do you think major media outlets should be reporting "breakthrough" cases, hospitalizations, and deaths? Do my simpleminded calculations help or hinder? How do we increase numeracy?

• • •

Missing some Tweet in this thread? You can try to force a refresh

This Thread may be Removed Anytime!

Twitter may remove this content at anytime! Save it as PDF for later use!

More from @jonlevyBU

Jon Levy

@jonlevyBU

26 Feb

Let me make sure I understand - with data (below) showing that #COVID19 is no longer declining in MA, with the more infectious variant increasing, and the goal to open schools in April, we are opening restaurants on Monday at full capacity with musical performances? #mapoli

@EricTopol

Here are some recent studies (graphics borrowed from @EricTopol) which reinforce risks associated with in-person dining 👇

At least we prioritized vaccinating food service workers to make sure we protected this vulnerable population before reopening.

Oh, wait...

Read 6 tweets

Jon Levy

@jonlevyBU

7 Feb

1. Just celebrated 20 yrs as a professor this week, so I’ve been thinking about life in academia.

I have only one piece of advice - know yourself, use that to figure out what success means to you, and find people who will help you achieve that success.

Let me explain... 🧵

2. You will meet a number of potential mentors along the way. Some will be lousy. Some will be good but will guide you toward their notion of success. The great ones will ask you first what you want to achieve, and strategize with you about how to get there.

3. There are so many dimensions of academia. Do you want to devote most time to teaching? Research? Administration? Why? And what would it mean TO YOU (even if not to others) be successful in each area?

Read 11 tweets

Jon Levy

@jonlevyBU

22 Aug 20

1. There has never been more focus on ventilation in school buildings. This leads to the obvious question - what do we actually know about ventilation in schools, and how important it is?

THREAD

2. First, a primer. Many studies use indoor CO2 to characterize ventilation, because the primary source in a school building is people breathing, and the outdoor concentration is relatively stable.

3. A recent review article by Fisk reported that ALL studies in schools found average CO2 of over 1000 ppm, a sign of a poorly ventilated space (outdoor CO2 is roughly 400 ppm).

onlinelibrary.wiley.com/doi/full/10.11…

Read 9 tweets

Jon Levy

@jonlevyBU

16 Aug 20

1. This has been a week of contentious meetings, angry online debates, and complex conversations about school re-opening.

One key topic: how risky is it to return to school? The literature on risk perception may help to increase understanding and improve communication.

THREAD

2. I think the brilliant work of Paul Slovic, as described in his article "Perception of Risk" in 1987, makes it clear how we are talking past each other when sifting through data on the health risks associated with returning to school.

science.sciencemag.org/content/236/47…

3. Slovic found that there was a large gap between how risky people thought different activities were and what experts thought. But his key insight was this is not because the public was ignorant - it was because people saw risk as more than just expected number of fatalities.

Read 9 tweets

Jon Levy

@jonlevyBU

8 Aug 20

1. I’ve been thinking a lot about our tendency for “bright line” decision making and what that means for how we approach school reopening decisions this fall...a little wonky, but indulge me, since I think it matters...

THREAD

2. Epidemiologists are quite familiar with this phenomenon, as p = 0.05 is treated as the magical bright line by some. If p = 0.049, it is significant and “real”. If p = 0.051, it is insignificant and “not real”. Many have written on this topic ⬇️

amstat.tandfonline.com/doi/full/10.10…

3. How does this connect to #COVID19? One place is the 6-foot “rule”. It is treated (by some) as a magical bright line. 5’11” - deep trouble. 6’1” - no risk, all is well.

The reality, of course, is that there is a continuum of exposure risk.

Read 11 tweets

Jon Levy

@jonlevyBU

11 Jul 20

1. What do exposure science and risk assessment, 2 core methods of environmental health, tell us about how to think about opening K-12 schools this fall?

THREAD

2. First, some definitions - exposure science involves the study of real-world contact with and uptake of toxic materials than can cause heath effects.

nature.com/articles/75004…

3. Risk assessment estimates the nature and probability of adverse health effects to people who may be exposed to toxic materials. And risk management tries to figure out what to do to reduce the risk.