Some people started following me after we published how heterogeneity affects the so-called "herd immunity threshold" (HIT). This remains needlessly controversial, so here's a summary of the discussion (or lack thereof) so far. 1/N
The controversy is unwarranted because it is common knowledge in mathematical epidemiology that the threshold 1-1/R0 is a feature of the simplest model, which assumes every individual is equally connected to everyone else (i.e. "homogeneous mixing"). 2/N
The ~60% value for the herd immunity threshold is an upper bound based on those assumptions and the estimate of the reproductive number R0=2.4. Realistically, contact networks are heterogeneous, i.e. some people have more contacts than others. That immediately means HIT is lower.
This doesn't say how much lower it is; that depends on the variance of the distributions of either contacts (exposure) or susceptibility (biological), but any analysis that looks at serological prevalence alone takes homogeneity assumptions for granted 4/N
Now there was a suggestion that the shift from HIT=1-1/R0 could go either up or down with different contact structures; that is not actually true: heterogeneity will always decrease HIT compared to the "homogeneous mixing" model. 5/N
What is true is Poisson connectivity is often considered baseline in the study of networks, because that's what you get when you randomly assign number of contacts to individuals based on some population mean.
It may be baseline, but it is not homogeneous.
It may be baseline, but it is not homogeneous.
I understand worries around politically rather than scientifically-motivated work, as well as cherry-picking or taking results out of context to downplay the severity of #COVID19. I made very clear the assumptions and limitations I can see in our work. 7/N
What I don't understand is why @ScienceMagazine would admit to not publishing accurate science out of fear it'd be misrepresented by some groups (actually I kind of do), especially when it was available on medRxiv, and misrepresentation is rampant anyway
That is probably a reason why our work had trouble being published in the same journal, despite the shitshow of low-quality, flawed, and biases studies being published at record rate about #Coronavirus/#COVID19 9/N
So our work is still not published in a peer-reviewed journal, and maybe it doesn't matter that much. It's been viewed over 100,000 times and discussed widely. It brought the idea of the effect of heterogeneity in population immunity into the discussion.
That said, I wish researchers in the field would step up and peer-review our work post-publication on @pubpeer or equivalent. I'd appreciate help getting a few unbiased reviews: @mlipsitch @joel_c_miller @Caroline_OF_B @nataliexdean @CT_Bergstrom
I somewhat doubt it, but it would be interesting to have a preprint on @medrxivpreprint with more impact that a @ScienceMagazine
