Coverage of disease modelling tends to focuses on population-level 'what could happen next?' scenario analyses and predictions. So I wanted to highlight some early COVID-19 insights that you've probably heard about, but may not realise modelling groups were involved in... 1/
An early (and widely reported) suggestion that the COVID outbreak was much bigger than the 41 initial cases reported in China in early Jan came from analysis by @MRC_Outbreak of cases exported to other countries. bbc.co.uk/news/health-51… 2/
A similar analysis by @AshTuite @BogochIsaac et al in Feb flagged that Iran probably had a large undetected outbreak, and that global containment of COVID would therefore be unlikely. medrxiv.org/content/10.110… 3/
Early estimates of the risk that infections would appear in new countries provided useful (and rather prescient) situational awareness, e.g. this by @alexvespi and colleagues datastudio.google.com/u/0/reporting/… 4/
The incubation period for COVID is widely quoted at 5-6 days - we know this in part thanks to modelling groups at Hopkins (annals.org/aim/fullarticl…), RIVM (eurosurveillance.org/content/10.280…) and Hokkaido (mdpi.com/2077-0383/9/2/…) 5/
Early evidence that COVID-19 was capable of sustained human-to-human transmission also came from modelling studies, e.g. biorxiv.org/content/10.110… 6/
There's growing evidence transmission can happen before people have clear symptoms, again based in part on analysis from modelling teams, e.g. ijidonline.com/article/S1201-… / cmmid.github.io/topics/covid19… 7/
Early identification of substantial transmission in Washington State came from phylogenetic modelling by @trvrb et al bedford.io/blog/ncov-cryp… 8/
Several studies from modelling groups have adjusted for delays between symptoms and death, and under-reporting of cases, to estimate that 1-1.5% of symptomatic COVID-19 cases were fatal in China. medrxiv.org/content/10.110… / eurosurveillance.org/content/10.280… / nature.com/articles/s4159… 9/
It's important to have rigorous critiques and reviews of modelling studies, but equally critics should be careful about making sweeping statements about disease models, when the modelling contribution to the evidence base has been wider and more nuanced than many realise. 10/10
