Useful analysis by @bencowling88 @gmleunghku et al of changes in case definition during the early COVID-19 outbreak in China, and what the underlying outbreak may have actually looked like medrxiv.org/content/10.110… 1/ 
In our early outbreak analysis (thelancet.com/journals/lanin…), there were clear differences between initial and later data in China, which is why we jointly analysed multiple datasets from within and outside Wuhan. This new study nicely illustrates why this was sensible approach. 2/
The authors also note reporting changes could have biased some early transmission estimates: "if we had estimated the exponential growth in the epidemic curve without accounting for the changes in case definitions, we would have substantially overestimated the growth rate" 3/
This could be issue in other locations too: "Our findings also suggest caution may be needed for analyses of the trajectories of epidemic curves elsewhere. Epidemics could appear to be growing faster than they actually are, because of rapid expansions in testing practices." 4/4
