Thread by @covidpath

Thread of commentary about the increased lag time between new cases and deaths:

1. About Simpson's Paradox, or why aggregating nationwide trends is misleading:

https://twitter.com/mbeckett/status/1278750652160634880

2. Understanding lead time bias

https://twitter.com/EpiEllie/status/1280305393516904448

3. The same pattern was seen in Iran

https://twitter.com/florian_krammer/status/1276326535901970432

4. Keep in mind that there are delays in reporting — this is why trends in general shouldn’t be deduced based on a handful of recent days.

https://twitter.com/Brief_19/status/1290545839744536576

5. I personally suspect that the strongest factor affecting how cases vs. deaths trends correlate is age of those testing positive. Check out this heatmap — until mid-July most cases in FL were in people under 50.

https://twitter.com/zorinaq/status/1290029180466298880

6. Another illustration of the reporting lag that leads to the need to 'backfill' data.

https://twitter.com/zorinaq/status/1293270181754544128

7. And it's simple to debunk the often-false argument that more cases are an artifact of more testing, by looking at the rate of change in tests vs cases.

https://twitter.com/sxbegle/status/1285176899719892993

8. Great thread synthesizing these points... however much one may hope otherwise, case surges bring fatalities in their wake.

https://twitter.com/firefoxx66/status/1296080742607355904

Same age heatmap patten in Spain now

https://twitter.com/drzoehyde/status/1304307145127505922

Same age pattern in the UK

https://twitter.com/Dr_D_Robertson/status/1305156278298324992