Thread by @dgurdasani1

Surprising to see @MAbsoud the senior author of the KCL study trying to explain the discrepancy between long COVID estimates from the very flawed Zoe KCL, and the nationally representative ONS study by critiquing the ONS without even understanding the ONS methodology.🧵

He says that ONS data are collected monthly & is retrospective so less valid. This isn't true- ONS collects data on long COVID every week for the first 5 wks & then monthly. Versus Zoe that depends on parents being motivated to report symptoms of a sick child for months.

It also assumes that symptoms resolved when parents stop reporting. How many parents would continue to report for 6 months or a year (ONS shows that sadly many children do have continuing symptoms for a year)- especially when they aren't even asked directly about common symptoms

ONS asks directly about 29 key symptoms, many of which were put forward after discussion with patients. Not only does the Zoe study not include many key common symptoms of long COVID, it also didn't consider it's relapsing & remitting nature.

I'm completely astonished that researchers who've conducted this research are rationalising their estimates based on flawed design by critiquing methodology of one of the most robust nationally representative studies, when they don't seem to even understand the methodology.

Surely publishing results that are lower than another study that was fairly robust it its methodology should mean at the very least that there should be explanations for why they are seeing differences- explanations that are based on an accurate assessment of the limitations.

It's very clear that the Zoe study bias will very likely underestimate incidence of long COVID and duration- researchers should be honest about this rather than flagging non-existent flaws in much better designed studies - which they clearly don't understand the methods for.

Link here:
theguardian.com/world/2021/aug…