A preprint, an un-published non-peer reviewed paper, uploaded to medRxiv, has looked at community prevalence of SARS-CoV-2 in England.
Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:
“Huge learnings from this well-designed and very well analysed study.
“Methods first: The public’s initial willingness to take part was 41% of nearly 400,000 invited in England. The ultimate swab-result-rate was 31%. Participation was lower by males, those aged 18-24 years, those resident in London – but not to an extent that changed inferences. Let’s hope that the informativeness of these results encourages a still higher participation-rate for REACT’s new foray.
“Inter-laboratory calibration threw up some critical issues that lesser science would have been unaware of, and which were appropriately dealt with.
“Third, correction for sensitivity (and specificity) of the swab-test made a considerable difference to the estimated average daily number positive in May in England: 74,000 (95% CI: 63K to 86K) without correction; 115,000 (95% CI: 99K to 134K) after correction. Notice the non-overlapping confidence intervals – so that correction matters hugely.
“Public health learnings (1): Overwhelmingly important is REACT’s finding that 69% of swab-positives had been symptom-free for 7 days prior to testing positive (95% CI: 61% to 76%). Implication: asymptomatic infections need to be quantified within Test & Trace.
“Equally important is that recent (I assume) contact with a confirmed case increases my odds on swab-positivity to around 20 (95 CI: 14-36) and so, in May, would have increased risk from around 1 in 1000 to 20 in 1000. Let’s now consider the implications of both these learnings for Test & Trace.
“Test & Trace lacunae: In its first two weeks of operation (27 May to 10 June), Test & Trace reached 10,395 symptomatic index cases. Their household members (quarantined for 14 days) are likely to have numbered at least 10,000, of whom more than 140 could be expected to have become infected, but only 34 to 55 may have developed symptoms and sought testing.We should know how many of these quarantined household members did indeed test RT-PCR positive for SARS-CoV-2 during their quarantine period plus two days!
“In the same first two weeks, Test & Trace reached approximately 90,000 external recent close contacts of symptomatic index cases; and asked them to self-isolate. How many of these 90,000 self-isolators tested RT-PCR positive for SARS-CoV-2 during their quarantine period plus two days? I’m guesstimating somewhere in the range 300 to 700 quarantined external close contacts would have become symptomatic and booked a test. Why don’t we know?
“Asymptomatic positives, about which Test & Trace currently learns nothing, could be twice as many! Yet, it is the quarantine of these individuals that is, in principle and (thanks to REACT) indeed, one of the greatest potential benefits from Test & Trace.
“Notice that the likely number swab-test positives among those quarantined by Test & Trace in its first two weeks of operation hugely outnumbers those reported by REACT – but we learn about T&T only via REACT, which underscore the efficiency of good statistical method.
“Public health learnings (2): REACT estimates the adjusted odds – relative to non-key-workers – for “have I got it” as 2, 5 and 8 for key workers, healthcare workers and care home workers (404 participants only). REACT demonstrated that swab-test positive rate is similar for males and females but may be higher for Asians. Children aged 5-17 years do get infected but, as for ethnic minorities, greater precision about their adjusted odds requires REACT-2.”
Preprint: ‘Community prevalence of SARS-CoV-2 virus in England during May 2020: REACT study’ by Steven Riley et al. This work is not peer-reviewed.
All our previous output on this subject can be seen at this weblink:
Prof Sheila Bird: “Prof Bird, together with two of REACT’s statistician-authors, serves on the Royal Statistical Society’s COVID-19 Taskforce.”