AMONG LA tools that health officials used to combat covid-19, the effects of blockade, masks and vaccines were well studied. Testing, the impact of which is difficult to measure because it coincides with other pandemic control policies, has received less attention. But thanks to a sad natural experiment, that is now beginning to change.
Enjoy more audio and podcasts iOS or Android.
In September covid appeared to cease in southwest Britain. Overnight, the positive rate of PCR tests near Bath and Bristol fell from 3% to 1%. Nowhere else in England has such a decline been enjoyed.
In the middle of the month a local volunteer group said this apparent soaking could come from faulty tests. A few weeks later, the group and local scientists asked the UK Health Safety Agency (UKHSA) explore.
October 12 the UKHSA closed a laboratory run by Immensa, a testing firm. The agency said the company incorrectly told 43,000 infected people that they were virus-free. The cause of the error remains unclear. Immense declined to comment.
As soon as the laboratory was closed, the reported incubation case of the region soared. A spokesman for the prime minister said the laboratory errors had not caused this increase. On November 14, however, Thiemo Fetzer of the University of Warwick published an article showing that they probably did. It has not yet been reviewed by colleagues, but offers firm evidence that accurate testing is indeed slowing the spread of covid, letting infected people know they need to isolate.
To assess what could have happened if it weren’t for the snafu, Fetzer built a “synthetic control”: a group of areas whose previous rates of vaccination, testing and incubation cases and deaths matched those of the 13 hardest-hit regions. The difference was huge. From September 2 to October 12, the affected areas recorded 13,000 fewer positive tests than the control. Later, 21,000 more were recorded.
This implies that each faulty test may have caused 0.6-1.6 extra cases (the interval reflects uncertainty about how many people who tested positive received earlier false negatives). Based on the UK mortality rate, this translates to 400-1,100 deaths.
Surprisingly, this toll is not the highest Mr. Fetzer has attributed to technical failures. In 2020 a spreadsheet error prevented the UK’s statistical service from reporting 15,000 incubation cases to contact trackers. In an earlier study, Mr. Fetzer estimated that 1,500 people died as a result.■
Sources: “Measuring the epidemiological impact of a false negative: evidence from a natural experiment,” by T. Fetzer, 2021; Office for National Statistics
This article appeared in the section Graphic detail of the printed edition under the title “An Immense cock-up”