The Times: Statistics teach us to be sceptical about ‘Operation Moonshot’
Covid maths/statistics.
Did you get the all clear in your daily cancer check this week? Of course you did not have one. Even after decades of trying it has proved extremely difficult to find regular tests that do not do more harm than good by wrongly telling healthy people they are ill. These basic statistics explain why scientists are very sceptical about Boris Johnson’s “Operation Moonshot”.
Last night the prime minister laid out a vision where people get a “passport to mingle” by giving everyone in the country daily pregnancy-test-style coronavirus checks.
Pregnancy tests are highly accurate when you think there is a pretty good chance you have conceived recently. But give them every day to every woman in the country — and still more so to every man — and most positive results will turn out to be wrong.
Tests that are accurate when people have a good chance of having a disease because of their symptoms become inaccurate when given to people who are well. The more sensitive you make a test — because you do not want infected people slipping through the net and mingling with others — the more you increase the chances of wrongly flagging people who are not infectious.
If your chances of having coronavirus are already quite high — because you have a new continuous cough or fever, for example — then you can be confident that a positive test result is accurate.
But at present only about one in 2,000 people chosen at random has the virus. But even if this increased to one in 500 the Scientific Advisory Group for Emergencies (Sage) says that assuming a test picks up 80 per cent of true infections and gives the all clear 96 per cent of the time, this would mean that of the 4,152 people who test positive out of every 100,000, only 160 would actually have coronavirus.
The other 3,992 face being told to isolate needlessly for two weeks. Scaled up nationally, this is millions of people every day.
This basic mathematical point is why the NHS is very cautious about screening. People are offered screening for only four types of cancer, and even the most established breast-screening programme has been shown to harm more women than it helps. Despite years of hope, checks for prostate cancer and dementia have been ruled out because too many false positives would lead to needless treatment.
Sage estimates that by giving those with positive coronavirus results a separate second test, false positives could be cut to 40 per 100,000. But it cautions that finding the right combination of tests is “not trivial” and even at this level tens of thousands of people a day would be wrongly told to stay at home, risking overwhelming the Test and Trace system.
Given the £100 billion price tag, Sage questions whether the money would be better spent on improving the troubled contact-tracing system and giving payments to ensure that those told to isolate for two weeks actually do so — at present most do not.
Of course two weeks at home is not needless surgery, and given the costs of lockdown we might be willing to tolerate large numbers of false positives. But despite the prime minister’s hopes, it is still clear that if we do not get an effective vaccine, there is no easy path back to normality.
The full Times Article: Statistics teach us to be sceptical about ‘Operation Moonshot’