Testing can defeat the covid-19 pandemic and lift the lockdown | thearticle

Alexander Woolfson ’ s recent critique of testing for TheArticle is factually correct. However, the conclusion I draw from it is exactly the opposite of the one given by the author. The

Covid-19 test is called RT-PCR (reverse transcriptase polymerase chain reaction) and, as Woolfson states, it is highly specific. A positive result means that the patient does indeed carry

the virus. However the test is not very sensitive, in that in up to 30 per cent of cases where the result is negative, the patient is still carrying the virus and may go on to infect others.

The error rate is due to several factors, including procedural errors such as inadequately placed swabs. The NHS already repeats some tests to get the error rate down. For the purposes of

this discussion, let ’ s assume that the 30 per cent error rate cannot be reduced any further. The significance of a test is very different when it ’ s used in a clinical setting, as opposed

to deploying it for public health purposes. In the clinic it ’ s not acceptable to tell a patient that they have a 70 per cent chance of not having a cancer, say, or in this case of not

carrying the virus. Such a test is rightly rejected by most clinicians as it fails to meet the needs of patient  or clinician. The approach to testing at the population level is very

different. A failure rate of up to 30 per cent may be acceptable, as the strategy of testing, tracing and quarantining is statistical. For example, let’s say the virus is highly infectious

with a reproduction rate (R0) of 2.5: that is, every infected person manages to infect three others before recovering or dying (for comparison, the number for seasonal flu is just 1.3).

Then, if every cohort of infected persons were to be tested and 30 per cent were falsely labelled as free of infection, the transmission rate would still fall to 2.5 x 0.3 = 0.75. Anything

less than one means that Covid-19 infects fewer people with each cycle, and so the epidemic fades away. If large numbers of people are already infected (as was the case at the start of the

UK ’ s lockdown), the test, track and quarantine approach is not enough. The 30 per cent of people who might test negative can _sustain_ (not grow) infection rates at high though declining

levels for an extended period of time. The human cost makes this both medically and socially unacceptable. The statistical approach means that if, via a lockdown strategy, you can get the

number of people carrying the infection down to low enough levels, then you can tolerate a degree of error in your testing programme. I would assume that the authorities would want to keep

some restrictions in place, such as maintaining social distancing and delaying the opening of large events, to keep the R0 down even further, at least until they knew that they had matters

under control.  It is important to note that this approach to testing is not just theory. South Korea has succeeded with this approach and managed it without a lockdown or even forbidding

international travel. Germany has run a big testing programme and is now ready to start easing its lockdown. Experience trumps theory, so we should learn from their success. The Woolfson

piece relies heavily on a Mayo Clinic study that was released pre-publication and is still subject to peer review. Such reports have to be treated with caution. The study shows that RT-PCR

tests produce false negative results in the 10-30 per cent range. This is true and widely known. The Mayo article then asserts: “The president of the region of Madrid has predicted that 80

per cent of Madrid ’ s 6.5 million residents will become infected by Covid-19. If the entire population was tested, of the anticipated 5.2 million infected individuals, 520,000 people would

be falsely classified as free of infection. Even with less widespread testing or targeted testing among random samples, the number of false negative tests could be massive.” This is a

decidedly dodgy extrapolation. Firstly, even in the worst epidemics such as smallpox or even the Black Death, only a fraction of the population carry the live infection at any one time and

these people go on to die or recover. The number of people carrying a live infection at any particular moment is always a fraction of the population. Even if 80 per cent of the population

were to eventually contract Covid-19, only a minority would express the live virus; the rest would test negatively because they would have recovered. Secondly, the whole purpose of the

lockdown is to ensure that 80 per cent don’t get infected. The mayor of Madrid was speaking about an unconstrained epidemic. He is quoted out of context. The approach to testing advocated

here suggests how we could go about stopping the epidemic in its tracks and ending the lockdown. The Peto brothers, Julian and Richard, are two of the country ’ s top statisticians and have

published together. Many years ago I did a course in medical statistics under Richard Peto (now Sir Richard), the Professor of Statistics and Medical Epidemiology at Oxford, and met his

brother Julian. I found them both frighteningly impressive. Julian Peto, a Professor of Epidemiology at the London School of Hygiene and Tropical Medicine, has come up with a way of beating

the virus that I believe just might work. He proposes to throat swab and PCR test everyone in the country once a week. Those who test positive would be temporarily quarantined. Even with 70

per cent accuracy, this programme would push the R0 under 1 and drive the epidemic to extinction. Professor Peto believes that the 10 million tests a day required can be done with the help

of all our university labs working together and scaling the operation. He suggests trialling the idea in a single city to see if works. The proposal might be expensive and logistically

challenging, but it ’ s a heck of a lot cheaper than keeping the whole nation under house arrest.