COVID-19 symptoms are a poor marker for the actual infection

Tests of large population groups are being carried out all over the world to find out how many people are infected with COVID-19. But the results are difficult to interpret. For example, a comprehensive British population study showed that symptoms of COVID-19 are a poor marker for whether the SARS-CoV-2 virus is actually present. 

The study has just been published in Clinical Epidemiology, and the person behind it is Irene Petersen, visiting professor at the Department of Clinical Epidemiology at Aarhus University, who is also employed as Professor of Epidemiology and Health Informatics at UCL, University College in London.

This is one of the largest population studies of its kind in the world.

"Each week, the Office for National Statistics collects data from randomly selected households in England, Wales and Northern Ireland, and the participants are at the same time interviewed, while SARS-CoV-2 test samples are collected from them," explains Irene Petersen. A total of 36,061 persons were tested during the two month period in the spring from which the researchers have utilised data. Nursing home residents and hospitalised patients were not included in the study.

Three quarters were symptom-free on the day of the test

In connection with the test, 625 of the tested individuals complained of symptoms that could be due to COVID-19. This was before they knew the results of the test, which subsequently showed that only 115 of 36,061 were found to be positive. Conversely, only 27 of the 115 had symptoms on the day of the test.

The remainder – i.e. 88 test subjects corresponding to 76.5 per cent – were asymptomatic. In other words, they showed no or only mild symptoms.

"I was surprised that the figures showed that more than three quarters didn’t have symptoms on the day they were tested. It’s possible that some of them later developed symptoms, although we don’t know," says Petersen.

A number of other studies around the world have also shown a large proportion of asymptomatic cases among those who have tested positive. For example, a study of 13,000 Icelandic inhabitants which showed that 43 out of 100 who had tested positive were asymptomatic. 

Just as surprisingly, Irene Petersen's analysis of data from the large population study showed that of the 115 who were tested positive, only 16 reported symptoms such as coughing, fever and/or loss of sense of smell and taste. 99 did not – corresponding to 86 per cent. 

"I had expected a greater correlation between symptoms and test results," says Petersen.

Should consider new test strategies

According to Irene Petersen, the results should lead to considerations about future testing strategies, because the current tests focus on those who have symptoms which delays the possibility of detecting sources of infection.

"What we argue for is that there should be greater consideration given to what are called monitoring tests, because we know that some groups are more vulnerable than others. For example, this could be testing everyone who works at an abattoir or a test of all university students before they return home to spend Christmas with their family," says Petersen.

"Some people might say that we can't afford to do this, but it’s possible to make use of what’s known as a pooled test, where you pool several samples and test five at a time, for example. This can be done if the incidences are relatively low as is the case with COVID-19. In the large population survey, this figure was only 0.32 per cent of the 36,061 tested. Using a pooled test means you can save four out of five tests, because you only switch to testing each individual who took the test in question if the test is positive.”

"There are no perfect solutions, but I believe that this type of test strategy would be able to identify some of the healthy infection carriers who we’re not reaching with the current test strategy," says Irene Petersen.

The research results – more information 

• The study is a population study based on publicly available data from the Office for National Statistics Coronavirus (COVID-19) Infection Survey. Data has been used from the period 26 April to 27 June 2020.

• The analysis was carried out by Professor Irene Petersen and Professor Andrew Philips, both of whom are employed at UCL, University College London.

• The study has no special external funding.

• Link to the scientific article: doi.org/10.2147/CLEP.S276825

Contact

Professor of Epidemiology & Health Informatics Irene Petersen
UCL, University College London and
Aarhus University, Department of Clinical Epidemiology (visiting professor)
Mail: i.petersen@ucl.ac.uk
Mobile: (+45) 9125 1038.