A web article by members of the board of Science Forum COVID-19.
This article supplements and expands on the article in Dagens Nyheter 2020/10/01.1
Sweden’s FHM and the press were stating most of the summer and into early September that our case numbers were going down and were lower than many other countries in Europe. But tests are too easily manipulated and in Sweden the test rate does not match the death rate when we compare to our neighbours in September.
For example, read: Coronavirus Case Counts Are Meaningless, by Nate Silver.
A better indicator to compare infection rates across countries might be death rates.
The rate of deaths in Sweden, albeit low compared to the spring has remained high when compared to our neighbors in Norway and Finland. And now for the last 4 weeks the number of positive tests has been increasing steadily (see Figure 1).
If we compare to our Nordic neighbors, we see the reported case numbers for the duration of the disease as reported by FHM to Our World in Data (Figure 2). Here it looks as if again we are increasing but the reported cases are not so different. But, is this accurate?
If deaths are the one statistic that cannot be hidden (as they can do with tests), then we should compare these.
From September 1-14 (later dates are inaccurate as there is a lag in reporting deaths as reported in the DN article), the number of COVID-19 deaths is:
Sweden: 27 (10.3 million population)
Denmark: 9 (5.7)
Norway: 1 (5.4)
Finland: 1 (5.5)
Adjusting for population, this would seem to indicate that Sweden has an infection rate 1.5 times that of Denmark and 14 times that of Norway or Finland. This is a somewhat different picture from that in Figure 2.
So, what is happening with the testing in Sweden that could make the issue look so much better than it is?
There are several possible reasons.
1) If you get COVID-19 symptoms, the best time to do the test is 0 – 3 days after onset (or even 1 day before), when the viral load is the highest. This drops sharply after that. In Stockholm, hot spot now, there is often a minimum 6 day wait to get tested. This is true in other areas as well, although perhaps to a lower degree. In an early report published May 13 in the journal Annals of Internal Medicine,3 it was shown that the probability of obtaining a false negative result decreases from 100% on Day 1 of being infected to 67% on Day 4. The false negative rate decreased to 20% on Day 8 (three days after a person begins experiencing symptoms). They also found that on the day a person started experiencing actual symptoms of illness, the average false negative rate was 38%. In addition, the false negative rate began to increase again from 21% on Day 9 to 66% on Day 21. These tests are getting better, but the viral load changes with the time course of the infection and this leads to a need for testing on the correct days, which is not being done in many places.
2) In Sweden different commercial tests are used. Most laboratories use Roche’s Cobas system. A major testing facility at the Karolinska Institute, however, uses the Chinese BGI Biotechnology’s tests. At very high PCR cycles false positive tests were noticed. Therefore, fewer PCR cycles were decided to be used in the COVID-19 tests. This, however, makes the tests less sensitive, and could miss some asymptomatic patients with low levels of virus excretion. This is the reason 3900 positive tests were reclassified as negative and removed from the positive test count in August. As other countries in Europe have not followed this path, this means our number of positive tests in Sweden is likely artificially low when compared across Europe.*
*Updated December 3, 2020 to reflect the current testing strategies.
3) We currently have a large number of self tests being done in Sweden leading to a large number of false negatives. In Sweden, 40 % of tests are now self-tests. Many European countries do not allow these self-tests as there is too much chance for error.
4) In the self-tests there is another problem. The instructions from FHM, the manufacturer, and the instructions for use in the regions of Sweden differ. The end result is that patients are adding an extra saliva dip AFTER swabbing their nose and throat. This differs from the instructions for which the test was validated where only the nose/throat swab was performed. FHM’s own self-test study said to also test saliva but that test should be performed separately from the throat/nose swab. Mixing saliva together with the throat/nose swab goes against both the FHM’s self-test study instructions and the test manufacturer’s validated test method. This extra step might lead to dilution which could possibly result in false negatives. This is even more likely when considering the less sensitive test cycles as indicated in point 2.”
5) In many laboratory tests in Sweden, not all, they seem to swab only the throat and not the nose, which can give up to a 30% false negative rate. 6) As the testing capacity in the test centers is overloaded, they have cut back testing, especially for children who should now only test if symptomatic, but most children show little or no symptoms, but they can still transmit. This leaves out another large group of infectious individuals. We also test much lower numbers of children age 0-19 than our neighbors.
So, what does this all mean?
All of these test irregularities can add up. If each of these situations leads to an average 25% increase in false negative tests compared to what is done in our neighboring countries, then we have only 18% accuracy in actually predicting the infection rate. Or in other words, we have more than 5 times the infection rate than assumed when comparing to other countries. This is similar to the difference in death rate between Sweden and Denmark, but not Finland and Norway.
When compared to the difference in death rate of 1.5-fold and 14-fold relative to Denmark, or Norway and Finland, respectively, this simple calculation seems to argue we have considerably more positive cases in Sweden than are being reported and it is because of the testing procedures in Sweden leading to many more false negative results. This is important to consider when comparing case rates between countries.
Andrew Ewing, Professor of Chemistry and Molecular Biology, University of Gothenburg. Member of the Swedish Academy of Sciences.
Emil J Bergholtz, Professor of Theoretical Physics, Stockholm University.
Nele Brusselaers, Associate Professor of Clinical Epidemiology, Karolinska Institutet.
Vetenskapsforum Covid-19 arbetar brett med information och upplysning om den pågående pandemin och till den relaterade ämnen och områden. Våra mål är att minska sjuklighet och dödlighet i Covid-19, att minska de kroniska sjukdomstillstånden hos dem som överlever Covid-19. I föreningen har vi kompetens inom bland annat virologi, biologi, molekylärbiologi, epidemiologi, infektionssjukdomar, lungsjukdomar, matematik, statsvetenskap, psykologi, etik och riskforskning. Ordförande i föreningen är Professor Emeritus Anders Vahlne, mer information på https://vetcov19.se/