2 November 2020
Internal discussion document: Covid testing at IST Austria
Apart from social distancing, hygiene, masks and a strict stay home when sick policy, a population-wide diagnostic regime is an essential pillar in the fight against Covid. Since the beginning of the pandemic we are in constant discussions internally, with the Ministry of Health and with other institutions if/when a Covid monitoring program makes sense for IST. Below is some background information and some arguments we deem relevant.
SOME BACKGROUND ON COVID TESTING
Types of diagnostic tests: broadly speaking there are three categories of tests
Detecting virus RNA: the widely used test is PCR based. It is the most sensitive test because it amplifies virus RNA (via DNA). Many alternative RNA detection tests are in earlier stages of development/application (many are Cas9-based). The PCR method as such is very sensitive. The main source of false negatives results (somebody has the virus but tests negative – see below) is not failure to detect the RNA in a sample but the sample itself. The gold standard to take the sample is to take a deep nasal swab (uncomfortable, has to be done by trained personnel). Alternatively pharyngeal swabs, gurgling or spitting is used. Sensitivity is probably lower with gurgling, pharyngeal swabs and spitting but there are no conclusive data yet. For screening purposes, sometimes pooled tests are used (sensitivity even lower). In general, PCR tests are resource intensive, relatively slow and expensive. Many alternative RNA detection tests are isothermal, hence cheaper and faster – but less sensitive.
Detecting virus antigens: “Rapid antigen tests” They detect virus-derived non-DNA antigens. These tests are relatively new and numerous products enter the market. Such tests are quick and cheap (“like a pregnancy test”). Like in RNA testing the sample has to be taken by trained medical personnel. As this test does not amplify the antigen, sensitivity is currently lower.
Detecting virus antibodies: a blood test that detects the humoral (antibody based) immune response a person developed against the virus. This can only be positive with a time-delay, usually when the person is not infectious anymore. In other words: It tells you if you have had the virus. Antibody tests have to be interpreted with much care.
- If a person has antibodies, it is likely that the person has had an infection (although there are some cross-reactivities with other Corona viruses).
- If a person has no antibodies, it does not necessarily mean that that this person did not have the virus because humoral immune responses a) can be minimal or absent when the infection was weak or asymptomatic and b) taper off with time.
- A person who had the virus and has no antibodies can still be immune because there is cellular immunity (T cells) that can protect.
- If a person has antibodies it does not mean that the person is 100% immune, as not all antibody responses neutralize the virus.
In other words: if you tested positive for antibodies you cannot forget about hygiene, distancing and mask.
Diagnostic testing vs. monitoring:
As a flawed testing regime naturally does more harm than no testing, testing underlies strict quality control.
- Diagnostics is strictly in the hands of the medical system. Diagnostics requires a workflow that is fully accredited. This includes: sample taking, sample transport, the actual labwork, documentation, reporting to patient, doctor and health authorities etc.. IST could not get accredited for diagnostic testing.
- What Vienna Biocenter etc. are doing is not diagnostics but a procedure termed “monitoring”. In case of a positive result in this screen, the person is alerted (anonymous to the testing institution) and can (but does not have to) then turn to the healthcare system and get a diagnostic test. Monitoring also needs accreditation (but at a lower level).
Some facts about RNA-based testing:
While RNA testing is the gold standard in diagnostics it is by no means sufficient to tell you that you do not have the virus. The false negative rate is substantial. Epidemiologically, false negatives are a big problem (while false positives – which do happen – are unpleasant for the individual but not so much for the population). False negative numbers are still controversially discussed but this is probably the biggest meta-study on the topic – published in “Annals of Internal Medicine”:
“…the probability of a false-negative result in an infected person decreases from 100% (95% CI, 100% to 100%) on day 1 (the day you are infected) to 67% (CI, 27% to 94%) on day 4. On the day of symptom onset, the median false-negative rate was 38% (CI, 18% to 65%). This decreased to 20% (CI, 12% to 30%) on day 8 (3 days after symptom onset) then began to increase again, from 21% (CI, 13% to 31%) on day 9 to 66% (CI, 54% to 77%) on day 21…”
There are some studies that are more “optimistic” about false negative rates in PCR tests, but the bottom line is: RNA (and antigen) testing is a snapshot with high failure rate and to capture an event, testing has to repeated with high frequency. The (often used) strategy to release a symptomatic person with typical symptoms from quarantine because one test was negative is epidemiologically very risky – as for almost all diseases, diagnostics should not be based on one parameter.
Quantitative considerations: There are no conclusive data yet about the quantitative correlation between 1) being PCR positive 2) being positive in an antigen test 3) being symptomatic and 4) being contagious. There is no conclusive study suggesting that asymptomatic spreaders are major drivers of the pandemic. There is some hope (but no good data yet) that the “rapid antigen test” pulls out the problematic cases because it might have a detection threshold that is similar to the threshold required for virus transmission. Similarly, quantitative PCR tests that measure the RNA load might make predictions on how infectious a person is.
Independent of the type of test there are two general categories:
- Targeted testing: Testing people with symptoms and testing people who were in direct contact with a positive person.
- Random testing: Randomly testing asymptomatic people.
It is important to note that any test can only be done on a voluntary basis. With current legislation, IST could not “force” employees to get a Covid test.
What we could and could not do at IST
- What IST could not do: Targeted testing (symptomatics and direct contacts) is prohibited at IST because a) symptomatic persons and anyone who was in contact with a positive person should not come to campus (strict “stay home when sick” policy) and b) targeted testing needs to be diagnostic testing.
- What IST could do: Monitoring asymptomatic persons (volunteers) as done e.g. at Vienna Biocenter in their monitoring program (see above). This could mean: a) offer tests to everybody and b) offer tests to select subgroups. This could be a lottery or a group under special acute or chronic risk (like an extended contact group, a team, a floor, a building, an age-group etc.).
What the epidemiologists say
The general recommendation of epidemiologists is against random testing, see e.g. the German Robert Koch Institute:
The relevant citation: “Von einer ungezielten Testung von asymptomatischen Personen wird aufgrund der unklaren Aussagekraft eines negativen Ergebnisses (lediglich Momentaufnahme) in der Regel abgeraten.” “We advise against broad random testing of asymptomatic persons because of the low indicative value of a negative result as it constitutes only a snapshot.”
It is needless to say that random testing does not scale as it is indefinitely resource intensive – every human would need to be tested at least twice a week (better every day). Hence, the strategy recommended by the majority of epidemiologists is to only do targeted testing and restrict random testing of asymptomatic persons to specific high risk groups like infection clusters, health workers etc.. It is not common practice at hospitals to randomly test all employees.
DISCUSSION OF SOME FREQUENTLY USED ARGUMENTS
1) By testing every IST employee (at least) twice a week we can identify pre-symptomatic and asymptomatic spreaders
This is a very strong argument for random testing and only conclusive data on this issue will determine if this pro outweighs other cons. We have to constantly weigh this argument against others. Current state: although asymptomatics can have high viral load, there is no indication that asymptomatics are strong drivers of the pandemic.
2) We have biology labs, could we PCR test ourselves?
True in principle: it is feasible to copy paste the workflow of other places or establish our own. But note: currently we neither have the lab-equipment nor the personnel nor the allowance (accreditation). Also consider that the actual labwork is the easier (and cheaper) part of the whole workflow, while the main challenge is sampling, logistics, quality control, reporting etc.. IST would have to invest in instrumentation and personnel.
3) Testing is expensive, but if we do it ourselves it is cheap
This is not correct. We are a highly powered and top-equipped publicly funded research institution. If we perform routine tests with highly repetitive workflows, a full cost calculation (how much does the tax payer have to pay for each test) will tell us that our costs per test are very high.
4) Rapid antigen tests are quick and easy, could we offer these?
Yes. However, it is too early to say if such tests are an adequate tool for testing asymptomatic persons. And asymptomatic persons are the only ones we would want to have on campus as a strict “stay-home when sick” policy is a very strong tool (and we can afford to implement it at IST – see also last point 12).
5) If we would offer testing, everybody would /should participate
We know that participation at Vienna Biocenter is relatively high – but far from complete. One might speculate that IST will be lower because in general awareness seems highest in the life sciences. If participation is incomplete (as mask wearing is incomplete at IST), there will be a strong sampling bias in the test results: the more aware and more careful people are more likely to test regularly – we might repeatedly test the people who are epidemiologically not relevant and miss the relevant ones.
6) Testing vs. home office
Regular testing collides with home office. E.g.: if we would test theorists or administration employees that can also do their work from home, we would invest institute funds in order to limit the damage caused by non-compliance of our employees (something difficult to justify). Moreover, testing on campus would dis-incentivize home office.
7) We should test at IST because the interactive nature of scientific work makes us a risk group
IST employees are low risk because: on average we are very young; the nature of the work does not require physical interaction (as opposed to schools, health care or professions with customer contact etc.. Known superspreading events happened in settings that are not required for the work we do); much of our work is highly suitable for home office; our work is not system relevant in an acute sense; our buildings have high air change rate; as we are a scientific institution, awareness and overall compliance is above average.
8) With monitoring we can prevent that a floor or building or the institute is shut down
This is only true under the assumption that we have very high on-campus transmission. Cases coming in from outside we cannot prevent. These will always co-fluctuate with Vienna / Lower Austria. If there is a major outbreak we will have to follow authorities anyways and testing will not help to prevent shutdown.
9) If we test, who would carry the costs
IST is completely tax payer funded. The institute receives federal funds to fulfill its mission to do cutting edge research etc.. It is within the discretion of the institute’s leadership that a (limited) part of this money can be dedicated to the health and well-being of employees (childcare, gym, pub etc.). Hence, costs would be carried by the taxpayer and would have to be deduced from our research budget. Alternatively, employees would have to pay.
10) Potential effects of Covid monitoring on employees well-being
A negative test creates a feeling of safety and thus might have a calming effect on otherwise nervous institute members. On the flip side: every feeling of safety is false as false-negative rates are high and tests are only snapshots. This false feeling of safety might undermine the only known and proven strategies (distance, hygiene, masks, stay home when sick) that prevent actual virus transmission.
11) Potential effects of monitoring on compliance regarding masks, hygiene, distance, stay home policy
Note that we could only do screening of asymptomatics but never diagnostics. When we offer free testing on campus there is a danger that these procedures are confused. Simply speaking: there would be an incentive for somebody with symptoms (or somebody who had contact with a positive person) to come to IST and get a free, quick and uncomplicated test. Here, theremote location of IST (public transport) might be an exacerbating factor.
12) As a scientific institution we should be role models for fighting Covid and set up a structure that provides maximum safety
This is true – especially when it comes to adequate behavior. When it comes to testing this has a down side: Free testing for everybody does not scale. Testing everybody twice a week would be logistically, resource-wise and financially completely impossible for a whole country. If we make testing free at IST we would put us in a very privileged position – and we would do so with taxpayer money.
One could argue that we use resources that others might need more urgently. This is a serious argument especially as a) our employees are on average low risk (point 7) and b) IST can be very generous with home office as the possible decrease of productivity has proven to be acceptable so far (as opposed to a school, hospital, manufacturing company and even a university). In other words: whenever there is a high risk, employees can work from home. As others do not have this possibility one could argue: if IST invests in testing, it should rather test critical infrastructures like schools, hospitals etc.. but not its own employees.
General remarks on the spending of tax payer money
What we can do:
fund research and innovation. Whenever there is an idea with scientific content and motivation we can provide the funds and the internal support to bring this idea to fruition. It is almost by definition that this idea (and the project lead) has to come from a scientist, not an administrator. It is the same as for almost any grant that is submitted to an outside funding agency: it has to come from a scientist. When it comes to Covid testing this means: if we want to monitor the institute we could do so (and internally fund it) if this is a pilot-project to implement a scientifically novel approach (a novel test, a parallelization of known workflows, a screening for virus mutants… this is what they do at Biocenter and CeMM). Whenever we internally fund projects like this, it is good practice that every such project (be it a Covid testing project or a technical solution for our shuttle bus usage) goes through rigorous internal evaluation. Hence the AVID program we set up at IST.
What needs rigorous scrutiny:
use institute funds to finance employee benefits. If a testing program does not fit the slot of “research and innovation” because we outsource it (e.g. to a company or also a peer institution like Biocenter) we are no different from any other tax payer funded entity. This means, we would have to justify why it is specifically important that we receive (or allow us) such special treatment but not others (employees of a sewage plant, nursing home, hospital or school).