In Switzerland, authorities say they test only clear cases because they are afraid to run out of testing kits. But is this the best sampling strategy? Shouldn’t we define a strategy that maximises the information we get from the available tests? In epidemiological research where shortage of funds is very common, we use randomized sampling if we don’t require individual level information.
Why are we testing?
Tests can be medically motivated. They help decide about treatment strategies, including differential diagnosis, and whether a patient can be released from the isolation ward. But a test can also be epidemiologically motivated, to gain information that helps prevent the spread of the virus, such as when it makes sense to check the presence of the virus in contacts without symptoms, and to understand the distribution of the virus in the general population.
What testing strategies can we apply?
The testing strategy needs to be primarily driven by the motives to do the testing. Different motives allow for different sampling strategies in function of the availability of test kits. I see three potential strategies. Here a rough layout of the concepts.
Strategy 1: There is no shortage of tests
- Test all potential patients with clear symptoms and test them several times.
Medical doctors will learn whether the case is this virus-related. while scientists will learn about the link between virus load and disease progression, and how long a person remains infectious after the last symptoms.
- Test all contacts of cases and their contacts at regular intervals
This will allow doctors to early identify potential new patients, while researchers will understand the time from contact to spreading to disease, and whether there are spreaders without obvious symptoms (and if yes how many).
- Test as many as possible with minor symptoms
This will give an idea of the number of cases that may get overlooked, thus it is more than to just calm hypochonders who run to the doctor for every little issue.
Strategy 2: There are just enough tests
- Apply strategy 1 but do repeat tests only in a random selection
As the number of cases increases, it is not necessary to test everybody multiple times to understand patterns. This can easily be done in a random selection of the patients and the contacts. Test are only needed when individual level decisions must be made.
Strategy 3: There are not enough tests
- Fully test only those cases where the result are central for the medical strategy.
Very clear cut cases don’t necessarily need a test. A person without any known lung disease can be assumed to be a new case if the symptoms are very evident and no other cause identified. You can still take a sample and deep-freeze it for testing at a later moment, when you have sufficient test kits. However, not all cases will be clear, especially amongst those with existing lung diseases. Those may require a differential diagnosis and the results of the test will be central for the treatment strategy and whether the case requires a place in an isolation ward.
- Continue testing a random selection of the groups defined in strategy 1.
Not using all tests for the serious cases will allow to keep track of the virus distribution and to be able to see changes in patterns that tell about the effectiveness of preventative strategies. It can also make sense to collect samples for storage (e.g. deep freezing) for later testing, when the peak is over or when more tests become available.
Learn from testing
Finally, for all strategies it will be important to learn from the testing so that we can better define who needs to be tested when, and what strategies work best in our battle against the virus. This can be done most effectively when the data is shared openly and rapidly so that many smart brains can look at these questions.