“We have a simple message for all countries: test, test, test.” The famous phrase was pronounced by Tedros Adhanom, the Director-General of the WHO, on March 16, when there were barely 6,500 deaths from Covid-19 in the world. Today, getting dangerously close to a million dead, we know that this message has not turned out to be as “simple” as Adhanom promised.
The tests are complex
The complexity of testing in Covid-19 is obvious. Diversity of tests, with different capacities to contribute to correct decision-making, that is, with different diagnostic yields (even within the same test families), with different indications and complex interpretations that, in part, depend on the characteristics and the candidate’s exposure history. To which are added the discrepancies on who and when to perform each of the possible tests.
Of course, the complexity of the tests seems, at this stage of the pandemic, a minor problem if we take into account the confusion generated by the testing strategies that different countries (and in Spain, different Communities) are using.
In Spain and outside of Spain, there are many who speak about evidence without any scientific basis (and in most cases, without knowing what they say). They even postulate –or carry out in their respective fields– t“massive” states with one or another proof, or claim evidence “at source” as the solution to all our ills. That is when they are not wanting to raise isolates because a test has been negative.
Confusion. And the mainstream media, apparently unable to understand that the use of diagnostic tests depends on the test used, on whom it is used and when it is used, contribute notably to the confusion.
How to interpret the tests that detect the presence of the virus
We basically have two types of tests. In the first place, those that detect the presence of the virus (or any of its components) in nasopharyngeal, oropharyngeal exudates, or saliva, as the PCR waves antigen testing (Although the latter are not yet available in Europe, at least with sufficient diagnostic validity to be used in clinical practice).
A positive result in these tests (even with many nuances due to the presence of false negatives and positives for remains of the virus without infectious capacity), can be interpreted as that this person (whether symptomatic, presymptomatic or asymptomatic) is infected at that time. And, at that time, it has the ability to infect other people. It must be isolated and its close contacts drew and isolated.
A negative result is more difficult (much, much more difficult) to interpret. The sensitivity of PCR is around 70% (it tests positive in 70% of people who are infected at that time). And this assumes that there will be false negatives, that is, people with COVID who test negative on PCR. Much more if the test is performed before 5 days after infection when the patient is still in the incubation phase.
In any case, this type of evidence, and despite its many shortcomings, is the essential instrument for the diagnosis of Covid-19, and for isolation decisions.
How to interpret tests that detect antibodies to the virus
Second, we have the tests that detect the antibodies against SARS-CoV-2 that have been generated by people who are in an advanced stage of the disease or who have already had it. Antibody tests use blood samples, no exudates, or saliva. And they can be performed in the laboratory (using sufficient sensitive and specific ELISA or chemiluminescence techniques) or by lateral flow techniques.
The latter is popularly known as rapid tests (confusingly, since antigen tests are also “fast”). And it turns out that its sensitivity and specificity, when known (manufacturers have tried to thwart independent validation studies), is much lower. In a systematic review they offered a mean sensitivity of 66%, still highly variable for the different commercial kits, and with validation studies in many deficient cases.
Although it depends on each commercial kit, these tests can detect essentially two types of antibodies: IgM that begins to become positive around the tenth day after infection and disappear around 3-4 weeks; and IgGs, which become even later positive, but persist for months and indicate past infection and, perhaps, immunity. Immunity is still unknown: We do not know for how long and several cases of reinfection have been described in people with circulating IgG.
Antibody tests They are valuable in conducting serological studies and, in some cases, can help clinicians to determine the time of evolution of a specific patient. But are not useful for diagnosis, for screening, to make isolation decisions, or to decide to perform a PCR in people with antibodies (even if they are IgM).
How to interpret the results (and when to isolate)
Sensitivity (ability to identify sick people as positive) and specificity (ability to identify healthy people as negative) are characteristics of each test. Although they can vary significantly even in the same family.
PCR is considered a mean sensitivity of 70% and a specificity of 95%. But –and this is essential in the assessment of diagnostic tests– the interpretation of a result depends on the probability that the patient had of being ill before performing the test, the prevalence of the disease in the population in which we are performing tests.
Let us suppose a patient with symptoms of Covid-19 and suggestive pulmonary radiology. Your probability of having Covid-19 could be greater than 90% (out of every 100 patients like this 90 will have COVID and 10 will not). A PCR type test, the appropriate one in this patient, will confirm the presence of disease (99% of the positives will have Covid-19). But 27 of the 90 patients will have a negative result.
A negative test does not rule out the disease (74% of the negatives will have COVID), so all patients with symptoms, regardless of the test result, must always be isolated and their contacts traced and also isolated.
Let us now assume close contact with this asymptomatic patient at the time of having a PCR. Your chance of having Covid-19 could be around 33% (a third of close contacts will develop the disease). In these circumstances, a positive PCR will continue to confirm the presence of disease (88% of those who are positive will have COVID), but 14% of people with a negative test will also have COVID.
It is a high enough proportion to isolate all contacts, even if they are asymptomatic and regardless of the test result (even knowing that the majority of those who are negative will not develop the disease). This is one of the essential strategies to control the epidemic, and if the identification of contacts or their isolation fails, the transmission will continue to grow.
Screening in the general population, would it help?
Screening in the general population (or in the so-called “at-origin” tests) is a very different situation from the previous ones. They are people without symptoms and who have not had close contact with patients. Even in places with high incidences, such as Madrid at this time, there are hardly any sick people. In the case of Madrid, about 340 people for every 100,000 inhabitants in a week; below 0.35%, although to avoid person fractions we will assume a 1% prevalence.
In this case, for every 100,000 people screened, 5,600 would test positive, but only 700 of them would be true positives. While another 4,900 (false positives) would be isolated without having the disease.
On the other hand, 300 patients (false negatives) would not be isolated. In this case, what we call negative predictive value is high since the vast majority (99.7%) of people who test negative would not have the disease. But there will be people with false-negative results who, having the infection, will believe that they are not infected and will contribute to its transmission.
With the current prevalence in Spain, much less than 1%, the results of indiscriminate tests (in populations, teachers, etc.) must be even worse, generating huge volumes of false positives while escaping detection by tests a substantial part of contagious patients. Confusion again.
In fact, if the transmission were high enough that population screening made sense, we should skip testing and go straight to general lockdown.
On the other hand, any population screening is a public health intervention that requires exhaustive planning, with clear objectives, with the necessary material and personal resources, and with a prior evaluation that demonstrates that its expected results in terms of health and disease control pandemics are favorable enough for the investment that is made.
The frivolity with which some and others propose interventions of this type shows irresponsibility that can only be explained by the infrequency with which true accountability is practiced in all social spheres.
Confusion is expensive
Confusion is not free. It generates discomfort (irritated patients bitterly claiming the inappropriate test at the wrong time from their family doctors), uncertainty (queues of people taking a test whose result no one can reasonably interpret), waste (resources used without generating value), low-quality care (delays in tests that are really useful), more infections (false negatives that are not isolated) and economic losses to families (false positives that must be isolated).
The use of diagnostic tests is complicated enough. Better not to complicate it more from ignorance or populism.
Salvador Peiró. Researcher, Health Services Research Area, FISABIO SALUD PÚBLICA, Fisabio.
Ildefonso Hernández Aguado. Professor of Preventive Medicine and Public Health, Miguel Hernández University.
This article was originally published on The Conversation.