How accurate is RT-PCR in diagnosing Covid-19?

According to WHO, "test, test and test" is the key strategy for containing the Covid-19 pandemic at the moment. Considering the significance of quick diagnosis, scientists throughout the world continued their effort to develop an easy, economic and effective testing method for diagnosing the disease. The methods used so far for diagnosis could be divided into two major categories: RT-PCR tests and serological tests.

In its interim guideline given on March 2, 2020, WHO suggested nucleic acid amplification tests (NAATs) for routine diagnosis of Covid-19 infection and cited RT-PCR as an example. As of now, this is the method practised most commonly and is regarded as the gold standard for diagnosing the infection. What is the basis of such a high reliance on this method and how valid is it? You might have found reports saying that someone tested negative once and positive later in a second test. This kind of flawed test results appeared as a huge problem for physicians in dealing with Covid-19 throughout the world. Even Dr Li Wenliang, the Chinese doctor who first made the world aware of this disease and eventually contracted the virus himself and died of it, tested negative several times before finally receiving a positive result. A case study published in the May 2020 issue of Japanese Journal of Radiology shows that a 34-year-old person tested negative four times for Covid-19 in RT-PCR test before ultimately testing positive.

Basically, two factors are taken into account when assessing the effectiveness of a diagnostic test: specificity and sensitivity. The term "specificity" refers to the ability of the test to eliminate "false positive" results—that is,identifying the infection caused by other viruses of the corona family (such as those causing common cold, SARS or MERS) as Covid-19. From this perspective, the efficiency of RT-PCR is almost 100 percent. If a person tests positive in RT-PCR, there is almost no chance for the result to be false positive. "The good news is that the tests appear to be highly specific: If your test comes back positive, it is almost certain you have the infection," writes Harlan M Krumholz, MD, a professor of medicine at Yale and director of the Yale New Haven Hospital Center for Outcomes Research and Evaluation (The New York Times, April 1). This is the reason for such a high enthusiasm for RT-PCR as a diagnostic test for coronavirus infection. The result can only be false positive if the sample of a non-infected person is contaminated with that of an infected individual.

The term "sensitivity" indicates the ability of the test to detect the virus and minimise "false negative" result—that is, failing to detect the virus when it is actually present in a patient's sample. The performance of RT-PCR is not that good in this respect. There are reports of 15-30 percent "false negative" results, as the sources vary. So, if you test negative, you cannot be sure that you did not contract the infection. This kind of error with Covid-19 diagnosis may cause serious consequences. On the one hand, the infected individual may be deprived of necessary care; on the other hand, he/she may socialise with others considering themselves "not infected", keeping on transmitting the disease to others.

Now, what to do if this kind of false negative result turns out to be a regular issue? "A lot of my patients who have symptoms, who I clinically think have Covid-19, are testing negative," says Dr Alain Chaoui, head of Congenial Healthcare, a North Shore practice with 50,000 patients across five locations in Massachusetts, USA. He further says that he is advising all of his patients who test negative but appear to be infected to assume they have the disease. He tells them to quarantine themselves until they experience no symptoms for at least 72 hours (The Boston Globe, April 2). Dr Harlan M Krumholz also shared a similar sentiment: "For now, we should assume that anyone could be carrying the virus. If you have had likely exposures and symptoms suggest Covid-19 infection, you probably have it—even if your test is negative." (The New York Times, April 1). At the University of Massachusetts Memorial Medical Center in Worcester, epidemiologist Dr Richard Ellison said that the hospital was retesting patients who received a negative test result but have symptoms of the novel coronavirus, in an attempt to cut down on false negatives (The Boston Globe, April 2). In some cases, lacking tests or not trusting the results, doctors have turned to chest X-rays or CT scans to diagnose patients by looking for signs of infection in the lungs. (Bloomberg, April 11)

You may ask what lies behind this drawback of RT-PCR in terms of sensitivity. Is it a lacking of the instrument or the testing method or does it result from a faulty sample? Experts are of the opinion that if reagents are fine and the test is run properly, this method should get 100 percent mark in terms of sensitivity as well. This test is "actually really good," says Jeff Pothof, chief quality officer at UW Health, the academic medical centre and health system for the University of Wisconsin-Madison. "So good that if we can capture a single strand of RNA, we can get a result" (Slate Magazine, April 6). So, the real problem likely lies not in the lab but in the samples sent for testing. If there is no virus in the sample, what will the PCR detect?

The defect in the sample may arise at any stage during sample collection, transport, preservation or processing. Experts particularly point to faulty sample collection. Here, one consideration is at what stage of infection the sample is being collected from the patient. It has been found that many patients tested negative first, despite having overt symptoms like cough and fever, and later tested positive in a second test. It is possible that your infection was in an early stage when the sample was taken first and so you tested negative. Actually, Covid-19 appears to have a pretty long incubation period and, being a new disease, there is a dearth of data about when in the course of the infection a test is most likely to be positive. However, Dr Omai Garner, an assistant professor of pathology and laboratory medicine at the University of California, Los Angeles (UCLA), had a suggestion: "Right now, the test appears to be most effective in patients who have been showing symptoms (specifically a fever and cough) for three to five days." No one knows how well it detects the virus in people who are asymptomatic, or how reliable it is after a patient has begun to recover (Slate Magazine, April 6).

Another, and probably more important, consideration is how adequately the sample has been collected. Most tests rely on a nasal swab that penetrates deep into the pharynx, the mucous membrane behind the nose and mouth. It's not an easy job and can be difficult even for a trained health worker. It's an invasive procedure that often causes patients to squirm. It is possible that the swab is not getting all the way into the nasopharynx. With a shortage of staff to conduct such widespread testing, in many cases people not typically trained to do so are collecting samples. Besides, people administering the test may be rushing from patient to patient. So, you should not get surprised if some of the swabs come back blank with no virus in it. When a swab with no virus on it arrives in the lab, it doesn't matter how sensitive the RT-PCR is. "If the sample is junk, just to be blunt, you're not going to find anything," says Nam Tran, an associate clinical professor at the University of California, Davis. (Slate Magazine, April 6)

It's evident from the above account that adequate sample collection at the right time can have a huge role in the accuracy of test results. In fact, a successful diagnostic test begins at sample collection because improper sampling can result in an inadequate specimen, rendering the total process useless (starting from sample collection to laboratory testing) and causing a wastage of time and resources. Many people who are actually infected might be incorrectly diagnosed as not infected, and may keep spreading infection in the community. So, it is important not only to increase the testing capacity but also appoint adequate skilled and trained manpower for sample collection. According to newspaper reports, there are a large number of trained medical technologists in the country who are ready to get involved in the job. The government has already decided to appoint a good number of them, but perhaps we should consider appointing more in order to respond to the demand of the time.

Dr Mohammad Didare Alam Muhsin is a professor of pharmacy at Jahangirnagar University.