Accuracy of the COVID-19 Test

Victor Luca, 1-Jul-20


Influenza is the Italian word for “influence” and came to be used for the disease because people believed that the position of the stars influenced the appearance of the deadly sickness. Today the SARS-CoV-2 virus that causes the COVID-19 disease, and is related to influenza, has certainly had an influence on us. Luckily we now know more about such diseases and have added many new sophisticated tools to our medical science arsenal. The genetic code of the SARS-CoV-2 virus was sequenced almost as soon as SARS-CoV-2 made its influence on us felt. With this information medical researchers developed a test for the virus and are now working vigorously on a producing an effective vaccine. In this article I want to address the important issue of testing.


I will start with the sweeping statement that there is no known scientific test that is 100% accurate or certain. This applies to the disciplines of physics, chemistry, medicine and many others. Just as in all our daily lives, there is uncertainty in almost everything we do and measure.


Let’s say that you wanted to measure your height. How would you do it? You would probably stand against a wall, put a mark at the level of the top of your head with a pencil and then measure the height from the mark to the floor with a tape measure. The accurate measurement of your height seems simple but it depends on how straight you are standing, how you estimate the top of your head and how you transcribe your position to the wall surface. Then if your tape measure is graduated in centimeters, the smallest graduation on a typical tape measure is the millimeter. You could repeat this measurement of your height multiple times and you would probably get slightly difference values. Maybe you would get a spread of values like 1495, 1500 and 1505 mm. The average of the three values would be 1500 mm and the uncertainty would be plus or minus 5 mm (0.3%). If you wanted your height to a better accuracy you would need to use a more sophisticated procedure and tools.


Determining if a person has the COVID-19 disease is more complicated than measuring your height but the issue of accuracy is still important if you want to get a good idea of whether a person is infected or not. The gold standard test for the COVID-19 disease involves the detection of appreciable quantities of SARS-CoV-2 virus in your mucosal body fluids by detecting the virus genetic material, the ribonucleic acid (RNA). By appreciable quantities of virus I mean sufficient amounts of the virus to be able to give you the disease and be detected.


The first step in the testing protocol is to take a throat (left; oropharyngeal swab) and/or nasal (right; nasopharyngeal) swab so that the virus is transferred to the swab itself. You will all have seen this sampling process being performed on your televisions many times and maybe even experienced it for yourselves. It is now known that nose swabs detect the COVID-19 virus more frequently than throat swabs which is why you don’t see throat swabs being taken very often.


The accuracy of a medical test is determined by measuring two things. The sensitivity and the specificity.


Let’s suppose that we have 100 people that we know to be infected with the SARS-CoV-2 virus. If the test is 90% sensitive, then 90 of those 100 people with the virus will test positive although we know all of them to be infected. These detections are called true positives. The test accurately gives a positive result 90% of the time; positive accuracy. But what of the 10 people who returned a negative test? We call these results false negatives because we know that all 100 are infected but for various reasons the test just didn’t detect the virus in 10 of them.


If on the other hand a test is 90% specific, it will correctly identify 90% of people who are not infected. However, 10% of people who are not infected will test positive for the virus and receive a false positive. These folk don’t have the virus but falsely test positive.


So sensitivity measures positive accuracy; specificity measures negative accuracy.

You will have heard the test for COVID-19 disease being referred to as RT-PCR which stands for Reverse Transcription-Polymerase Chain Reaction. It is too complicated to describe here how this test actually works at the molecular level. Suffice it to say that the test detects the SARS-CoV-2 virus RNA with very high accuracy. RNA is the molecule that gives the ability of the virus to replicate. Under ideal laboratory conditions the RT-PCR test is a great test. Recent scientific results have shown that five types of RT-PCR tests achieved 100% sensitivity on positive samples, and at least 96% specificity on negative samples.


In a real world or clinical setting these values of sensitivity and specificity are for various reasons greatly reduced. For the RT-PCR test, the clinical or real-world sensitivity has been found to fall in the range 66 – 80%. Probably most of the uncertainty is introduced in the sampling step. False negatives could arise if the swab didn’t go in deep enough, or for long enough, or if the test was performed too early or too late in the course of the infection and there just wasn’t enough virus present at the time? Serious error can also be introduced if the sample sits around too long before being tested and the virus denatures, or in other words, the RNA and other components of the virus break down.

Although the RT-PCR test has its limitations (sensitivity of about 75%) it is extremely important in estimating the spread of infection. We are lucky to have such a test, but as I stated at the outset, nothing in this world is perfect or absolutely certain. When it comes to controlling infection, the RT-PCR test is a wonderful tool. However, it is not the only tool in the testing arsenal. Recent results show that X-ray Computed Tomography (X-ray CT) of the lungs can also be useful. This diagnostic tool relies on irradiating yourself with X-rays using expensive scanning equipment as shown below.


X-ray CT machine used to scan the lungs for COVID-19 infection.


In terms of controlling disease spread, reliance on a test that is 66 – 80% sensitive does represent something of a chink in the armor because those 1/3 of people that don’t test positive but are infected will slip through the cracks and go on to infect others. Because of the infectious nature of the COVID-19 disease we can easily find ourselves back at square one. Vigilance and robust testing is therefore required.


A peculiarity of the SARS-CoV-2 virus is that it can be present in quite a large number of people without them showing any symptoms; they are asymptomatic. That there can be a large number of asymptomatic people was known early on in the COVID-19 pandemic from data from Iceland and the town of Vò in Italy, a town of about 3,000 people. In Vò the virus was eventually completely eradicated.


Baring a few recent slipups in terms of allowing infected people coming into the country to slip through the cracks, we in NZ have done a pretty good job of getting rid of the virus through effective social distancing and testing. In contrast, what we are now seeing occur in places like the United States and Brazil emphasizes the serious consequences of poor leadership, impatience and complacency and an emphasis on money and politics as opposed to the health of people.


Whilst we have effectively gotten rid of COVID-19, we are not out of the woods yet because many countries are still struggling to get this disease under control, including our neighbors in Australia. These countries could still let COVID-19 spiral out of control. As long as we in NZ continue not to interact with these countries we are OK. Opinion will vary on whether this is a serious restriction. Personally, I would rather be in the position we are in than the position in which the United States, Brazil and others currently find themselves.


We are presumably all very much aware that viruses don’t always stay the same. They change or mutate over time (called genetic drift). Influenza is a classic example. After decades of battling influenza, which is in many ways similar to COVID-19, the struggle continues. Every year the virus changes slightly and the vaccine needs to be modified. It is like a war that never ends.


This brings me to the subject of vaccines, which like the COVID-19 test, are not likely to be perfect either. All vaccines are not created equal and some are better than others. The best vaccines would protect the vast majority of people forever and be safe to yse. They would offer close to 100% protection. Whilst a frantic international effort is currently underway to develop an effective and safe vaccine, and some of the 130 or so vaccines that are in the development pipeline are showing promise, we need to manage our expectations. Making a vaccine is not like making popcorn, it takes time and the dedication of teams of scientists and stringent testing to ensure it is safe and the end result may not be perfect. To date the fastest we have ever produced an effective vaccine is four years (mumps and measles). There is still no vaccine for HIV/AIDS after 30 years of effort.


So although we are likely to break all records this time, and by a long way, it would be very optimistic to expect an effective vaccine to be developed by next year. Let’s keep our fingers crossed. In the meantime, let’s be careful and grateful for what we have been able to achieve so far. Let’s also be very careful about who gets on our waka.

In the meantime, let’s hope like hell that the Tricorder we saw all those years ago on Star Trek turns up sooner rather than later.


7 views0 comments

Recent Posts

See All