Quidel (QDEL) is down over 15% from its close on Friday (8/7) on no obvious corporate news. I believe the sell-off may be due to this NY Times article, which indicates Ohio Governor Mike DeWine received a Quidel rapid antigen test and received a false positive result (he was confirmed to be negative by two subsequent molecular tests). Questions about the test’s accuracy adds a new angle to my existing short thesis on the stock. Please reference that article for background on Quidel and its coronavirus-related diagnostic tests.
Bears on the sell-side and buy-side have questioned the accuracy of Quidel’s rapid antigen SARS test, but I did not make it part of my short thesis because sacrificing accuracy in exchange for speed and convenience seems like a rational choice in a world where more accurate molecular testing is capacity constrained and quite slow. It was easy for me to believe that Quidel would be able to sell everything it could manufacture, even with lower accuracy. But this high-profile incident could conceivably give pause to states or institutions planning to use the test as a way to resume economic activity. It could also conceivably put some pressure on pricing. I am not necessarily putting a high probability on either of these two outcomes, but I am slightly increasingly the probability that one or the other could occur. In my opinion, this explains the recent pressure on shares.
Based on a study included in the test’s package insert (n=209), the Sofia SARS Antigen test should have a sensitivity of about 97% and a specificity of about 100%, which is very high. Sensitivity is the probability of getting a positive test result if the patient actually has the disease. Specificity is the probability of getting a negative test result if the patient truly doesn’t have the disease. But Governor DeWine’s test results suggest the specificity of test is not actually 100% in the real world.
That the real world doesn’t resemble controlled laboratory results should not come as a surprise to anyone familiar with diagnostic test statistics. 209 test samples in a controlled setting is probably not a sufficiently large sample to understand how the test will perform in a population of millions in less-than-ideal conditions. If we take Quidel’s Sofia flu assay as an example of a test that has been used in the real world for much longer, academic studies of sensitivity/specificity have been all over the map. One study has sensitivity and specificity both in the 80% range. Another has sensitivity in the 40/50% range and specificity near 99%. The lesson from all of this is that it is simply very difficult to know for sure how accurate a diagnostic test really is in the real world.
With such widely varying results, the more realistic question is how useful a test can be even with reduced sensitivity and specificity. Imagine we have 100 people who need to be tested for COVID-19 to see if they can return to work. Using a molecular test, we get accurate results, but we might have to wait up to two weeks if there is a test backlog. Instead, let’s assume we use the rapid antigen test and that it has 80% sensitivity/specificity in the real world. If 100 people are given the test, 10 of which actually have COVID-19, about 26 people would be told they are positive for the virus (80% of the 10 who actually have it plus 20% of the 90 people who don’t). 74 people would be told they don’t have they virus, although 2 of these 74 actually would have the virus. The 26 people who tested positive would then be given the more accurate molecular test, which might take a week to weed out the 8 true positives from the 18 false positives. We would send 74 people back to work immediately.
Based on this example, it’s not clear whether using a slow molecular test as the first-line test is the way to go (everyone has to wait), or whether it is better to use a rapid antigen test as the first-line test and then reflex the positive results to the molecular test (74% get to return to work immediately but 2% have the virus and might spread it). It’s still conceivable to me that many states and institutions will want to use a rapid test, and thus Quidel’s sales trajectory will stay strong until a vaccine is available. However, the high-profile challenge to the test’s specificity increases the probability that some states and institutions will simply stick with molecular methods, in my view.
Lastly, it is possible that pricing could come under pressure. If competitor tests prove to be more accurate, Quidel could be compelled to lower its price. Also, if Medicare contractors reduce their reimbursement rate for the test because of its apparent lower accuracy, Quidel would likely need to lower the price it charges distributors. Medicare contractors set reimbursement rates that are widely followed by other insurers (including for patients not covered by Medicare). The Medicare code currently used for the rapid antigen test appears to be 87426. Medicare often moves slowly to set national prices, and disclosures about reimbursement are limited. However, this press release suggest reimbursement might be around $50 for the antigen test, which currently provides a healthy spread between Quidel’s estimate test price of ~$20. This spread would need to close materially before Quidel would need to reduce its price, but it’s not out of the question.
The bottom line, in my opinion, is that a widely distributed vaccine is still the most likely path to a profitable short trade. However, questions about accuracy slightly increase the probability that near-term revenue could disappoint.
Disclosure: I am/we are short QDEL. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.