Rapid coronavirus test, commonly used in U.S., may miss infections in some situations

A rapid test for the novel coronavirus that was unveiled by President Trump on the White House lawn and has already been used hundreds of thousands of times can falsely produce negative results in patients who are infected, according to clinicians and laboratory experts.

The 13-minute test, developed by Abbott Laboratories, a major U.S. health care and medical device company, usually still works properly, the company and clinicians say. But Abbott’s initial guidance on how to administer the test — guidance approved by the Food and Drug Administration — could lead to an unspecified number of false negative results. Abbott says changing how samples are collected should fix the problem.

The test “performed equivalent to the other platforms with patients that had high and moderate loads of virus,” said Alan Wells, a pathologist who directs the clinical laboratories at the University of Pittsburgh Medical Center. “However, with lower loads of virus, a large fraction of these patients were not detected as positive.” Wells ran a validation of the test compared with stored samples that had been tested on other platforms.

In recent weeks, clinicians have expressed growing concerns about false negative results in testing for the coronavirus, known as SARS-CoV-2. Much of the problem has been blamed on issues with how samples for the main laboratory tests are collected. 

Since Abbott’s test, run on its “ID Now” platform, was granted emergency use authorization by the FDA on March 27, the company has shipped 600,000 tests. They are being used at CVS and Walgreens drive-through testing locations, to test nurses and police officers in Detroit, and at urgent care centers in Chicago. They are also being used in hospitals to test surgical candidates to determine if they should be operated on and what level of protective equipment physicians should wear.

The test can be used in two different ways. In one method, a swab that has been rubbed against the nose, throat, or the area deep in the back of the nose is inserted directly into the ID Now platform, a toaster-sized machine. That approach is still seen as effective. 

But the FDA cleared the test to be used in another way: The swabs could be put in a liquid solution — known as “viral transport media” — and tested, which is the standard way for processing many other kinds of tests and usually allows samples to be stored longer.

In a letter sent to health care providers Wednesday, obtained by STAT, Abbott said that the test should not be used in the solution because it could lead the test to give a negative result when a patient is actually infected. The company said it will work with the FDA to change the language of the product’s package, and intends for changes to be made by the week of April 20.

The letter says the issue was discovered as the new customers conducted validation tests.

In a statement, Abbott said that when that when the collection method is not used, “the test is performing as expected and we have confidence in the performance of the test.” It also stated that the ID Now platform is designed for and typically used in “urgent care clinics, physicians’ office labs, emergency departments, and retail settings,” where collection with transport media would not occur.

There is little doubt that when the ID Now coronavirus delivers a positive result, that means a patient is infected with the coronavirus. The question is how secure patients and doctors should feel with a negative result.

Wells, the pathologist, said that the University of Pittsburgh Medical Center is now not using the ID Now to rule out whether patients are carrying SARS-CoV-2, but will consider using the machine to identify patients to enroll in clinical trials. For testing surgery patients, it is using another testing machine, from Cepheid, that can return results in 45 minutes.

On an April 8 webinar by the Society for Maternal Fetal Medicine, one of the participants, Emily Miller, an OB-GYN at Northwestern, said her institution had a similar experience with the Abbott ID Now test, saying that its sensitivity (the ability to not miss infections) was “clinically not useful” for determining if pregnant women were infected.

“So we have not moved forward with it,” Miller said. “But I think the lesson there is it’s incredibly important to validate whatever test it is with your existing population and make sure you’re troubleshooting before you’re actually using that information clinically because I shudder to think — the path we would have gone down.”

Neither Miller nor Northwestern’s medical school responded to an email requesting comment.

Michael Misialek, a pathologist at Newton-Wellesley Hospital in Massachusetts, which is part of the Partners Healthcare system, said that he had run his own validation of the ID Now Covid-19 test and that the machine’s results compared favorably to other diagnostic devices. However, Misialek said, there was one sample where there was not agreement. He decided that the result could be related to over-dilution — the issue Abbott later warned about. His hospital took to sending the swabs dry, not in viral transport media.

“I think the important point is all these assays have been emergency release by the FDA,” Misialek said. “So they haven’t gone through the traditional rigor of studies, and validations and certainly not extensive comparisons. But right here, at Newton-Wellesley and my colleagues across Partners, we haven’t seen anything to date that worries us.”

He currently uses both the Abbott ID Now device and the Cepheid one when his hospital has a patient with Covid-19 symptoms and a decision needs to be made about whether they need to be admitted, or to which part of the hospital. Supplies of test kits for both devices are limited, he said. 

That means that health care professionals collecting samples must know, one day at a time, whether they need to collect dry swabs, for the Abbott device, or swabs in transport media, for the Cepheid one. Abbott has said that by June, it will be able to produce 2 million tests a month, about a fifth of the total number of Covid-19 tests currently being run in the U.S.

“It really goes week by week, and we will look at our supply, estimate what the usage has been and be in touch with the inpatient teams, letting them know where we are and what we’re going to use or not,” he said.

Does anyone know how many patients the diagnostic tests miss?

“I’ll tell you, my pathologist colleagues are asking the same question,” Misialek said. “And I don’t think anybody has a good sense of what the clinical sensitivity of not only the Abbott [test], but all these tests are.“

Gideon Gil contributed reporting.