There has been recent discussion of the lack of accuracy of COVID-19 tests — both virus (rRT-PCR), which has been widely regarded as “the gold standard”, and antibody assays.
“The largest study on coronaviruses testing to date estimated a rate of 41% false negatives on RT PCR diagnostic tests used in China.”
The issue with PCR is not with the methods themselves, but the degree of human error that creeps in due to the complexity of the workflow — including pre-analytical sample handling, which can account for up to 68% of errors according to a recent paper, with overall inaccuracies of up to 41%. Variations in the accuracy of provisionally-approved EUA kits add to the issues.
Here’s some front-line input from a Healthcare Scientist Practitioner at the NHS in the UK:
PCR is — from my own experience — an absolute bitch of a tool to use. Its a pain in the ass to do correctly without missing out a step or contaminating your results. …what constitutes positive and negative can be altered by the operator. Also…it requires a certain level of starting viral DNA — which a truly infected person may not have. This is true not just for coronavirus testing but also for STI testing — false negatives for things like herpes are alarmingly high because you have to swab at peak infectivity. Most people don’t attend GUM clinics until the peak has passed. Its understandable to mistrust any PCR result and qPCR is even worse. That’s not to say most results are false, just that given the nature of PCR and the fact it’s reliant on the operators not committing a human error AND when you consider how many samples are being processed per day — its reasonable to suspect false negative and false positive reporting.
To illustrate the point raised about when the sample is taken, here’s a chart showing how accuracy is affected by the point during the infection when the specimen is taken. If we are to effectively control, manage and prevent the spread of COVID-19 — and the predicted increasing incidence of similar potential pandemics — perhaps we need to take a serious look at the entire testing approach.
Accurate results for both PCR-based (NAAT) virus testing and antibody (serology) testing depend on the point when samples are taken during the clinical progression of COVID-19 — and perhaps other viral diseases.
So how do we fix this?
Well, obviously, serology (antibody) testing is best done post-infection — and probably after symptoms have subsided. No real problem there, the reasons for that kind of testing are mostly to learn about acquired immunity, to find antibodies for research or to donate blood for treatment of others.
But testing for the virus is different. That is a fundamental part of stopping the spread, protecting people and beginning early treatment (once the most effective treatments are proven). How can we make molecular testing effective?
The answer is two-fold. First, it is clear from the chart above that, like it or not, it is basically pointless to test anyone who is asymptomatic. That also is not too much of a problem, if the latest statement from the World Health Organization (WHO) is accurate and it is indeed rare for an asymptomatic person to pass on a COVID-19 infection. So we only test those with symptoms.
The second part of the answer has to do with addressing the inherent complexities of PCR-based testing that cause unacceptable rates of inaccuracy. This one of those opportunities for technology to step up and deliver. Laboratory software (LIMS — Laboratory Information Management Systems, or LIS — Laboratory Information Systems) can control the testing process workflow so that it is virtually impossible — or at least significantly less likely for lab workers to miss out steps or make other procedural mistakes. Also, interfaced instruments rather than manually acquired results eliminate transcription errors. In short, the numbers and what is at stake make the use of a LIMS or LIS with those capabilities essential. No longer can it be acceptable to rely on paper and spreadsheets.
LIMS/LIS vendors like LabLynx, Inc. have begun offering their well-established (20 years), reliable and HIPAA-compliant system in a configuration especially optimized for COVID and other epidemic/pandemic work. That version of their LIS is called COVIDLiMS. It is very good at letting labs specify workflow sequences and steps, and mandate the completion of each in the correct order. Even better, they have also introduced a sample information collection kit (aptly named C-SIC Kit: COVID-Sample Information Collection Kit). The C-SIC Kit automates everything possible at point of sample collection and prints barcoded labels for the specimens. The labels are then scanned at the lab and all of the necessary physician, collector, patient and sample data are called up from the system’s secure cloud-based database.
The C-SIC Kit links automatically with COVIDLiMS, which in turn is also interfaced to PCRs and other instruments, creating a fully automated closed system that provides maximum protection against errors. The system also can offer a patient portal where they can be screened, exchange messages and get their results, as well as a physician web portal. For labs who have a suitable LIS already, the C-SIC Kit and portals can be interfaced with their existing system. Accurate testing during a predicted increasing frequency of contagious disease outbreaks is a global issue, for both public health systems and businesses/organizations. This is the time for laboratories to get the bugs out of molecular testing and for businesses to institute protective testing programs that will help ensure continuity and reduce risk.