What’s Going On With COVID Testing?


It is quite possible the world has been driven mad by COVID.  Theories about the virus and its impacts are created by feverish minds about as rapidly as they are destroyed by reality. The latest casualties are Indian natural immunity and the value of seroprevalence studies as some marker of population level resistance to spread of the virus that escaped Wuhan.  There will surely be many more narratives that implode before this story ends. The complex ever changing epistemological landscape of all things COVID underscores the difficulty of translating the acquisition of knowledge into decisions — a topic that has been the subject of debate long before this pandemic afflicted the world. The scale of errors made through action or inaction by those given powers they should never have been given have been dumbfounding. Time and time again it has been those experiencing the pandemic at street level that have taken the body blows dealt by the virus and adjusted accordingly.  It was an ophthalmologist in Wuhan that sounded the alarm about the unusual pattern of cases, and it was bureaucrats that attempted to stifle spread of this news.  It was the head of the sequencing lab in China that uploaded a sequence of the new viral genome to a public server instead of waiting endlessly for permission from some higher up.  The NBA decided to cancel its season before the CDC thought it was a good idea, and it was citizens that rushed to buy n95 when Anthony Fauci was recommended against one.  

This isn’t unexpected. There are no professionals who manage pandemics of this scale because they happen once every hundred years.  The experts anointed to manage this aren’t gifted athletes who have honed their skills by shooting ten thousand free throws, they are mostly titled suits whose job is to confirm the biases of a large enough audience to be called back over and over again.  In this world, it is the plugged in art dealer who is right more often than wrong on all things COVID because the ‘expert’ scientists are busy using science as a political tool to play up the existential threat of a brain/lung/heart eating virus that can only be conquered by prophylactic restrictions on basic rights designed with uber eats ordering zoomocrats in mind. It’s no accident that the policy designers happen to be zoomocrats. 

The stories of COVID exposes the widening gap that social theorist Thomas Sowell wrote about a half century ago between those with first hand knowledge and decision makers. Sowell warned that clueless intellectuals making decisions would not only threaten our economic and political efficiency, but ultimately threaten our very freedoms because a knowledge based society would be replaced with an elitist, abstract, utopic vision of what society should be like.

The gap between knowledge and decision makers is best demonstrated through stories of the pandemic. This particular first hand story comes from a geriatrician in Philadelphia.

The Philadelphia Nursing home story, Fall 2020

It was a tired disbelief that swept through the air as another “slowdown” was announced in Philadelphia mid-November.  The inferno of the spring had given way to a quiet summer and early fall, but a rising number of positive cases left our public health overlords no choice.  New restrictions and warnings were planned, and the public schools that had been partially open were to be shuttered in an all virtual mode for the time being.  

This was an unfortunate surprise to many who hoped society had learned to live with the virus in some type of a tense Korean peninsula style standoff.  Rising case positivity was the buzzword used to describe a pandemic that once again threatened to spin out of control and overwhelm hospitals.  But this particular surge had a different flavor than the spring.

The Spring was a particularly brutal time for residents of nursing homes.  In some states like Pennsylvania, most of the deaths due to COVID were in Nursing Homes.

Prompted by the Spring devastation in nursing homes, the State introduced a host of rules designed to make nursing homes safer.  Many of the mandates mirrored what medical teams had been asking for in March : Universal testing of residents and staff and the ability to cohort COVID positive patients being chief among them.

But something strange happened after mass testing got underway.  There were lots and lots of positive results. Unlike March when symptomatic patients that were COVID positive rapidly decompensated and ended up in critical care units, the large majority of the patients that tested positive now seemed perfectly fine.  While the Spring saw scores of patients that died, this wave was marked by much less morbidity despite the much larger number of positive patients.  

Explaining this fall paradox has much to do with how we track the spread of pandemics in 2020.  For the first time in human history disease tracking is happening using PCR, a truly wonderful technology that is, unfortunately not without its limitations.  There is simply no escaping trade-offs.

The major problem is that it is simply too sensitive.  Any doctor can tell you that the same disease process affects patients in a variety of different ways that are usually beyond the scope of what humans can explain.  Coronary disease, cancer, and infectious diseases are indolent affairs in some and raging inferno’s in others.  Traditional tracking of pandemic spreads was by tracking how hospitals and mortuaries filled.  The yearly influenza waves in the United States gave rise to the U.S. Outpatient Influenza-like Illness Surveillance Network (ILINet) — a network comprising of ~3000 health care providers that report to the CDC on outpatients seen with a flu like illness, and hospitalization surveillance of laboratory confirmed influenza admissions (FluSurv-NET). 

COVID19, however, has been primarily tracked by PCR, and as evidenced by the nursing home protocol implemented in late May, this is happening in completely asymptomatic individuals.  Understanding why requires a little review of PCR.  PCR works by using small strands of genetic material (primers) designed to be complementary, and highly specific to the target of interest.

 False positive PCR tests can result from primers designed for COVID19 not being 100% specific to SARS-COV2, primers annealing to themselves and getting amplified, or sample contamination.  Unfortunately, it is not easy to ascertain exactly what the false positive rate is because testing in the United States is incredibly widespread and performed by a variety of labs using different primers and different processes.

But while some very smart experts in PCR have some reliable concerns about the current process, there are a fair number of real world datasets in low prevalence areas that suggest a low false positive rate.  The Office for National Statistics (ONS) in the United Kingdom reported 159 out of 208,730 samples were positive between July and September.  Even if every single positive test was a false positive that would still suggest a false positive rate of 0.08%.  Since the virus is known to be circulating the actual false positive rate is likely to be significantly lower than this.  The Aussies down under provide another laboratory because they have largely managed to reduce infection rates to near zero.   They also do a fair number of tests. Since mid-May they have been doing  20,000 – 60,000 tests per day in a country of 25 million people.

This is true in America as well.  Despite the continuous massive testing in a background of very little coronavirus there were remarkably few positive cases until recently when the Delta surge began.  Add to this bevy of data a recent report from Wuhan on the testing of 10 million residents over 15 days that found only 300 asymptomatic cases and it seems fairly likely that there is a high likelihood a positive PCR translates to the presence of Sars-CoV2.  

What the Wuhan paper doesn’t do, however, is provide any comfort to those looking for meaning in positive PCRs.  All 300 residents that tested positive were asymptomatic and there were no positive tests among 1174 close contacts that were traced.  This suggests a certain percentage of positive PCRs are asymptomatic, and don’t transmit virus to others.  In essence, these are false positives because they neither bother the carrier of these viral particles or threaten anyone who comes into contact with these carriers.

The achilles heel for PCR is that it doesn’t differentiate between live and dead virus, it simply is a fantastic way of picking up minute amounts of viral RNA.  Interestingly, the number of PCR cycles required for the test to be positive may provide some clues to the live-dead problem.

Jaafar and colleagues published a research letter in Clinical Infectious Diseases, that sought to correlate Cycle threshold (Ct) values with the ability to grow the virus in culture from a sample of almost 180,000 patients tested for Sars-Cov2.  At Ct of 25, up to 70% of patients are positive in culture, and at Ct of 30 this drops to 20%.  At Ct of 35 , a value commonly and widely used to report a positive result, less than 3% of inoculated samples grew in culture.

Figure 4. Cycle Threshold (Ct) values correlate with the ability to grow virus in culture.  At Ct values > 34, almost no PCR positive samples grow in culture, suggesting PCR positivity beyond this value doesn’t indicate live virus

The problem of course is that this represents one validation study in one lab.  There are 2 million tests per day being performed in the United States in a variety of different labs and on a variety of different platforms.  The FDA Emergency Use Authorization (EUA) which these tests operate under do provide information on the Ct and Limits of Detection (LoD), but provide no guidance on whether labs should be calling a positive result beyond the Ct at which the LoD was noted.

Curious about positive PCR tests my geriatrician friend was seeing in asymptomatic patients, he called the local private lab that was managing the testing for a long term care facility, to inquire about the Ct of these patients.  It turns out they were sending their specimens to Labcorp.  A few days later he was on the phone with one of the Medical Director’s of Labcorp who was sympathetic, but unhelpful. They were doing hundreds of thousands of tests per day, they didn’t have the manpower to track down a handful of patients for every doctor that called.  And no, he also couldn’t produce a general breakdown of Ct numbers for their positive tests.

So in this strange world we live in where testing companies LabCorp and Thermo Fisher did a combined $3 Billion in revenue for COVID testing alone in one quarter, it is impossible in realtime to sort out what percentage of positive tests represent patients with an active infection who can transmit the virus to others.  The stakes are not small. Many localities took steps to shut down local businesses and public schools after reaching a certain arbitrary threshold of positive cases. 

To be clear, the problem isn’t that cases defined by PCR are meaningless, and have no value, it’s that PCR based case spikes may not always mean death and destruction follows.  Recall, that the point of massive testing to curb pandemics was to head off spikes in hospitalizations and death.  But if a significant portion of the positive cases are of the non-infectious variety, mitigation strategies that are harsh by nature may have pain without the gain.  

The public health expert who isn’t walking the halls of the nursing homes doesn’t know any of this.  His ignorance of the relevant local knowledge blinds him to the truth just as the economist Friederich Hayek predicts in his seminal essay describing the necessary ineptness of any central planner faced with complex systems.

The epidemiologist, the bureaucrat and the virologist COVID expert on twitter would come to the natural conclusion that since an avalanche of deaths followed the spike in cases in the Spring of 2020, cases rising even more steeply in the late Fall of 2020 would portend more death that would require societal mitigation measures. It’s only someone that was in the nursing homes observing what was actually happening in nursing homes that would observe the difference in acuity of positive cases between the two surges and consider a different outcome.

There were other matters of concern in the fall – hospitalizations were rapidly rising as well.  But this, yet again, serves as an important reminder that data aggregators fall far short of those actually involved in the clinical care of patients on hospital wards.  When hospitalization rose rapidly in the Spring, elective surgeries were canceled and entire floors were converted to become COVID critical care units because the need for beds was immense.  The daily toll of deaths in the Spring in Philadelphia, while never close to New York City’s gruesome die-off of up to 1,000 deaths per day was still ugly.

But what wouldn’t be immediately obvious to the data scientist is that hospitalizations in the Spring of 2020 in Philadelphia were much different than hospitalizations in the Fall of 2020.  Doctors were more seasoned, less panicked, and had therapeutics available to them in the Fall they did not have in the early months of the pandemic.  The combination of more seasoned doctors and less ill patients meant that we never heard about a looming ICU bed shortage, or a ventilator shortage in the Fall despite the fact the hospitals never shut off elective surgeries.  

In stark contrast, the city of Philadelphia, spooked by rising cases and hospitalizations  announced a series of restrictions which included no indoor dining at restaurants, capacity limits at retail stores and religious institutions, telework for office workers, a ban on indoor gatherings of any size, outdoor gatherings limited to 10 percent occupancy with a cap of 2,000 people, and no youth or school sports. Colleges, universities, and high schools were also only allowed to offer online classes. In effect, the Philadelphia central planners stopped the world again because they didn’t know positive cases in the fall were different from positive cases in the Spring, or that hospitalizations in the fall were nothing like hospitalizations in the Spring.  

Not suprisingly, the feared second wave that was going to make the first spring wave pale in comparison never materialized.  There was an uptick in deaths, but it was nothing like the first wave.(Figure 6)  Those under the impression it was the reimposition of restrictions that impacted the second covid wave would be wrong.  A harder lockdown was attempted in early March in Philadelphia before COVID hit the city hard.  This wasn’t the somewhat complicated slowdown of the Fall, it was a complete shut down that kept everyone possible at home.  But COVID wasn’t stoppable, and no doubt the proximity to fleeing New Yorkers didn’t help.  A month later, hospitals in Philadelphia were converting regular units to intensive care units and straining from the surge.  To summarize, a hard lockdown initiated almost a month prior to the spring surge didn’t abort what was to come.  Why would anyone confidently  believe a considerably softer lockdown in the Fall would have had a significant impact on the Fall surge?

Hospitals did not respond to this particular second wave by closing elective surgeries because they didn’t need to.  There were no extra intensive care units, or new ICU staffing protocols that were instituted.  The bedside clinicians, and the hospitals with plenty of skin in the game were right. The Intellectual class, basing their decisions on utopic, hyperbolic visions of the effects of their little pronouncements, as usual, were wrong. 

Figure 6. Philadelphia hospitalizations in the Fall did not have the same mortality as the Spring, because patients admitted in the Fall of 2020 were not as sick.  Important, because hospitalizations were suggested as a metric for lockdown measures

The pandemic should humble us precisely its final effects vary in relation to a bewildering array of constantly shifting variables: City size, density, hospital capacity, pcr cycle thresholds, the age of a population, ambient temperature, presence of multigenerational households.  It is stupid to attempt to model this because its impossible to predict which locale will turn into Delhi and which locale will turn into Philadelphia. Even worse is to bark authoritatively about what single restriction from the governor’s or President’s desk caused the pandemic to either accelerate or decelerate. The best we can hope for is nimble, decentralized local units with plenty of skin in the game that can react with alacrity to changing circumstances on the ground. This means letting small business owners decide themselves whether to open or close, and it also means giving parents the right to decide whether to send their children to school or not. Fire the professional pandemic handlers. That may be the only path out of the madness afflicting the globe.

Anish Koka, MD is a cardiologist based in Philadelphia.

Leave a Reply