Here’s some incredible data on PCR tests demonstrating a radically excessive lab practice that generates false positives. I’m almost tempted to say we’d do just as well using a thermometer and the coffee ground test. Open a coffee tin and take a sniff. Can you smell the distinct aroma of the grounds? If not, and if you have other common symptoms, there’s a decent chance you have an active COVID infection. That test is actually in use in some parts of the globe!
The data shown below on PCR tests are from the Rhode Island Department of a Health and the Rhode Island State Health Lab. They summarize over 5,000 positive COVID PCR tests (collected via deep nasal swabs) taken from late March through early July. The vertical axis in the chart measures the cycle threshold (Ct) value of each positive test. Ct is the number of times the RNA in a sample must be replicated before any COVID-19 (or COVID-like) RNA is detected. It might be from a live virus or perhaps a fragment of a dead virus. A positive test with a low Ct value indicates that the subject is likely infected with billions of live COVID-19 viruses, while a high Ct value indicates perhaps a handful or no live virus at all.
The range of red dots in the chart (< 28 Ct) indicates relatively low Ct values and active infections. The yellow range of dots, for which 28 < Ct <= 32, indicates possible infections, and the upper range of green dots, where Ct > 32, indicates that active infections were highly unlikely. It’s important to note that all of these tests were recorded as new COVID cases, so the range of Ct values suggest that testing in Rhode Island was unreasonably sensitive. That’s broadly true across the U.S. as well, which means that COVID cases are over-counted by perhaps 30% or more. And yet it is extremely difficult for subjects testing positive to learn their Ct values. You can ask, but you probably won’t get an answer, which is absurd and counterproductive.
Notice that the concentration of red dots diminished over time, and we know that the spring wave of the virus in the Northeast was waning as the summer approached. The share of positives tests with high Ct values increased over that time frame, however. This is borne out by the next chart, which shows the daily mean Ct of these positive tests. The chart shows that active infections became increasingly rare over that time frame both because positive tests decreased and the average Ct value rose. What we don’t know is whether labs bumped up the number of cycles or replications to which samples were subjected. Still, the trend is rather disturbing because most of the positive cases in May and the first half of June were more likely to be virus remnants than live viruses.
It’s also worth noting that COVID deaths declined in concert with the upward trend in Ct values. This is shown in the chart below (where the Ct scale is inverted). This demonstrates the truly benign nature of positive tests having high Ct values.
This is also demonstrated by the following data from a New York City academic hospital, which was posted by Andrew Bostom. It shows that a more favorable “clinical status” of COVID patients is associated with higher Ct values.
It’s astounding that the U.S. has relied so heavily on a diagnostic tool that gets so many subjects wrong. And it’s nearly impossible for subjects testing positive to obtain their Ct values. Instead, they are subject to self-quarantine for up to two weeks. Even worse, until recently there were delays in reporting the results of these tests of up to a week or more. That made them extremely unhelpful. On the other hand, thecoffee ground test is fast and cheap, and it might enhance the credibility of a subsequent positive PCR test, if one is necessary … and especially if the lab won’t report the Ct value.
The PCR test has identified far too many false infections, but it wouldn’t have been quite so damaging if 1) a reasonably low maximum cycle threshold had been established; 2) test results had not been subject to such long delays; and 3) rapid retests had been available for confirmation. The cycle threshold issue is starting to receive more attention, quite belatedly, and more rapid tests have become available. As I’ve emphasized in the past, cheap, rapid tests exist. But having dithered in February and March in approving even the PCR test, the FDA has remained extremely grudging in approving newer tests, and it persists in creating obstacles to their use. The FDA needs to wake up and smell the coffee!
My pre-Thanksgiving optimism about a crest in the fall wave of the coronavirus has been borne out for the Midwest and Mountain states in the U.S. These regions were the epicenter of the fall wave through October and most of November, but new cases in those states have continued to decline. Cases in a number of other states began to climb in November, however, contributing to a continuing rise in total new cases nationally. Some of these states are still in the throes of this wave, with the virus impacting subsets of the population that were relatively unscathed up till now.
My disclaimer: COVID is obviously a nasty virus. I don’t want to get it. However, on the whole, it is not a cataclysm on the order of many pandemics of the past. In fact, excess deaths this year will add just over 10% to projections of total deaths based on a five-year average. That level puts us in line with average annual deaths of about twenty years ago. And many of those excess deaths have been caused by our overreaction to the pandemic, not by the virus itself. As my endocrinologist has said, this is the greatest overreaction in all medical history. Unfortunately, a fading pandemic does not mean we can expect an end to the undue panic, or pretense for panic, on the part of interventionists.
This post will focus largely on trends in newly diagnosed COVID cases. I have been highly critical of our testing regime and COVID case counts because the most prominent diagnostic test (PCR) falsely identifies a large number of uninfected individuals as COVID-positive. However, case numbers are widely tracked and it’s fairly easy to find information across geographies for comparison. Deflate all the numbers by 30% if you want, or by any other factor, but please indulge me because I think the trends are meaningful, even if the absolute level of cases is inflated.
I’ll start with the good news and work my way down to states in which cases are still climbing (all of the following charts are from @Humble_Analysis (PLC)). The first chart is for the Great Plains, where cases peaked a little before Thanksgiving and have continued to fall since then. That peak came about six weeks after it began in earnest and cases have faded over the last four weeks.
Next we have the Mountain states, where again, cases peaked around Thanksgiving, though Idaho saw a rebound after the holiday. You’ll see below that a number of states had a distinct drop in new cases during the week of Thanksgiving. There was somewhat of a pause in testing during that week, so the subsequent rebounds are largely due to a “catch-up” at testing sites, rather than some kind of Thanksgiving-induced spike in infections.
Back to the Mountain region, the peak came an average of about six or seven weeks into the wave, but the duration of the wave appears to have been longer in Montana and Wyoming.
Here are the Southern Plain states, where cases plateaued around Thanksgiving (though cases in Missouri have clearly declined from their peak). In this region, case counts accelerated in October after a slow climb over the summer.
The situation is somewhat similar in the Midwest. where cases have generally plateaued. There were some post-Thanksgiving rebounds in several states, especially Tennessee. The wave began a little later in this region, in mid- to late October, and it is now seven to eight weeks into the wave, on average.
Here are the Mid-Atlantic states, which may be showing signs of a peak, though North Carolina has had the greatest caseload and is still climbing. These states are about seven weeks into the wave, on average.
The Northeast also shows signs of a possible peak and is about seven weeks into the wave, except for Rhode Island, which saw an earlier onset and the most severe wave among these states.
And finally we have the South, which is defined quite broadly in PLC’s construction. It’s a mixed bag, with a few states showing signs of a peak after about seven weeks. However, cases are still climbing in several states, notably California and Florida, among a few others.
Oregon and Washington were skipped, but they appear as the Pacific NW in the following chart, along with aggregations for all the other regions. Maine is Part of the “Rural NE”, which was also skipped. The fall wave can be grouped roughly into two sets of regions: those in which waves began in late September or early October, and those where waves began in early to mid-November. The first group has moved beyond a peak or at least has plateaued. The latter group may be reaching peaks now or one hopes very soon. It seems to take about seven weeks to reach the peak of these regional waves, so a late December peak for the latter group would be consistent.
Justin Hart has a take on the duration of these waves, but he does so in terms of the share of emergency room (ER) visits in which symptoms of COVID-like illness (CLI) are presented. CLI tends to precede case counts slightly. Hart puts the duration of these waves at eight to ten weeks, but that’s a judgement call, and I might put it a bit longer using caseloads as a guide. Still, this color-coded chart from Hart is interesting.
If this sort of cyclical duration holds up, it’s consistent with the view that cases in many of the still “hot” states should be peaking this month.
Aggregate cases for the U.S. appear below. The growth rate of new cases has slowed, and the peak is likely to occur soon. However, because it combines all of the regional waves, the duration of the wave nationwide will appear to be greater than for the individual regions. COVID-attributed deaths are also plotted, but they are reported deaths, not by date of death (DOD) or actual deaths, as I sometimes call them. Deaths by DOD are available only with a lag. As always, some of the reported deaths shown below occurred weeks before their reported date. Actual deaths were still rising as of late November, and are likely still rising. However, another indicator suggests they should be close to a peak.
A leading indicator of actual deaths I’ve discussed in the past now shows a more definitive improvement than it did just after Thanksgiving, as the next chart shows. This is the CLI share discussed above. An even better predictor of COVID deaths by actual DOD is the sum of CLI and the share of ER patients presenting symptoms of influenza-like illness (ILI), but ILI has been fairly low and stable, so it isn’t contributing much to changes in trend at the moment. There has been about a three-week lead between movements in CLI+ILI and COVID deaths by DOD.
(The reason the sum, CLI+ILI, has been a better predictor than CLI alone is because for some individuals, there are similarities in the symptoms of COVID and the flu.)
The chart shows that CLI peaked right around the Thanksgiving holiday (and so did CLI+ILI), but it remained on something of a plateau through the first week of December before declining. Some of the last few days on this chart are subject to revision, but the recent trend is encouraging. Allowing for a three-week lead, this indicates that peak deaths by DOD should occur around mid-December, but we won’t know exactly until early to mid-January. To be conservative, we might say the latter half of December will mark the peak in actual deaths.
The regional COVID waves this summer and fall seem to have run their course within 10 – 12 weeks. Several former hot spots have seen cases drop since Thanksgiving after surges of six to seven weeks. However, there are several other regions with populous states where the fall wave is still close to “mid-cycle”, as it were, showing signs of possible peaks after roughly seven weeks of rising cases. The national CLI share peaked around Thanksgiving, but it did not give up much ground until early December. That suggests that actual deaths (as opposed to reported deaths), at least in some regions, will peak around the time of the winter solstice. Let’s hope it’s sooner.
Successive waves within a region seem to reach particular subsets of the population with relatively few reinfections. The 10 – 12 week cycle discussed above is sufficient to achieve an effective herd immunity within these subsets. But once again, a large share of the vulnerable, and a large share of COVID deaths, are still concentrated in the elderly, high-risk population and in care homes. The vaccine(s) currently being administered to residents of those homes are likely to hasten the decline in COVID deaths beginning sometime in January, perhaps as early as mid-month. By then, however, we should already see a decline underway as this wave of the virus finally burns itself out. As vaccines reach a larger share of the population through the winter and spring, the likelihood of additional severe waves of the virus will diminish.
Lest there be any misunderstanding, the reasons for the contagion’s fade to come have mostly to do with reaching the effective herd immunity threshold within afflicted subsets of the population (sub-herds). Social distancing certainly plays a role as well. Nearly all of that is voluntary, though it has been encouraged by panicked pronouncement by certain public officials and the media. Direct interventions or lockdown measures are in general counter-productive, however, and they create a death toll of their own. Unfortunately, the fading pandemic might not rein-in the curtailment of basic liberties we’ve witnessed this year.
Let’s get one thing straight: when you read that “hospitalizations have hit record highs”, as the Wall Street Journal headline blared Friday morning, they aren’t talking about total hospitalizations. They reference a far more limited set of patients: those admitted either “for” or “with” COVID. And yes, COVID admissions have increased this fall nationwide, and especially in certain hot spots (though some of those are now coming down). Admissions for respiratory illness tend to be highest in the winter months. However, overall hospital capacity utilization has been stable this fall. The same contrast holds for ICU utilization: more COVID patients, but overall occupancy rates have been fairly stable. Several factors account for these differing trends.
Admissions and Utilization
First, take a look at total staffed beds, beds occupied, and beds occupied by COVID patients (admitted “for” or “with” COVID), courtesy of Don Wolt. Notice that COVID patients occupied about 14% of all staffed beds over the past week or so, and total beds occupied are at about 70% of all staffed beds.
Is this unusual? Utilization is a little high based on the following annual averages of staffed-bed occupancy from Statista (which end in 2017, unfortunately). I don’t have a comparable utilization average for the November 30 date in recent years. However, the medical director interviewed at this link believes there is a consensus that the “optimal” capacity utilization rate for hospitals is as high as 85%! On that basis, we’re fine in the aggregate!
The chart below shows that about 21% of staffed Intensive Care Unit (ICU) beds are occupied by patients having COVID infections, and 74% of all ICU beds are occupied.
Here’s some information on the regional variation in ICU occupancy rates by COVID patients, which pretty much mirror the intensity of total beds occupied by COVID patients. Fortunately, new cases have declined recently in most of the states with high ICU occupancies.
Resolving an Apparent Contradiction
There are several factors that account for the upward trend in COVID admissions with stable total occupancy. Several links below are courtesy of AJ Kay:
The flu season has been remarkably light, though outpatients with symptoms of influenza-like illness (ILI) have ticked-up a bit in the past couple of weeks. Still, thus far, the light flu season has freed up hospital resources for COVID patients. Take a look at the low CDC numbers through the first nine weeks of the current flu season (from Phil Kerpen):
There is always flexibility in the number of staffed beds both in ICUs and otherwise. Hospitals adjust staffing levels, and beds are sometimes reassigned to ICUs or from outpatient use to inpatient use. More extreme adjustments are possible as well, as when hallways or tents are deployed for temporary beds. This tends to stabilize total bed utilization.
The panic about the fall wave of the virus sowed by media and public officials has no doubt “spooked” individuals into deferring care and elective procedures that might require hospitalization. This has been an unfortunate hallmark of the pandemic with terrible medical implications, but it has almost surely freed-up capacity.
COVID beds occupied are inflated by a failure to distinguish between patients admitted “for” COVID-like illness (CLI) and patients admitted for other reasons but who happen to test positive for COVID — patients “with” COVID (and all admissions are tested).
Case inflation from other kinds of admissions is amplified by false positives, which are rife. This leads to a direct reallocation of patients from “beds occupied” to “COVID beds occupied”.
In early October, the CDC changed its guidelines for bed counts. Out-patients presenting CLI symptoms or a positive test, and who are assigned to a bed for observation for more than eight hours, were henceforth to be included in COVID-occupied beds.
Also in October, the FDA approve an Emergency Use Authorization for Remdesivir as a first line treatment for COVID. That requires hospitalization, so it probably inflated COVID admissions.
In addition to the above, let’s not forget: early on, hospitals were given an incentive to diagnose patients with COVID, whether tested or merely “suspected”. The CARES Act authorized $175 billion dollars for hospitals for the care of COVID patients. In the spring and even now, hospitals have lost revenue due to the cancellation of many elective procedures, so the law helped replace those losses (though the distribution was highly uneven). The point is that incentives were and still are in place to diagnose COVID to the extent possible under the law (with a major assist from false-positive PCR tests).
Improved Treatment and Treatment
While more COVID patients are using beds, they are surviving their infections at a much higher rate than in the spring, according to data from FAIR Health. Moreover, the average length of their hospital stay has fallen by more than half, from 10.5 to 4.6 days. That means beds turn over more quickly, so more patients can be admitted over a week or month while maintaining a given level of hospital occupancy.
The CDC just published a report on “under-reported” hospitalization, but as AJ Kay notes, it can only be described as terrible research. Okay, propaganda is probably a better word! Biased research would be okay as well. The basic idea is to say that all non-hospitalized, symptomatic COVID patients should be counted as “under-counted” hospitalizations. We’ve entered the theater of the absurd! It’s certainly true that maxed-out hospitals must prioritize admissions based on the severity of cases. Some patients might be diverted to other facilities or sent home. Those decisions depend on professional judgement and sometimes on the basis of patient preference. But let’s not confuse beds that are unoccupied with beds that “should be occupied” if only every symptomatic COVID patient were admitted.
Finally, here’s a little more information on regional variation in bed utilization from the HealthData.govweb site. The table below lists the top 25 states by staffed bed utilization at the end of November. A few states are highlighted based on my loose awareness of their status as “COVID “hot spots” this fall (and I’m sure I have overlooked a couple. Only two states were above 80% occupancy, however.
The next table shows the 25 states with the largest increase in staffed bed utilization during November. Only a handful would appear to be at all alarming based on these increases, but Missouri, for example, at the top of the list, still had 27% of beds unoccupied on November 30. Also, 21 states had decreases in bed utilization during November. Importantly, it is not unusual for hospitals to operate with this much headroom or less, which many administrators would actually prefer.
Of course, certain local markets and individual hospitals face greater capacity pressures at this point. Often, the most crimped situations are in small hospitals in underserved communities. This is exacerbated by more limited availability of staff members with school-age children at home due to school closures. Nevertheless, overall needs for beds look quite manageable, especially in view of some of the factors inflating COVID occupancy.
Marc Boom, President and CEO of Houston Methodist Hospital, had some enlightening comments in this article:
“Hospital capacity is incredibly fluid, as Boom explained on the call, with shifting beds and staffing adjustments an ongoing affair. He also noted that as a rule, hospitals actually try to operate as near to capacity as possible in order to maximize resources and minimize cost burdens. Boom said numbers from one year ago, June 25, 2019, show that capacity was at 95%.”
So there are ample beds available at most hospitals. A few are pinched, but resources can and should be devoted to diverting serious COVID cases to other facilities. But on the whole, the panic over hospital capacity for COVID patients is unwarranted.
We have a false-positive problem and even the New York Times noticed! The number of active COVID cases has been vastly exaggerated and still is, but there is more than one fix.
COVID PCR tests, which are designed to detect coronavirus RNA from a nasal swab, have a “specificity” of about 97%, and perhaps much less in the field. That means at least 3% of tests on uninfected subjects are falsely positive. But the total number of false positive tests can be as large or larger than the total number of true positives identified. Let’s say 3% of the tested population is truly infected. Then out of every 100 individuals tested, three individuals are actively infected and 97 are not. Yet about 3 of those 97 will test positive anyway! So in this example, for every true infection identified, the test also falsely flags an uninfected individual. The number of active infections is exaggerated by 100%.
But again, it’s suspected to be much worse than that. The specificity of PCR tests depends on the number of DNA replications, or amplification cycles, to which a test sample is subjected. That process is illustrated through three cycles in the graphic above. It’s generally thought that 20 – 30 cycles is sufficient to pick-up DNA from a live virus infection. If a sample is subjected to more than 30 cycles, the likelihood that the test will detect insignificant dead fragments of the virus is increased. More than 35 cycles prompts real concern about the test’s reliability. But in the U.S., PCR tests are regularly subjected to upwards of 35 and even 40-plus cycles of amplification. This means the number of active cases is exaggerated, perhaps by several times. If you don’t believe me, just ask the great Dr. Anthony Fauci:
“It’s very frustrating for the patients as well as for the physicians … somebody comes in, and they repeat their PCR, and it’s like [a] 37 cycle threshold, but you almost never can culture virus from a 37 threshold cycle. So, I think if somebody does come in with 37, 38, even 36, you got to say, you know, it’s just dead nucleotides, period.“
Remember, the purpose of the test is to find active infections, but the window during which most COVID infections are active is fairly narrow, only for 10 – 15 days after the onset of symptoms, and often less; those individuals are infectious to others only up to about 10 days, and most tests lag behind the onset of symptoms. In fact, infected but asymptomatic individuals — a third or more of all those truly infected at any given time — are minimally infectious, if at all. So the window over which the test should be sensitive is fairly narrow, and many active infections are not infectious at all.
PCR tests are subject to a variety of other criticisms. Many of those are discussed in this external peer-review report on an early 2020 publication favorable to the tests. In addition to the many practical shortfalls of the test, the authors of the original paper are cited for conflicts of interest. And the original paper was accepted within 24 hours of submission to the journal Eurosurveillance (what a name!), which should raise eyebrows to anyone familiar with a typical journal review process.
The most obvious implication of all the false positives is that the COVID case numbers are exaggerated. The media and even public health officials have been very slow to catch onto this fact. As a result, their reaction has sown a panic among the public that active case numbers are spiraling out of control. In addition, false positives lead directly to mis-attribution of death: the CDC changed it’s guidelines in early April for attributing death to COVID (and only for COVID, not other causes of death). This, along with the vast increase in testing, means that false positives have led to an exaggeration of COVID as a cause of death. Even worse, false positives absorb scarce medical resources, as patients diagnosed with COVID require a high level of staffing and precaution, and the staff often requires isolation themselves.
Many have heard that Elon Musk tested positive twice in one day, and tested negative twice in the same day! The uncomfortable reality of a faulty test was recently recognized by an Appeals Court in Portugal, and we may see more litigation of this kind. The Court ruled in favor of four German tourists who were quarantined all summer after one of them tested positive. The Court said:
“In view of current scientific evidence, this test shows itself to be unable to determine beyond reasonable doubt that such positivity corresponds, in fact, to the infection of a person by the SARS-CoV-2 virus.”
I don’t believe testing is a bad thing. The existence of diagnostic tests cannot be a bad thing. In fact, I have advocated for fast, cheap tests, even at the sacrifice of accuracy, so that individuals can test themselves at home repeatedly, if necessary. And fast, cheap tests exist, if only they would be approved by the FDA. Positive tests should always be followed-up immediately by additional testing, whether those are additional PCR tests, other molecular tests, or antigen tests. And as Brown University epidemiologist Andrew Bostom says, you should always ask for the cycle threshold used when you receive a positive result on a PCR test. If it’s above 30 and you feel okay, the test is probably not meaningful.
PCR tests are not ideal because repeat testing is time consuming and expensive, but PCR tests could be much better if the number of replication cycles was reduced to somewhere between 20 and 30. Like most flu and SARS viruses, COVID-19 is very dangerous to the aged and sick, so our resources should be focused on their safety. However, exaggerated case counts are a cause of unnecessary hysteria and cost, especially for a virus that is rather benign to most people.
I hope readers share my compulsion to see updated COVID numbers. It’s become a regular feature on this blog and will probably remain one until infections subside, vaccine or otherwise. Or maybe when people get used to the idea of living normally again in the presence of an endemic pathogen, as they have with many other pathogens and myriad risks of greater proportions, and as they should. That might require more court challenges, political changes, and plain old civil disobedience.
So here, then, is an update on the U.S. COVID numbers released over the past few days. The charts below are attributable to Don Wolt (@tlowdon on Twitter).
First, reported deaths began to creep up again in the latter half of October and have escalated in November. They’ve now reached the highs of the mid-summer wave in the south, but this time the outbreak is concentrated in the midwest and especially the upper midwest.
Reported deaths are the basis of claims that we are seeing 1,500 people dying every day, which is an obvious exaggeration. There have been recent days when reported deaths exceeded that level, but the weekly average of reported deaths is now between 1,100 and 1,200 a day.
It’s important to understand that deaths reported in a given week actually occurred earlier, sometimes eight or more weeks before the week in which they are reported. Most occur within three weeks of reporting, but sometimes the numbers added from four-plus weeks earlier are significant.
The following chart reproduces weekly reported deaths from above using blue bars, ending with the week of November 14th. Deaths by actual date-of-death (DOD) are shown by the orange bars. The most recent three-plus weeks always show less than complete counts of deaths by DOD. But going back to mid-October, actual weekly deaths were running below reported deaths. If the pattern were to follow the upswings of the first two waves of infections, then actual weekly deaths would exceed reported deaths by perhaps the end of October. However, it’s doubtful that will occur, in part because we’ve made substantial progress since the spring and summer in treating the disease.
To reinforce the last point, it’s helpful to view deaths relative to COVID case counts. Deaths by DOD are plotted below by the orange line using the scale on the right-hand vertical axis. New positive tests are represented by the solid blue line, using the left-hand axis, along with COVID hospitalizations. There is no question that the relationship between cases, hospitalizations, and deaths has weakened over time. My suspicions were aroused somewhat by the noticeable compression of the right axis for deaths relative to the two charts above, but on reviewing the actual patterns (peak relative to troughs) in those charts, I’m satisfied that the relationships have indeed “decoupled”, as Wolt puts it.
Cases are going through the roof, but there is strong evidence that a large share of these cases are false positives. COVID hospitalizations are up as well, but their apparent co-movement with new cases appears to be dampening with successive waves of the virus. That’s at least partly a consequence of the low number of tests early in the pandemic.
So where is this going? The next chart again shows COVID deaths by DOD using orange bars. Wolt has concluded, and I have reported here, that the single-best short-term predictor of COVID deaths by DOD is the percentage of emergency room visits at which patients presented symptoms of either COVID-like illness (CLI) or influenza-like illness (ILI). The sum of these percentages, CLI + ILI, is shown below by the dark blue line, but the values are shifted forward by three weeks to better align with deaths. This suggests that actual COVID deaths by DOD will be somewhere around 7,000 a week by the end of November, or about 1,000 a day. Beyond that time, the path will depend on a number of factors, including the weather, prevalence and immunity levels, and changes in mobility.
I am highly skeptical that lockdowns have any independent effect in knocking down the virus, though interventionists will try to take credit if the wave happens to subside soon for any other reason. They won’t take credit for the grim lockdown deaths reaped by their policies.
Despite the bleak prospect of 1,000 or more COVID-attributed deaths a day by the end of November, the way in which these deaths are counted is suspect. Early in the pandemic, the CDC significantly altered guidelines for the completion of death certificates for COVID such that deaths are often improperly attributed to the virus. Some COVID deaths stem from false-positive PCR tests, and again, almost since the beginning of the pandemic, hospitals were given a financial incentive to classify inpatients as COVID-infected.
It’s also important to remember that while any true COVID fatality is premature, they are generally not even close to the prematurity of lockdown deaths. That’s a simple consequence of the age profile of COVID deaths, which indicate relatively few life-years lost, and the preponderance of co-morbidities among COVID fatalities.
Again, COVId deaths are bad enough, but we are seeing an unacceptable and ongoing level of lockdown deaths. This is now to the point where they may account for almost all of the continuing excess deaths, even with the fall COVID wave. It’s probable that public health would be better served with reduced emphasis on COVID-mitigation for the general population and more intense focus on protecting the vulnerable, including the distribution of vaccines.
The CDC changed its guidelines on completion of death certificates on April 5th of this year, and only for COVID-19 (C19), just as infections and presumed C19 deaths were ramping up. The substance of the change was to broaden the definition under which death should be attributed to C19. This ran counter to CDC guidelines followed over the previous 17 years, and the change not only makes the C19 death counts suspect: it also makes comparisons of C19 deaths to other causes of death unreliable, since only C19 is subject to the new CDC guidance. That’s true for concurrent and historical comparisons. The distortions are especially bad relative to other respiratory diseases, but also relative to other conditions that are common in mortality data.
The change in the CDC guidelines was noted in a recent report prepared for the Florida House of Representatives. It was brought to my attention by a retweet by Justin Hart linked tothis pieceon Andrew Bostom’s site. Death certificates are divided into two parts: Part 1 provides four lines in which causes of death are listed in reverse clinical order of events leading to death. Thus, the first line is the final clinical condition precipitating death. Prior clinical events are to be listed below that. The example shown above indicates that an auto accident, listed on the fourth line, initiated the sequence of events. Part 2 of the certificate is available for physicians or examiners to list contributing factors that might have played a role in the death that were not part of the sequence of clinical events leading to death.
The CDC’s change in guidelines for C19, and C19 only, made the criteria for inclusion in Part 1 less specific, and it essentially eliminated the distinction between Parts 1 and 2. The following appears under “Vital Records Criteria”:
“A death certificate that lists COVID-19 disease or SARS-CoV-2 as a cause of death or a significant condition contributing to death.”
How much difference does this make? For one thing, it opens the door to C19-attributed deaths in cases of false-positive PCR tests. When large cohorts are subject to testing — for example, all patients admitted to hospitals — there will always be a significant number of false positives even when test specificity is as high as 98 – 99%.
The elimination of any distinction between Parts 1 and 2 causes other distortions. A review of the Florida report is illustrative. The House staff reviewed almost 14,000 certificates for C19-19 attributed deaths. Over 9% of those did not list C19 among the clinical conditions leading to death. Instead, in those cases, C19 was listed as a contributing factor. Under the CDC’s previous guidelines, those would not have been counted as C19 deaths. The Florida House report is conservative in concluding that the new CDC guidelines inflated C19 deaths by only those 9% of the records examined.
There are reasons to think that the exaggeration was much greater, however. First, the Florida House report noted that nearly 60% of the certificates contained information “recorded in a manner inconsistent with state and national guidance”. In addition, almost another 10% of the fatalities were among patients already in hospice! Do we really believe the deaths of all those patients whose diseases had reached such an advanced stage should be classified as C19 fatalities? And another 1-2% listed non-C19 conditions as the immediate and underlying causes.
Finally, more than 20% of the certificates listed C19 alone as a cause of death despite a range of other contributing conditions or co-morbidities. This in itself may have been prompted by the change in the CDC’s guidelines, as the normal standards often involve a “comorbidity” as the initial reason for hospitalization — in that case a clinical event ordinarily listed in Part 1. The high rate of errors and the fact that roughly two-thirds of the deaths reviewed occurred in the hospital, where patients are all tested and often multiple times, raises the specter that up to 20% more of the C19 deaths were either erroneous and/or misclassified due to false positives.
(An exception may have occurred in New York, where an order issued in March by Governor Andrew Cuomo to return C19-positive residents of nursing homes (including suspected C19 cases) back to those homes, The order was made before the change in CDC guidelines and wasn’t rescinded until later in April. There is reason to believe that some of the C19 deaths among nursing home residents in New York were undercounted.)
All told, in the Florida data we have potential misclassification of deaths of 9% + 9% + 2% + 20% = 40%, or inflation relative to actual C19 deaths of up to 40%/60% = 67%! I strongly doubt it’s that high, but I would not consider a range of 25% – 50% exaggeration to be unreasonable.
We know that reports of C19 deaths lag actual dates of death by anywhere from 1 to 8 weeks, sometimes even more. This is misleading when no effort is made to explain that difference, which I’ve never heard out of a single journalist. We also know that false positive tests inflate C19 deaths. The Florida report gives us a sense of how large that exaggeration might be. In addition, the Florida data show that the CDC guidelines inflate C19 deaths in other ways: as a mere contributing factor, it can now be listed as the cause of death, unlike the treatment of pneumonia as a contributing factor, for instance. The same kind of distortion occurs when patients contract C19 (or have a false positive test) while in hospice.
There is no doubt that C19 led to “excess deaths” relative to all-cause mortality. However, many of these fatalities are misclassified, and it’s likely that a large share were and are lockdown deaths as opposed to C19 deaths. That’s tragic. The CDC has done the country a massive disservice by creating “special rules” for attributing cause-of-death to C19. If reported C19 fatality rates reflected the same rules applied to other conditions, our approach to managing the pandemic surely would have inflicted far less damage to health and economic well being.
We’ve known for some times that COVID-19 (C19) follows seasonal patterns typical of the flu, though without the flu’s frequent antigenic drift. Now that we’re moving well into autumn, we’ve seen a surge in new C19 case counts in Europe and in a number of U.S. states, especially along the northern tier of the country.
The new case surge began in early to mid-September, depending on the state, and it’s been coincident with another surge in tests. From late July through early October, we had a near doubling in the number of tests per positive in the U.S. An increase in tests also accompanied the previous surge during the summer, which claimed far fewer lives than the initial wave in the early spring. In the summer, infections were much more prevalent among younger people than in the spring. Vitamin D levels were almost certainly higher during the summer months, our ability to treat the virus had also improved, and immunities imparted by prior infections left fewer susceptible individuals in the population. We have many of those advantages now, but D levels will fade as the fall progresses.
As for the new surge in cases, another qualification is that false positives are still a major testing problem; they inflate both case counts and C19-attributed deaths. In the absence of any improvement in test specificity, of which there is no evidence, the exaggeration caused by false positives grows larger as testing increases and positivity rates fall. So take all the numbers with that as a caveat.
How deadly will the virus be this fall? So far in Europe, the trends look very promising. Kyle Lamb provided the following charts from WHO on Twitter yesterday. (We should all be grateful that Twitter hasn’t censored Kyle yet, because he’s been a force in exposing alarmism in the mainstream media and among the public health establishment.) Take a look at these charts, and note particularly the lag between the first wave of infections and deaths, as well as the low counts of deaths now:
If the lag between diagnosis and death is similar now to the spring, Europe should have seen a strong upward trend in deaths by now, yet it’s hardly discernible in most of those countries. The fatality rates are low as well:
As Lamb notes, the IFRs in the last column look about like the flu, though again, the reporting of deaths and their causes are often subject to lags.
What about the U.S.? Nationwide, C19 cases and attributed death reports declined after July. See the chart below. More recently, reported deaths have stabilized at under 700 per day. Note again the relatively short lags between turns in cases and deaths in both the spring and summer waves.
Clearly, there has been no acceleration in C19 deaths corresponding to the recent trend in new cases. Northeastern states that had elevated death rates in the spring saw no resurgence in the summer; southern states that experienced a surge in the summer have now enjoyed taperings of both cases and deaths. But with each season, the virus seems to roll to regions that have been relatively unscathed to that point. Now, cases are surging in the upper Midwest and upper mountain states, though some of these states are lightly populated and their data are thin.
A few state charts are shown below, but trends in deaths are very difficult to tease out in some cases. First, here are new cases and reported deaths in Michigan, Wisconsin, and Minnesota. There is a clear uptrend in cases in these states along with a very slight rise in deaths, but reported deaths are very low.
Next are Idaho, Montana, North Dakota, and South Dakota. A slight uptrend in cases began as early as August. Idaho and Montana have had few deaths, so they are not plotted in the second chart. The Dakotas have had days with higher reported deaths, and while the data are thin and volatile, the visual impression is definitely of an uptrend in deaths.
The following states are somewhat more central in latitude: Colorado, Illinois, and Ohio. There is a slight upward trend in new cases, but not deaths. Illinois is experiencing its own second wave in cases.
Out of curiosity, I also plotted Massachusetts, Pennsylvania, and New Jersey, all of which suffered in the first wave during the spring. They are now experiencing uptrends in cases, especially Massachusetts, but deaths have been restrained thus far.
The upshot is that states having little previous exposure to the virus are seeing an uptrend in deaths this fall. The same does not seem to be happening in states with significant prior exposure, at least not yet.
There are major questions about the reasons for the lingering death counts in the U.S.. But consider the following: first, the infection fatality rate (IFR) keeps falling, despite the stubborn level of daily reported deaths. Second, deaths reported have increasingly been pulled forward from deaths that actually occurred in the more distant past. This sort of “laundering” lends the appearance of greater persistence in deaths than is real. Third, again, false positives exaggerate not just cases, but also C19 deaths. Hospitals test everyone admitted, and patients who test positive for C19 are reimbursed at higher rates under the CARES Act; Medicare reimburses at a higher rates for C19 patients as well.
We’re definitely seeing a seasonal upswing in C19 infections in the US., now going on five weeks. In Europe, the surge in cases began slightly earlier. However, in both Europe and the U.S., these new cases have not yet been associated with a meaningful surge in deaths. The exceptions in the U.S. are the low-density upper mountain states, which have had little prior exposure to the virus. The lag between cases and deaths in the spring and summer was just two to three weeks, and while it’s too early to draw conclusions, the absence of a surge in deaths thus far bodes well for the IFR going forward. If we’re so fortunate, we can thank a combination of factors: a younger set of infecteds, earlier detection, better treatment and therapeutics, lower viral loads, and a subset of individuals who have already gained immunity.
There are a bunch of nice graphs below summarizing the course of the coronavirus (C19) pandemic in different countries, as well as their policy responses. The charts are courtesy of Kyle Lamb, who has been an unlikely (in my mind…) but forceful voice regarding the pandemic over the past few months. I’m sorry if the resolution in some of the charts is poor, but I hope you can click on them for a better view.
The data reported in the charts goes through September 12. The first few charts below are “mirror charts”: they show newly diagnosed C19 cases by day on top, right-side up; on the bottom of each chart are C19-attributed deaths, but the vertical axis is inverted to create the “mirror effect”. The scales on the bottom are heavily stretched compared to the top (deaths are much smaller than cases), and the scales for different countries aren’t comparable. The patterns are informative nevertheless, and I’ll provide per capita deaths separately.
Let’s start with the U.S., where the early part of the pandemic in the spring was quite deadly, while the second, geographically distinct “wave” of the pandemic was less deadly. It looks bad, but the high number of deaths in the spring was partly a consequence of mismanagement by a few prominent government officials in the Northeast, most glaringly Governor Andrew Cuomo of New York. The full pattern for the U.S. combines different waves in different regions. The overall outcome to-date is 622 deaths per million of population.
Then we have charts for (deaths/mil in parens): the UK (628), Italy (591), Spain (653), France (467), Germany (114), the Netherlands (364), and Switzerland (240), which all have had second waves in cases, of but hardly any noticeable second wave in deaths, at least not yet:
And finally, we have Sweden (576), which had many deaths during the first wave, but very few now. Overall, to-date, Sweden has faired better than the U.S., Spain, the UK, and Italy — not to mention Belgium (870), for which I don’t have mirror charts.
There are several points to make about the charts:
First, the so-called second wave this summer has not been as deadly as the virus was in the spring. The U.S. is not an exception in that regard, though it did have more C19 deaths than the other countries. The count of U.S. deaths in the summer was partly due to C19 false positives under a much heavier testing regime, as well as “death laundering” by public health authorities that looks suspiciously like a politicization of the attribution process: C19 deaths over the summer have been well in excess of what would be expected from C19 hospitalizations and ICU admissions. It’s also evident that deaths are being reallocated to C19 from other natural causes, as this chart from The Ethical Skeptic shows (compare the bright line for 2020 to the (very) dim but tightly clustered baselines from prior years):
Second, most of the charts for Europe (not Sweden) show a late summer escalation in cases, though cases in Spain and Germany appear to have crested already. If an uptrend in deaths is to follow, it should become noticeable soon. Thus far, the wave certainly looks less threatening.
Finally, it’s noteworthy that Sweden’s early experience, which was plagued by mismanagement of the virus’ threat to the nursing home population, later transitioned to a dramatic fading of cases and deaths. There has been no late summer wave in Sweden as we’ve seen elsewhere. This despite Sweden’s far less stringent non-pharmaceutical interventions (NPIs). Sweden’s deaths per million of population are now less than in the US, the UK, Italy, Spain, and Belgium, and most of those differences are growing.
All of the other countries discussed above have had far more stringent lockdown policies than Sweden, and at far greater economic cost. The following charts show some cross-country comparisons of an Oxford University index of NPI stringency over time. It combines a number of different dimensions of NPIs, such as mask mandates, restrictions on public gatherings, and school closures. The first chart below shows the U.S. and the UK contrasted with Sweden. The other countries discussed above are shown in separate charts that follow.
In the U.S., there has been tremendous variation across states in terms of stringency due to the federalist approach required by the U.S. Constitution, but overall, the Oxford measure for the U.S. has been broadly similar to the UK over time, with the largest departures from one another at the start of the pandemic.
The stringency of NPIs over the full pandemic depends on their day-by-day strength as well as their duration at various levels. One could measure stringency indices and deaths at various points in time and produce all kinds of conflicting results as to the efficacy of NPIs. On the whole, however, these charts suggest that stringent NPIs hold no particular advantage except perhaps as a way to temporarily avoid overwhelming the health care system. Even the original “flatten the curve” argument acknowledged that the virus could not be avoided indefinitely at a reasonable cost via NPIs, especially in an otherwise free society.
Note that most of these countries eased their NPIs after the initial wave in the spring, but several remained far more stringent than Sweden’s policies. That did not prevent the second wave of cases, though again, those were far less deadly.
As Jacob Sullum writes, and what is increasingly clear to honest observers: lockdowns tend to be ineffective and even destructive over lengthy periods.
A working paper from the National Bureau of Economic Research finds that four different “stylized facts” about the growth in C19 deaths are consistent across countries and states having different policy responses to the virus. The authors say:
“… failing to account for these four stylized facts may result in overstating the importance of policy mandated [non-pharmaceutical interventions] for shaping the progression of this deadly pandemic.“
Here’s Bill Blain’s discussion of the inefficacy of lockdowns. And here is Donald Luskin’s summary of his firm’s research that appeared in the WSJ, which likewise casts extreme doubt on the wisdom of stringent NPIs.
The virus is far from gone, but this summer’s wave has been much more docile in both Europe and the U.S. There are reasons to think that subsequent waves will be dampened in many areas via the cumulative immunity gained from exposure thus far, not to mention improvements in treatment and knowledge regarding prophylaxis such as Vitamin D supplements. Government authorities and their public health advisors should dispense with the pretense that stringent NPIs can mitigate the impact of the virus at a reasonable cost. These measures are constitutionally flawed, impinge on basic freedoms, and look increasingly like government failure. Risk mitigation should be practiced by those who are either vulnerable or fearful, but for most people, particularly children and people of working age, those risks no longer appear to be much worse than a bad year for influenza.
The tremendous increase in testing for COVID-19 (C19) this summer was associated with an increase in cases. Most of these tests were so-called PCR tests with samples collected via deep nasal swabs. More testing did not fully explain the increased case load, but false positives (FPs) still accounted for a substantial share. That’s especially true in light of the decline in positivity rates, which reflected a decline in the actual prevalence of active infections. FPs also account for a substantial share of the deaths attributed to COVID, which are obviously cases of false attribution. If a test for C19 is positive, it will be listed on the death certificate.
COVID Case Inflation
The exaggeration of confirmed cases due to FPs is more substantial as the prevalence of active infection declines. That’s because the share of true positives in the tested population declines, while the share of false positives must rise due to the greater share of uninfected individuals in the population.
Now, as the contagion is waning in former hot spots, there is a danger that FPs create the impression of persistence in the case counts. That’s costly not just for those incorrectly diagnosed, but also in terms of medical resources, for communities subject to excessive public intervention, such as inappropriate lockdowns, and in terms of the fear promoted by these inaccuracies.
There is evidence that testing standards under CDC guidance are so broad that a large number of inactive, non-infectious cases are being flagged as positives (see the chart above for the intuition, as well as the graphic at the bottom of this post). The tests sometimes amount to a coin flip when it comes to evaluating positives; some of the positives might even come from non-novel coronaviruses such as the common cold! This paper by Andrew N. Cohen, Bruce Kessel, & Michael G. Milgroom – CKM) questions the guidance of public health authorities on testing more generally. From the abstract (my emphasis):
“Unlike previous epidemics, in addressing COVID-19 nearly all international health organizations and national health ministries have treated a single positive result from a PCR-based test as confirmation of infection, even in asymptomatic persons without any history of exposure. … positive results in asymptomatic individuals that haven’t been confirmed by a second test should be considered suspect.”
False Positive Math
When I wrote about “The Scourge of False Positives” in July. I noted that a test specificity of 95% implies that 5% of uninfected individuals will falsely test positive. Unfortunately, that still produces a huge number of FPs when testing is broad. That’s NOT a good reason to avoid broad testing; it just means that positive tests should be confirmed by another test. (In this case, two tests with the same specificity reduce a 5% false positive rate to 0.25%.That’s why fast, cheap tests are necessary for confirmation.
Again, exaggerated case counts due to FP’s become more severe as a contagion wanes. That’s because FPs become an increasingly large share of positive test results and overstate the persistence of the virus. If active infections fall to 1% of 750,000 daily tests, or 7,500 true cases, the 5% specificity implies 37,125 FPs: true positives would be only 17% of positive cases. Much worse than a coin flip! And again, which cases are infectious?
How Bad Are FPs, Really?
This recent research, also authored by CKM, explains the reasons why FPs are usually an issue in the real world, despite the tests’ reportedly perfect reactivity to anything other than the virus’ genetic fragments. CKM find that the median FP rate in their sample of “tests of tests” was 2.3%. That means 23 out of every 1,000 uninfected people tested will test positive.
If that seems small to you, suppose the true prevalence of active infection in a population is 4%. If 1,000,000 people are tested and there are no false negatives (unlikely), then 40,000 infected people will be identified by the test. However, another 22,000 uninfected people will also test positive ((1,000,000 – 40,000 infected) x 0.023). That means the number of positive tests will be inflated by 55%. They’ll all receive some form of treatment or ordered into quarantine.
Expanded Testing and FPs
This summer, the volume of daily tests increased from about 150,000 a day in early April to more than 750,000 a day in July. That’s a 400% increase, but the true prevalence of active infection in the expanded test population during the summer was almost certainly lower than in the spring. Suppose active infections fell from 10% of the test population in the spring to 5% in the summer. That means the daily number of “true positives” would have risen from 15,000 to 35,000 in the expanded test population (and again I assume no false negatives for simplicity). The number of FPs, however, would have risen from 3,105 to 16,445. Therefore, FPs would have accounted for 40% of the increase in “confirmed” cases between spring and summer.
False COVID Deaths
FPs are also inflating COVID death counts. PCR tests are routinely given at hospital admission for any cause, and even after sudden death, especially as the availability of tests increased late in the spring. This subset of the tested population will certainly have its share of FPs. If such a patient dies, regardless of underlying cause, it might well be attributed to COVID-19 as it will still appear on the death certificate. The same has occurred in the case of traffic fatalities, suicides, and other sudden deaths.
The FP problem also plagues tests of seroprevalence, which determine whether an individual has had the virus or is cross-protected against the virus by antibodies acquired via non-novel coronavirus infections. The consequences of these antibody FPs can be serious as well, because it means a positive test might not ensure immunity. As the exposed share of the population increases, however, the FP share of antibody tests is diminished.
As long as testing is required, dealing with FPs (and false negatives, of course) requires repeated testing, as CKM state unequivocally. And the tests must be fast to be of any use. The current testing regime must be overhauled to prevent false positives from costly impositions on the lives of uninfected patients, consuming unnecessary medical resources, making unrealistic assessments of cases and deaths, and unnecessary suspensions of normal human social activity and liberty.
In advanced civilizations the period loosely called Alexandrian is usually associated with flexible morals, perfunctory religion, populist standards and cosmopolitan tastes, feminism, exotic cults, and the rapid turnover of high and low fads---in short, a falling away (which is all that decadence means) from the strictness of traditional rules, embodied in character and inforced from within. -- Jacques Barzun