(Editor’s note: The following is excerpted from a chapter in the recent book, “Unmasked 2020: Colorado’s Radical Left Turn and a Warning to America”)
The response to the coronavirus in Colorado is a tale of faith and devotion. Not toward the people of Colorado or God, but to a handful of academics and a model they created. It is a tale of hubris in that the model, and the assumptions baked into it, became sacrosanct. As such, the predictive outputs of the model drove Colorado’s responses, unchallenged by anyone or anything. Conflicting opinions and actions were literally considered deadly and conflicting data was simply ignored. As a result, businesses were shuttered, 340,000 Coloradans lost their jobs and immeasurable pain was inflicted, largely to avoid the model’s fantastic doomsday scenarios.
The model was infallible in that it was designed such that any outcome would fall within its range of predictions and any model-generated value, such as a measure of social distancing, would be used as the sole measure of Coloradans’ obedience to orders. So, for example, if hospitalizations climbed, the model would say it’s because social distancing dropped to some value, such as for example, 45%. The 45% wouldn’t be derived from a study of behavior, it was simply what the model said must have happened to produce the hospitalization increase. The model’s self-validation limited criticism (how do you argue against 45%?) and enabled it to carry more and more influence over policy.
Early Panic
The model was born in panic and developed in desperation. Images of makeshift Wuhan hospitals, the news of the first COVID-19 death in the U.S., and the announcement of Colorado’s first case on March 5, 2020, set the backdrop for Colorado Governor Jared Polis’ declaration of a state of emergency on March 11. Three days later, Colorado saw its first COVID-19 death and Polis held a press conference to discuss Colorado’s strategy. Standing before an easel supporting a “Flatten the Curve” poster, Polis explained that a goal of Colorado’s response was to slow the spread of COVID-19 such that Colorado’s hospitals would not be overwhelmed, which would lead to rationing of care as was being experienced in Italy.
It is important to note that the idea behind flattening the curve was not to avoid significant numbers of hospitalizations, but to spread out their occurrence such that no one is denied access to healthcare due to lack of available resources. Likewise, deaths are not avoided by flattening the curve, they are only delayed. Polis basically acknowledged this when he lamented, “Under some projections most Americans will likely get coronavirus.”
Meanwhile, a paper released on March 16 from Imperial College London gained worldwide attention for its prediction of 2.2 million deaths in the United States if we did nothing in response to the coronavirus. Colorado’s share of such a disaster would be about 39,000 fatalities. For reference, Colorado’s portion of the 2017-2018 flu season’s 61,000 U.S. deaths was likely around 1,000.
Also on March 16, the Governor’s Expert Emergency Epidemic Response Committee (GEEERC) held a meeting where they discussed Colorado’s response to the coronavirus. The committee discussed possible responses and as a first step recommended the closing of bars, restaurants, gyms, theaters and casinos. A few hours after the meeting, Polis did just that. Two days later, Polis would close schools and ban gatherings of more than 10 people. GEEERC also discussed additional closures that would be implemented in the following days. The committee was not working with a lot of information, with one member declaring, “What we’re embarking on here is evidence-informed guesswork.”
Modeling
It was during this early period that the COVID-19 Modeling Group (CMG) was formed and began preparing estimates of infections, hospitalizations and deaths due to COVID-19 and feeding that information to the Polis administration. The group was staffed with professors and post doctorate fellows with one person from the Colorado Department of Public Health and Environment (CDPHE) who had earned a PhD in 2019. Over the next several months the CMG would produce multiple reports on the estimated damage caused by COVID-19 in terms of hospitalizations and fatalities.
What the group lacked was any sense of the collateral damage their recommendations would do to society.
In a March 27 press conference, Governor Polis discussed the CMG’s first publicly released report and its prediction of 33,200 deaths by June 1, 2020 if Coloradans did not immediately begin social distancing. This was a horrifying prediction, particularly when considering that, at the time, Colorado had suffered only 31 deaths from COVID-19. However, the number did align well with the Imperial College London prediction for the US.
What was just as concerning as the numbers being presented was how they were being presented. Aside from the occasional lip service to the model being an estimation of the real world, the model was generally treated as gospel by all government officials and the media. This was troubling since the model’s main input used to project future trends was the percentage of social distancing practiced by Coloradans, and that value was derived from the model itself. This was done by basing Coloradans social distancing performance on hospitalizations that occurred a few days later.
This aspect was explicitly described, for example, in CMG’s July 16 report in which they stated, “We note that due to the approximately 13-day lag between infection and hospitalization, we are currently able to estimate social distancing through late June.” In other words, once they see how many people are hospitalized, they can tell you what must have been the social distancing level of 13 days ago. This isn’t necessarily a bad feature of a model. Indeed, to make models as accurate as possible, it is reasonable to continually update them with the latest information.
However, such results should at least be put to the smell test. For example, on May 29, the CMG reported that during the Stay at Home order Coloradans socially distanced at a rate of 80% (0% would be no social distancing and 100% would be no contact whatsoever). Then when the Stay at Home order was lifted for much of Colorado, the model suggested Coloradans increased their social distancing to 85%. It seems unlikely that Coloradans, in a time where hospitalizations were plummeting and restrictions were easing, would actually increase their reclusiveness. Yet the data goes unchallenged because the model is deemed infallible.
It is this reverence for the model and the underlying assumptions used in creating it that did the most damage to Colorado. As noted, the originally released model data showed Colorado suffering 33,200 deaths by June 1 and this number aligned well with the Imperial College London study. However, where the Imperial College London study was subsequently panned for being poorly crafted and wildly overly pessimistic, subsequent CMG predictions went the other way. In their April 6 report, the CMG, inexplicably and without explanation, changed their June 1 “0% social distancing efficacy” prediction to 73,162 deaths with total deaths of over 80,000 by January 1, 2021! This value was criticized and after the April 6 report, subsequent reports no longer predicted fatalities, although the underlying assumptions remained basically unchanged.
The 80,000 potential fatalities was wildly out of step with other epidemiologists and with real-life results and evidence from around the world. 80,000 deaths out of Colorado’s population of 5.8 million is a population mortality rate of 1.4%. To put that in perspective, if the entire world saw that mortality level, COVID-19 would kill 109 million people. In the U.S., that rate would translate into over 4.5 million deaths. Nobody but the CMG was predicting such causalities. At the 1.4% population mortality rate, Sweden, where they never went into lockdown and never closed primary schools, would suffer 140,000 deaths. As of July 30, 2020, Sweden had experienced 5,739 COVID-19 fatalities and was seeing around 1 new COVID-19 death a day. None of this seemed to be on the radar of the CMG.
While the world was taking in new information and updating predictions, the CMG was doggedly sticking to their March predictions and not allowing any outside information to alter their path. On March 30, Polis noted that Colorado needed 9 or 10 thousand ventilators. Colorado hit its peak ventilator usage of 471 on April 16. Colorado spent tens of millions of dollars to have the Army Corps of Engineers convert the Colorado Convention Center into a makeshift hospital that never saw a single patient. Yet, none of these misfires caused the CMG to rethink their assumptions.
Fundamentally, the CMG’s model has four key control measures: social distancing, mask wearing, identification and isolation of cases, and contact tracing. Everything else seems to have remained basically unchanged from March through the end of July (the time of this writing). That is, the underlying assumptions of infection fatality rates, basic (unmitigated) reproduction number, and asymptomatic rates seem to be unwavering. Furthermore, the model does not seem to contemplate any percentage of society having an innate or preexisting resistance to infection.
And therein lays the root problem of the CMG’s self-validating model: massive death if we do nothing, and the only significant variable used is social distancing.
(Mis)Using the Model to Guide Policy
To understand how the model became gospel to Governor Polis, controlling his response, you must understand Polis. He has a tendency to latch onto a study or claim that supports his world view and follow it while disregarding any evidence to the contrary. For example, he based claims that state provided pre-school is a good investment based on tiny studies that claimed kids who attend pre-school are less likely to serve prison sentences later in life. Yet Polis completely disregards large Head Start studies that show the effects of preschool dissipate during elementary school. In other words, Polis selects experts that support his agenda and world view and ignores other sources. This appears to be happening with respect to COVID-19.
The threat from the coronavirus put Polis in a position to declare a state of emergency and take on tremendous power. On March 25, Polis issued a state-wide Stay at Home order scheduled to last through April 11. Given the magnitude of the number of predicted deaths from the model, the escalating number of hospitalizations and the expected limited duration of the order, there seemed to be little consideration of conflicting values and identifiable trade-offs. Indeed, in confronting the COVID-19 threat, Polis appeared to give the recommendations of the CMG experts and their models an unprecedented monopoly over predicting the future.
Governor Polis would maintain his state of emergency through the summer of 2020, never seriously considering the trade-offs involved – a far cry from the deliberative policy process one would hope our government would undertake when enacting such widespread and oppressive orders.
As noted, in early April the CMG upped the ante by reporting that the lockdowns were preventing 80,000 deaths. Thus all of Polis’ decisions were supported by the belief his actions were the demarcation between life and death for tens of thousands. Under these conditions, over time the model outputs became sacrosanct, for they were the source of his power and the justifications for his actions. If the models were wrong, that would mean Polis’ actions were wrong. You can see this progression in Polis’ press briefings. Early in the pandemic, the model was sometimes described with a degree of humility. As time went by, the model’s outputs were treated more and more like iron-clad representations of the real world.
This combination of infallible quasi-scientific models and the government’s responsibility for staving off massive fatalities had tremendous negative effects. Foremost is that there may have been up to one non-COVID-19 death caused by the lockdown for every two to four COVID-19 deaths. During the height of the Stay at Home orders, emergency rooms across the country saw massive declines in patients. ER doctors, for example, reported seeing less heart attack and stroke victims and the ones they were seeing were more serious than normal. People who would have normally gone to a doctor or ER due to trauma were delaying or skipping the trip for fear of catching COVID-19. Many health care professionals were actively telling people to stay away from doctors and ERs if at all possible. In the U.S., these delays likely caused tens of thousands of non-COVID-19 fatalities. Moreover, missed or delayed treatments for cancer and other ailments and reduction of screenings may have negative ramifications for years to come.
The massive number of predicted deaths justified reactions that were catastrophic:
- destroyed 340,000 jobs in Colorado,
- contributed greatly to the intensity of social unrest after the horrific death of George Floyd,
- destroyed countless businesses both large and small,
- likely caused numerous deaths of despair due to suicide and substance abuse,
- likely enabled significant increases in domestic abuse and child abuse,
- delayed the education of over a million of Colorado’s children,
- delayed elective surgeries and medical treatment for thousands of people,
- created a huge budget shortfall diminishing public services and benefits,
- diminished the quality of life of millions of Coloradans.
None of the above “side effects” worked their way into the CMG’s modeling.
The sheer number Polis and the CMG believed would die if Colorado returned to normal also led to a fundamental change in Colorado’s response: they completely abandoned “Flatten the Curve” and moved to “Hunker Down until a Vaccine.” This was made explicit in Governor Polis’ June 15 press conference where he described the “Protect Our Neighbors” phase of Colorado’s COVID-19 response as, “the phase that is the way we have to live until there is a vaccine or cure.”
In truth, they were trying to win an unwinnable battle.
Dr. Rachel Herlihy, whom Polis called the state’s top epidemiologist, has said that we “must continue to be vigilant if we are going to defeat this virus.” Flatten the curve said that you can’t defeat coronavirus, you can only delay it to prevent healthcare system overload. But now Colorado’s top epidemiologist was declaring the goal of defeating coronavirus when over 50,000 Coloradans had tested positive for the virus. By the end of July, Polis was speaking of the effort to “crush this virus” and all semblance of rationality was gone.
Few commentators or news reports have pointed out the obvious: No one voted for this, not even the state’s elected legislators. Only one Coloradan made the decision to hunker down until a vaccine is available – Governor Polis. He made the decision and Colorado was thrown into perpetual semi-lockdown with a government official exercising perpetual emergency powers — justified by the 80,000 deaths he was allegedly preventing.
When you are preventing 80,000 deaths, almost anything can be justified, such as Stay at Home orders where, for the first time in our state’s history, healthy people are told they can’t leave their homes under threat of fines and imprisonment for up to one year.
The threat of 80,000 deaths can justify basically calling people (and neighbors and coworkers) “murderers” just because they want to eat at a restaurant on Mother’s Day. In response to such an illegal celebration of motherhood, Polis claimed the lack of adherence to his social distancing orders can make the difference “between going on living or suffering a particularly agonizing, painful and lonely death.” This is absurd since Polis’ own claimed policy is meant to suppress the virus to a manageable level, which implies allowing a certain number of people to die from COVID-19. Again, it is Polis’ hubris that allows him and him alone to determine what risk level Colorado should maintain. His determination of acceptable levels of death is noble, while the people’s is selfish and murderous.
Since Polis and the CMG believe there is an almost unlimited number of fatalities that will befall Coloradans if they open up too much, they have created an atmosphere of constant fear that the next surge is right around the corner. Instead of fostering resolution to weather a storm, they placed the blame for any surge on a disobedient population — typically Republicans and conservatives who value individualism above collectivism.
Treating Colorado citizens like impetuous teenagers, Polis frequently warned them that if they didn’t change their behavior, punishment would follow. Regarding lockdowns, in early April Polis warned,24 “If there is any way to safely end it sooner, then we will. And likewise, if Coloradans aren’t staying at home and the numbers of the dead and dying continue to increase, then it could go longer.” In late April, he proudly displayed a graphic at a news conference that read, “LET ME BE CLEAR: IF PEOPLE DON’T DO THEIR PART — We will go back to Stay at Home.”
This tactic creates a no-lose political dynamic for Polis. If hospitalizations and deaths go down, he will claim it was his leadership that saved the day. If they go up, Polis will simply blame the people for not doing what they were told. And here’s the most insidious part: the model will back him up either way. Since the model is simply a formula that takes hospitalizations experienced as an input and spits out a social distancing level as an output, any rise in hospitalizations will be attributed to a theoretical fall in social distancing levels. So, if hospitalizations go up, the model will say it must be because social distancing dipped; then Colorado’s czar will blame the people and institute another restriction.
Many of Colorado’s coronavirus response shortcomings were displayed in late June and July when, after two months of decreasing hospitalizations, the pace of reduction slowed and then hospitalizations began to increase. At the time, COVID-19 hospitalizations occupied a little over 1% of Colorado’s hospital beds. Colorado’s healthcare system was not under any threat of being overwhelmed; yet the CMG sprang into action and produced a new, official projection predicting the path of hospitalizations.
Figure 1. The projected trajectory of COVID-19 hospitalizations if Colorado social distancing levels remained at levels seen in May after the full state transition to Safer at Home (5/9-5/26, 89% red line), in June (5/27-7/1, 66% blue line), or at present estimated levels (6/21-7/1, 41% yellow line).
True to form, the health department’s graph showed a frightening scenario of an uncontrolled surge in hospitalizations. Also true to form, the prediction turned out to be a massive overestimation and almost immediately the rise slowed and hospitalizations began to fall again.
Figure 2. Graph of the COVID-19 Modeling Group projected trajectory of hospitalizations (from Figure 1) overlayed with the actual number of people hospitalized with COVID-19 from the CDPHE showing real world divergence from the CMG projection.
Also true to form, Polis took the opportunity to issue more mandates, including the reclosing of bars and mandating that all Coloradans over ten years of age wear masks when in indoor public places. Originally, flatten the curve was about avoiding overwhelming hospitals, but by mid-July, a slight uptick from COVID-19 patients occupying 1% of Colorado’s hospital beds drove Polis to increase restrictions. All perspective was lost.
With respect to masks, they went from not even being mentioned as a mitigation tool in early April to being absolutely essential in July. Indeed, Polis has declared, “The modeling shows that if we want to avoid a catastrophic breach of our health care system, and if we want to re-open our economy to a greater degree, then masks are absolutely essential.” Yet the CMGs modeling shows masks have a tiny effect in relation to the effect of social distancing. It is hard to see masks being essential when the entire modeled effect of mandating them can be hidden within the noise associated with social distancing. This could lead one to suspect that masks have become more of a political issue than a public health issue.
Finally with respect to masks, in true Polis fashion, he attempted to sell the masks as good for the economy, citing a study that declared that wearing masks was worth trillions of dollars to the economy, despite the fact that wearing a mask is a constant reminder that disease is just outside of your lips, and but for a piece of cloth, you are dead. A mask is worn out of fear, and fear generally is not economically stimulative.
Another effect of the thinking that there were tens of thousands of potential deaths if the state opened up is that COVID-19 cases became elevated in importance to an irrational level. Remember, flatten the curve was all about keeping hospitals from being overwhelmed. Accordingly, hospitals should have been the driving force in policy. However, under the idea that increases in the number of cases leads to proportional increases in the number of hospitalizations, Polis looked to cases to inform decisions.
However, as the pandemic wore on, the relation between cases and hospitalizations changed. One reason for the phenomenon was that as testing capacity increased beyond the level needed for those with serious symptoms, more and more people with mild or no symptoms were able to be tested. Since Colorado was now detecting more cases that were not in need of hospitalizations, the percentage of positive results needing hospitalizations plummeted. Also, as society opened up, many more relatively young people were catching COVID-19, who also rarely needed hospitalization.
These twin factors of testing healthier people and more younger people catching COVID-19 meant that the relationship between cases and hospitalizations had shifted radically around May and June. Below is a graph of Colorado Department of Public Health and Environment (CDPHE) case and hospital data. The number of cases found March through May coincided with hospitalized numbers in excess of 800 hospitalizations. However, a similar number of cases detected in July and August accompanied about a quarter of the cases seen in March through May.
Figure 3. 3 Day Average of COVID-19 Cases in Colorado by Date Reported to the State
However, this did not stop Polis and the CDPHE from using cases throughout the epidemic as a driver of policy even though after May Colorado’s hospital capacity wasn’t threatened. Throughout the COVID-19 mess, Polis and the CDPHE used a system where any two cases in a facility or non-household group constituted an outbreak. Such a low threshold enabled Polis to rely on a continuous stream of outbreaks to maintain public anxiety. As of August 20, 2020 the number of outbreaks listed by CDPHE was 561. Calling two or more cases an “outbreak” also had the effect of frightening Coloradans and pushing them toward more radical reactions in response. Instead of a calming force, Polis and the CDPHE provided a constant stream of alarmism.
For example in August, Fort Lupton High School closed the entire school for two weeks when two students were tested positive for COVID-19. This while the number of hospital beds in Colorado with COVID-19 patients was less than 2% of Colorado’s capacity and Colorado was seeing two to five COVID-19 deaths per day. Moreover, it was an “outbreak” among high school students for whom catching COVID-19 is less lethal than catching the flu; far less lethal. The risk to high school students of dying from COVID-19 is minuscule, and for teachers under 65, the risk of death is comparable to or less than the risk from catching the flu.
Speaking of schools, despite a continuous stream of hopeful contemplations such as the Trump Administration’s “15 Days to Slow the Spread” and Governor Polis’ own plans for reopening, when President Trump began to talk in July about reopening schools, Jon Samet, the head of the CMG, immediately accused Trump of political motivations. Samet basically accused Trump of using the opening of schools to bolster his reelection chances, writing, “The Trump Administration and others have intertwined school reopening with politics, and the symbolism of a return to signaling ‘normalcy’ in the fall at the time of the November election (enough said).” This was in a letter where Samet himself discussed the benefits of returning to school, and in the same month that the CDC was announcing that the benefits to children of returning to school outweighed any risks. But if Trump says that, well it couldn’t be because he wants kids to go back to school and families to have a more normal life.
Summary
The story of Colorado’s handling of the coronavirus is one of hubris: hubris on the part of Polis acting as protector of Colorado and treating its citizens like reckless children who must be ordered around for their own good; and hubris on the part of the COVID-19 Modeling Group who assumed coronavirus lethality to be far greater than most experts, and never wavered from that belief despite contradictory evidence from Colorado and around the world.
The CMG created an un-falsifiable model that essentially counted hospitalizations and spat out a social distancing value. The value was then treated as inviolable – even if it made no sense. And if the real-world outcomes were not satisfactory, the model’s “scientifically” derived threat of massive fatalities was used to justify emergency orders and/or to terrify and intimidate Coloradans into obedience.
Governor Polis, who is always ready to extol experts who support his world view, went all in and treated the model as gospel. No countervailing information was allowed into his realm. Armed with the righteousness of ostensibly saving tens of thousands of lives, Polis seized power and unilaterally suspended, changed and created laws by Executive Order as he saw fit. The fact that his actions led to hundreds of thousands of Coloradans losing their jobs, a million children booted from schools, and immeasurable suffering foisted on Coloradans, was all justified and sustained by the exaggerated predictions of his anointed soothsayers. His orders were to be obeyed under threat of being ordered to stay at home or of having livelihoods destroyed.
Put simply, if Coloradans did not do as King Polis demanded, he would take away their stuff and put them in time-out.
Karl Dierenbach is an engineer, attorney and writer living in Centennial.
