We R0 New York City

Originally, it was not my intention to write a position paper on epidemiological models. It was my intention to answer a couple of quick questions and concerns posed by my Clients, and colleagues. However, the more I explained, the more involved the answer became.

So, to be brief, I'm going to focus just on SARS-CoV-2, and the work by the Imperial College (London, UK) COVID-19 Response Team (ICLCRT). Everything I say here should be considered within that context.[1] 

To begin with, many epidemiologists and other scientists have been very critical about the model developed by the ICLCRT and upon which many countries based their response decisions (including the US). I have been very critical of the ICLCRT, but NOT because of the flaws in the model.

In truth, from the perspective of a modeler, the ICLCRT model was a thing of beauty, a work of art and really pretty cool and there was nothing wrong with the model, per se. As I have said before "All models are wrong, and some models are useful." In one sense, models can never be flawed in the same way a mathematical equation is flawed; only the interpretation of models or their applications can be flawed. That is because models are toys to be played with, and compare the outcomes of imaginary "what if" scenarios. As such, one has to be careful about placing an appropriate amount of faith in a particular model.

 Some models, such as those that simulate the motion of planets around the sun, are very stable and can be used to make confident predictions when imputing unknowns (such as predicting the path of a satellite when one varies the velocity or the mass of the satellite).

 In the case of the ICLCRT, there was an inappropriate intersection between good science and public policy, and the resulting collision of those two fields resulted in a wreck of both.

 To be succinct - models are toys. Models are war games, and video games developed and played by mischievous practitioners who should begin the development of their model by asking the "what if" question. "What if the population density is X instead of Y?" What if the case mortality rate is X instead of Y?" What if I build a wall along the Mississippi and prevent all travel west of that wall?" This is good science.

 Bad science is when the modeler starts with: "I have a pet theory and I'm going to develop a model that proves my theory." This is what happens in the Global Warming community and this is why the models are laughable and not just bad science, but pathological science. 

 Where the ICLCRT went wrong wasn't with regard to the technical aspects of their model, it was their inappropriate dabbling in public policy and the misrepresentation of the confidence in the interpretation of the validity and rigor of their model, and their presentation of the various scenarios their model played out.

 It is true that, in accordance with good modeling, the ICLCRT provided an honest evaluation of the uncertainties of their model in their discussion of the model. But then, almost inexplicably, they ignored their own precautions and ended up providing public policy "recommendations" as though their predictions had absolute certainty, and even ignoring their own concerns that the two modes of addressing the pandemic would possibly have catastrophic effects on societies, and may not even be possible to implement in free societies.

 TERMS

So first a couple terms need to be presented:

 Case - a "case" is merely an individual who is infected. It doesn't mean the person is sick, or will become sick, or how sick, or a statement of resolution or death or prognosis. If one is infected, one is a "case." 

 Incidence rate - the absolute number of cases, per population, per specified time frame. Since there is a slope associated with the parameter, it is a "rate." 

 Prevalence - This is the absolute number of cases after a specified time. This is not a rate, it is an absolute value. The ICLCRT did not address "length bias" which is a distinction between "incident cases" (those cases discovered through diagnosis) and "prevalent cases" (those cases discovered through screening).

 R0 - the value known as "R naught" is the "basic reproduction number." It is the number of cases caused by the transmission of the virus by one infected person. Thus, an R0 of "3" means that one person gives the infection to three people, and each of those three give the infection to three people, and those three give it to three each, and so forth.   Where R0 is less than one, the transmission of the infection regresses.

 In their Fifth Report (which was really cool), the ICLCRT established a probable R0 of 2.15 (95%CI: 1.79-2.75). In their March 26, 2020 report, the ICLCRT revised its March 16 estimate of the R0 upwards to between 2.4 and 3.3; and in a March 30 report on the spread of the virus in 11 European countries, the researchers put it somewhere in the range of 3 to 4.7. I have recently seen a stupid meme on the internet suggesting that the R0 is 26. Actually, I wish the R0 was discovered to be 26, since that would imply a much greater prevalence and therefore a much lower case mortality rate.    

 MOI - the multiplicity of infection. This is the ratio of infectious particles available to infect susceptible (and permissive) cells. The MOI is independent of (but associated with) the infectious particle density (which, in absolute terms is more important than the MOI).

 ASSUMPTIONS

All models begin with asking an appropriate question and then filling in the unknowns with assumptions. Sometimes those assumptions are known to be wildly off-base, but if you don't have any inputs, you have to start somewhere.

 The ICLCRT did not reinvent the wheel, instead they used known models from 2005 and other modifications for influenza virus transmission (which is similar in virtually all respects to the transmission of the SARS-CoV-2 virus). In fact, an earlier version of the Imperial model, estimated that SARS-CoV-2 would be about as severe as influenza in necessitating the hospitalization of those infected.

The ICLCRT were honest and acknowledge the limitations of their model:

"With currently available data, precise estimates are not possible and different methods do not give concordant results, but very wide confidence intervals cover a realistic range of values."

This is about as honest as you can get, and the layman's translation is: "Our models don't agree with other models and even our own model doesn't agree with itself, but since nobody knows the real answer, we think this is pretty good stuff."

OK - but this is not the sort of confidence I would want if I'm proposing that a government take actions by implementing illegal civil restrictions on a nation of 327 million people (the US) or 65 million (the UK), or change a behemoth economy. 

In various other places in the report, the ICLCRT appropriately identified the serious limitations with their assumptions and their model.

TARGET POPULATIONS

The ICLCRT focused primarily on the US, and the UK, and admitted that the US was much harder to model because the US is not an homogenous society but an hodge-podge of different peoples, different cultures, different climates, different eco-systems, different geographies, and so forth.

Herein lies the first two problems. The initial data inputs for the model were based on observations from China. China is not the UK. China is not the US. China is as different from the US as chalk is from cheese. Similarly, New York City, USA is not Northumberland, England, and Northumberland, England is not Bill, Wyoming. But in the ICLCRT model, the ICLCRT presumed (in one part of their model), that Wuhan, China and Bill, Wyoming were the same (in other model scenarios, the ICLCRT played with breaking out the various US States to see what those scenarios looked like - see below).  

Next is my use of the word "behemoth." The US economy is a behemoth - a huge, living, dynamic monstrous creature. Change one tiny thing in the chaotic network of the US economy and the butterfly effect may throw a million people out of work for two years.  

By comparison, make even a HUGE mistake in a model parameter or input, and an error is noticeable in a millisecond and correctable in under 30 seconds, and no one gets hurt.

Therefore, making huge policy decisions regarding the US economy, with KNOWN deleterious effects based on whimsical model parameters was not only unwise, but the modelers who made the suggestions left the realm of good science, and entered the center-stage of bad decisions.

When the State of Colorado made similar models, they left the model to speak for itself. To their credit, they allowed the model, with all its errors show the outputs without any attempt to validate the outputs. That's what good modelers do.

But it gets worse. The ICLCRT modelers made assumptions they knew were not realistic, and at the beginning of their discussion they even described those unrealistic parameters:

"There are a number of limitations to this analysis. The exponential growth model does not account for a reduction in transmission due to public health interventions, such as travel bans, and quarantine measures. These estimates do not reflect the situation in areas under quarantine and are more strongly influenced by epidemic dynamics near the time of origin. The model does not account for geographic structure and this analysis included many samples from outside of Wuhan City and outside of mainland China."

But by the end of their report, they failed to reinforce those limitations, and used language that implied greater confidence in their model than was warranted by the model.

But it gets worse still because the ICLCRT model presumed that no intervention would occur, no public health response would be taken; there would be no governmental or spontaneous public response. That is, the people and governments of the world would behave like slow walking zombies, and stumble unabated into disaster zones without the slightest cognizance or awareness that anything was wrong. The ICLCRT knew this was an unrealistic parameter and they said so in their discussion.

More troubling, they considered the potentially disastrous effects their recommendations would have on civil rights, and they even stated that it is unlikely the US population would allow such Communist Iron-Fist dictates to be imposed on them. The ICLCRT brushes those concerns aside and state:

"We do not consider the ethical or economic implications of either strategy here, except to note that there is no easy policy decision to be made. Suppression, while successful to date in China and South Korea, carries with it enormous social and economic costs which may themselves have significant impact on health and well-being in the short and longer-term. Mitigation will never be able to completely protect those at risk from severe disease or death and the resulting mortality may therefore still be high. Instead we focus on feasibility, with a specific focus on what the likely healthcare system impact of the two approaches would be. …"

HAMMERS and NAILS

If you are a hammer, everything is a nail. 

 If one is a cladist modeling the evolutionary pathway between three animals, one will always end up with a "tree." That is because the model only "knows" how to draw trees, and never asks if a tree is the appropriate association. Thus if one inputs the homologies between a table, a car and a dog, the model will draw an evolutionary tree showing how the one evolved from the other. 

 Such is the case with the ICLCRT model in that the developers specifically designed the model to represent a very pessimistic scenario - NOT a realistic, likely scenario. There is nothing wrong with that, per se, in fact, it's one of the reasons we use models. With a model, we can safely kill millions of people and nobody gets hurt, and then re-run the model again under different conditions and see if we get different results.

 However, the modelers considered exclusively two possibilities: (A) Mitigation: this approach tries to slow (but not stop) the spread of the virus over time; that is "flattening the curve", - the same number of illness and deaths will be tolerated, but spread out over a longer period of time. (This is the issue I discussed in another article, click here.) This is the myth (now, more of an intentional misrepresentation) of the "Stay-at-home-save-a-life" nonsense.

 The other, more Draconian strategy is "suppression," which aims to reverse epidemic growth, reducing case numbers to low levels and maintaining that situation indefinitely. In my opinion, "suppression" may be marginally successful in extremely restrictive environments (on board military ships at sea, high security prisons, Military Dictatorships where you can just shoot and kill anyone who violates your quarantine, etc.), but would otherwise be repugnant to free societies who value civil liberties.

 The ICLCRT model only proximally addressed the MOI or the density of infectious particles in a crowd during the transmission of the disease when they put all that under the R0. It is for this reason that the models entirely failed to predict realistic transmission virtually everywhere. Because the model presumed that the R0 in Wuhan, China would be the same as the R0 in Bill, Wyoming. (The R0 in Wuhan, China may very well be the same as the R0 in New York City - the two societies are similar in many respects and both even have Communist governors).

 The one-size-fits-all ICLCRT model translated to a "this-size-fits-nobody."

RECOMMENDATIONS

The model was then over sold using language that was NOT supported by the model, or even the acknowledged limitations of the model:

"We therefore conclude that epidemic suppression is the only viable strategy at the current time. The social and economic effects of the measures which are needed to achieve this policy goal will be profound. Many countries have adopted such measures already, but even those countries at an earlier stage of their epidemic (such as the UK) will need to do so imminently."

To be VERY clear - there is NOTHING, ABSOLUTELY NOTHING in the ICLCRT model or discussion that would lead one to "…conclude that epidemic suppression is the only viable strategy at the current time. " This is a conclusion that an intellectually honest person cannot possibly find in the ICLCRT model or its parameters, or even in its assumptions.  

In fact, the ICLCRT were neither the only players at the table, nor necessarily the best. Utilizing a stochastic transmission model that was custom fit to the on-going COVID-19 outbreak, Hellewell and Abbot concluded that: 

“…highly effective contact tracing and CASE isolation is enough to control a new outbreak of COVID-19 within 3 months.”

We knew in March of 2020 that the conclusions by the ICLCRT were NOT true. Sweden and Japan and other countries were wise enough to consider alternatives and not read the ICLCRT report as the only possible path. I'm confident they understood that the ICLCRT had gone way out on an unsupportable limb and had absolutely no basis for confidence in the model (as admitted by the ICLCRT itself).

Several people ridiculed me when, at the beginning of this event, I proposed intentionally exposing our "front-liners" (law enforcement, health care, EMS, Fire) to the virus in a controlled fashion.  In fact, Government officials had previously considered allowing the disease to spread while protecting the oldest in society, because large numbers of infected people would recover and provide herd immunity for the rest of the population. At that time, the opposing advisory teams had discussed suppressing the pandemic by social distancing, but at that time, officials were worried that this would only lead to a bigger second outbreak later in the year (and they were right). US and the UK are now worried about this second wave whereas Sweden and Japan are not. (Obviously long term consequences in Japan remain more uncertain than, say, Sweden).

This is all the more confusing since even ICLCRT member Neil Ferguson stated his hope that countries would follow the example of South Korea, which has managed to impose a less rigid version of social distancing and by conducting high volume testing and tracing the contacts of those infected.

CONCLUSIONS

So - here is what we know (and it is not much different than what we knew at the beginning of February 2020).

1) There is no science now, and there never was any science to support the current fad of social distancing as it is being practiced.

 2) There is not now, and never was ANY valid medical or scientific justification for "stay-at-home" orders implemented by national or local governments.

 3) The deployment of "social distancing" as practiced in the UK and the US (and other countries such as Ireland), have had exclusively detrimental effects on those societies without any known or demonstrable benefit in controlling the spread of SARS-CoV-2.

 4) There is not now, and there never was any valid scientific justification to support the current fad of general "face mask" usage.

 The concept of The Tenth Man, again has raised its ugly head (click here.) Well respected scientists and medical practitioners with opposing opinions have been shouted down by the pack of hounds baying on a false trail, and ignored by the media because they would not provide a ripping good horror story of doom and gloom. 

 And now, as a result, we have healthy people running around like idiots wearing cloth bandanas around their faces falsely thinking they are being protected (or even more ludicrous, believing they are protecting others), we have the quarantine of healthy people, and collapsing economies. (I have addressed the respirator issue elsewhere - click here.)

 All of these myths and misconceptions based on a model, a scientific toy, a numbers game, and the misrepresentation of the same.

(A PDF version of this discussion can be found here: https://www.forensic-applications.com/misc/We_R0_New_York_City.pdf I reserve the right to make corrections and edits to this discussion without notice. CP Connell)

REFERENCES

[1] Neil M Ferguson, Daniel Laydon, Gemma Nedjati-Gilani et al. Impact of non-pharmaceutical interventions (NPIs) to reduce COVID-19 mortality and healthcare demand. Imperial College London (16-03-2020), doi:https://doi.org/10.25561/77482

Other COVID-19 discussions by CP Connell:

 How to Peddle Backward - What happened to the 2020 Flu Epidemic? A summary of the US Crude Mortality Rate's refusal to cooperate with the popular narrative.

 WHO thought this was a good idea... (Comments regarding the December 1, 2020, "Mask use in the context of COVID-19".)

 The Failing Mask Cure Aid a review of Bundgaard H, Bundgaard JS, Raaschou-Pedersen DET, et al, "Effectiveness of Adding a Mask Recommendation to Other Public Health Measures to Prevent SARS-CoV-2 Infection in Danish Mask Wearers, A Randomized Controlled Trial" (Ann. Int. Med. Nov 18, 2020, https://doi dot org/10.7326/M20-6817).

  Don't be a Maskhole, Karen A review of Zeng N, Li Z, Ng S, Chen D, Zhou H, Epidemiology reveals mask wearing by the public is crucial for COVID-19 control. (Medicine in Microecology, https://doi.org/10.1016/j.medmic.2020.100015):  

 Masks, and the new Doctor Schnabel von Rom: Review of Stadnytskyi V, Bax CE, Bax A, Anfinru P, The airborne lifetime of small speech droplets and their potential importance in SARS-CoV-2 transmission (Approved by PNAS May 2020: https://www.pnas.org/cgi/doi/10.1073/pnas.2006874117)

Pathological Science - Zhang et al and the PNAS: Zhang R, Annie Y Zhang L, Wang Y, Molinae M: Identifying airborne transmission as the dominant route for the spread of COVID-19 (fast-tracked through the PNAS on June 11, 2020)

Defacing Mask Science - Rossettie S, Perry C, Pourghaed M, Zumwalt M, "Effectiveness of manufactured surgical masks, respirators, and home-made masks in prevention of respiratory infection due to airborne microorganisms" The Southwest Respiratory and Critical Care Chronicles 2020;8(34):11–26

Masks - Don't look behind the curtain: Review of Vivek Kumar, Sravankumar Nallamothu, Sourabh Shrivastava, Harshrajsinh Jadeja, Pravin Nakod, Prem Andrade, Pankaj Doshi, Guruswamy Kumaraswamy "On the utility of cloth facemasks for controlling ejecta during respiratory events "

 Size matters! A Brief Description of filtering mechanisms and size.

Materials v. Masks: A review of Konda A, Prakash A, Moss GA, Schmoldt M, Grant GD, Guha S "Aerosol Filtration Efficiency of Common Fabrics Used in Respiratory Cloth Masks" (American Chemical Society, April 2020)

"Junk Science: In Favor of Community Face Masks - a return to Lysenkoism" A review of: Jeremy Howard, Austin Huang, Zhiyuan Li, Zeynep Tufekci, Vladimir Zdimal, Helene-Mari van der Westhuizen, Arne von Delft, Amy Price, Lex Fridman, Lei-Han Tang, Viola Tang, Gregory L. Watson, Christina E. Bax, Reshama Shaikh, Frederik Questier, Danny Hernandez, Larry F. Chu, Christina M. Ramirez, Anne W. Rimoin Face Masks Against COVID-19: An Evidence Review NOT PEER-REVIEWED | Posted: 13 May 2020

Wishful Science - A review of Anna Davies, BSc, Katy-Anne Thompson, BSc, Karthika Giri, BSc, George Kafatos, MSc, Jimmy Walker, PhD, and Allan Bennett, MSc Testing the Efficacy of Homemade Masks: Would They Protect in an Influenza Pandemic? (Disaster Med Public Health Preparedness. 2013;7:413-418)

If Manikins Could Fly… A Review of Eikenberry SE, Mancuso M, Iboi E, Phan T, Eikenberry K, Kuang Y, Kostelich E, Gumel AB "To mask or not to mask: Modeling the potential for face mask use by the general public to curtail the COVID-19 pandemic" (Infectious Disease Modelling 5 (2020) pp. 293-308)

Review of Cheng VC, Wong S, Chuang V, So S, et al "The role of community-wide wearing of face mask for control of coronavirus disease 2019 (COVID-19) epidemic due to SARS-CoV-2" (Journal of Infection April 30, 2020;16:13)

Gassed Masks! Reactivation of viruses and deoxygenation during mask wearing.

Masking the Truth - A discussion of aerosols and droplets

We R0 New York City - A discussion of the basic reproduction number.

The epidemic of ignorance: Lessons from "Flattening the Curve" April 14, 2020

Think Tanks! The Dangers of Group-Think April 13, 2020