Covid-19 Forces of disruption Infodemic Op-Ed Pandemic Public Health Trust

Don’t believe the Death toll predictions until we have better Data

[...] don’t trust death toll studies, watch out for bad information overload, don’t trust poor data, in fact—question everything you see, read and hear about Covid-19-- treat coronavirus as unprecedented, look around you and learn, be vigilent if you live in a megacity, no matter where on the globe, and start planning for the next wave.

I usually side with scientists, policy makers and parents. Each role can only be carried out well with trust. Personally, I have missed a steady hand in government in the US, in the UK, in Norway, and in the EU. In the beginning of the coronavirus crisis, I was indeed missing the voice of statistics over semi-educated guesses. But do you know what’s worse than a hunch? Presenting speculative numbers disguised as science is far worse, and it is happening now, affecting millions of lives.

Death toll studies are inaccurate

Instead of clear public health crisis communication from national leaders, over the past few months, collectively, around the world, our elected leaders have given lip service to a small avalanche of well-meaning studies on the death toll of coronavirus, one more wrong than the other. These studies aren’t off by plus or minus five percent, the accepted lower end confidence interval for statistical reasoning. No—they are wrong by a substantial margin, by thousands of lives, by weeks and months, confusing countries and cities around the world.

There’s bad information overload

To make matters worse, corporations (I won’t name names) have put up their own Covid-19 “observatories” running their own independent tally of geographically specific infection rates, death rates, and the like, using a myriad of good and questionable sources, in combination. The result is what I would call bad information overload (think systemic, fake news). The complexity is that it gets rehashed from otherwise reliable sources, such as national media, governments, and trusted, brand name corporations. It renders the WHO’s task of clear, global, public health emergency communication (see WHO’s Covid-19 info page) near impossible to execute. The new challenge is that we are fighting an infodemic from within the ranks, from good people who are smart, mean well, and are in power. This has, arguably, rarely, if ever, been seen before at this magnitude.

We are misusing numbers

At that point, we are not in the realm of scientific use of statistics, we are misusing numbers—and for what? What matters now is clear communication: people will die, stay calm, take care of your kids, don’t visit the elderly, stay home, kickstart medical equipment manufacturing and prepare to volunteer at hospitals. That’s already a mouthful, and people are not getting that either. Complex messages of hope and actionable advice don’t play well in the ether. Death and negativity do. That’s a shame.

We trusted poor data

First, we trusted public sector data from China (take that in) about the disease only attacking the elderly, mostly hitting men and about the mortality rate being low (with some caveat that since Chinese men smoke they die quicker, which may be true, but may or may not impact their Coronavirus mortality at this point). To debunk that quickly: the virus can attack healthy people any age.

As a result, the UK implements a “wait and see”, laissez-faire approach to coronavirus eventually even flirts with the upside of herd immunity without going through the trouble of national vaccinations. The US does nothing, except banning flights from China. Italy gets overwhelmed by sheer surprise. Why Spain also is in trouble, is more complex to explain but they certainly didn’t look up at the right time.

Second, on 17 March, Imperial College issues the Impact of non-pharmaceutical interventions (NPIs) to reduce COVID19 mortality and healthcare demand and arguably comes to an “alarmist’ conclusion: now we needed to worry, because unmitigated and following the “herd immunity approach”, 2.2 million people globally could die after contracting the virus of which 250,000 people in the UK (and 510,000 if no action). This leads the UK and Scandinavian countries (with the exception of Sweden) to implement social distancing, already several weeks too late, as the disease is already raging in Italy. The US does very little and New York, the world’s most important megacity, apart from London, is hit hard.

On March 24, Oxford University’s Evolutionary Ecology of Infectious Disease group, one of the most prestigious in the world, universally admired in the UK, issues a report claiming coronavirus may have already infected half of UK population starting in January, claiming this is a good thing which means the country is beyond the peak and is nearing herd immunity. This spurs a debate of whether the social distancing is too strict. Fact check, there is no data yet on the duration of immunity (if any) after contracting Covid-19. For SARS, a similar coronavirus, one study says it lasted around three years.

But even if there was some sort of immunity by now in the UK population, the role of London as a global megacity would ensure the impact of this choice not to contain the virus (until it was too late) will surely be felt around the world. Just watch the news from Manila, Jakarta, Mexico City and Mumbai in the weeks and months to come (I’ll issue a back-of-the envelope analysis of superspreader risk on a megacity-by-megacity basis soon–the analysis will be transparent and you can change the values of that simple analysis on your own and get to your own result, if you wish).

On March 27, a University of Washington study estimates 81,000 U.S. deaths from coronavirus. This study starts to change the tone in the US, and policy makers now start to parrot more alarmist rhetoric. Correspondingly, on March 29, Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases, appearing on CNN’s “State of the Union”, has the courage to make the estimate that between 100,000 and 200,000 could die in the US from coronavirus, without stating his parameters or reasoning. He does it, not because he is acting as a scientist, but because he feels he has no choice but to scare people, as the public face of the disease in the US, and increasingly, worldwide. He is visibly uncomfortable, doing so, at least, that, thanks for the effort. The earlier, planned messaging has not won the day—with the White House or with the American people. I get the justification. Just know one thing: whether the numbers are in the right range nobody knows, not even Fauci.

Policymakers spread fear through a lack of transparency

On March 31, US President Trump says White House estimates now are between 1.5m and 2.2m deaths in the US “without mitigation” and up to 240,000 even with mitigation. Little is known about how the White House came to those numbers. Dr. Deborah Birx, the Coronavirus Response Coordinator for the White House Coronavirus Task Force, as reported in NBC news, showed a single slide and said they were combining the Imperial College London model with a half-dozen other models from leading epidemiology teams around the world.

This is not encouraging news in terms of credibility. Pooling random models together quickly is a recipe for data error and statistical inference issues. Moreover, the lack of specificity and transparency makes it impossible to fact check. Which studies? What are the assumptions? What is the margin of error? The only ethical rule with numbers is, if you are going to use them, know what you’re doing. If not, stick to what you know. Numbers are powerful and have consequences.

Then, there’s the 2019 The White House study (that’s a year ago) estimating that a pandemic flu could kill up to half a million Americans and inflict as much as $3.8 trillion in damage on the economy. We will leave aside that its suggested remedies, e.g. insurance and marginally quicker vaccines for the pandemic flu, are laughable with what we now know about what it would take to stem a global disease even more deadly than in their scenario. However, if thee projected US death toll in a pandemic was 500,000 a year ago, it is most certainly not less today!

We didn’t question our sources

What is my objection to all these reports (and I’ve only mentioned a few)? First, they cannot all be right—these reports contain wildly different assertions and predictions. As such, estimates, projections, models and guesses that don’t take into account testing of an adequate number of people to have a statistical sample (of the UK, US or any other) population) are merely indicative. They should be used to form a research hypothesis not to inform a city or country, much less to direct world response.

A model is not better than the data it is based on, or at least only slightly better (with advanced data science methods some errors and inaccuracies in data can be corrected for with synthetic data sets, e.g. simulated data, for example). In the US, there is little or no testing data. First there was literally zero, and as of 30 March, one million are tested, as reported by USA Today. One million when the virus has been spreading for months and, should we believe the Oxford study, 50% of the population could be infected or about to? You can’t make data out of that, even with advanced data science.

We still don’t treat this as unprecedented

Whatever the disease mortality rate ultimately will compute to be, it’s clear that no report, no public health official, has a clear idea about very much related to this disease. It is unclear exactly how infectious the disease is. Even top hospitals have seen significant staff infection rates, which is highly unusual. They don’t know what the mortality rate is right now, or what it will be, and how it will differ between countries, megacities, towns, and countryside—nor about the extent to which it will adversely affect healthy middle-aged men (or perhaps not)? I’m waiting for that one. I’m also curious as to when (if at all) my mother can, again, walk around freely on this planet without the fear of infection. We all have questions. But answering questions before we have data is not doing anyone a favor.

Journalists report numbers without context and commentary

Journalistic reporting on these surveys have been so lacking in scientific awareness that it’s almost unbelievable. Where are all the science journalists? Let me answer that question: there is just a few handfuls of college degree or certificate programs focused on science journalism in the whole world. This cannot continue.   

Policymakers struggle with messaging

Now, what about the US President, in 24 hours, going from stating that 100,000 dead “would be a very good job” to expecting up to 240,000 dead, and it “could have been more” the day after. The body bags are piling up in our hospitals and are visible to those who wish to see. What purpose is there in bringing up these numbers at this point?

A human life is a human life, young or old and talking about it in this haphazard way is not just poor statistics but a poor moral choice. If numbers are to be such a thing, it would be safer to state that the US projected death toll is “somewhat likely” to be” in the range” of 50,000-750,000 (or more) depending on many factors we cannot fully control, but we are trying to keep it as low as possible, since every life counts, and here is how you can help. Any more certainty, and you would fail statistics 101.

History will judge all these number-mongers poorly. Why? Their margin of error (rarely mentioned) is likely to go beyond the +/- 50 percentage points range which makes such studies moot. As a comparison, it is worse than if a local TV-station’s election survey consisted in asking a sample of 10 random people on the street “who will win the presidential election” and expected it to be predictive of the end result. You’d be better off rolling the dice.

How can you interpret or even report on an advanced statistical model that has life altering consequences and policy impact upon the world population if you don’t know what you are talking about? Because you can. But why don’t we say something? Because we are paralyzed by fear, which is what governments around the world want us to be right now. It’s quite convenient.

In stark contrast to spending time politically interpreting epidemiological number crunching, the more relevant immediate questions now are:

(a) from psychology: when will social order break down due to isolation so that the consequences of that isolation (suicide and an uptick in domestic and societal violence, or even the fact that our kids are woefully alone in their educational journeys and away from all the daytime activities they know and have ever experienced before) outweigh the public health savings of social distancing and quarantine, and;

(b) from the study of economy: when will our society be brought so far to its knees that if we don’t restart production, we will enter a Great Depression, or worse? A severe depression would also take lives through what we know about systemic shocks, including the consequences of unemployment, mass starvation events among the poor, and a chronic lack of medical assistance. Few other questions matter now.

Let’s create a mass medical technology race to test a meaningful sample of the world population for the disease, and by that I mean hundreds of millions of people globally. Let’s then learn about this disease over the next three months, improving our data and the worldwide response in a more reasonable time frame, and based on established public health practice. The time for playing around with poor data must now be over.  

We are uncomfortable when facts don’t make sense

Why is this back-and-forth with numbers happening? My best take is that nobody has a better alternative and seek comfort in numbers, which seem final, and induce fear, which paralyses critics.

It seems epidemiologists and policy makers alike feel it is better to go with a false confidence in “facts” than admitting that policy responses around the world -including public health advice—is being made up in the fly. There is power in numbers. True numbers don’t lie. Only people do. Right now, there is absolutely no reason to trust death toll predictions, it simply isn’t based on trustable data. It could be far better than predicted and it could also be far worse. We just don’t know.

We already need to plan for the second wave—and next time

The current crisis will now run its course. We have lost control. The only thing that matters is to try to slow the crisis locally, in each megacity of the world, which is where the millions of people live who are the most affected, so we can share medical resources globally over time. Next, we need to avoid that something like this is allowed to run unchecked for so long the next time. For that, we need to be aware of six things:

  • One, pandemic outbreak risk is largely related to population density which fosters a multitude of superspreader events, like mass gatherings, public transportation congestion and a general impossibility of maintaining social distancing. Short of stopping pandemics at the source of the outbreak (best), we need to stop it from reaching big cities with 10 million people or more, or at least, we need to isolate it to as few as possible of those cities.
  • Two, the general quality of the health system predicates health outcomes. Preparation and training help avoiding mistakes and equipment stockpiles help handle surges. As I pointed out in my 2014 op-ed in Fortune, Ebola: The dark side of globalization, we cannot live with failed states. They pose too many risks, to us and to themselves. In fact, according to the Global Health Security Index, there are hundreds of countries that have significant shortcomings in their health security, so this task will take decades to fix.
  • Three, innovation can help mitigate the situation, but it is no cure and even incremental innovation takes time.
  • Four, we can learn a whole lot from a crisis, to better deal with what comes next, but we need to start learning this instant, even if it still hurts.
  • Five, the world’s forty something megacities, those with more than 10 million inhabitants need to be subject to mandatory quarantine immediately. Next time around, whether it is the second wave in the fall, a later mutation or another virus, we need to quarantine all megacities in a sequence that matches the disease picture, but proactively. This staggered world response only creates different “flattened curves” and random spikes and will take years to resolve.
  • Six, there are winners in any scenario. Now’s the time to sit down and consider how you can turn this crisis into something positive—in your personal life, for your family, and for others.

In conclusion, don’t trust death toll studies, watch out for bad information overload, don’t trust poor data, in fact—question everything you see, read and hear about Covid-19– treat coronavirus as unprecedented, look around you and learn, be vigilent if you live in a megacity, no matter where on the globe, and start planning for the next wave.

Lastly, as a parent, be honest with your kids that this is unprecedented. If you hide this event from them, they will find out the true impact some day from somebody else. They should of course not watch the news, in fact it is doubtable if you should even watch the news right now, but they should get to ask questions and you should explain it as it is. We are going down. Full stop. But, as a family, and as the human race, we will be okay, I think, but I cannot prove it to you right now.

Leave a Reply

%d bloggers like this: