It’s time we start asking the hard questions

Image adapted and modified by D&D

Pressing the panic button and sending countries to lockdown, might have been well justified a couple of months back: a hardly understood virus, a WHO’s estimate of a 3.4% fatality rate, and doomsday scenario theoretical models coming out of Imperial College forecasting over half of million deaths in the UK and two million deaths in the US . As the pandemic spread globally, new studies and data have furthered our understanding of the COVID-19 virus, and have challenged and often debunked some of the original assumptions.

Lockdown strategies, under an emblematic ‘Putting people’s health first’, are revealing to be myopic, largely overlooking both the long-term repercussions and indirect consequences of such policies across dimensions like non-COVID-19 mortalities, mental health, economic hardship, quality of end of life, and many many others. Balancing out these different dimensions — surely a very sensitive and delicate topic — has been brushed aside, and the scrutiny and hard questions averted.

Before we start planning for immunity passports, giving away our privacy to contact tracing apps, or committing to year-long lockdown policies, we need to step out of the panic and re-evaluate the situation in light of the new data and learnings, to figure out whether we need to readjust course and whether that panic is still warranted. In times of great challenges, citizens need to unite and commit behind a plan and strategy, however this should by no means lead to an abdication from citizens civic right and duty to debate, scrutinise and ask the hard questions, especially for the experts amongst us. We have not done enough of that, now is a good time to start, it is not too late, and we should not let political correctness get in the way of starting that debate.

Initial Assumptions May Not Hold True Anymore

New information coming to light over the past few weeks is starting to challenge a lot of the assumptions underlying the models and policies put in place. ( The first two points were covered in more depth in my recent article, with references to the studies )

• The population infection rate is much higher than initially thought. Multiple independent small scale studies in Europe and the US, using both conventional and antibody tests converge on this observation. Most cities seem to have an infection rate roughly in the 5–20% range.
• The fatality rate is being revised to well below 0.5 %. This is a direct consequence of the first point. Some studies are even bringing it down to as much as 0.1–0.2%. Surely this is good news and potentially revises the original WHO estimate by an order of magnitude.
• First infections seem to have started much earlier than thought. There has been a widespread assumption that the first assumption coincided with the first confirmed cases. Studies in different countries are showing that the epidemic might have started much earlier. A study coming out of Italy is showing that the epidemic might have started in January in Italy, a different study is pointing to the first cases in Spain dating to February, and a recent article is suggesting that the first infection in California might have happened in January if not earlier.
• More than 50% of the infections occur in nursing homes and hospitals A new study coming out of Italy based on a sample of cases registered in April said 44.1% of infections occurred in nursing homes and another 24.7% spread within families. A further 10.8% of people caught the virus at hospital and 4.2% in the workplace. Notice the high infection rate during lockdown coming from family infections.

The new understanding of the virus that we are gaining through some of the numbers presented above also means that the assumptions based on which models were built need to be updated, and the models themselves revised. Otherwise, these models and their forecasts are nothing but obsolete.

Theoretical models are not accurate

In 2005, The Guardian wrote an article about scientists sounding the alarm about an imminent global Bird Flu pandemic that would cause as much as 150 Million fatalities worldwide, and that the best case scenario would result in 7.4 million casualties. Dr. Neil Ferguson ( the same scientist predicting the half million fatalities in the UK alone if no measures are introduced ), predicted in this same article that the worst case scenario fatality number could go as high as 200 million. How many fatalities happened that year ? Less than a thousand.

This is not to discredit scientists, but just to say that it is very hard, even for the most gifted ones, to build accurate models of complex real-life phenomena. Scientists will often disagree, but from that debate usually emerges a much better understanding of the phenomena we are trying to model.

Regardless of what model ends up being adopted, the model itself needs to be validated and constantly iterated on, by collecting new data, and fine tuning parameters based on the discrepancy of the model forecast and the observations. We have yet to see a revision of models and forecasts on which initial measures have been based.

Lockdown strategy and hospital capacity

The lockdown strategy was never going to be a one shot solution to the epidemic, all the lockdown strategies assume successive on and off phases till herd immunity is achieved either through vaccination or infection. Long lasting complete eradication in a one shot lockdown is not feasible. The primary aim of the lockdown hence, has always been to prevent the overflow of the hospitals, while buying everyone some time to find a solution to the problem.

So where are we in terms of hospital capacity? As part of its COVID-19 outbreak response, The UK decided to build 7 critical care temporary hospitals, called the Nightingale hospitals . The largest one built in London with a capacity around 4000 beds. The London facility has remained largely empty, peaking at less than 2% capacity, as reported by the Telegraph. The article also reports that ‘tens of thousands of NHS hospital beds remain unoccupied amid the coronavirus crisis due to the efforts to free up space. Figures from the national NHS operational dashboard show that 40.9% of the 91,600 general acute beds were unoccupied last weekend — nearly four times the normal amount at this time of year’ . The same situation also happened across different states in the US, as reported in this article.

The situation begs for a couple of questions to be asked: Why was the hospitalisation capacity needed so much overestimated ? Where did we go wrong in the model, and what were the wrong assumptions ? And last, but definitely not least, if there is no danger of the hospital system collapsing, especially given the additional spare capacity provided by the nightingale hospitals, why are we further preventing vital population immunity and prolonging the problem?

Limited scientific proof of the efficiency of the lockdown strategy

Yes this might sound a bit controversial, but do hear me out on this one. Scientifically, an improvement in a measured outcome is not proof that the applied intervention is working. This is why scientific studies and drug approval trials use controlled experiments: in these settings there are two identical groups except that one, the experimental group, receives a treatment while the other, the control group, doesn’t. The control group provides a baseline that lets us see if the treatment has an effect ( check this tutorial if you want to read more about controlled experiments ). The closest there is to using a similar setup to quantify the impact of lockdowns, is to compare countries that have put this measure in place to countries that haven’t e.g. Sweden. Also, any scientist wishing to quantify the impact of an intervention will choose controlled experiments, when possible, over complex theoretical models every single time. The models can then be used to better understand, and attribute the result of the intervention to different factors.

The overwhelming majority of the journalists attempting this comparison, have compared the current fatality rates between Sweden and, for example, the UK to assess which strategy is better, this approach is fundamentally flawed. The comparisons should happen only after the full extent of both strategies have run their course in a year’s time, as mentioned earlier, lockdown strategies involve multiple on off cycles whereas Sweden’s approach does not assume the need for these multiple phases.

Now let’s look at some assumptions when lockdowns were put in place, that have not shown to be true:

• Need for hospital beds far exceeding current capacity if no lockdown. Sweden has a lower number of beds per 1000 inhabitants than the UK ( source) ,2.6 versus 2.8, yet even with no lockdown strategy in place, their healthcare systems did not get overwhelmed. Even in the UK, these limits have not been reached and some newly built nightingale hospitals have even remained unused.
• Catastrophic number of deaths if no lockdown. If we were to do a back of the envelope approximation and take the initial UK government estimates of half a million deaths if not strict lockdown is put in place and translate that to a per capita number. The equivalent for Sweden would be around 76,000 fatalities, yet till now the number has not even reached 3000, and is already decreasing ( 25x less ).
• Achieving herd Immunity would kill a large portion of the population before it is reached. Sweden is currently an important data point that disproves this theory, so far. In a recent interview the Swedish state epidemiologist, Anders Tegnell, says that the numbers show that the amount of people already immune to the virus in Stockholm has exceeded 25%, with a death toll of far lower than the doomsday scenario predicted by the Imperial College publications. The Swedish estimation is that they are weeks away from achieving herd immunity( source ).
• Lockdowns would reduce the effective reproduction number Rt to below one. (for an intro about Rt read this article ). Numerous data points indicate that this number was already in decline well before lockdown measures were put in place. This regularly updated page tracking Rt across different states in the US shows an already decreasing metric well before lockdown measures were put in place — with the rate of decrease hardly affected by the measure. Also, a study by the Robert Koch Institute in Germany, shows an Rt value below one, before the lockdown measures were put in place.

On that last point, epidemiologists seem to be overlooking the poorly understood intrinsic dynamics of flu viruses that create cyclical like phases and end up, like the seasonal flu, self-limiting to a 5–20% infection rate (comparable to the latest data on coronavirus, Cf. article), without having to infect the whole population despite an R0 estimate that is well above 1.

Hybrid vs Homogeneous Policies

Policy makers have mostly considered homogeneous policies applied across the population. Hybrid approaches, i.e. applying different measures for different segments of the populations (based on their vulnerability profile) have not been considered in depth, despite the many advantages that they can offer. A study coming out of CMU and University of Pittsburgh details the merits of such an approach. Furthermore having a very accurate model to determine the population at risk ( 1.5 Million people in the UK have already been identified as such ) and making sure that they are shielded from infection, would allow for the largest chunk of the population to carry on an, as normal as possible, life without having to stop the economy, while building herd immunity. Why have these scenarios not been widely modeled, and such policies largely overlooked ?

We need to go beyond epidemiology to decide on the right course of action

Lockdowns have indirect repercussions and long term consequences that go beyond the averted COVID-19 fatalities. These consequences span multi dimensions: economical, lifestyle, health and mental health considerations, etc… . Deciding how to weigh these different dimensions when choosing a course of action requires asking some hard questions, which I present a few of in the next section. For now let’s just focus on the obvious dimension: population mortality.

Decreased population mortality estimated from lockdown measures put in place, will be counterbalanced by long term consequences that are yet to unravel and that scientists and journalists are already warning about:

• Access to healthcare: averted access to medical attention, as the system focuses on the COVID-19 patients. Care homes have already reported seeing a rise in deaths not linked to coronavirus as hospitals refuse to take ill residents. A recent article by a medical doctor in NYC pointed out that the number of ambulance runs decreased by almost 50% since the lockdown was put in place, and that a growing number of people dying at home from heart conditions, strokes and bacterial infections. In the UK, the number of patients attending A&E fell from 477,000 this time last year to 221,000 in the last week ( article ).
• Economic hardship: a scientific study concluded that the 2008 crisis increased cancer mortalities by 260k+ deaths in OECD countries alone, the number is estimated to be more than 500k+ worldwide.
• Increase in domestic violence victims
• Increase in suicides. Areas in the US are already seeing an increase in suicide related helpline calls ( article )

One should also note that lockdowns can also have an indirect positive impact on lowering mortalities linked to air pollution, traffic, hazard at work, dangerous sports, etc…

Though, the analysis on mortality I have done here is by no means exhaustive, it does show that such decisions require experts in the room that span multiple disciplines, and whose inputs are as valuable as the epidemiologist’s. I am not arguing here that adding up the numbers will tip the balance for one side or another, I am just advocating that no sound decision on the course of action can be taken without undertaking a complete analysis of the direct and indirect consequences of these policies; and I have still to see one coming from any government who has implemented a lockdown strategy.

Furthermore, the impact of such measures, if no full scope assessment as I am suggesting in this section is done, can even be more extreme in developing countries. A very interesting study coming out of India shows the crash in the rural provision of health services due to the COVID-19 measures suggesting ‘that a serious public health crisis is already brewing, with the potential to erase gains made against a number of diseases over decades.’ and whose negative impact would far outweigh the COVID-19 projected fatality numbers.

Ethical dilemmas and hard questions

As I have touched on in the previous section, the outcomes of any such policies and its impact on society is heavy and multidimensional. Weighing outcomes across different dimensions requires us to ask some hard questions, which seems to me that as a society we have been so far avoided doing.

• Fatalities vs Long Term traumas. For people with dementia, the pandemic is an absolute nightmare, creating long term irreversible damage to the mental health and health of the elderly population ( Economist article ) . For the younger population, the situation is as dangerous, COVID-19 lockdown measures and economic hardship could cause a parallel child abuse epidemic ( article ). How are we balancing one against the other ?
• Quality of life versus life expectancy. What is life ? Is a beating heart, the life we are trying to preserve ? In his excellent book, Being Mortal, the best selling author and doctor Atwul Gwande says : ‘Our ultimate goal, after all, is not a good death but a good life to the very end.’ emphasising that for people in older age, quality of life trumps sheer longevity. If quality of life becomes our primordial concern, then, suddenly, it’s not just the quality of life of the vulnerable people that should be factored in in any decision, but that of the whole population.
• On cognitive dissonance and preventable deaths. In the UK alone, every year, smoking causes 78,000 deaths and alcohol adds another 7,500 fatalities. In the US the numbers are 480’000 for smoking, and another 88,000 for alcohol-related illnesses. Why haven’t governments put a ban on alcohol and cigarettes if every life is precious — the lives saved could be comparable to the projected number of COVID-19 fatalities when lockdowns are in place, yet at a minute fraction of the economical cost. Yes, every life is precious, but governments claiming that we should do all that is within our capabilities to save every person is a sign of cognitive dissonance. So, what is the ‘threshold’ number of fatalities, or the criteria that make governments willing to bring a whole economy to halt ?
• Economic cost of each quality adjusted life year. Economists have a measure know as Quality-Adjusted Life-Year (QALY), which captures both the quality and quantity of life, and is used in assessing the value of medical interventions ( for more info, check the wikipedia definition ). In the UK, the National Institute of Health and Care Excellence, NICE, places a value of up to £30,000 per QALY ( source ) when measuring value for money in relation to public health interventions. Put in other terms, NICE estimates that we should not be spending more than £30,000 to add one year of perfect health to one person. A recent article, presented an in depth analysis of the situation in the UK: The government has decided to invest as much as £500,000 per life saved, which estimates that they are estimating adding another 16.5 years of quality life for each life saved ( the average of someone dying from COVID-19 is 79.5, and UK life expectancy is 81 ). There again seems to be cognitive dissonance when it comes to dealing COVID-19. Why are we valuing the QALY for COVID-19 much higher that other conditions ? Or put differently, why is government willing to pay more to extend the life of a COVID-19 patient as compared to patients with other conditions ?
• People having a say. What mental health cost are people vulnerable to COVID-19 willing to pay, themselves just to avoid death ? I am particularly speaking about the older populations, grand-parents over 80 and 90 years old. As this opinion piece in the telegraph points out: If you are, say, 85 years old and you have a choice between possibly catching the virus or living out what might be your final year of life without ever seeing your grandchildren again, which would you go for? I would surely opt for seeing my grand kids, as would the author of the article. If a large fraction of the vulnerable population would make the same choice, is the lockdown policy still as justified ?

These are all tough points and hard questions, which don’t necessarily have clear cut answers, but that is no reason to overlook them. We should be able to have a debate about them, far from panic, and the society needs to be able to have a say. Furthermore, as pointed out by the Nuffield council on bioethics : in a time where we are in it together, we shouldn’t lose sight, nor have to remind governments of the basic principles of good democratic governance, especially pertaining to transparency and engagement in decision-making.

The panic needs to stop

Panic and anxiety at every level in our society needs to stop — not only for the sake of the overall mental health of a country but equally so that clarity in decision making and thinking subdue this highjack of our cognitive capacities. Population wide anxiety directly impacts every component of the society: poor decision making by policy makers, hoarding behaviour in supermarkets leading to shortages, increased pressure every single day for health professionals leading to overly aggressive use of ventilators , and lower immunity in both patients and non-patients, which increases fatality risk.

Pseudo-scientific, and data-naive journalism has majorly contributed to this general level of anxiety through its sensationalist headlines based on a very liberal interpretation of the data.The scope and impact of this reporting is by no means less harmful than the campaigns of fake news and misinformation that western governments have been trying to fight only a couple of months back. I have published earlier an article about scientifically sound basic guidelines to interpret ( and report ) COVID-19 data, that I invite journalists and news reporters to check before writing their next article.

Where this leaves us

In these highly uncertain times, we should all stand united. United in supporting each other, united in keeping each other safe, but also united in our quest for truth. Choosing the scientific way of addressing the pandemic does not mean putting scientists on a pedestal. The scientific way is an inquisitive, iterative, constant learning approach, where no hypothesis is sacred and no assumption is unquestionable. Science is inclusive, not exclusive. There has been so far enough data and experiments run around the globe to warrant that we start asking some hard, albeit uneasy, questions to seek answers and re-evaluate our policies, if needed. The strategy that we are adopting needs to be as multi-dimensional as are the threats posed by this pandemic, and should not discard mental health, wellbeing, quality of life and long term ramifications of anxiety and economic hardship. Keep calm and stay safe.

Follow me on @DonRiachi