Tuesday, 22 December 2020

Two Failures Of Covid Science - And How To Do Better Next Time

While there have been obvious achievements by Covid science these should not obscure the very significant failures that have also occurred, such as around the politicisation of scientific advice and the delay in rolling out vaccine programmes. These failures may have allowed hundreds of thousands of avoidable deaths, as well as extending and worsening the social and economic impact of the epidemic on billions of people. Nevertheless, the point of identifying such failures is not to allocate blame, but rather to plan for how to do things better in future. 

I. Politicised Advice

The science of this novel corona virus - the facts about how it works and what works against it - were in a state of flux for a great part of 2020. However, grappling to understand a new phenomenon by gathering evidence and using it to identify promising theories and using those theories to ask more specific empirical questions and so on is just science working at its best as a discipline of dialectical empirical inquiry, albeit at a highly accelerated pace. Even though some interim results of those empirical investigations turned out to be significantly mistaken, that is not a failure of Covid science because it is not the kind of error we should seek to avoid. 

The failure lies in the way the shifting state of scientific knowledge of Covid was communicated to the general public and governments. Scientific advisers frequently expressed opinions with great certainty - such as about the (in)effectiveness of masks or the risks of mass-transit or in-person schooling - which varied greatly from one moment or country to the next. (E.g. the British virus can travel 50cm further than in mainland Europe, somehow.) The reasonable inference from such mutually contradictory claims is that many of these scientific advisers were not properly and competently representing the state of scientific knowledge. A further reasonable inference is that, if scientific advisers had done their basic job better, governments would have been more likely to adopt more effective policies and the general public would have been more likely to trust the underlying science and therefore comply with them. Instead publics around the world increasingly perceived Covid science as inherently politicised and directed at political objectives such as controlling our behaviour, rather than as an independent global inquiry concerned with understanding the facts. 

For example, on the masks issue, scientific advisers in some countries seem to have decided that because stocks of personal protective equipment were limited and therefore best allocated to healthcare workers, it was part of their job to reduce public demand by telling a misleading story about how masks wouldn't be of any use to them anyway. Thus scientists improperly took on political tasks and in doing so undermined their independence from the political sphere and hence the credibility of their future scientific pronouncements, since that depends the idea that facts are facts regardless of how convenient or inconvenient they might happen to be. This collapse in credibility was apparent when they switched tack and started advising everyone to wear the same masks they had said were useless.

For another example, in many countries there was no fully worked out plan for which scientists' advice should be heeded in this kind of emergency (exceptions like S. Korea and Taiwan prove the rule). As a consequence we saw scientists jockeying for influence on government policy. Scientists are prone to self-aggrandisement like anyone else: they think they understand things best and should get to decide things. There is a self-selection effect here, since those scientists who are most drawn to power are the ones who seek out positions as advisers in the first place, and will be most alert to opportunities to extend their influence. Thus, in a crisis governments are faced with numerous apparently qualified scientists offering advice, from which they can choose the analysis and corresponding policy menu most congenial to them. This appears to have happened in the UK, where Johnson's government postponed introducing a shutdown while it dallied with pseudo-scientific theories of herd immunity and behavioural fatigue. The outcome is that those scientists most concerned to get close to power will be the most willing to say the kind of thing that politicians would like to hear, and will also be least likely to risk their relationship to power by publicly correcting those politicians. Thus is science corrupted by politics - and often without politicians even intending it!

Of course, the problem of the politicisation of Covid science is not only due to the failings of scientists. Many governments exerted themselves to bend Covid science to their political interests and will. China treated Covid and the facts about it as a state secret (and later also as state propaganda), locking up scientists who tried to share vital information with global science networks. In Russia, numerous scientists working on Covid have developed 'Falling-out-of-a-high-window' syndrome. Nor is this limited to traditionally authoritarian regimes. In America, Trump attempted to promote an alt-science of Covid that supported his re-election campaign, managing in the process to collapse the reputations and functioning of institutions like the CDC. In such conditions scientific advisers may believe that their morally best course of action is to go along with the political corruption of the facts (nodding along when the president muses about higher rates of testing causing higher positivity rates, etc) because only by doing so can they remain close enough to power to influence policy for the better. And they may worry that if they get fired someone even more suborned by political interests will take their place.

Preventing such failings in future requires better attending to the institutional fences and bridges between science and politics. On the one hand there should be fences that protect both science and politics from depredations by self-interested actors from the other side. In democracies, scientific institutions (like the CDC) should be more systematically protected from political interference so that they are not vulnerable when a crisis occurs. Scientific advisers should more clearly abjure political tasks and they should be accountable to and preferably selected by the relevant scientific communities, not politicians. (Elsewhere I have written further about the relationship between truth and democracy.) 

In non-democracies independent high quality science is at much greater risk. There the best we can do is to support international scientific institutions like the WHO. If properly funded and defended from political interference, such international institutions can play a very important role in setting benchmarks for best scientific practise, sharing the latest credible research findings, and managing the development and fair global distribution of tests, treatments, and vaccines (thereby including poorer countries otherwise left out). However, before Covid the WHO's budget was a pathetically small $4.4 billion (and 10% of that came from one private donor - the Gates Foundation). This left it highly vulnerable to the (geo)politics of the pandemic when it could have been a counterweight to the politicisation of Covid science at the national level. For example, the WHO had to pretend to believe everything the Chinese government told it in order to have access to the crucial data it controlled, but doing so undermined its credibility with everyone else.

II. Vaccine Delays

Recent technological improvements in vaccine science have been extraordinary. It is astonishing to consider that the Moderna vaccine just granted emergency use authorisation by the FDA was designed back in January, more than a week before America's first confirmed Covid case and a month before the first American death (Timeline). The US government prepaid to start mass-production before trials were even complete. But then it lingered in the (accelerated) vaccine development pipeline for 11 months while millions of people were catching and dying of the disease around the world and many billions more were suffering through the economic and social catastrophes of shutdowns. (See further David Wallace-Wells

Imagine the suffering that could have been averted if this vaccine had been made available earlier in the process, before all trials were complete. For example, even if production lines were only operating at a trickle of their full capacity, that might still have allowed the most vulnerable groups to be protected months ago - the over 85 year olds, for example, who make up only 1% of the population but 1/3 of Covid deaths (in rich countries). Why wasn't that possible?

The answer seems to be - in part - a quirk of medical ethics, a field of applied ethics that has been institutionalised within Western medical science. (It is probably no coincidence that two defiantly anti-Western states lacking such constraints, Russia and China, started vaccinating people months ago)  Two standard ethical distinctions seem particularly relevant to the Covid vaccine delay. 

  1. Actions are distinguished from inactions and considered especially morally laden. Hence, it is generally considered morally worse to be the cause of a harm, than to sit on your hands while something much more terrible happens.

  2. Patients in need of treatment for some condition are distinguished from non-patients. Justifying allowing a treatment for patients requires showing that it will achieve the most net good relative to other options, even if it also creates harms of its own (such as chemotherapy). However, since non-patients do not have a medical problem (yet!) there are no benefits to set against the potential harms of a vaccine and so preventative treatments face a very high bar to . 

Both of these are generally quite reasonable. They combine to put a safety brake on the runaway utilitarianism (and arrogance) that doctors and researcher are otherwise prone to, of the kind that would rationalise killing healthy people to use their organs to save more lives. They can even be given their own 'rule utilitarian' defence in terms of supporting the long-term credibility of medicine and hence the total number of lives it can save over time. Nevertheless, when applied to the development of vaccines against a deadly epidemic, their built in fixation with risk aversion resulted in mass death. 

Firstly, it is one thing to reject an entirely utilitarian moral analysis; it is quite another to blind oneself completely to the relative costs and benefits of different approaches. If your medicine leads to the death of one person who had a severe reaction to it but saves millions of other people from dying that is just morally better (much much better!) than the opposite. Any moral rule which doesn't allow you to see that is an absurd rule to be following. Putting a brake on excessive utilitarianism is surely one goal of medical ethics, but not the only or even primary goal! (And if the reason you are attracted to such a rule is its emphasis on what consequences are and aren't your fault, then you are guilty of moral cowardice and should remove yourself from the business of (telling other people how to make) life and death decisions.)

Secondly, the patient-nonpatient distinction already fails to make sense in many medical contexts and has been challenged for decades (e.g. by Geoffrey Rose). Populations at high risk of disease (such as heart disease or malaria) should be considered 'almost patients' and offered preventative treatment on the same principle that people in cars should all be offered seatbelts and airbags. In future this 'almost patient' status should certainly be extended to populations ravaged by epidemic.

There was an evident absurdity in the same medical ethics bureaucracy allowing Emergency Use Authorizations for experimental treatments for Covid symptoms on the basis of minimal evidence of efficacy and safety (such as convalescent plasma, remdesevir, and Trump's beloved hydroxychloroquine) while demanding exponentially higher standards of safety and efficacy, backed by evidence from lengthy Randomised Controlled Trials, for vaccines against the same disease. Randomised Controlled Trials are the gold standard for testing truth questions: whether we can be completely certain that a medicine is very effective and safe. But such certainty comes at a price in human lives that must be justified in the same currency. Medicine is a practical science in which truth questions are only a means to the end of better helping people: they do not matter in themselves. The processes and epistemic standards for answering them must always be subordinate to that real goal and adapted to the needs of the particular case. 

Remember also that this is all about what medications are permitted to be offered to people, on which they remain free to make their own decision depending on their own values and attitudes to risk. Medical science is always in some sense a gamble. Medical ethics requires respecting its subjects by making that gamble as clear to them as possible and allowing them to make their own choices. It is true that vaccines are different from treatments in important respects, e.g. there are public health dimensions to consider as well as the individual risks and benefits that autonomy concerns. Nevertheless, it is hard to see how this can support requiring vaccine developers to make such a strong distinction between the risks that people may be permitted to take in order to avoid catching a widespread deadly disease and the risks they may take in treating that disease after catching it. 

It is clear that medical ethics is not fit for purpose in the case of vaccines against ongoing epidemics (and perhaps elsewhere too). Ethicists need to dramatically rethink standard analytical frameworks, decision protocols and bureaucratic bottlenecks for cases characterised by a trifecta of urgency, scale, and medical need. This may have immediate benefits for other similar cases (such as the neglected tropical diseases), which tend to be in poorer countries and whose neglect had therefore previously failed to get much attention.


Both the failures I have explored at some length are not problems with science per se, but with the functioning of the science-policy industrial complex. Science is an incredibly powerful empirical technology for finding out what is, and what works. When Covid appeared, tens of thousands of researchers and hundreds of institutions spun on a dime and directed their extraordinarily rigorous inquiries to understand this new virus, the disease it causes, and what interventions would best prevent and treat it. But the power of science to affect the world for the better depends on institutional links whose goals, character and robustness should be the focus of much more serious attention. And it would be better if we could get started on that now, before we land in our next crisis.