From coronavirus to climate change, scientific discussion has been stifled by the ‘settled science’ trope. As a long-term analyst and careful critic of medical papers it annoys me. In 2000 I began a column in the journal of the British Society for Rheumatology (conveniently titled Rheumatology) trawling the journals for interesting, unusual or bizarre pieces of research, which I would dissect for my readers. Five years as a columnist honed my critical eye.
One cornerstone of medical research is the null hypothesis. You set up an idea and try to find something that knocks it down. A parallel concept is that of the Black Swan, where a plan is derailed by an unexpected event. In each case a single observation or fact which disposes of the hypothesis is sufficient for its abandonment. Bill Stott encapsulated this in a brilliant cartoon in Punch some years ago.
Various doctors have described how one of their teachers would tell them that, after five years, half the things they had been taught would be shown to be wrong, and the problem was that you didn’t know which half. But it underpins the argument that medicine – and by extension science – is not a static subject and is always subject to change. There are endless examples of how scientific consensus has been disrupted by a contrary observation which pulls the whole house down. Some are well-known, some less so. Here are a few. As you will see there are different mechanisms and causes.
Settled science said the Sun went round the Earth until Galileo and Copernicus made observations that proved it did not. Both were lone voices and suffered for their beliefs. Settled science said that cholera and puerperal sepsis were considered miasmic conditions until John Snow conducted a controlled trial proving the former was water-borne and Ignac Semmelweis conducted a similar one for the latter. Snow’s work took some time to become accepted, while Semmelweis’ work took longer, no doubt in part because of the way he presented it and attacked his disbelieving colleagues.
Before 1921, settled science did not know the cause of diabetes. Then Banting and Best discovered insulin, though Paulescu in Romania was the first. So settled science then began to understand that insulin lack caused diabetes. But then it became apparent that there were distinct types of diabetes and that insulin lack was not the only cause; insulin resistance could occur. Then it was found that insulin lack might be caused by an autoimmune reaction destroying the insulin-producing cells in the pancreas. Then the hormones glucagon and later leptin were discovered. Diabetes was no longer due to failure to produce insulin alone, and diabetes became as confused as it had been at the beginning of the 20th Century. No settled science here.
Some infections of unknown cause were found to be due to viruses – but it was not until it was possible to see these microbes that cause and effect could be adduced. Similarly, the understanding of inflammatory processes and the realisation that white blood cells held a key role in disease transformed rheumatology management, but, just as with diabetes, the exploration of the inflammatory response changed the science. It was not just white cells; these now were subdivided into T-cells, B-cells and macrophages, and then the T-cells got further divided into T-helper cells, T-suppressor cells and T-null cells, and then the inflammatory chemical cascade ran amok with the discovery of all sorts of interferons and interleukins, the number of which latter is unclear (search Google using “how many interleukins are there” and you will find varying estimates from 30 to over 60). I have sat through many lectures describing the inflammatory cascade and the slides become more complex by the week. In these examples, advances in science – here immunology, biochemistry and microscopy – were key to unravelling mysteries and taking down hypotheses.
In the 1970s some research pointed to the possibility that rheumatoid arthritis was triggered by infection with the infectious mononucleosis virus. This hypothesis was confirmed in a couple of laboratories, only to be knocked flat by the discovery that the original results were due to contamination. So, one may argue that the revelation of faulty research led to a change in understanding (or belief). Belief also caused patients with back pain to be put to bed, often in hospital and in spinal traction. After all, who could argue with the concept that such pain should be treated with rest? I didn’t; sometimes I had four or five back pain patients in my ward. We had always done it, we were taught to do it and so we went on doing it. It was settled science. But in fact no-one had done any research to compare bedrest against ambulant treatment and when they did – if ever there was a medical U-turn, that was it.
Settled science relies nowadays on clinical trials. But have these been properly conducted? In other words, is the conclusion justified? Was the sample size sufficient to prove the case? Was randomisation properly done? Have the correct statistical tests been used? Have confounders been accounted for? Have all the data been properly analysed? Were there any conflicts of interest that could have influenced the publication of results? Have the conclusions been fairly presented? Have the data been open to outside inspection? Statistical teaching in medicine is patchy and I certainly would not profess to be a statistics expert, but others are and I can rely on them.
There were numerous trials of anti-inflammatory drugs in the 1970s which had too small sample sizes, and several which concluded that drug A was similar in efficacy to drug B, although re-analysis merely proved that the trial had failed to show a difference, which is not the same. Many trials used the wrong statistical tests and were improperly randomised or excluded specific groups. Patient selection for trials? Benoxaprofen (Opren) was deemed safe but had never been tested in older people who, it transpired, metabolised it differently and in whom it was far from safe. Similarly a preparation of oral gold (for rheumatoid arthritis; injectable gold was the mainstay of therapy in the 1970s) had pharmacokinetic data showing two patterns of drug clearance judged by its half-life in the body, but the sample size was small. Only one patient showed the second pattern of delayed clearance – but this was 15% of the sample, yet the analysis ignored it. Extrapolate to a population and you can see the problem; as it turned out, the population did show such a dichotomy and the delayed clearers had a higher incidence of side-effects due to drug accumulation. Some trials excluded dropouts from analysis and thus failed to assess properly why subjects failed to complete the trial. Conflicts of interest? Some trials were and are conducted by drug companies or by researchers receiving large grants from them, and negative results are often concealed. Positive results are great, but when looking at the risk-benefit analysis it is important to compare like with like. Thus, if a statin trial shows a 50% reduction in cardiac risk, what is that a 50% reduction of? If the absolute risk is 5% then a 50% reduction drops the risk to 2.5%, which is not huge. Often the risk is presented in absolute terms and compared with the benefit in relative terms, which is scientifically dishonest. And all too often the raw data from such trials are not offered up for independent analysis, with the feeble excuse that it is commercially confidential. All of these unsettle settled science.
The explosion of genomics has led to the unravelling of numerous medical mysteries, in particular in hereditable disorders and inborn errors of metabolism. Fifty years ago this would have been unthinkable, but science progresses, we find out more things – and we find things that undermine current opinion, as Einstein did with gravity and new space telescopes have done with quasars, black holes and other studies of the universe. If we did not find those things and realise that science can change, we would still, perhaps, think the Earth was at the centre of the Universe.
In today’s context we have the twin issues of SARS-CoV-2 and climate change. In the first case we have been led to believe that settled science proves that lockdowns and masks work and that the virus was of natural origin. In each case unpicking of the data, some of it concealed for whatever reason, undermines all three conclusions. We were bamboozled into believing that settled science showed that there would be an explosion of deaths, failing to see that the conclusion was not based on facts but on prophecy. We were told that settled science proved that vaccines were safe and effective, but when doubters asked for the evidence, this was largely absent, and our belief that settled science had proved the safety of m-RNA vaccines was destroyed by analysis proving significant DNA contamination. There is a hypothetical risk to this, but that risk remains unquantified if indeed it exists at all, but it is unsettling to think there might be a significant risk. Would it not be safer to do some long-term studies before exposing a global population?
The debate on climate change has also relied on models, and settled science activists have brushed aside the GIGO axiom – garbage in, garbage out. They have also failed to quantify the relative contributions of fossil fuel burning, volcanic eruptions, solar storms and sunspot activity, changes in the Earth’s magnetic field, deforestation and water diversion and failed to understand the effect of urbanisation. In plunging headlong into the promotion of so-called renewable energy they have also fallen into another common trap – that of proclaiming the benefits while failing to analyse what could possibly go wrong. If, as I was told this week, five of 50 luxury cars of the same marque had to be returned to the saleroom for replacement of their batteries after less than two years, what are the costs both to purchaser and manufacturer? If spontaneous combustion risk causes insurance premiums for such vehicles to become unaffordable, what is the future, let alone the likelihood that demands on electricity supply will become unmanageable, so the cost of recharging goes up and also becomes unaffordable? And this set of risks excludes any analysis of whether the climate is really changing as dramatically as the prophets foretell, which factual evidence seems to contradict.
When sceptics raise all or any of these contradictions they are disparaged as vaccine deniers, climate change deniers and dangerous fanatics. If there is doubt, then say so. There is nothing wrong with saying you don’t know. While there are undoubtedly some proponents of wacky conspiracies lurking in the undergrowth, the huge majority of sceptics have merely analysed the data in a scientific way and unravelled the settled science consensus. And then the settled scientists themselves become deniers because they are too rigid to change their minds despite the revelation of incontrovertible facts.
These few examples lead to the inescapable conclusion that there is no such thing as ‘Settled Science’. Such science may be informed by dogma, belief, poor research, inadequate information and even fraud. New investigative techniques, re-analysis of past work, repeats of trials in the light of new knowledge all confirm that science may change. ‘Settled Science’ is an oxymoron. Let’s leave it behind.
Dr Andrew Bamji is a retired Consultant Rheumatologist and author of Mad Medicine: Myths, Maxims and Mayhem in the National Health Service.