Abstract
Four years ago, we did not use the words “Fake News” so much, although it has probably existed since humans started exchanging information and has been used ever since. Fake or Fact? This is a crucial distinction to make whenever we act while basing our action on information. Natural sciences have given us the instruments to distinguish facts from errors and when using these instruments experimentally, facts are circumstances that can be verified. Three out of many instruments are a standardized design, a well-fitting control, and statistics. Since humans plainly lack a reliable “feeling of truth”, these instruments are designed to ultimately help us detect facts. The undeniable achievements, not just in medicine, are innumerable and culminate in the popular phrase on t-shirts: “Got Polio? Me Neither. Thanks Science”. It is not negotiable: Facts are not optional, not just a user-defined “construction”; they are facts because they can be tested, and they have no “alternative”. The alternative to a fact is falsehood.
Two distinctions need to be made: The term “Fake News” describes in effect a deliberate hoax with the explicit, mostly political aim to spread uncertainty, mimicking reputable news. There is certainly a fundamental difference between fake news and scientific studies of poor quality. To distinguish Fake News from scientific facts is particularly important and in essence the competence of the reader rather than of the originator of a report. Secondly, the term “scientific fact” is not unproblematic since we (the scientists) concede a 5% error margin. This means that at least 5% of all that we publish, regardless how well a study is done, is, in fact, false. This situation is one of the biggest strengths of science: That we all – and always – know that we could be wrong. A good scientist will always accept that whatever we prove or demonstrate only mirrors the current level of errors and omissions. This greatest strength is also a fatal weakness when confronted by fake news: If we, the scientists, do not even know the truth – the truth may be arbitrary and becomes what someone chooses to believe (1).
Unfortunately, we cannot simply translate the above-mentioned phrase into “Got Migraine? Me Neither. Thanks Science”. We are coming a long way, but although our knowledge about the pathophysiology and treatment of headache is tremendous compared to what we knew just 50 years ago, we have not truly understood migraine. As of yet. The fact (sic) that we have, with triptans and CGRP antibodies, not one but two highly specific medications specifically designed for migraine is one out of many a triumph for headache science. All this is the result of strict adherence to scientific rules and Cephalalgia, the official journal of the International Headache Society, is not just a member of the Committee on Publication Ethics but also strongly recommends that authors follow the Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals formulated by the International Committee of Medical Journal Editors (ICMJE) (2). Standards for reporting involve, for example, that authors should follow STROBE (observational studies), CONSORT (randomized controlled studies), ARRIVE (animal research), and MOOSE (systematic reviews/Meta-analyses). Cephalalgia also requires that studies on headache and cervical pain adhere to the terminology and criteria within the ICHD (3) and studies on facial pain adhere to the terminology and criteria within ICOP (4). The combination of a standardized method of reporting (5) and an unequivocal classification of what type of headache is investigated (6,7) are the basis on which the headache community ensures that any data reported in – let us assume America – is verifiable; that is, reproducible in Europe and Asia. Other quality issues are a sufficiently high number of participants and adequate statistics. Possibly even more important is transparency – to note how experiments have been conducted without leaving out information. This requires that the corresponding author takes all the responsibility for the correctness of a study. All of this has only one purpose: to allow others to reproduce (or not) the findings of a given study. Moreover, by reproducing (or not), we build the fundament of what we think is true.
If reproducibility is the key to eliminating type I and type II errors; that is, to differentiate between correct and wrong findings, the question is how to ensure this. A possible answer can be found in Open Science. One key aspect of open science is open access publishing (8), and Cephalalgia welcomes initiatives like Plan-S (9). Another key aspect of open science is equally important; namely, availability of source data. The combination of clearly described methods and the availability of source data is the key to reproducing scientific findings. Cephalalgia strongly supports and looks favorably at submissions offering the deposition and availability of source data. And, thirdly, what becomes more and more important in science is the pre-registration of studies in a specified database such as, for example, at clinicaltrials.gov. To preregister studies requires authors to frame a hypothesis, declare effect size and what methods and statistics will be used, and define the nature and number of the intervention(s) as well as the control group.
Readers and reviewers are then in the position to compare the methods and hypothesis in the preregistration and the final manuscript. Every statement in the preregistration can be changed and adapted and, since this is time-locked, it is transparent as to when during the process this has been done. All this is common knowledge in pharmacological studies, but it is still rarely used in basic science. Why exactly is this the case? Perhaps because basic scientists were, until now, not forced to think about it. Publishing the hypothesis and methods before starting the experiment is actually a rather clever way to convince reviewers and future readers of valuable work. Researchers may answer that they do not know what they will find and thus do not want to tie themselves down. The answer is that to falsify a hypothesis is as valuable as to verify it, and serendipity findings may prove to be particularly beneficial. As an editor, I follow the philosophy that the fact that something is published does not prevent the reader from thinking. What we should end is publishing innumerable underpowered papers with findings that contradict all other studies in this area and simply state that “future” studies should be done. How much better is it to conduct a study, and rather than publishing it directly, to take these results as hypothesis generating and preregister a replication of this study. Findings that have been preregistered and reproduced by the same group are very likely to be replicated by an unrelated group as well. They are also more likely to be true. As a headache scientist once stated: “You do not want to publish just in a journal, what you want is your findings to be published in textbooks.”
If we require the pharmaceutical industry to be transparent and publish exactly what hypothesis they had and what type of methods they used, and not to leave out any content that could question the conclusions, we should do the same with clinical and basic science. Let us all defend the integrity of what science is based upon. If we all follow this, it becomes more likely that we may one day wear a t-shirt with the slogan: “Got Migraine? Me Neither. Thanks Science”.
