How to stop pounding water in a mortar
Doctors write "about nothing"
Nadezhda Markina, <url>Most of the articles in medical scientific journals are nonsense.
This is the opinion of John Ioannidis, professor of medicine at Stanford University, which he published in the open journal PLOS Medicine. His first review article on this topic, Why Most Published Research Findings Are False, appeared in 2005 and collected more than 1 million "clicks". In the second article, How to Make More Published Research True, published on the 10th anniversary of the journal PLOS Medicine, the author suggests measures that will help correct the situation and restore confidence in medical publications.
Actually, the reasoning and conclusions, as well as the proposed measures, relate not only to articles on medicine, but also to all other scientific publications. Their number is huge. The author cites data that in 1996-2011, more than 15 million scientists around the world published more than 25 million scientific articles. However, large and useful discoveries are much rarer.
Many of the published dependencies and phenomena are actually generally incorrect or greatly exaggerated. The application of the acquired knowledge in practice requires a long time and is not always effective.
Why is that?According to the author, 85% of scientific research in the field of medicine leads to incorrect or useless results.
Among the reasons he cites insufficient statistical reliability of the results, exaggeration of small effects, attempts to achieve the expected result, resourcefulness of scientists in interpreting data, as well as prejudice and conflict of interests of different scientific groups.
The interests of an academic career, the organization of work at universities and research institutes, the practice of reviewing articles in journals - all this sometimes encourages scientists to spend time getting results that no one needs and certainly do not help treat patients.
To change the situation, the author believes, it is necessary to reconsider the principles on which an academic career is built. Many scientific institutions date back to the Middle Ages (as academic titles) or to the XVII century (as a professional community and scientific journals), and it is unknown how optimal they are in modern society.
Standards and statisticsAccording to the author, effective intervention is required to increase the reliability and effectiveness of scientific research.
Some of the risk factors will not go away, but the rest can be changed. To avoid prejudice and conflicts of interest affecting the results, research should be conducted in large collaborations, as transparently as possible and according to standardized rules.
But do not go to the limit. For example, you can completely get rid of mistakes by rejecting all works with at least some preliminary expectations, leaving only those tasks that no one cares about solving, and uniting all scientists within the framework of one research methodology. The number of errors will drop to zero, as will the number of studies themselves. So the solution should be pragmatic, implementable and ideally verifiable for effectiveness.
The same data presentation, sample preparation, research protocols and software create the possibility of reproducible results. This approach is common in molecular genetic research. The use of the same standards in clinical trials, according to the author, can significantly increase their reliability. Registration in the database of randomized clinical trials increases their transparency.
The introduction of the best statistical methods can reduce the number of unreliable results in such "inaccurate" specialties as epidemiology, psychology and economics, the author believes. Perhaps, as a result, they will shift in reliability towards the physical sciences. In the case of a supposed scientific discovery, higher data reliability thresholds should be used.
ReviewingThe article review system needs to be improved.
For example, it is known, the author writes, that most reviewers are scientists with a modest track record, and most outstanding scientists do not review grant applications themselves. The practices of nepotism, sexism and unjustified conservatism seem to be widespread. Conscientious reviewing is not encouraged in any way. Superficial reviewing does not greatly improve the quality of articles and even skips frankly incorrect works. At the same time, the reviewer has no incentives to work in good faith.
Revision of the "scientific currency"The author proposes to revise the traditionally established system of incentives and incentives for scientists.
Works can be rewarded for their publicity, attracting money, applicability of results or for profitability. These wishes are not always compatible, the author notes. For example, if publications and grants are required, then scientists are interested in continuing research, even if there is no real progress.
Private investors need patents and profits, respectively, applicable inventions will flourish, but there will be barriers to the dissemination of information. The influence of corporations can turn the whole science into a division of advertising departments – through publications in influential journals, prestigious public promotions and public experts who receive salaries in these departments.
"The two main "currencies" in science are publications and grants. They acquire scientific benefits – career and power, and scientific titles and power increase the "wealth" of their owner. The real "exchange rate" in different areas may vary greatly. Administrative power, personal connections and lobbying further distort the picture," Ioannidis writes. "In such a situation, there is a selection of those who fit better into the system, publish modest and/or non–reproducible studies, control peer review, breed bureaucracy, lobbying and behind-the-scenes maneuvers."
The current system values publications, grants, academic titles and power. Higher-class scientists publish more and have more grants. However, at the very top of the administrative ladder (university rectors, etc.), the level of publications is usually low. Perhaps this is due to the fact that they are clever lobbyists and easily bypass more worthy candidates who could make more reasonable decisions. But it is also possible that they spend their lives managing the bureaucratic machine so that their colleagues can safely do science.
The author suggests two types of reform of the remuneration system in science. At the first proposed reform, it is proposed to reward not the publication as such, but links to it and the use of the results. This will be used for effective treatment, diagnostic tests, etc., confirmed in clinical studies. It is worth paying extra for the exchange of scientific data, for qualified review and for educational activity.
The second approach is even more radical. Grants, awards and administrative posts themselves are evaluated negatively. In other words, great opportunities must be accompanied by much greater scientific success, otherwise the amount will go into negative. Scientific titles are also evaluated negatively and, apparently, self-destruct – scientists will get rid of negative "points". In this provocative scenario, scientists will not recruit grants and positions, because they will lose in the "scientific currency".
"It is not known how much the effectiveness of the current research system can be improved at all," Ioannidis sums up. "However, the existing state is so ineffective, because something can almost certainly be done. There are many different scenarios, ideally it would be good to test them in practice."
Portal "Eternal youth" http://vechnayamolodost.ru24.10.2014