Yes, it is intentional misleading. The authors and editors seem to agree.
When one finds problems in the statistical analysis of a paper, serious enough to invalidate the conclusions drawn, what should the reader do?
It is in the right spirit of science that you assume this might be because of oversight. To err is human and researchers are humans to begin with. If a reader notes serious problems in a paper, it is necessary to point it out with an expectation that the authors and/or editors respond. Science would welcome two types of responses. (i) They disagree with you. What you perceive as serious mistake is not really a mistake and they have a sound justification for what they did. If this is the case they need to justify sufficiently elaborately. Often the differences of opinion may not resolve. In that case both the sides need to be made transparent for the reader. That is the responsibility of the editor. (ii) In case the authors realize that there is some problem with their analysis and inferential logic, they need to correct the analysis or clearly state the limitations of the inference by publishing a correction to the paper. Both these responses are completely in the spirit of science and should be welcome.
There are two more possibilities that are not really in the spirit of science. (iii) The third possibility is that the authors neither have a sound justification, nor the readiness to correct themselves, but the editors realize the gravity of the problem and decide to retract the paper. This happens quite often but not so easily. Retraction is treated as serious damage to the prestige and reputation by the authors, their institutes, editors and the journals. Therefore they try to avoid, postpone or cover up the problem. (iv) The forth possibility is that the conclusions drawn from the flawed analysis were published with a deliberate intention to mislead the readers. If this is so, they will certainly not publish any correction because it goes against the very purpose of publishing. They will not have any justification to what they did because it was misleading anyway.
Either for the third or the fourth reason, editors are reluctant to take any action or just keep on delaying the action until the paper is old enough and readers have lost interest in reading any correction, even if published. By the time popular science perception has accepted the misled direction and then it is tough to change it. The publication of correction is a low key event, the purpose of deliberate misleading is served until then. If the editors do not take any action the sleuths still have an option of posting the cross questions in platforms such as PubPeer. If the authors think they are not wrong, they can publish a rebuttal on PubPeer. If they accept the problem they can publish a correction. If nothing of this happens we are left with the conclusion that the authors as well as editors deliberately intend to mislead the readers.
In certain fields of science misleading has a great benefit. The field of medicine, in particular, is prone to this because of the millions of dollars of possible profits involved. After spending huge amounts on developing a drug, if a clinical trial does not show it to be sufficiently effective, accepting the result leads to huge losses for the pharma industry. In that case making an impression that the drug is effective is necessary for the business. And it is not very difficult to mislead the medicine community, because either they do not understand the scientific method and inferential logic, or they simply do not have the time to waste in being careful about what they accept as science. People in academia certainly have the expertise but have no motivation. Their career progress depends upon how many high impact papers they themselves publish. Hardly any credit goes to exposing frauds in the field. The peril of being busy in making a successful career is that a lot of pervert science gets published and nobody cares.
What are the most common ways employed in misleading people in the field of clinical trials? Particularly clinical trials for life-style related disorders. There seem to be a few common tricks repeatedly used.
- Multiple statistical comparisons without correction: Statistical inferences are probabilistic. There is always a small chance that your inference can be wrong. When you do hundreds of statistical tests in a single study, at least a few of them turn out to show “significant” results. But this could be just a result of having done a large number of tests each one with a small chance of being wrong. This is a well recognized problem in statistics. Solutions have been suggested, which are not free of problems. But not having a good solution is not a justification for hiding or disowning the problem itself. Most clinical trials in the field of lifestyle related disorders typically use two strategies to take advantage of doing multiple tests and then hide or disown the problem.
- a. Register a large number of clinical trials. Not every trial gives results you want. Publish the ones in which you get favorable results, don’t publish those that give inconvenient results. In the field of type 2 diabetes clinical trials registered on https://clinicaltrials.gov/ , only about one third of the completed trial results have been made public. Two thirds remain silent about what they found.
- b. A given trial looks at a large number of outcomes, then categorize by sex, age groups, BMI groups and other possible subgroups. So the total number of tests typically performed are in hundreds and sometimes even in thousands. Then they prominently report the ones that are significant in the direction that they expect. It is almost guaranteed that a few will turn out to be significant by chance alone. This is enough to start beating drums that the drug is effective.
- Selective reporting and different format of reporting: Just as some of the tests turn out to be significant in the expected direction, a few are significant in the other direction. They do one of the two things when this happens. They either just don’t report the inconvenient ones or report them in a way that makes a misleading impression. For example they would use indices of relative risk reduction (such as odds ratio, hazards ratio) while reporting desirable effects of the drug; in contrast they use absolute risk reduction indices while reporting undesirable effects of the drug. Absolute risk reduction generally turns to have smaller looking numbers than relative risk reduction. So the reader is made to think that the good effects of the drug are large and bad effects small.
- Convenient subgroups and subtotaling: The subgroups are made by convenience and their results are selectively reported. For example they may make totals of many subcategories when that gives more expected picture, but report the subcategories separately when that gives a more convenient picture. The adverse events where they get expected results, are called “serious adverse events”, the ones in which results go in the other direction are called non-serious adverse events. This classification is not accompanied by any clear definition of the word “serious”. There are no standard guidelines on how to report statistics and in effect they take a systematically misleading path.
These are common tricks used for painting a picture that the drug is effective. Then they concoct more ways to mislead in a context specific manner. Very intelligently twisting data they reach conclusions that have been already pre-decided. Sometimes, raw data are available or are made available on request. I analyzed the raw data in many such cases and my analysis did not support their conclusions at all. The published conclusions could not be reached without cherry picking or twisting the data in some way or the other, and there are so many ways in which data can be twisted.
I am giving below the details of three clinical trials published in very reputed journals where we detected serious flaws in the published analysis, we wrote to authors and editors and ultimately made our concerns public on PubPeer. To begin with we were open to all the four possibilities. The authors could have counter-argued to show that they were not wrong; they could have published the corrected versions; if authors did neither, editors could have retracted the papers. But none of these things happened. Both authors and editors kept mum and did nothing except some promises that they will consider the cross questions seriously. From the response (or lack of it) of authors and editors we could differentiate between the above four possibilities. After careful scrutiny in all the three cases we are compelled draw a conclusion that both authors and editors in all these cases clearly intend to mislead people. Perhaps this is quite representative of clinical trials. We have more reasons to believe that a significant proportion of clinical trials have been systematically fooling people all the time.
The first case is that of a paper in Lancet Diabetes and Endocrinology. Our pubpeer comment on it and the prior correspondence with the editors is at these two links.
The Pubpeer comments were published in August 2024. The authors had ample time to respond, but they didn’t.
The second case is about a paper in PLOS Medicine. This paper was clearly hiding inconvenient results although the raw data could reveal them. We wrote to the editors in Aug 2024 who initially responded positively and asked clarifications from the authors. We have no idea whether the authors responded because the editors suddenly stopped all correspondence. Ultimately we published our comments on pubPeer in February 2025 to which the authors have not responded. Links to the correspondence with PLOS medicine editors and the pubpeer comments respectively are here.
Recently with some colleagues we started looking at the GLP-1RA trials about which there is much brouhaha. They are being projected as wonder drugs effective against anything that you name on earth. Looking at raw data shows that these clinical trials suffer the same set of problems which make their inferences invalid. On one of the papers published in Nature medicine and another on NEJM both from the SELECT trial we wrote comments on PubPeer in February 2025 and the authors haven’t responded as yet.
Interestingly to the letter in response to PLOS Medicine paper and the PubPeer comments on the Nature medicine paper after requesting correction we wrote, “However, corrections need not be made if the misleading is intentional since the purpose is served.” Both the authors did not make any correction, nor counter-argued in defense of their analysis in any form. This is a clear admission that the misleading was intentional. Since the editors also did nothing, it is clear that even editors of such prestigious journals are interested in deliberately misleading people.
Perhaps intentional misleading is common across clinical trials. Earlier I published my analysis of multiple clinical trials related to type 2 diabetes. There were many responses to this article, all in the public domain, mostly supporting our arguments. Not a single response was from authors of the original papers (https://www.qeios.com/read/IH7KEP ).
If anyone has any doubt about intentional misleading, please contact the authors of these papers or editors of the respective journals for confirmation.
April 11, 2025
My science, My way 
Dear Dr. Watwe
I retired as Scientist ( Atmospheric Radar and Instrumentation) and I keenly read your articles and I also believe that the many times publications do not report the factual details. However, now days some of the publishers are insisting on keeping their data and software programs used in public domain to verify their claims.
But there are many challenges in accessing the data and verification of the results.
However, my intention of writing this is mail to understand if there is any way of controlling Type 2
Diabetes scientifically. There is tons of tons literature scattered on the internet what to eat or not eat
Or what exercises to do.
I just wanted to hear from your research experience , is there any way to effectively control it.
I am sorry for asking such layman questions but I am sure you would give me unbiased and more scientific approach to the diabetes problem.
With Warm Regards
-Anil Kulkarni
LikeLike
Sugar control has lost its relevance. Clinical trial data increasingly show that sugar control is not correlated with arresting diabetic complications. Glucose by itself is not dangerous. Complications are. But physical fitness is the key to avoid complications. Maintaining the required fitness for you age keeps you away from diabetic complications independent of sugar.
LikeLike