Our manuscript was rejected by BMJ Evidence Based Medicine. The paper was examining data from 6 large clinical trials, evaluating the trial design, statistical rigor and inferential logic used to reach conclusions. We also reexamined the methodology used in published meta-analyses of these trials. The result was astonishing. Together there was no evidence that regulating glucose reduces the incidence of diabetic complications. This is important because glucose regulation is the main, perhaps the only focus of diabetes treatment at present. Our analysis showed that the entire line of treatment which has currently something like a trillion dollar turnover is without a sound evidence base. The preprint of Medrxiv is here https://www.medrxiv.org/content/10.1101/2022.08.02.22278347v1.
The journal rejected the manuscript within about 6 hours of submission saying that it did not achieve “a high priority score”. They say the rejection is not based on the quality of the paper. Only on a judgment of priority. This has interesting and far reaching implications for the field of type 2 diabetes on the one hand and for the science publishing system on the other.
What our analysis mainly found was the following
- All the papers have multiple and serious statistical flaws. The main being that when a large number of statistical comparisons are made, some are bound to come out to be significant by chance alone. For this a correction called Bonferroni correction needs to be made. None of the trials do this. If we apply this correction to their data, nothing remains statistically significant. Bonferroni correction is more conservative. Therefore we also suggested an alternative based on simulations, but even with this approach, none of the claimed benefits of treatment turn out to be really statistically significant.
- The clinical trial that is believed to have shown the benefits of sugar regulation, the UKPDS does not have a placebo control. Other trials that have placebo or blinding of some kind do not show as many benefits as the UKPDS. Therefore the assumed positive effects of glucose regulation are likely to be placebo alone. Then there is a second level of placebo in trials with surrogate end points such as glucose. The feeling that my glucose is in better control is likely to exert a placebo effect at a different level and none of the trials has appropriate controls for this.
- Even if we assume that the marginal benefits are true, it cannot be inferred that it was because of sugar normalization. Insulin has so many other functions in the body. Some of the anti-diabetic drugs also have other sites of action independent of glucose. So the inference that these marginal benefits are because of sugar normalization has no support.
- The magnitude of difference is so small that it is clinically meaningless. Even if we take only the favorable results and assume them to be true, 10 patients will have to be treated for 25 years each in order to prevent one diabetic complication in one person.
All this is crystal clear from the data and it is high time we give up glucose normalization as the main focus of diabetes treatment. But beliefs appear to matter more than data in medicine. Since the analysis showed something against their belief, how can they publish it? Whatever the quality of data, analysis and arguments!!
Now since I have no formal career in science, the rejection will not affect me. On the contrary I am more delighted for having one more sample to understand how the secretive editorial machinery works. The rejection was so fast that the chance that anything in the manuscript would have been read seriously is simply out of question. The important thing to be read is only from where the paper comes. If the authors and their affiliation is obscure, there is no need to read anything further.
But what is more interesting is the reason given for the rejection. By saying, this was not a high priority issue, the journal admits that either type 2 diabetes is not a high priority disease or being statistically sound is not at a high priority, being critical about the nature of evidence is not a high priority issue or questioning a line of treatment based on absence of supportive evidence is not a high priority concern for the journal called EVIDENCE BASED MEDICINE.
The actual correspondence is pasted as it is below.
Dear Dr. Watve,
Manuscript ID bmjebm-2022-112095 – “Does sugar control arrest complications in type 2 diabetes? Examining rigor in statistical and causal inference in clinical trials.”
I write you in regards to the manuscript above.
We are sorry to say that we are unable to accept it for publication, as it did not achieve a high enough priority score to enable it to be published in BMJ Evidence-Based Medicine. We have not sent this manuscript for external peer review as in our experience this is unlikely to alter the chances of ultimate acceptance. We are keen to provide authors with a prompt decision to allow them to submit elsewhere without unnecessary delay.
Our decision may be disappointing, especially in view of the lack of a detailed critique. This decision must be based not only on quality but also timeliness and priority against other subject areas.
Thank you for considering BMJ Evidence-Based Medicine for the publication of your research. I hope the outcome of this specific submission will not discourage you from the submission of future manuscripts.
Dr. Juan Franco
Editor in Chief, BMJ Evidence-Based Medicine
|Dec 8, 2022|
Dear Dr. Franco
Thanks for your prompt response. I just have one request. I would like to have your consent to quote your reply in any article/blog/comments on preprints.
Let me also tell you in what context I would like to quote your reply.
Our paper pointed out many statistical and inferential flaws in a series of clinical trials on glucose normalization treatment to prevent diabetic complications. The nature of the flaws is such that their conclusions become completely invalid. The current practice of type 2 diabetes treatment becomes questionable. All this is supported by sound analysis of data from systematically selected 6 large clinical trials.
Your response says our arguments “did not achieve a high enough priority score” which implies that
1. Being statistically sound is not a high priority for you.
2. Being critical about the evidence base is not a high priority for you.
3. Questioning a line of treatment based on absence of supportive evidence is not a high priority for you.
Kindly let me know your consent to quote your email in this context.
I believe in complete transparency of the editorial process and therefore expect cooperation from you in this regard.
(Dr. Milind Watve)
जुस्ते हक़ की रहगुजर में जो सियाही है, मेरी है
उस मजाज़े आराइश में तेरा ही बस हो तो क्या है
The darkness on the path to truth is my homeground. If the dazzling lights in the rest of the world are under your command, why should I care!!
I received no reply to this later.