Prolific data fabricator

Mohammed Rezk from Menoufia

Dr Mohammed Rezk is Assistant Professor of Obstetrics and Gynecology at Menoufia University in Egypt. He has 71 publications on Google scholar, including 17 RCTs and 34 cohort studies. My colleagues and I first became aware of his work through two trials.

• Rezk M, Emarh M, Masood A, Dawood R, El-Shamy E, Gamal A, et al. Methyldopa versus labetalol or no medication for treatment of mild and moderate chronic hypertension during pregnancy: a randomized clinical trial. Hypertens Pregnancy. 2020;39(4):393-8.

• Salama M, Rezk M, Gaber W, Hamza H, Marawan H, Gamal A, et al. Methyldopa versus nifedipine or no medication for treatment of chronic hypertension during pregnancy: A multicenter randomized clinical trial. Pregnancy Hypertens. 2019;17:54-8.

Despite being allegedly different trials both recruited over the identical time period in the same hospital, and many result values were either identical, or within a single digit of each other – see PubPeer comment (here). No satisfactory explanation was forthcoming, and each journal eventually published an expression of concern (click here and here).

There was another unusual feature. In Rezk 2020, 69/70 categorical variables were even numbered, and in Salama 2019, all 75 were even. The probability of this happening in real data is extremely low. The same even number excess is a feature of many of his other papers (e.g. here. Tabulated here).

We were not the first to notice this. Retraction Watch recently reported that in 2017 John Carlisle, the famous fraud buster, had noticed an excess of even numbers in a trial submitted by Dr Rezk to Anesthesia (click here). He declined to publish the paper, got an evasive reply to his request for original data, and no reply at all to his request for Menoufia university to investigate. Others independently noted an excess of numbers in multiples of 5 or 10 (click here and here).

Rezk was not only undeterred – he has published 30 more papers since being called out by Carlisle – but he doesn’t seem to learn either. At least nine of these later papers contain the same highly improbable excess of even numbered categorical variables.

Last year we reviewed all 51 of Dr Rezk’s clinical papers (here or preprint). At least 35 have problems, of which to date four have been retracted, and a further eight have a published expression of concern.

If all 35 were retracted, he would be 16th equal on the Retraction Watch leader board (click here). With 51 he would make 9th place.

Jim Thornton