Imprimir Republish

good practices

Statistical inconsistencies

Statistical methods can be useful in helping investigate scientific misconduct, indicated a study by Mark Bolland, a professor at the School of Medicine of the University of Auckland, New Zealand, published in the journal Neurology. He examined 33 clinical studies of bone disease therapies conducted by a group led by physician Yoshihiro Sato at the Mitate Hospital in Tagawa, Japan and found numerous inconsistencies. The characteristics of the groups of people chosen to participate in the trials were much more similar than would have occurred by chance.

The studies had impressive results, showing that among those receiving the therapy, 78% were less likely to break a hip than the control group. Studies by other groups showed less favorable results, indicating that the risk of facture was between 0% and 40%. The choice of Sato’s studies as a target was not coincidental. In June 2016, three articles by the Japanese physician were retracted by the journal Neurology. Sato admitted having fabricated data in the fraudulent papers, which reported on the effects of therapies to reduce hip fractures in patients who had suffered a stroke and in Parkinson’s patients.