乌鸦传媒

A team of statisticians has found evidence of scientific misconduct in up to 33 osteoporosis studies published in several journals.

Three of those papers were published by Neurology and retracted this summer after author , of Mitate Hospital in Tagawa, Japan, admitted to the journal that "the work reported...was fraudulent." Sato was the lead author on 79% of the studies.

, of the University of Auckland in New Zealand, and colleagues, applied a statistical technique -- comparing the distributions of P-values for between-group differences for several measures -- to the 33 papers, revealing "improbable" similarities in treatment groups, they .

Their systematic review also found "remarkably positive" outcomes and unlikely recruitment speeds, they said.

Bolland explained that in a randomized trial, the composition of the randomized groups is determined by chance, so any baseline differences between the groups should follow the expected chance distribution.

But if that distribution isn't consistent with the expected distribution, it suggests randomization wasn't successful -- and in the Sato studies, "the baseline variables in the randomized treatment groups were strikingly similar, and the distribution of differences in the baseline variables was very inconsistent with the expected distribution ... [suggesting] a systematic failure of randomization in these trials," Bolland told 乌鸦传媒.

, a nephrologist and statistics expert at Yale University in New Haven, Conn., who wasn't involved in any of the studies, said the statistical approach is sound and that the findings are of interest. While there could be other explanations for that besides outright fraud, he said, the numbers are unusual.

"The group are too similar," he said. "You'd expect more differences."

Bolland and colleagues also found "remarkably positive" outcomes: the studies had very low mortality and study withdrawals despite substantial comorbidity, they reported.

For instance, there were very large reductions in hip fracture incidence, regardless of intervention (RR 0.22, 95% CI 0.15 to 0.31, P<0.0001), which greatly exceeded those reported in meta-analyses of other trials.

The researchers were also "remarkably productive," publishing 33 randomized controlled trials over 15 years, and these involved a large number of older patients with substantial comorbidity recruited over very short periods. This included recruiting 500 female Alzheimer's patients in 2 months; 280 men over 65 with hemiplegic stroke in 2 months; and 374 women over 65 with acute hemiplegic stroke in 4 months, Bolland and colleagues reported.

"Combined with the other various analyses we performed, and the problems identified in our systematic review, this suggests the trial results are unreliable," he said.

Bolland said the analysis arose out of a 2012 conversation with colleague , of the University of Aberdeen in Scotland, who had, while doing systematic reviews, come across three trials by Sato that had identical results. She asked Bolland about the studies given his expertise in osteoporosis.

"Even a superficial viewing of several of the trials ... raised a number of questions, so we decided to investigate further by systematically reviewing all the published trials of this group," Bolland told 乌鸦传媒.

They submitted their work to Neurology in December 2015, and in an , journal editor-in-chief , noted that the review process took so long because the findings "had implications for authors and institutions, for other journals and editors who published work by these authors, and for clinicians, guideline committees, and policymakers who rely on the validity of these findings."

The journal had three statisticians who regularly review for Neurology take a look at Bolland's paper. Gross subsequently notified other editors whose journals had published papers by Sato and colleagues, including JAMA and the Journal of Bone and Mineral Research.

He also spoke with Sato's institution, and ultimately retracted the three studies published in Neurology at the request of Sato, who absolved his co-authors of any wrongdoing.

Gross cautioned, however, that fraud "in an individual paper may be difficult to detect. One cannot conclude that any one study in the analysis is, or is not, fraudulent."

He added that Neurology will continue to use "rigorous statistical review to detect fraud with the goal of maintaining the highest possible standards in publishing."

Bolland said that if investigations are being undertaken by Sato's institution, he hopes the results "will be made public so that everyone can understand what has taken place, and also whether the results of the numerous other trials and publications by these researchers are reliable or not."

Comment