I knew I smelled something fishy about this paper when I read and commented on it last year. Now, Chest reports they're giving this study a burial at sea, after the authors could not produce actual data supporting the trial.
To help set the record straight: Chest published a retrospective study in March 2011, "Chest Tube Drainage of Transudative Pleural Effusions Hastens Liberation From Mechanical Ventilation." The paper purported to show an improvement by >3 ventilator days in ventilated patients with transudative pleural effusions into whom chest tube(s) were inserted. It was non-randomized and the authors' findings were surprising, at least to me.
Turns out, that's because their data didn't really exist, apparently. (Or as Chest put it, they were "incorrect.")
Some astute reader pointed out to Chest that these authors published the same data set in 2007 in a meeting abstract which they called "Pigtail Catheter Drainage of Transudative Pleural Effusions Hastens Liberation From Mechanical Ventilation." In that one, they presented their study as a randomized trial, which it apparently was not. Chest also published the meeting abstract. (Awkward, sure, but you try vetting thousands of manuscript submissions every year, and even more abstracts, and see how many lemons you let through.)
Editors at Chest called the authors to the carpet. The authors' response:
"Dude, where's my data?"
Chest's retraction states the authors could not produce a data set supporting the 2011 article. The authors' institution, Maimonides Medical Center, did its own internal investigation in response. Maimonides found only the old data set, according to Chest, who retracted both this article and the 2007 abstract (although if you click through to the 2011 article, there's no mention of its retraction on the current web version).
Ivan Oransky and Adam Marcus's influential blog Retraction Watch were all over this one. They spoke with the lead author, Yizhak Kupfer, associate director of pulmonary and critical care at Maimonides, who told RW:
"The committee that reviewed the data felt that because we didn’t have a computerized generated randomization code, we shouldn’t have called it randomized. We think that [it was properly randomized], but we had to go along with the committee’s [recommendation]."
Huh? Properly randomized? To all appearances it was an observational, retrospective case-control study -- as they described it themselves on clinicaltrials.gov.
Kupfer also told Retraction Watch a "computer error" was responsible for their using the data from the 2007 abstract in the 2011 manuscript:
When we wrote it, unfortunately we used the wrong data set. Because of that mistake, we had to retract the later paper.
The "wrong" data set? So was there a right one? From what I can see from the sidelines, none the authors could produce, or were willing to share.
Kupfer told RW that Chest has banned his group from publishing with them for three years, and that his group has hired a research coordinator to help them avoid such, ahem, oversights in the future.
Notice of Retraction: Use of Incorrect Data Set and Inability to Verify Results Due to Missing Data in “Chest Tube Drainage of Transudative Pleural Effusions Hastens Liberation From Mechanical Ventilation” (Chest. 2011;139:519-523)