Ob-gyn journal pulls pregnancy test paper for undeclared conflict of interest, other problems
A little background: Earlier this year, Laurence Cole, an academic obstetrics specialist at the University of New Mexico in Albuquerque, published a paper in the AJOG looking at the wide variability in the hormone human chorionic gonadotropin, or hCG, during pregnancy (
we’d link to the article, but the journal has pulled it, so here’s the abstract on Medline).
Cole runs — or did run until recently, more on that in a moment — the USA hCG Reference Service at the university, which purports to be the only lab in the world that can measure all forms of hCG. He has published extensively in this area of research, with at least 125 papers to his name on the subject, according to a Medline search. One of his websites puts the figure at 246.
So Cole was definitely a known quantity to the journal editors when he submitted his manuscript and when it was published online in February of this year. That’ll be more important in a bit. But first, the retraction notice:
The article titled “Individual deviations of human chorionic gonadotropin concentrations during pregnancy” was published in March 2011 (Am J Obstet Gynecol 2011;204:349.e1-7). The Journal received an allegation that the author failed to declare conflict of interest and that the assays that were used for determinations of human chorionic gonadotropin (hCG) and luteinizing hormone (LH) were flawed. In accordance with the published guidelines of the Journal, the initial inquiry process was conducted to investigate the allegation.
The inquiry committee of 3 independent experts in the field reviewed all related documents and the author’s response to questions that had been raised and concluded that the author violated the guidelines of the Journal’s specific inappropriate acts in the publication process.
It’s unusual, in our experience, for journals to take investigations upon themselves. Most wait for the authors’ institutions, often for years, and then are stymied when those institutions refuse to make the results public.
But wait, there’s more — much, much more. In fact, the retraction reads like an article itself, with an outline, subheadings and internal citations. Under the heading “Issue of Conflict of Interest,” the notice states:
The panel concluded that an undisclosed conflict of interest existed. Charles & Dwight Co (C&D) is the manufacturer of First Response home pregnancy and home ovulation tests, and the study findings might have influence on these products and other similar products. The author acknowledged that he had contract research with funding for C&D from 2004-2008. The urine samples that were used for the study that was reported in the article came from his previous studies, which had been funded by C&D under the contract; the study period was July-December 2009. The panel concluded that that fact clearly constitutes a conflict of interest that should have been disclosed. Failure to disclose conflicts of interest is a serious violation of publication principle/guidelines
That’s followed by several paragraphs, which we bring you below, under the rubric “Scientific Issues”:
Rationale of the study
The author states “It is well-known that total human chorionic gonadotropin (hCG) levels in serum and urine samples vary tremendously between individuals. In fact pregnancy hCG concentrations vary among individuals so greatly that hCG measurements are of little use in dating a pregnancy.” Other investigators have shown repeatedly that hCG measurements are highly variable. There is no credible scientific reason given for conducting the study.
The author used Siemens Immulite 1000 for both hCG and LH measurements in urine. The validation information shown is for hCG assay and should have included LH, especially when the urine sample might not be a suitable matrix for Siemens Immulite LH assay. One major point of the study was linking hCG levels to the LH surge. The product information clearly states that the Immulite Analyzers are for the qualitative measurement of hCG in serum and for strictly qualitative determinations in urine. The procedure for urine samples is designed solely as a screen for pregnancy, with the use of a 30-mIU/mL cutoff.
The limit of sensitivity was not identified in a typical manner. When asked to respond to this issue the author responded that no official sensitivity was set and that “no patient samples were detected for clinical purposes, just research patient samples.” The panel concluded that the sensitivity should have been set for the study, that the linearity should be over the range of the tested samples and/or dilutional standards, and that the impact of dilution should have been included.
There is a discussion regarding the significant variation in creatinine concentrations, yet no correction was exercised for the impact of urinary concentration. This might be the source of the variation of the rise within a single individual pregnancy and not an intrinsic variability as proposed.
Assessment of the implantation date from ovulation is questionable because it would be biologically impossible for implantation to occur as early as 3 days after the LH surge or 2 days after ovulation. The calculation likely was compromised by the lack of sensitivity at the lower end of the assay for LH and by the fact that the Immulite is not suitable for LH determination in urine. The article indicates that the author validated the Immulite for quantification of hCG in urine, but not of LH in urine.
Based on the inquiry report and recommendations, the Editorial Board concluded that the author committed specific inappropriate acts in publication: (1) failure to disclose the conflict of interest and (2) methodologic flaws and obfuscation.
The notice — which we have to commend for thoroughness, at least to a point, as we’ll make clear — ends with a ditty on transparency in scientific publishing (hear, hear!), and the ills that lack of disclosure can cause to the public trust.
The editors might have stopped themselves there, but they felt compelled to add a few more lines:
Also, it is the responsibility of authors to identify and validate methods that are used in a study and to report with clarity so as not to cause any confusion for readers. The Editorial Board decided to retract the article and to sanction the author appropriately. Although we regret that the article was published before we learned about and investigated the inappropriate acts, the editors and publishers hope that the readers will appreciate that the decision to retract the article reflects the commitment of the Journal to
This all leaves us wondering: Did Cole submit a manuscript that wasn’t worthy of publication, as the retraction notice contends? Did he fail to identify his research methods? Was there truly “no credible scientific reason given for conducting the study”? Were there serious methodological gaps in the paper?
If the answer to any of those questions is “yes,” where were the reviewers? And why did the journal accept the paper in the first place?
There must be something else going on here. Methodologic flaws aren’t good, but many studies are imperfect, and the notice suggests they’re all “inappropriate acts in publication.” They might be errors, certainly, and sloppiness but they don’t rise to the threshold of “acts” — unless there’s more to the “obfuscation” claim that we don’t know.
To that point, what “appropriate sanctions” did Cole receive? The vaguely menacing construction is far from the sort of transparent approach that, say, the American Society of Microbiology chose to take when dealing with Naoki Mori. We’ve tried to contact the editors of the AJOG and will update this post when we learn more.
We will note one more thing about this notice. Although it’s ostensibly from the two top editors of the journal, Thomas Garite and Moon Kim, both at the University of California, Irvine. the heavy-handed language comes from the editorial board. The implicit “we’re the adults in the room” presence of the ed board isn’t exactly a signal of confidence in the stewardship of its main agents on the publication.
Meanwhile, we’re certain there’s more to this story. We tried to reach Cole at his office, but a voice mail message says he is on leave until further notice. The message also states that the hCG lab is closed and won’t be conducting any tests for the same amount of time (although the website says it will reopen Dec. 1).
Update, 6:30 p.m. Eastern, 11/10/11: Commenter Neuroskeptic points out that a PDF of the paper is in fact available on the site. It didn’t seem to be when we checked earlier, but we’re glad to correct this post and the update with strikethroughs.