Ethics: September 2009 Archives

Sometimes even Mark and David Geier can flog only a partial confession out of data.

The father-son duo, along with frequent coauthor Janet Kern, were able to publish yet another study implicating at least some kind of link between mercury (this time from maternal dental fillings) and the severity of childhood autism. But the authors missed on what would generally be construed as the primary endpoint of their study. The negative finding is buried in the latest issue of Acta Neurobiologiae Experimentalis (Warsaw), a journal with an impact factor in the 25th percentile.*

For anyone familiar with the Geiers' previous publications, there is little shock value here. Evidently the authors remain incapable of disinterest. The introduction of the article preferentially cites highly questionable data linking mercury in dental amalgam with various bodily dysfunctions generally and autism spectrum disorder (ASD) specifically. And in consistent form, the Geiers cite their own previously published data to bolster their preformed ideas.**

The study methods are vague and fraught with potential bias: They write, "The present study looked at 100 qualifying participants who were prospectively recruited from patients presenting for outpatient genetic consultations at the Genetic Centers of America." Mark Geier is president of the "Centers," which are evidently located in the basement of his residence in Silver Spring, MD. The study is also described as blinded, although it's not clear who is blinded to what. The subjects, born either from 1990 to 1999 or between 1990 and 1999, were "previously diagnosed" with autism or pervasive development delay (PDD) by a "trained professional." As far as the reader knows, this term could refer to an autism specialist, a general pediatrician, a chiropractor, a naturopath, or a plumber. The study protocol itself is described as being approved by the Institutional Review Board (IRB) of the Institute of Chronic Illnesses, another brainchild of Mark Geier. The makeup and objectivity of this IRB has been called into serious question.

The primary finding of the study was that there was no statistically significant difference between the mean number of maternal dental fillings during pregnancy and the severity of autism. After adjusting for age, sex, race, and geographic residence, mothers of children with autism (defined as "severe") had 4.6 fillings, and mothers of children with ASD (defined as "mild") had 3.1 fillings (P = .0946). (Of course, the arbitrary distinction of severe and mild autism on the basis of the prestudy diagnoses, one of which includes the term "spectrum," is highly problematic.)

Undaunted, Geier et al attempted to define a threshold number of maternal dental fillings that significantly elevated the risk of "severe" autism. After performing a "logistic regression analysis," they concluded that the odds of "severe" (vs "mild") autism more than tripled if the number of maternal dental fillings during pregnancy exceeded 5; however, the risk was only statistically significant with 8 or more fillings. The "threshold" finding is the only conclusion cited in the article's abstract.

The most profound limitation (and there are several...arguably many) of this study is the lack of any data from a control population—namely the mean number of dental fillings in mothers whose children do not have autism.

* For comparison purposes, the NEJM has an impact factor of greater than 99%.

** In the opening paragraph, Geier et al write, "As of mid-2008, the US Food and Drug Administration (FDA) has declined to classify the medical-device safety of amalgams used in dentistry [emphasis added]." Given the article's recent publication date, this statement suggests that Geier et al had been shopping this article around for some time. The FDA issued its final regulation on dental amalgam in July. The mercury component of dental amalgam was reclassified from Class I (low risk) to Class II (moderate risk).

Ghostwriting and honorary authorship—the practice of adding, for example, the name of a laboratory head to lend cachet or credibility to an article—remain common in top medical journals, including the prestigious NEJM and the sanctimonious JAMA. This conclusion is based on recent survey results that were analyzed by JAMA investigators and presented last week at the Sixth International Congress on Peer Review and Biomedical Publication in Vancouver.

After submitting an online survey to the authors of 900 research, review, or editorial articles that were published in the Annals of Internal Medicine, JAMA, The Lancet, Nature Medicine, the NEJM, or PLoS Medicine during 2008, Wislar et al reported the following, preliminary results (author response rate, 70%).

Authorship Type	2008, % of Articles	1996, % of Articles
Honorary	26 (range, 16-39)	19
Ghost	8 (range, 2-11)	11
Both	2	2

When compared with survey results from 1996, these data indicate that the rate of honorary authorship has increased, while the rate of ghost authorship has decreased slightly. (The meeting abstract does not provide data on the statistical significance of these temporal differences.)

The rate of ghostwriting was highest in the NEJM (11%) and lowest in Nature Medicine (2%). The reverse was true for honorary authorship: Nature Medicine, 39%; NEJM, 16%. Ghost authorship was significantly more common with original research articles (12%) than among reviews (6%) or editorials (5%). Honorary authorship was reported frequently with all article types: original research, 31%; reviews, 24%; and editorials, 22%.

The JAMA investigators found no authorship differences between journals that require contribution disclosures and those that do not.

The results of the survey will presumably be published in JAMA at some point in the near future. The anchor author of the abstract, JAMA Editor-in-Chief Catherine DeAngelis, is presumed to be an actual contributing author and not an honorary author.

Two days ago, I posted a somewhat snarky response (although this is a blog) to the NYT's expose of Forest's 2004 marketing plan for Lexapro (which was drafted in 2003). Contained within the company's leaked, abridged marketing plan was a proposal to use "reporters" from selected psychiatry journals, including "CNS News"—which was probably meant to be written as "CNS Spectrums"—to cover Lexapro data at major psychiatry meetings. (CNSNews is a conservative, mainstream, online news source.) The reports would be included as supplements in the journal and provide continuing medical education (CME) credit.

CNS Spectrums is a monthly neuropsychiatric journal published by MBL Communications, Inc, which offers some CME-certified articles in conjunction with its accredited partner, the Mount Sinai School of Medicine. The general reputation of the journal among clinicians is that of a "throwaway" publication. To wit, I used to receive unsolicited, free issues of CNS Spectrums by mail; I'd glance at them and then throw them away. Nevertheless the journal, despite its lackluster reputation, is included in the National Library of Medicine's PubMed database.

So as a follow-up exercise (and because I'm slightly nuts), I examined the articles that made it into the supplemental issues of CNS Spectrums during 2004. The objective was to determine to what extent Forest's Lexapro marketers realized their described plan, at least with respect to this particular tactic. My methods consisted of 1) a PubMed search within the confines of the year 2004 and the search term "CNS Spectr"[Journal]; 2) a directed search of the CNS Spectrums web site; and 3) a review of clinical supplements listed at the CNS Spectrums web site.

Here are the results:

At least by my search, there is no evidence that a Forest-sponsored article made it into a supplement of CNS Spectrums during the 2004 calendar year. In 5 cases, however, the supplements could not be accessed by using the search function provided at the publication's web site** or through the PubMed link (when provided). Nevertheless, in these cases, it is unlikely that Forest sponsored the supplements (ostensibly to promote the antidepressant Lexapro), given the designated topics—for instance, Alzheimer disease, antipsychotic use, or bipolar disorder.

There is also no evidence that any of these supplements were certified for CME, given the absence of designated learning objectives and other ACCME-required language. Although in 2 cases, the supplements were supported by an "unrestricted, educational grant" (both from GlaxoSmithKline). Whether any or all of these 2004 articles, all of which have clinician authors, were ghostwritten by "reporters" is just about anybody's guess. 

2004 Supplement	Topic	Sponsor	CME	Faculty Disclosures
February	Rapid-cycling bipolar disorder (panel discussion)	AstraZeneca	No evidence	No
April	Anxiety disorders (panel review)	UCB Pharma	No evidence	Yes
June	Mood and anxiety disorders (4 articles)	GSK	No evidence*	Yes
July	Alzheimer’s disease (5 articles)	No data**	—	—
August	Bipolar disorder in women (1 article)	GSK	No evidence*
August	Antipsychotic-associated hyperprolactinemia (1 article)	No data**	—	—
August	Anxiety disorders (1 article)	No data**	—	—
September	Antipsychotic use (4 articles)	No data**	—	—
October	Psychosis/schizophrenia (4 articles)	No data**	—	—
November	Bipolar disorder (3 articles based on data presented at the 2004 APA meeting)	None indicated	No evidence	No

* Although funded by an "unrestricted, educational grant."
** An "error" occurred when attempting to access the article at the CNS Spectrums web site. The error message advises contacting the web designer, which has the unfortunate name of spinindustry.com.

A review of listed clinical supplements at the CNS Spectrums web site reveals only 1 that was sponsored by Forest: "Bridging the Clinical Gap: Managing Patients with Co-occurring Mood, Anxiety, and Alcohol Use Disorders." Published in April 2008, the supplement consists of 5 articles (including an introduction), none of which appear to offer CME credit. Faculty disclosures are provided with each article, and "editorial assistance" is acknowledged by name (Eileen McGee, Marsha Kellar, and Joyce Waskelo) and company (Hudson Medical Communications, which is described as a promotional firm at the web site of its parent company). The acknowledgement appears to be an attempt at editorial transparency and to undermine accusations of ghostwriting. But given the suspicion that McGee, Kellar, and Waskelo actually drafted the articles, they should have been defined as coauthors.

In the 5 articles, "escitalopram" (ie, Lexapro) is mentioned exactly 4 times in the text bodies and in conjunction with other antidepressants (eg, citalopram [Celexa], fluoxetine [Prozac], and paroxetine [Paxil]).

ACCME = Accreditation Council for CME; APA = American Psychiatric Association.