Conceived and designed the experiments: JMM. Performed the experiments: JMM. Analyzed the data: JMM. Contributed reagents/materials/analysis tools: JMM. Wrote the paper: JMM.
The author previously received research and/or salary support from the Cannabinoid Research Institute, a research division of GW Pharmaceuticals (
Previous research has shown that academic physicians conflicted by funding from the pharmaceutical industry have corrupted evidence based medicine and helped enlarge the market for drugs. Physicians made pharmaceutical-friendly statements, engaged in disease mongering, and signed biased review articles ghost-authored by corporate employees. This paper tested the hypothesis that bias affects review articles regarding rimonabant, an anti-obesity drug that blocks the central cannabinoid receptor.
A MEDLINE search was performed for rimonabant review articles, limited to articles authored by USA physicians who served as consultants for the company that manufactures rimonabant. Extracted articles were examined for industry-friendly bias, identified by three methods: analysis with a validated instrument for monitoring bias in continuing medical education (CME); analysis for bias defined as statements that ran contrary to external evidence; and a tally of misrepresentations about the endocannabinoid system. Eight review articles were identified, but only three disclosed authors' financial conflicts of interest, despite easily accessible information to the contrary. The Takhar CME bias instrument demonstrated statistically significant bias in all the review articles. Biased statements that were nearly identical reappeared in the articles, including disease mongering, exaggerating rimonabant's efficacy and safety, lack of criticisms regarding rimonabant clinical trials, and speculations about surrogate markers stated as facts. Distinctive and identical misrepresentations regarding the endocannabinoid system also reappeared in articles by different authors.
The findings are characteristic of bias that arises from financial conflicts of interest, and suggestive of ghostwriting by a common author. Resolutions for this scenario are proposed.
The epidemic of obesity began, as many modern epidemics do, with a reclassification. In 1998, the number of overweight and obese individuals in the USA swelled instantaneously by 37 million, when a NIH task force redefined overweight as a body mass index (BMI) ≥25 kg/m2, and obesity as BMI ≥30
In 2004, researchers from the Centers for Disease Control (CDC) reported that obesity caused 400,000 deaths in the year 2000
Two new anti-obesity drugs, rimonabant (Acomplia®, Sanofi-Aventis) and taranabant (Merck), work by a new mechanism: blockade of the cannabinoid 1 (CB1) receptor, an integral part of the endocannabinoid system (ECS)
Four rimonabant-in-obesity (RIO) RCTs, all funded and conducted by Sanofi-Aventis, have been published, although 25 RCTs of rimonabant for the treatment of obesity and diabetes are completed or underway
One taranabant clinical trial has been published in the peer-reviewed literature
In addition to RCTs and meta-analyses, clinicians base rational EBM decisions upon CME presented by fellow physicians. Clinicians must participate in CME to fulfill licensure requirements, making them a “captured audience” for corporate-sponsored messages. Pharmaceutical corporations routinely seed CME with review articles that promote their products, thereby further unraveling EBM
In the past few years, several CME review articles of rimonabant have been published. Some authors of the review articles also served on the NIH obesity task force, the NAASO board, and coauthored RIO publications. One rimonabant review article was presumably ghostwritten because the authors were listed as “editors,” without identifying a primary author, and a Sanofi-Aventis copyright appeared in small print on the back cover of the supplement
Review articles were identified through a MEDLINE search using the keywords endocannabinoid AND obesity AND rimonabant, limited to articles published prior to 2007 (when the FDA reviewed rimonabant). To be included in the analysis, a review article had to meet the following criteria:
ample information (≥2 paragraphs) describing the ECS system and obesity;
ample information (≥1 paragraph) describing obesity and rimonabant;
authored by a USA physician who received financial support from Sanofi-Aventis.
Because the study aimed to uncover identical bias by different authors, only one article by each author was analyzed; the earliest published article was analyzed and subsequent publications were excluded. Evidence of real or potential conflicts of interest (i.e., financial support in the form of honoraria, education support, research funding, or identification as a consultant or speaker) was obtained by searching Google using each physician's name combined with Sanofi-Aventis. Biases and misrepresentations were measured by three methods:
Review articles were analyzed with a validated instrument for monitoring bias in CME
The “RIO bias tally” searched articles for biased statements or inappropriate omissions that originally appeared in the RIO publications (see
Articles were scanned for recurrent themes and for identical misrepresentations about the ECS written by different authors.
A. | Conflict of interest was |
B. | Commercial interest was clearly present ( |
C. | Valid, credible evaluation of peer-reviewed evidence-based medicine (EBM) was |
D. | The author did |
E. | The data were presented in an unbalanced manner, and some outcomes were favored over others (i.e., data were presented that favored one company's products over another's). |
F. | Published sources were identified for evidence reported. |
G. | The data presented in the program were incomplete or framed in a biased fashion. |
H. | Rival drugs for treatment of obesity were not mentioned ( |
I. | Trade names of the drug were used ( |
J. | If unapproved uses of drugs were discussed, the author informed the audience of this according to current guidelines. |
K. | The paper does not contribute to the best interests of patients. |
L. | The paper promotes marketing of drug knowledge. |
M. | This program enhances medical knowledge. |
1. | Disease mongering: the ECS requires pharmacological blockade because it induces detrimental effects: overfeeding, obesity, diabetes, hyperlipidemia, and/or hepatic steatosis. |
2. | Weight reduction from rimonabant was described as “appreciable,” ”large,” “dramatic,” etc. |
3. | Rimonabant's reduction of high-density lipoprotein (HDL) cholesterol was highlighted, while no mention was made of its inability to lower total cholesterol or LDL. |
4. | Adverse effects were not mentioned or were described as “mild,” “transient,” “well tolerated,” or “slightly greater than placebo.” |
5. | The external validity (or generalizability) of the RIO trials went unquestioned. |
6. | Methodological weaknesses (internal validity) in RIO trials went unmentioned; or if high drop-out rates were mentioned, they were justified as “typical of obesity trials.” |
7. | Relative contraindications other than depression were not mentioned. |
8. | Inappropriate surrogate markers went unquestioned. BMI of ≥25 was used as a surrogate marker for adiposity and an accurate predictor of mortality. |
9. | Competing drugs were not mentioned or mentioned only in a way that highlighted adverse effects. |
10. | No mention was made of rimonabant potentially counteracting drugs or other therapeutic interventions that augment the ECS. |
Eight review articles were identified that met inclusion criteria
The Takhar CME bias instrument demonstrated bias in all the review articles (
Takhar bias scale item | Eight review articles (citation numbers from Reference section) | RIO bias tally item | |||||||
A | 4 / 1 | 3 / 2 | 4 / 2 | 1 / 2 | 4 / 2 | 4 / 2 | 4 / 2 | 1 / 2 | 1 |
B | 3 / 2 | 3 / 2 | 3 / 2 | 4 / 2 | 3 / 2 | 3 / 2 | 4 / 2 | 4 / 2 | 2 |
C | 3 / N/A | 3 / 2 | 3 / 2 | 3 / 2 | 3 / 2 | 3 / 0 | 3 / 2 | 3 / 2 | 3 |
D | 3 / 2 | 2 / 1 | 3 / 1 | 4 / 2 | 2 / 0 | 1 / 1 | 4 / 0 | 3 / 1 | 4 |
E | 4 / 2 | 2 / 2 | 3 / 2 | 4 / 2 | 3 / 2 | 2 / 2 | 4 / 2 | 3 / 2 | 5 |
F | 2 / 2 | 2 / 2 | 2 / 2 | 3 / 2 | 2 / 2 | 2 / 2 | 2 / 2 | 2 / 2 | 6 |
G | 4 / 2 | 3 / 2 | 3 / 2 | 4 / 2 | 4 / 2 | 4 / 2 | 4 / 2 | 4 / 2 | 7 |
H | 4 / 1 | 2 / 2 | 2 / 2 | 4 / 2 | 4 / 2 | 2 / 2 | 4 / 2 | 3 / 2 | 8 |
I | 3 / 2 | 3 / 1 | 3 / 0 | 3 / 2 | 3 / 2 | 3 / 0 | 3 / 2 | 3 / 1 | 9 |
J | N/A | N/A | N/A | N/A | N/A | N/A | N/A | N/A | 10 |
/ 2 | / 2 | / 2 | / 2 | / 2 | / 2 | / 2 | / 2 | ||
K | 2 | 2 | 2 | 3 | 2 | 2 | 3 | 2 | |
L | 2 | 3 | 4 | 4 | 3 | 4 | 4 | 4 | |
M | 2 | 3 | 3 | 4 | 3 | 3 | 4 | 3 | |
Takhar mean | 3.0 | 2.6 | 2.9 | 3.4 | 3.0 | 2.8 | 3.6 | 2.9 | |
1.78 | 1.8 | 1.7 | 2.0 | 1.8 | 1.5 | 1.8 | 1.8 | RIO bias mean |
Each article was scored with the Takhar bias instrument (items A to M in the first column, see
Recurrent visual and textual themes emerged from the eight review articles. Graphics from Sanofi-Aventis promotional materials
All the authors of rimonabant review articles held academic positions, many at prestigious institutions. They typified “medical opinion leaders” sought by pharmaceutical corporations to sign ghostwritten articles
The RIO bias tally identified ten nearly-identical industry-friendly statements or inappropriate omissions in articles written by different authors (
Rationally choosing the best medication, like other sorts of clinical decision-making, has increasingly relied upon EBM. Thus whoever generates EBM, by funding RCTs, meta-analyses, and CME, may bias clinical decision-making regarding pharmaceuticals
Financial conflicts of interest also bias clinical practice guidelines and FDA decisions. An analysis of 44 clinical guidelines revealed 87% of panelists received financial support from pharmaceutical companies, yet only two of the guidelines disclosed panelists' financial conflicts
In summary, financial conflicts permeate the system and are by no means limited to corporations referenced in this article, such as Merck, Parke-Davis, Pfizer, Sanofi-Aventis, and Wyeth-Ayerst. On balance, pharmaceutical corporations do good work and aid in humanitarian efforts. For example Sanofi-Aventis provides artemisinin at cost to malaria-endemic countries
While this paper was under review, Merck halted taranabant RCTs, and Sanofi-Aventis removed rimonabant from the European market. The FDA rejected rimonabant after data submitted by Sanofi-Aventis revealed adverse effects in RIO trials that went unreported in RIO publications