The standard biomedical paradigm for drug treatment of disease follows a straightforward algorithm. A disease is identified and diagnostic criteria are developed. Promising medications are tested using the scientific method, especially randomized controlled trials (RCTs), to determine safety and efficacy. The results of these studies are presented to the scientific community in peerreviewed journals and scientific debate occurs.
If a drug is approved by the FDA, the results of the trials and the conflicting scientific opinions are presented to physicians. Practicing psychiatrists must evaluate the results for their own patients and decide whether the benefits of a medication outweigh its shortcomings.
In this article, we discuss evidence that the pharmaceutical industry influences this paradigm in ways that encourage sales of their products and that are not always in the best interests of the patient.
Expanding the indications: new diseases
An investigative article in the British Medical Journal by Moynihan and associates1 describes how drug companies developed an explicit campaign to promote shyness as an important and treatable disorder, successfully shaping medical and public opinion about social phobia. A senior pharmaceutical industry executive involved in the promotion of social phobia in Australia acknowledged that marketing efforts exaggerated the prevalence of disease, and recognized the inherent conflict of interest when a drug company sponsors publicity about a disease for which it markets a treatment, as his company did with social phobia.2
An article in the Pharmaceutical Executive (a trade journal) discusses generally the making of a new disease, and how new markets can be created by defining a new disease (such as the metabolic syndrome).3 Another way to expand indications for a drug (and thus the market) is to lower the threshold of normal, rendering more people ill and in need of medication. Some critics think that pharmaceutical companies have succeeded in medicalizing risk factors or normal experiences, thus making otherwise healthy people think they need certain drugs.4
Pharmaceutical companies have successfully expanded the FDAapproved uses of various selective serotonin reuptake inhibitors (SSRIs) to include a variety of psychiatric diseases, such as generalized anxiety disorder, obsessive-compulsive disorder, premenstrual dysphoric disorder, posttraumatic stress disorder, and bulimia nervosa.5 Trials for many offlabel uses such as premature ejaculation, migraine headache prophylaxis, and neurocardiogenic syncope are conducted and offlabel uses are often suggested.6 Drug company efforts to get patients on treatment have been motivated by a variety of reasons. Although there is no doubt that some patients have benefited from the promotion of treatment for these syndromes, drug company sales have certainly increased.
In Australia, where direct-toconsumer advertising (DTCA) is not permitted, the value of campaigns promoting diseases can be clearly seen. At the same time that Merck’s hair growth drug finasteride(Drug information on finasteride) was approved in Australia, a public relations firm hired by the manufacturer launched a vigorous campaign about the dangers and tragedy of baldness. Eventually, newspapers published articles about the emotional trauma associated with hair loss, including expert opinions that losing hair could lead to emotional problems, harm to mental well-being, and loss of job prospects. These articles did not reveal that the studies they cited were funded by Merck, and that the experts quoted had been supplied by the company’s public relations firm.1
Industry funding in the peerreviewed published literature
For many practicing psychiatrists, the results of peerreviewed, published RCTs are an essential element in making treatment decisions about drug therapy. A systematic review of all clinical trials published in 4 leading, peerreviewed psychiatry journals from 2001 to 2003 found that pharmaceutical sponsorship was common and was statistically associated with outcomes that favored the sponsor.7 Of the 162 RCTs published during this time, 70% were funded by pharmaceutical companies and 46% had at least 1 author who had a financial conflict of interest either through employment or by having received direct financial support from a pharmaceutical company. Studies with at least 1 author with a conflict of interest were about 5 times more likely to report positive results than were nonindustry sponsored studies.7
This article can only point to a strong correlation between industrysponsored studies and the reporting of positive results but cannot establish causation. A summary of the types of design and reporting modifications that have been used in industrysponsored psychopharmacology trials is presented in the Table.8
Clinical trials research on atypical or secondgeneration antipsychotics (SGAs) provides an example of this association between a trial’s sponsorship and its outcome.9 Of 59 RCTs studying SGAs, those sponsored by the vendors of SGAs were significantly more likely to find them superior to firstgeneration antipsychotics than were nonindustry-funded RCTs (P = .02). Furthermore, studies with first authors employed by the industry showed a strong trend (P = .05) toward favoring SGAs over conventional antipsychotics, compared with articles whose first authors were independent of industry.
Could the industry-sponsored trials simply have been bigger, better, and more able to identify a true superiority of the sponsored drug? Two lines of evidence argue against this possibility. First, the researchers who reported this difference in outcome found a trend (P = .07) toward higher overall RCT quality score in the studies that were not industry funded. The second line of evidence is indirect. A review by Hogan and associates10 of studies comparing one cholinesterase inhibitor with another for Alzheimer disease (AD) concluded that, All studies were funded by pharmaceutical companies, coauthored by their employees, and reported results that favoured the sponsor’s product. Similarly, industry-funded studies that compared 2 long-acting inhaled corticosteroids and found a difference between them, found evidence in favor of the sponsor’s product in 37 of 38 of the peer-reviewed published RCTs.11
Summary of examples of design and reporting modifications in industry-sponsored trials8
|Technique used||Examples from the literature|
|Using doses outside the usual range||Haloperidol at a dose of 20 mg was the comparator in studies of EPS with second-generation antipsychotics30|
|Altering the dosing schedule||Twice daily amitriptyline(Drug information on amitriptyline) was the comparator in evaluation of daytime sleepiness (vs paroxetine(Drug information on paroxetine))31|
|Using self-serving measurement scales||“Worst EPS score” rather than total or last observed EPS score was used to show risperidone(Drug information on risperidone) causes the same amount of EPS as placebo30|
|Selecting major findings and end points post hoc (secondary analyses or “data dredging”)||Although not the initial hypothesis of the trial, side-effect differences at 1 week were used to compare venlafaxine and fluoxetine(Drug information on fluoxetine)32|
|Masking unfavorable side effects||Specific questions about sexual side effects of SSRIs elicit a 73% positive response; change to unstructured open-ended questions drops the rate to 2%33,34|
|Repeatedly publishing the same or similar positive studies||Multiple publication of the same study (stand-alone or as pooled data) is discussed in the text and in an editorial35|
|Selectively highlighting findings||Ignoring placebo as being as efficacious as the drugs (and having fewer side effects) in the comparisons of 2 antidepressants for anxiety symptoms36|
|Editorializing for the sponsor in abstracts and conclusions||The study drug heralded as the “new first-line treatment” when data do not support this conclusion37|
|Publishing the obvious to emphasize a point for marketing||Publications of the risks of nortriptyline(Drug information on nortriptyline) in older adults with heart disease in comparative trials versus various SSRIs38,39; these publications add little to medical knowledge but are of great use for marketing|
|Touting nonsignificant but favorable differences and negating dropout differences statistically||A conclusion that paroxetine was better tolerated than clomipramine(Drug information on clomipramine) was based on insignificant differences40|
|Selection of participants and trial duration||Short trial duration of 8 weeks was used to compare olanzapine(Drug information on olanzapine) or risperidone versus chlorpromazine(Drug information on chlorpromazine) in treatment-resistant patients; chlorpromazine typically shows effects over a longer time frame41|
|Withholding unfavorable results||Withholding unfavorable results|
|Masking sponsorship||Ghostwriting is discussed in the text; disclosure of conflict of interest is now more commonly required|
EPS, extrapyramidal symptoms; SSRIs, selective serotonin reuptake inhibitors.