Objective: The literature continues to highlight the debate on the ethics and merits of trials sponsored by the pharmaceutical industry. This study attempts to determine the prevalence and outcomes of industry-sponsored trials involving clozapine, risperidone, or olanzapine.
Methods: We searched the literature from January 1, 1990, to December 31, 2001, to capture all eligible clinical trials involving clozapine, risperidone, or olanzapine. The primary outcome measured was the clinical outcome of industry-sponsored studies. Secondary outcome measures included the following parameters: disclosure of any sponsorship and financial support, author(s) employed by the industry, use of comparator drug(s) within the trial, sample size, blinding, and use of placebo.
Results: The database comprised 372 articles. Of these trials, 124 (33.3%) were sponsored by the pharmaceutical industry. In general, trials sponsored by Eli Lilly or Janssen had better research design than trials not funded by the pharmaceutical industry. With regard to authorship, more trials funded by Eli Lilly (74.6%) were coauthored by an employee of the company, compared with trials funded by either Janssen (23.3%) or Novartis/Sandoz (5.6%). Further, more trials sponsored by Eli Lilly reported positive outcomes (92.1%), compared with Janssen-sponsored trials (88.4%) and Sandoz/Novartis-sponsored trials (72.2%). No negative results were reported in any of the industry-funded trials.
Conclusions: One-third of the published clinical trials involving clozapine, risperidone, or olanzapine were funded by their respective manufacturer. The reported outcomes of the sponsored trials highly favour the manufacturer's product.
(Can J Psychiatry 2004;49:601-606)
Information on funding and support and author affiliations appears at the end of the article.
Clinical Implications
* More than ever, clinicians are developing partnerships with the pharmaceutical industry for the purpose of participating in clinical trials.
* Trials sponsored by the pharmaceutical industry are often of higher quality than publicly funded studies.
* Outcomes of pharmaceutical industry-sponsored trials highly favour their own product.
Limitations
* Although every effort was made to obtain all articles from the literature search, this was not possible.
* It is possible that disclosure of funding and conflicts of interest were underreported.
Key Words: clinical trials, conflict of interest, research support
Clinical trials are widely accepted as invaluable tools providing evidence-based guidance to practising clinicians. Ideally, when clinical trials are designed, investigators should pose scientific questions that primarily intend to make available better and safer treatments for a target population. When properly conducted, these clinical trials become the cornerstones for evidence-based medicine, which is the standard of care in most medical practices. Evidence-based medicine has been defined as "the conscientious, explicit and judicious use of current best evidence in making decisions about the care of individual patients" (1). Within this context, best evidence is derived from clinical trials (that is, external evidence) that have been critically appraised for validity and importance.
Since the introduction of the atypical antipsychotics into the clinical practice of psychiatry, clinical trials have increased in number, size, and of course, cost. With limited private funding available to conduct independent studies, it is not surprising that the number of clinical trials sponsored by the pharmaceutical industry continues to grow. It follows that there should also be an increase in the number of partnerships developed between the pharmaceutical industry and clinicians or academics. Although skepticism in regard to this alliance has been and continues to be expressed, the many benefits of the new reality have also been acknowledged. This study investigates the prevalence and outcomes of pharmaceutical industry-sponsored clinical trials involving the atypical antipsychotics clozapine, risperidone, and olanzapine.
Methods
We initially searched the literature on all 3 of these antipsychotics by subject and key word between January 1, 1990, and December 31, 2001. This covers the period from clozapine's introduction into the North American market (1990) to the time of our study (2002). Next, we exploded the phrase "schizophrenia and disorders with psychotic features" in Medline. This search was then limited to the different types of clinical trials (that is, phase I, phase II, phase III, or phase IV; controlled; multicentre; and randomized). We then set a final limit by human and English language and compiled all trials found within the search. We excluded review articles, letters, metaanalyses, and irrelevant studies that did not directly involve clozapine, risperidone, or olanzapine in the study protocol.
All journal articles were assigned a unique identifier code. A total of 614 articles were captured from 71 journals. Of these 614 articles, 414 met the criteria for the study. We obtained journal articles not available in our library through DOCLINE, the National Library of Medicine's automated interlibrary loan request routing and referral system. DOCLINE serves over 3200 US, Canadian, and Mexican medical libraries at no cost. Unfortunately, we were unable to retrieve 42 of the eligible trials via DOCLlNE. This left us with 372 trials to analyze.
The primary outcome measured was the clinical outcome of industry-sponsored studies. We reviewed and rated these sponsored articles with the use of a classification system that we derived a priori (Table 1). We classified the outcome of each study as positive, neutral, or negative with respect to the study drug of the sponsoring company. secondary outcome measures included the following parameters: disclosure of any sponsorship and financial support, authors employed by the industry, use of comparator drugs within the trial, sample size, blinding, and use of placebo.
To realize the objective of this study, we are justified in employing a qualitative approach that tends to be synthetic rather than analytic. Unlike deductive research, it does not start with preconceived notions or hypotheses. The intent is simply to determine the prevalence and outcomes of pharmaceutical industry-sponsored clinical trials. As such, we have refrained from doing statistical analyses, because we do not believe that there is any benefit in depicting significant differences among the various companies and (or) drugs.
Results
Of the 372 eligible (and obtainable) trials, 248 (66.7%) were conducted without financial assistance from the pharmaceutical industry (Table 2). Of the remaining trials, 63 (16.9%) were sponsored by Eli Lilly, 43 (11.6%) by Janssen, and 18 (4.8%) by Sandoz/Novartis. The cumulative sample sizes were 784 for the Sandoz/Novartis-sponsored trials, 6394 for the Janssen-sponsored trials, 37 365 for the Eli Lilly-sponsored trials, and 26 692 for the nonindustry-sponsored trials. Figure 1 gives the years in which these trials were published.
Our study found some differences with respect to study design. For example, some form of blinding in the protocol was found more often in trials sponsored by Eli Lilly (69.8%). Next were Janssen-sponsored trials (46.5%), nonindustry sponsored trials (28.6%), and Sandoz/Novartis-sponsored trials (22.2%) (Table 2). To give another example, a greater proportion of the Janssen- and Eli Lilly-sponsored trials included a placebo control in their design (30.2% and 27.0%, respectively), compared with nonindustry-funded trials (15.7%) or trials funded by Sandoz/Novartis (5.6%). Finally, the use of another atypical agent as a comparator was most often found in the Eli Lilly-sponsored trials (28.6%), followed by Janssen-sponsored trials (14.0%), nonindustry-sponsored trials (12.5%), and Sandoz/Novartis-sponsored trials (5.6%).
Apart from elements of research design, we also examined authorship and study outcomes. In this regard, trials sponsored by Eli Lilly were more likely to have authors affiliated with the industry (74.6%), compared with trials sponsored by either Janssen or Sandoz/Novartis (23.3% and 5.6% respectively). Further, trials sponsored by Eli Lilly reported more positive outcomes (92.1%), compared with Janssen-sponsored trials (88.4%) and Sandoz/Novartis-sponsored trials (72.2%). No negative results were reported in any of the industry-funded trials.
As mentioned in the methods section, we found the trials represented in the data set within 71 different journals. Since it is not practical (nor of any benefit) to list the number of trials published in each journal, we decided only to cite the journals that published more than 10 trials (Table 3). We found the most sponsored trials in The Journal of Clinical Psychiatry (n = 27), followed by The American Journal of Psychiatry (n = 17), and the Journal of Clinical Psychopharmacology (n = 17). Similarly, we also found the greatest number of nonsponsored trials in The Journal of Clinical Psychiatry (n = 26), followed by The American Journal of Psychiatry (n = 20) and Psychopharmacology (n = 19).
Discussion
With its primary responsibility being to generate profits for shareholders, is it possible for the pharmaceutical industry to design clinical trials objectively, let alone coordinate them and analyze the resulting data? This question has been debated intensely and remains a highly contentious issue. Without question, it is essential for the reader of any trial to know the source of its funding. However, equally important is knowing whether any conflicts of interest exist among the investigators and authors of the trial. In this regard, conflict of interest includes any financial or personal relationships with other persons or organizations that could inappropriately influence or bias their actions (2). Identified conflicts of interest include grant support, consulting to companies, membership on advisory boards, speaking engagements, gifts, samples, patent and royalty arrangements, authorship of ghostwritten manuscripts, conference travel, funding of continuing medical education, honoraria, and equity interest in a company. Although full disclosure of these types of relationships between investigators and the industry appears justified and essential, it is not always mandatory. Some even suggest that disclosure of any conflict of interest should be extended to include the author's institution, the reviewers of the study, and even the publishing editor (2). Failure to disclose any conflict of interest, such as financial support, would breach the publication requirements of many journals. Although our study was not designed to determine whether breaches occurred with respect to disclosing financial support, we were suspicious in some cases. For example, no funding was disclosed in 6 trials wherein an author was affiliated with a particular company.
As do other studies (3-5), our data indicate that trials sponsored by the pharmaceutical industry are often of higher quality than their publicly funded counterparts. In this regard, our data revealed that trials sponsored by Janssen and Eli Lilly were more likely to be blinded, to have a placebo control, and to have a larger sample size (data on file) than trials that were not sponsored by the pharmaceutical industry. Although these elements of study design are intended to reduce bias and increase the validity of the results, it is also recognized that improper use of a placebo control can increase the probability of producing results favouring the intervention. Halpern and Karlawish state that the industry may prefer to use a placebo control because such studies are frequently less expensive (requiring smaller sample sizes) and because regulatory agencies imply that interventions compared with placebo may be more favourably reviewed (6). The relatively low proportion of blinded studies sponsored by Sandoz/Novartis may be explained by the fact that clozapine was introduced as a treatment only for those patients who had not responded to prior therapy. For this reason, using placebo control subjects was not deemed necessary (or ethical), and patients likely served as their own control subjects. Obviously, the ethical issue of placebo use is beyond the scope of this manuscript; it has been addressed elsewhere (7).
Our data indicate that a relatively high proportion of industry-sponsored trials have outcomes that favour their own manufactured and marketed product. Eli Lilly shows the highest proportion of trials with positive outcomes (92.1%), followed by Janssen and Sandoz/Novartis (88.4% and 72.2%, respectively). These findings are consistent with the findings of other investigators who have also reported a particularly high proportion of industry-sponsored research reporting positive results (8-10). Perhaps explaining this observation in part is the fact that clinicians participating in industry-sponsored trials are in most instances under contractual obligations that may limit their access to and control of the data. Often, contracts between the industry and study clinicians contain a clause stipulating that the industry has the final decision as to whether the data will be submitted for publication (11). Regrettably, this arrangement has resulted in the delay and (or) the suppression of valuable clinical data, the extent of which we may never know (12-14). This calls into question the validity of evidenced-based treatment decisions that may be based on an incomplete record of published data. Clearly, it is not desirable that the sponsoring agency completely control the data, along with its analysis and interpretation.
The question of clinical equipoise also deserves comment. By definition, clinical equipoise involves an a priori state of uncertainty or disagreement among experts as to which of 2 or more interventions should be favoured for a particular population (15,16). In other words, to be faithful to the ethical doctrine of equipoise, a randomized controlled trial can only enrol subjects if there is considerable uncertainty about which of the trial treatments would benefit a patient most. Consequently, if true clinical equipoise exists, the outcomes of trials comparing an intervention and standard treatment would be 50/50. Unfortunately, owing to the a priori nature of clinical equipoise, it is extremely difficult (if not impossible) to retrospectively acertain whether this principle was upheld in the trials we assessed. Nonetheless, with plenty of data demonstrating the clinical advantages of the atypical antipsychotics over haloperidol (considering benefit and harm), one may query the a priori state of uncertainty of more recent study designs that continue to include haloperidol as an intervention supposedly representing standard of care. However, it is perhaps ironie that a recent study reported that olanzapine did not demonstrate clinically significant advantages, compared with haloperidol(17).
The results of our study need to be interpreted within the context of its limitations. First, we suspect that the disclosure of funding and the disclosure of authors' industry affiliations is underreported. second, most of the articles were rated by only a single investigator, and hence, bias is possible with respect to assessing the outcomes of the trials. However, a second, and sometimes a third, investigator reviewed the article whenever the trial outcome was questionable. Finally, even though every effort was made to obtain all articles identified from the literature search, this was not possible. As noted above, of the 414 eligible articles, we were unable to acquire 42 (10.1%) through DOCLINE.
Our intent in this paper is not to question the ethics of the pharmaceutical industry, nor is it to assume the integrity of the clinicians and academics who participate in clinical research. With reference to the latter, it is easy to list various factors that may lead researchers into unethical circumstances. These include ambition, financial gains, peer pressure, career opportunities, and the need to publish. Further, biases, errors, misunderstanding, and fraud have all been described in various clinical trials settings (18,19). On this point, it should also be noted that publicly funded studies are not exempt from biases perceived as unique to industry-funded trials (20). No doubt, the pharmaceutical industry will continue to influence not only the process of clinical science but also the practice of clinical medicine. Although skepticism exists regarding collaborative relationships between clinicians and industry, the fact cannot be ignored that this alliance or partnership has also contributed immense and immeasurable benefits to practitioners and patients alike. Let us not lose sight of the fact that industry-sponsored research has provided new classes of compounds which have significantly improved the survival rate and quality of life of countless individuals. Notwithstanding this, clinicians, academics, and the pharmaceutical industry are responsible for safeguarding and preserving the scientific integrity of medical practice by focusing on the patient's best interest.
R�sum� : La pr�valence et les r�sultats de l'industrie pharmaceutique - essais cliniques commandit�s utilisant la clozapine, la risp�ridone ou l'olanzapine
Objectif : La documentation pr�sente toujours le d�bat sur l'�thique et les m�rites des essais commandit�s par l'industrie pharmaceutique. Cette �tude tente de d�terminer la pr�valence et les r�sultats des essais commandit�s par l'industrie qui utilisent la clozapine, la risp�ridone ou l'olanzapine.
M�thodes : Nous avons recherch� la documentation du 1^sup er^ janvier 1990 au 31 d�cembre 2001 pour rep�rer tous les essais cliniques admissibles qui utilisaient la clozapine, la risp�ridone ou l'olanzapine. Le premier r�sultat mesur� �tait le r�sultat clinique des essais commandit�s par l'industrie. Les r�sultats secondaires mesur�s incluaient les param�tres suivants : la divulgation de tout commanditaire et appui financier, le ou les auteurs employ�s par l'industrie, l'utilisation de m�dicament(s) comparateur(s) durant l'essai, la taille de l'�chantillon, le double aveugle et l'utilisation d'un placebo.
R�sultats : La base de donn�es comportait 372 articles. Parmi ces derniers, 124 (33,3 %) �taient commandit�s par l'industrie pharmaceutique. En g�n�ral, les essais commandit�s par Eli Lilly ou Janssen avaient une meilleure m�thodologie que les essais non commandit�s par l'industrie pharmaceutique. En ce qui concerne les auteurs, davantage d'essais commandit�s par Eli Lilly (74,6 %) avaient un employ� de la compagnie comme co-auteur, comparativement aux essais commandit�s soit par Janssen (23,3 %) soit par Novartis/Sandoz (5,6 %). En outre, davantage d'essais commandit�s par Eli Lilly d�claraient des r�sultats positifs (92,1 %), comparativement aux essais commandit�s par Janssen (88,4 %) et Novartis/Sandoz (72,2 %). Aucun essai commandit� par l'industrie ne d�clarait un r�sultat n�gatif.
Conclusions : Un tiers des essais cliniques publi�s utilisant la clozapine, la risp�ridone ou l'olanzapine �taient financ�s par leur fabricant respectif. Les r�sultats d�clar�s des essais commandit�s sont extr�mement en faveur du produit du fabricant.
[Reference]
References
1. Sackctl DL, Rosenberg WMC, Gray JAM, Haynes RB, Richardson WS. Evidence based medicine: what it is and what it isn't. BMJ 1996;71-2.
2. Davidoff F, DeAngelis CD, Drazen JM, Hoey J, Hojgaard L, Horton R, and others. Sponsorship, authorship and accountability. JAMA 2001; 165:786-8.
3. Kaitin KI, Bryant NR, Lasagna L. The role of the research-based pharmaceutical industry in medical progress in the United States. J Clin Pharmacol 1993;33:412-7.
4. Kjaergard LL, Nikolova D, Gluud C. Randomized clinical trials in hepatology; predictors of quality. Hepatology 1999;30:1134-8.
5. Djulbegovic B, Lacevic M, Cantor A, Fields KK., Bennett CL, Adams JR, and others. The uncertainty principle and industry-sponsored research. Lancet 2000;356:365-8.
6. Halpern SD, Karlawish JH. Industry-sponsored research. Lancet 2000;356:2193.
7. Rolhman KJ, Michels KB. The continuing unethical use of placebo controls. N Engl J Med 1994;331:394-8.
8. Davidson RA. Source of funding and outcome of clinical trials. J Gen Intern Med 1986;1:155-8.
9. Cho MK, Bero LA. The quality of drug studies published in symposium proceedings. Ann Intern Med 1996; 124:485-9.
10. Bodenheimer T. Uneasy alliance-clinical investigators and the pharmaceutical industry. N Engl J Med 2000;342:1539-44.
11. Anonymous. The tightening grip of big pharma [editorial]. Lancet 2001;357:1141.
12. Blumenthal D, Campbell EG, Anderson MS, Causino N, Louis KS. Withholding research results in academic life sciences: evidence from a national survey of faculty. JAMA 1997;277:1224-8.
13. Luritsen K, Havelund T, Laursen LS, Rak-Madsen J. Witholding unfavourable results in drug company sponsored clinical trials. Lancet 1987;1:1091.
14. Vogel G. Publishing sensitvie data: who calls the shots? Long-suppressed study finally sees light of day. Science 1997;276:525-6.
15. Freedman B. Equipoise and the ethics of clinical research. N Engl J Med 1987;317:141-5.
16. Lilford RJ, Jackson J. Equipoise and the ethics of randomization. J R Soc Med 1995;88:552-9.
17. Rosenheck R, Perlick D, Bingham S, Liu-Mares W, Collins J, Warren S, and others. Effectiveness and cost of olanzapine and haloperidol in the treatment of schizophrenia: a randomized controlled trial. JAMA 2003;290:2693-702.
18. DeMets DL. Distinctions between fraud, bias, errors, misunderstanding, and incompetience. Control Clin Trials 1997;18:637-50.
19. Hagmann M. Cancer researcher sacked for alleged fraud. Science 2000;287:1901-2.
20. Dicersin K, Min YI. NIH clinical trials and publication bias. Online J Curr Clin Trials 1993. Doc 50.
[Author Affiliation]
Ric M Procyshyn, PhD1, Anthony Chau, BScPharm2, Patricia Fortin, MLS3, Willough Jenkins2
[Author Affiliation]
Funding and Support
Dr Procyshyn has been on advisory boards for Janssen-Ortho Pharmaceuticals, Eli Lilly, GlaxoSmithKline, and Astra Zeneca. He has received honoraria from Janssen-Ortho Pharmaceuticals, Eli Lilly, GlaxoSmithKline, and Astra Zeneca for providing continuing education lectures. He has received grant support from Janssen-Ortho Pharmaceuticals and Eli Lilly. He has received funding for conference travel from Janssen-Ortho Pharmaceuticals and Eli Lilly.
[Author Affiliation]
Manuscript received October 2003, revised, and accepted January 2004.
1 Research Psychopharmacologist, Division of Research, Riverview Hospital, Port Coquitlam, British Columbia; Adjunct Professor, Faculty of Pharmaceutical Sciences, The University of British Columbia, Vancouver, British Columbia.
2 Pharmacy Student, Faculty of Pharmaceutical Sciences, The University of British Columbia, Vancouver, British Columbia.
3 MSc Candidate, School of Health Information Science, University of Victoria, Victoria, British Columbia.
Address for correspondence: Dr RM Procyshyn, 422 Henry Esson Young Building, Riverview Hospital, 500 Lougheed Highway, Port Coquitlam, BC V3C4J2
e-mail: rprocyshyn@bcmhs.bc.ca
