The Financing of Drug Trials by Pharmaceutical Companies and Its Consequences: Part 1. A Qualitative, Systematic Review of the Literature on Possible Influences on the Findings, Protocols, and Quality of Drug Trials

Clinical drug trials funded by pharmaceutical companies yield favorable results for the sponsor’s products more often than independent trials do. This has been demonstrated by a number of studies in recent years (1–4). Various ways have been described in which pharmaceutical concerns exert influence on the protocol and conduct of drug trials, as well as on the interpretation and publication of their results (1, 3, 5, 6).

Two important reviews on this topic were published in 2003 (7, 8):

Bekelman et al. carried out a quantitative analysis of 37 studies to investigate the extent and the relevance of financial links between biomedical companies on one hand and academic institutions and scientists on the other (7). This systematic review showed widespread conflicts of interest in the shape of financial connections between scientists, academic institutions, and the pharmaceutical industry. Around one quarter of academic staff and two thirds of academic institutions had financial relationships with industry. Analysis of 8 review articles embracing a total of 1140 original articles (including randomized controlled trials [RCT], economic analyses, and retrospective cohort studies) revealed a statistically significant association between funding by biomedical companies and conclusions favorable to the pharmaceutical industry (summarized odds ratio [OR] 3.6, 95% confidence interval [CI] 2.6–4.9). Industry financing was also connected with limitations of publication rights and constraints on access to trial data.

In the second review, a systematic analysis of 30 publications, Lexchin et al. showed that drug trials financed by pharmaceutical companies are less likely to be published, but that those published more frequently yield positive results for the sponsors’ products than do independently funded studies (8). The quality of the methods employed (analyzed in 13 publications) in trials financed by pharmaceutical companies was not inferior to that in studies with other sources of funding.

Influence on many different aspects of a clinical drug trial can affect the results, from the purpose via the planning and conduct of the study all the way to its evaluation and publication (5).

The authors of the present systematic review set out to assess whether recently published studies reveal a connection between financing of drug trials by pharmaceutical companies and results favorable to these companies’ products. Part 1 investigates whether and, if so, how the type of funding affects study protocol and quality. Part 2 identifies and depicts the aspects of clinical drug trials that can be influenced by financial support from the pharmaceutical industry.

Methods
Systematic literature review
The criteria for the selection of systematic investigations into the effects of financing on study results were oriented on the above-mentioned reviews (7, 8). The literature search was carried out in the PubMed database. The most important MeSH terms and search terms, combined by means of Boolean operators, were “conflict of interest”, research support as topic”, clinical trials as topic”, “commerce”, “drug industry”, “authorship”, “publication bias”, and “financial support”.

The following limits, among others, were imposed: “publication date”, “clinical trial”, “meta-analysis”, and “randomized controlled trial”.

Searches were carried out on 26 May 2008 (for the period 1 November 2002 to 26 May 2008), 4 December 2008 (for 26 May 2008 to 3 December 2008), and 16 December 2009 (for 1 October 2008 to 16 December 2009).
Publications in all languages were included. Two of the authors (G. Schott, U. Limbach) independently selected suitable studies on the basis of the inclusion and exclusion criteria (see below). In the case of disagreement, consensus was achieved by discussion with a third author (K. Lieb). The reference lists of all selected articles were inspected for further relevant publications (9).

Inclusion and exclusion criteria
The criteria for inclusion were:

• Publication between 1 November 2002 and 16 December 2009 (i.e., directly following on from the period analyzed by Bekelman et al. [7])
• Description of the methods employed
• Provision of empirical data as to whether and how financing by pharmaceutical companies affects various aspects of clinical trials (e.g., protocol, conduct, results, conclusions, or publication).

Commentaries, editorials, and abstracts were excluded.

Some studies were found to combine clinical drug trials and analyses of, for example, surgical interventions, medical devices, or preclinical investigations. In such cases the study was included as long as the results of clinical drug testing were described separately.

Evaluation of the literature
The basic characteristics, purpose, and principal findings of each of the studies reviewed were recorded (eTable gif ppt). The quality of the studies’ methods was not systematically evaluated, because there is no validated, reliable instrument for quality assessment that can be applied to all of the various study designs (including systematic reviews, meta-analyses, case studies, and cross-sectional studies).

In the effort to ensure comprehensive portrayal of the various ways in which influence could be exerted, a descriptive approach was chosen; no hypotheses were advanced or statistical analyses performed. The studies were arranged by topic according to their results.

Results
The search of the PubMed database yielded 1705 publications. After inspection of reference lists, 57 publications that satisfied the inclusion and exclusion criteria were evaluated (Figure gif ppt). The results of this evaluation are described in Parts 1 and 2 of the present study.

The publications included were primarily studies performed with the expressed goal of comparing clinical trials funded by pharmaceutical companies with clinical trials that had not received financial support from such companies, e.g., with regard to the results or conclusions. These studies were accompanied by a number of publications that investigated the consequences of financing of a study by pharmaceutical companies. These included, for example, articles in which information from the files of the US licensing authority (Food and Drug Administration, FDA) was compared with data from publications in medical journals, and case studies on individual substances.

Two systematic reviews carried out for much the same reason as the present review were excluded on the grounds that their principal primary studies had already been included (10, 11). The results of the publications covered in the present review are based largely on clinical drug trials conducted principally in the Anglo-American countries. A few publications also analyzed studies performed in Germany. The most recent studies whose data were taken into account in the papers covered were published in November 2008. The publications evaluated concern various areas of medicine, but it cannot be completely ruled out that original data from a small number of drug trials were included more than once. A few studies dealt with several aspects.

Connection between type of funding and results of drug trials
Twenty-six of the 57 publications analyzed sought to ascertain whether the results and/or conclusions of drug trials depended on the type of funding or on financial conflicts of interest on the part of the authors (eTable).

These investigations dealt with various medical topics and disciplines (among them oncology [e1, e2], cardiovascular diseases [e3], and psychiatry [e4]) and featured different study designs (e.g., systematic review, meta-analysis, case study).

Altogether, 23 of these 26 studies came to the conclusion that there was a positive correlation between the financing of a study by pharmaceutical companies and/or conflicts of interest on the part of the authors and results or conclusions that were favorable to the sponsor. The statistical significance of this finding was investigated in 22 cases and confirmed in 20.

In 4 cases it was apparent that the findings were interpreted favorably towards the pharmaceutical concern that had funded the study, independent of the results (e5–e8).

In 3 investigations there was no clear association between the funding of a study by pharmaceutical companies and findings that favored the sponsors (e9–e11). In one of these, a urological study (e11), very few publications were analyzed (n = 24); furthermore, the author disclosed “Speakers’ Bureau—Pfizer” as a conflict of interest. Another study investigated the connection between the conclusions and the source of financial support in clinical trials that had appeared in 5 influential medical journals over a period of 20 years (e10). Most trials yielded positive results for the drug in question regardless of the funding source, but this study also revealed a trend over the course of time towards more positive findings in industrially financed trials than in trials supported by non-profit organizations (e10). The third study compared the results (but not the interpretations or conclusions) of clinical trials of drugs used in pain management, some of them long available as generics (e9).

Study protocol
Five of the 57 studies analyzed investigated whether funding by pharmaceutical companies affected the design of the study protocol (Table 1 gif ppt). The use of placebos was shown to be significantly more common in RCTs of drugs for psoriasis that were financed by such companies than in those with funding from other sources (e12). Moreover, several studies of treatment for premature ejaculation that were sponsored by a pharmaceutical company were found to have disregarded the relevant objective endpoint (e13). In an investigation of inhaled corticosteroids, significant differences in the frequency of adverse drug reactions (ADR) between the probands and the control group occurred only half as often when the study had been funded by the manufacturers (see also Part 2). The differences could be attributed wholly to the study design. For example, studies financed by pharmaceutical companies used lower dosages.

A further study revealed that an RCT of rofecoxib was a “seeding trial,” i.e., a clinical study that claims to be testing a scientific hypothesis but in reality has the purpose of making the drug known to prescribing physicians and thus increasing sales (e14). Examination of confidential internal documents that became public in the context of legal proceedings revealed that the study had been designed and conducted by the marketing department of the pharmaceutical company concerned. However, there were also signs of other irregularities. The target group of prescribers (family doctors) was determined at the outset, and only then was the study protocol formulated. Neither ethics committees nor participants were informed of the true goal of the study, namely promotion of the drug. To make this marketing as successful as possible and pitch the product to a large number of doctors, the study was carried out in 600 centers. Such a high number of centers would not have been required for the official purpose of the study (comparison of the gastrointestinal tolerability of rofecoxib and naproxen). Indications of increased cardiovascular morbidity among those taking rofecoxib were not pursued. The article describing the study was written by employees of the manufacturer of rofecoxib, but appeared under another researcher’s name. This was thus a case of guest authorship (see Part 2). The pharmaceutical company concerned investigated the marketing effect of the study, finding that participating physicians did indeed prescribe rofecoxib significantly more often than non-participants in its first 6 months on the market.

In contrast, differences in study protocol among investigations of atypical antipsychotics could not be attributed to the type of funding (e15). The protocols of 3 pharmaceutical companies and of studies not supported by the pharmaceutical industry were compared. Blinding, placebo controls, and use of an atypical antipsychotic for comparison were found more frequently in the protocols of two of the pharmaceutical companies than in those of the third company or the studies without industrial financing.

Connection between type of funding and quality of study methods
Altogether, 4 publications concerned themselves with the methodological quality of drug trials in relation to financial support (Table 2 gif ppt). Various instruments were employed to assess quality: for RCTs the Jadad scale (12) and a scale based on the CONSORT Statement (e16), both of which ask, for example, about blinding and randomization; for systematic reviews, the Oxman and Guyatt index, which for instance asks about search strategies and avoidance of selection bias (13).

In 2 investigations the quality of the methods of the clinical trials financed by pharmaceutical companies was comparable to or even better than when funding came from a sponsor without commercial interests (e16, e17), and in only one study did it tend to be worse (e18). Comparison of industrially financed meta-analyses and Cochrane Reviews showed that the latter displayed higher quality (e6); search strategies and randomization methods were described more often, for example, and literature searches were more comprehensive.

Discussion
The results of clinical drug trials that are funded by pharmaceutical companies or whose authors have financial conflicts of interest are favorable to the products of the sponsoring company far more frequently than studies whose funding comes from other sources. Furthermore, interpretation of the data in the conclusions of industrially financed trials more often favors the sponsor. This was shown by the present systematic review and analysis of investigations, published between 1 November 2002 and 16 December 2009, into various diseases, study types (e.g., RCTs and observational studies), and drugs. The results confirm the conclusions of 2 systematic reviews, both published in 2003, conducted with similar intent (7, 8). The principle of equipoise, i.e., uncertainty which of the alternative approaches benefits the patient most, forms the ethical foundation of clinical studies in which the probands receive various treatments (14). This principle seems to be violated in many studies funded by pharmaceutical companies.

There are numerous reasons why studies financed by pharmaceutical manufacturers more often yield positive results. Four investigations found evidence that pharmaceutical companies influence the study protocol to their advantage (e12–e14, e19), e.g., by more frequent use of placebos in control groups than in independently funded studies (e12). Although the responsible authorities sometimes demand placebo-controlled trials as a condition of licensing, they also request active controls (15). Further factors leading to higher frequency of results favorable to the sponsor in trials funded by pharmaceutical companies are described in Part 2 of this review.

Trials financed by pharmaceutical concerns displayed no signs of poorer methodological quality. On the contrary, two studies showed superior quality (e16, e17). It must be taken into account, however, that some factors that serve to assess the quality of the instruments used in a study were not determined, among them the clinical relevance of the target parameters. In oncology, for instance, there are currently major defects in the protocols of industrially sponsored clinical trials, e.g., deficiencies in the definition of patient-relevant endpoints and in the selection of suitable substances for the control arm of RCTs (16–19). Moreover, clinical trials in oncology are often discontinued after preliminary analysis (20), with the result that only a short time after the licensing of a drug its additional benefits and the safety of new substances can frequently no longer be evaluated, preventing any benefit/risk analysis (21).

This work was supported by funds from the Initiative for Health Services Research (Förderinitiative Versorgungsforschung) of the German Medical Association (Bundesärztekammer, BÄK). The 110th German Medical Assembly tasked the BÄK with investigating, in the framework of the Initiative for Health Services Research, the influence of sponsors on the scientific results of drug trials.

Conflict of interest statement
The authors declare that no conflict of interest exists according to the guidelines of the International Committee of Medical Journal Editors.

Manuscript received on 29 May 2009, revised version accepted on
23 February 2010.

Translated from the original German by David Roseveare.


Corresponding author
Dr. med. Gisela Schott
Arzneimittelkommission der deutschen Ärzteschaft
Herbert-Lewin-Platz 1
10623 Berlin, Germany
gisela.schott@akdae.de


@For e-references please refer to:
www.aerzteblatt-international.de/ref1610
eTable available at:
www.aerzteblatt-international.de/article10m0279