Medical specialists and pharmaceutical industry-sponsored research: a survey of the Australian experience

There are extensive ties between pharmaceutical manufacturers and clinical researchers.1 While engagement with industry is inevitable and has positive features, there is concern that these interactions may lead to a dependence on commercial funding for clinical research, which might influence the research agenda, the conduct of studies and the reporting of their results.2,3 Incomplete or biased reporting of the results of clinical studies is a particular concern, and has been highlighted by the recent controversies surrounding the harms of selective serotonin reuptake inhibitors in treating childhood and adolescent depression, and of cardiovascular toxicity with the selective cyclooxygenase-2 inhibitor, rofecoxib.4,5

Studies of the relationships between the medical profession and industry have tended to focus on academic medical centres in the United States (rather than on individual practising clinicians) or on the experience from other countries.6 There is evidence that the relationship between industry and researchers has changed in recent years, with commercially-sponsored clinical research increasingly being performed by contract research organisations (CROs) rather than academic medical centres, and with physicians in private practice playing a more prominent role in research.2,3

Despite interest in this important topic, few Australian data have been published.3 However, there have been calls locally for more public investment in clinical trials.7 As part of a study of relationships between Australian medical specialists and the pharmaceutical industry, we sought to characterise the extent, nature and consequences of their research collaboration.

Our methods were approved by the Human Research Ethics Committee of the University of Newcastle. We obtained 5000 names of medical specialists from an electronic file extracted from the 2002 edition of the Medical Directory of Australia (MDA),8 and stratified these by clinical subspecialty and geographical location. Specialists with minimal prescribing responsibility (eg, surgeons, anaesthetists) were excluded; general practitioners were not included.

We surveyed participants by means of a questionnaire developed in collaboration with three advisory groups — medical specialists, consumers, and staff from pharmaceutical companies. The questionnaire had 46 questions seeking information on all aspects of relationships between medical specialists and pharmaceutical companies. Topics covered included: the frequency and nature of interactions with the pharmaceutical industry; descriptions of industry-sponsored activities and gifts; the extent and nature of research funding (see below); details of advisory panel and consultancy roles; personal views and opinions on the value of these interactions; and personal demographic, professional and practice details. The sections covering industry research funding comprised 11 questions that allowed up to 43 response categories, including: who initiated the activities; the source of funding (company or CRO); the reasons for working with industry; the type of research (including stage of drug development); negative experiences, including drafting of reports by company staff rather than investigators, alteration of data (beyond the normal process of cleaning), concealment of relevant findings, inappropriate editing of reports, major changes to protocols, delay in publication of data, failure to publish key findings, and premature termination of trials. After answering these questions, respondents were invited to provide a brief account of the events that they had reported (without naming companies or products). This article will focus on the nature and outcomes of research collaborations with industry.

The questionnaire was mailed to 2253 specialists in September 2002.

Data were analysed with SAS statistical software.9 Conventional descriptive analyses (means and standard deviations [SDs], proportions, and 95% CIs) were calculated when relevant. We compared proportions using χ2 analysis; means were compared using t tests.

Of the 2253 specialists, 133 were unable to participate (for reasons such as having died or having changed professional role). Of the remaining 2120, 823 completed and returned the survey instrument (overall response rate, 39%). Respondents were fairly similar to the original sample in terms of geographical location and clinical specialty (Box 1). The average age of the respondents was 48 years (95% CI, 32–69 years), and, on average, they were 27 years (95% CI, 10–45 years) from graduation. Most (79%; 95% CI, 76%–82%) were male. We do not have data on the sex distribution of the mail-out sample, but data provided separately by the publishers of the MDA indicate that 4960 of 6422 doctors listed as practising internal medicine specialties (77%) were male. Respondents spent an average of 46% of their time in salaried hospital practice and 47% in private practice, and, on average, saw 20 patients and wrote 20 prescriptions daily.

Collaborative research between medical specialists and pharmaceutical companies has been responsible for many significant advances in clinical practice. In Australia, there are widespread research ties between the pharmaceutical industry and medical specialists, including those working largely or exclusively in private practice. This is in line with experience in the US, where there has been a shift from involvement of established researchers in academic centres to private research organisations.1,2 Our data also show the complexity of relationships between researchers and the pharmaceutical industry, with collaboration sometimes initiated by medical specialists, rather than the industry, and management often in the hands of CROs rather than the manufacturing companies themselves.

Most respondents in this study did not believe they had made ethical compromises during their research collaboration, and many expressed positive views of their relationships with industry. However, instances of questionable practice were reported. The two most frequent — premature discontinuation of trials and preparation of first drafts of research reports by industry staff — were considered appropriate when termination was in response to safety concerns and the first drafts of reports were generally meant primarily for internal circulation. More serious instances of possible research misconduct included delayed or non-publication of results and modification or concealment of data that negatively reflected on a drug product. Overall, 8.6% of respondents (equivalent to 21% of those with an active research relationship with industry) reported experiencing one or more episodes of potentially serious research misconduct. This finding is similar to the 19.8% of researchers at life sciences faculties in the US who reported delaying publication of results for (largely) commercial reasons.10 These results are consistent with abundant literature that has examined the risks of collaborations between industry and researchers, including conclusions biased in favour of products, restrictions on publication and data sharing, and financial interests that conflict with primary professional responsibilities.11 In our analysis, we did not categorise protocol changes as serious research misdemeanours, reflecting the comments of some of our respondents. However, other researchers have shown that in over 60% of trials one or more primary outcomes were changed, introduced, or omitted while the research was under way.12 Such changes may bias the reporting of results in favour of the drug being tested. As noted earlier, our respondents were generally tolerant of industry staff drafting reports for internal use, but others have pointed to the undesirability of ghost-writing of manuscripts for publication in journals, and editors have introduced guidelines designed to discourage this practice.2,13

Our findings have limitations. One is the low response rate (39%), although it should be noted that data in the MDA are updated voluntarily by practitioners, so it is likely that a number of the non-respondents had left active clinical practice.

Other limitations include reliance on self-report and our focus on a specific population. The views of non-respondents may also have differed from the views of respondents. In terms of demographic characteristics and clinical specialty, our respondents were fairly similar to all medical specialists listed in the original sampling frame, and the sample size was large (more than 800 respondents), giving us some confidence in the precision of our estimates. However, our reliance on self-report leaves open the possibility of selective or inaccurate recall; given the need to maintain anonymity, we could not independently assess the reliability of respondents’ disclosures. The study was restricted to medical specialists, limiting any generalisations that can be made about other medical populations, such as general practitioners.

Our data do not allow us to make an overall evaluation of industry-sponsored research; we accept that our questionnaire concentrated on the negative consequences of such studies, and did not ask respondents to describe their positive experiences. There are clear benefits from such research collab-orations with industry, but to quantify those was beyond the scope of this study.

Our results raise some significant concerns. The first is the apparent high level of industry-supported research being conducted by medical specialists working substantially or exclusively in private practice. We don’t have data on whether the involvement of private specialists has increased in recent years, but hard questions are now being asked about who oversees clinical research and the integrity of research publications and their authorship.13 We believe that professional organisations, institutions, and practitioners themselves, need to ensure that mechanisms are in place to ensure appropriate governance of clinical research when it is conducted substantially in the private arena.

The second issue concerns the examples of possible research malpractice, particularly delayed publication or failure to publish the results of clinical research. This clearly troubled some of our respondents, and has been the subject of widespread comment and action by the editors of major medical journals.13

Our findings underscore the necessity for ethics committees to oversee not only the conduct of research, but also the analysis and reporting of results, to ensure that the public has access to accurate data on the benefits and harms of modern medical treatments. However, it is doubtful that ethics committees have the resources to act alone in policing these activities. It is essential that clinical trials are registered at their inception and that pharmaceutical companies and researchers themselves adopt and adhere to guidelines for good publication practices.14,15