[Research results should be freely accessible!-- Case reports demonstrate obstacles to contact with drug industry].

Recommendations on the use of medical technologies should be based on best available evidence. Publication bias due to delayed publication or non-publication of negative studies leads to inflated effect sizes in systematic reviews. We carried out a health technology assessment on prevention and treatment of venous thromboembolism, and in order to minimize publication bias asked the six pharmaceutical companies with antithrombotic drugs registered in Sweden to supply us with results from unpublished clinical trials. The answers were returned after 4 to 60 weeks, and with only one exception, after multiple reminders by telephone, fax and e-mail. No relevant unpublished studies were reported to us although we were made aware of two published studies that our literature search had not identified. One company did not mention a large negative unpublished clinical trial with their drug which has remained unpublished due to conflict between the company and the steering committee. We conclude that the companies did not contribute much and that the process was very time consuming with little result. As non-publication of trial results is unethical, we propose agreements between industry and researchers that all studies be prospectively registered and made publicly available.