Article type
Year
Abstract
Background: The validity of systematic reviews on medical procedures and technologies may be limited by unpublished data (publication bias). Health technology assessment agencies try to overcome this problem by including commercial-in-confidence data in their assessments. If unpublished data cannot be presented in the assessment report due to confidentiality agreements, the impact of these data on the assessment's conclusion remains unclear. The German Institute for Quality and Efficiency in Health Care (IQWiG) negotiated an agreement on submission of unpublished data and publication of this information in its reports with the German Association of Research-based Pharmaceutical Companies.
Objectives: To evaluate the impact of including unpublished studies or unpublished information from published studies in systematic reviews.
Methods: The impact of unpublished industry data on the following criteria was investigated: number of relevant studies; quality rating of published studies; study results. The analysis was based on the assessments of rapid-acting insulin analogues for treatment of diabetes mellitus Type II and of montelukast for treatment of persistent asthma.
Results:
Insulin analogues: Three out of seven studies included had not been or only partially been published in journals (43%). The impact of unpublished information on the quality ratings of published studies was inconsistent. Unpublished information was partly inconsistent with published data. Montelukast: One out of 15 studies included had not been published in a journal (7%). The incorporation of unpublished data improved the quality ratings of published trials in the majority of cases. Information provided by the industry yielded additional data on published and unpublished endpoints.
Conclusions: The impact of unpublished information on drug assessments varies for different drugs. However, unpublished data may substantially affect an evaluation's results and conclusions. Analyses of further assessments performed in 2006 will be presented.
Objectives: To evaluate the impact of including unpublished studies or unpublished information from published studies in systematic reviews.
Methods: The impact of unpublished industry data on the following criteria was investigated: number of relevant studies; quality rating of published studies; study results. The analysis was based on the assessments of rapid-acting insulin analogues for treatment of diabetes mellitus Type II and of montelukast for treatment of persistent asthma.
Results:
Insulin analogues: Three out of seven studies included had not been or only partially been published in journals (43%). The impact of unpublished information on the quality ratings of published studies was inconsistent. Unpublished information was partly inconsistent with published data. Montelukast: One out of 15 studies included had not been published in a journal (7%). The incorporation of unpublished data improved the quality ratings of published trials in the majority of cases. Information provided by the industry yielded additional data on published and unpublished endpoints.
Conclusions: The impact of unpublished information on drug assessments varies for different drugs. However, unpublished data may substantially affect an evaluation's results and conclusions. Analyses of further assessments performed in 2006 will be presented.