Article type
Year
Abstract
Background: Conflicts of interest can influence the evaluation and presentation of scientific results. There are professional, political, and financial conflicts of interest related to the promotion of mammography screening.
Objectives: To examine if author affiliation influences the presentation of benefits and harms of mammography screening in scientific articles.
Methods: We searched PubMed for articles on mammography screening published in English in 2004. Articles were eligible if they mentioned benefits or harms, or both, related to mammography screening. Included articles were evaluated using a checklist. We noted whether specific items were mentioned and whether they were acknowledged, rejected, or downplayed, and related our results to author affiliation and source of financial support. First, two authors performed an unblinded evaluation independently and settled differences by discussion. Second, the articles were masked, and the third author performed a blinded evaluation of author affiliation separately from a blinded evaluation of the content.
Results: We found 145 eligible articles. Our preliminary analysis of the unblinded extraction shows that authors working with screening or in screening-related specialties are significantly less likely to mention, and more likely than other authors to reject or downplay, the harms of mammography screening, in particular the most important harms, 'overdiagnosis' and 'overtreatment'. This finding was confirmed by the blinded analysis. Both analyses showed that no authors unrelated to screening downplayed or rejected 'overdiagnosis', whereas this was the case in about half of the articles by authors related to screening. Contrary to this, the major benefit, a reduction in breast cancer mortality, was mentioned similarly often, regardless of author affiliation. Our final data analyses will be presented at the Colloquium.
Conclusions: Our results underscore the need for independent evaluation of medical interventions.
Objectives: To examine if author affiliation influences the presentation of benefits and harms of mammography screening in scientific articles.
Methods: We searched PubMed for articles on mammography screening published in English in 2004. Articles were eligible if they mentioned benefits or harms, or both, related to mammography screening. Included articles were evaluated using a checklist. We noted whether specific items were mentioned and whether they were acknowledged, rejected, or downplayed, and related our results to author affiliation and source of financial support. First, two authors performed an unblinded evaluation independently and settled differences by discussion. Second, the articles were masked, and the third author performed a blinded evaluation of author affiliation separately from a blinded evaluation of the content.
Results: We found 145 eligible articles. Our preliminary analysis of the unblinded extraction shows that authors working with screening or in screening-related specialties are significantly less likely to mention, and more likely than other authors to reject or downplay, the harms of mammography screening, in particular the most important harms, 'overdiagnosis' and 'overtreatment'. This finding was confirmed by the blinded analysis. Both analyses showed that no authors unrelated to screening downplayed or rejected 'overdiagnosis', whereas this was the case in about half of the articles by authors related to screening. Contrary to this, the major benefit, a reduction in breast cancer mortality, was mentioned similarly often, regardless of author affiliation. Our final data analyses will be presented at the Colloquium.
Conclusions: Our results underscore the need for independent evaluation of medical interventions.