It has become clear that conflicts of interest can influence scientific judgement and reporting of research results in the medical literature.1 In this issue of EBM, Rasmussen et al2 studied citation of systematic reviews of breast cancer screening3–5 that found smaller benefits and greater harms than did prior reviews. They then compared how the reviews were cited by general versus specialty journals. In most papers citing the reviews, the key, perhaps controversial, findings regarding overdiagnosis and breast cancer mortality were not discussed. But in those papers that did describe those findings, papers published in general medical journals were more likely than those published in specialty journals to explicitly agree with the findings about overtreatment and overdiagnosis, or to have cited the results without comment. Similarly, specialty journal articles were more likely than general journal articles to explicitly reject the review findings regarding breast cancer mortality or focus on subgroup results only. Rasmussen et al2 suggest that conflicts of interest might explain these findings.

Conflicts of interest may well be the explanation for Rasmussen et al's findings. But it should be made clear that they did not explicitly compare papers with authors who specifically disclosed conflicts to those in which authors did not disclose such conflicts. So the conclusion that conflicts of interest might lead to explicit rejection of findings not favourable to mammography are based on the assumption that specialty journals, their editors, authors and papers published in …