Background: Publication bias is related to aspects other than quality that can influence the viability of a publication. In general there is a greater probability of publishing studies with positive results not showing the totality of the avaliable data. Considering the number of reviews published in breast cancer in relation to the modern treatments, the assessment of publication bias in these studies was conducted.

Objective: To assess the methods used to avoid or consider risk of publication bias in systematic reviews published on breast cancer.

Methods: We conducted a eletronic search identifying systematic reviews of breast cancer published in Medline during the last five years. The data were extracted using a structured form and the studies were analyzed with calculation of frequency and 95% confidence interval.

Results: 185 reviews were included. A mean of 4 (± 2.1) databases were used in the included reviews. Of these, 37% (CI 95%, 30% to 44%) reported plan to used funnel plot analysis with 18% (CI 95%, 13% to 24%) of the total reviews using it. Only 30% (CI 95%, 24% to 37%) used other sources than eletronic databases.

Conclusion: There is an inadequacy in the planning and reporting of the assessment of risk of bias in systematic reviews in breast cancer.