Background: In breast cancer it is not uncommon for results of trials to be published as conference abstracts, well before the full paper appears in a peer-reviewed journal (if a paper appears at all). Although previous studies suggested instability of data reported in abstracts and substantial changes of results with subsequent follow-up (Hopewell, 2004; Marinovich, 2004), there is a strong temptation to use this data particularly when an intervention is novel or effective treatment for the condition of interest is not available. An additional complication in trials in early breast cancer is the need for extended periods of follow-up competing with the desire for answers now rather than later. This can lead to premature conclusions based on surrogate outcomes such as response to chemotherapy, rather than harder outcomes such as survival.

Objectives: To explore the impact over time of the inclusion of data from abstracts on a Cochrane review of preoperative (neo-adjuvant) chemotherapy for early breast cancer.

Methods: The specialised register of the Cochrane Breast Cancer Group includes the results of hand-searches of the major conference proceedings in breast cancer. The register will be searched for studies eligible for inclusion in the review, and copies of all articles and abstracts obtained.

Results: The results will be presented as a cumulative meta-analysis using the data available at selected time points. Emphasis will be placed on the type and amount of information available at each time point (such as rate of conservative surgery, tumour response rate, risk of local recurrence), and the reliability of any conclusions made along the timeline. This project is in progress. Results will be presented at the meeting.