Intended for healthcare professionals

Letters

ATAC trial: reporting interim results is not helpful

BMJ 2003; 326 doi: https://doi.org/10.1136/bmj.326.7402.1329-b (Published 12 June 2003) Cite this as: BMJ 2003;326:1329
  1. Heather Goodare (hm.goodare{at}virgin.net), chair,
  2. Clare Dimmer (clare{at}solvatec.demon.uk), secretary,
  3. Kathy Page (kathypage{at}btinternet.com), treasurer
  1. Breast UK (Breast-cancer Research Ethics and Advocacy Strategy), Horsham, West Sussex RH13 6DF
  2. Breast UK Waterlooville, Hants PO7 6LA
  3. Breast UK Lower Layham, Suffolk IP7 5NA

    EDITOR—We should like to raise some concerns about the ATAC (arimidex, tamoxifen alone or in combination) trial, which was reported on in the Lancet after only half the time stipulated in the protocol (2.5 years instead of 5).1 Ravdin has raised important points: “Early reporting rules can powerfully affect what information can be gleaned from a trial, particularly if they cause a trial to be reported when many of the patients have not completed therapy.”2 In spite of this, the “results” of the trial were reported in the United Kingdom's national lay press.3

    However, is arimidex really the best bet? Of course we still don't know as some of the more serious adverse events may not emerge until the five year mark is reached. The problems in patients with endometrial cancer who were taking tamoxifen took many years to become obvious. Why publish results at the halfway mark only? Does this not indicate a lack of respect for the participants?4

    Early reporting of the trial was certainly good for AstraZeneca, after losing patent protection for tamoxifen. But was it good for patients? Was it good for science? Publication of interim results can seriously mislead, as in the case of the infamous study of women attending the Bristol Cancer Help Centre.5 As Ravdin says: “It remains to be seen how many of the patients on the two remaining blinded arms will continue to take the therapy that they were randomised to.”

    References

    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.