Observer variation in recording clinical data from women presenting with breast lesions. Report from the Yorkshire Breast Cancer Group.

The degree of observer variation in recording 11 186 items of clinical data from 242 woman who presented complaining of a lump in the breast to a group of 10 surgeons was studied. Each women was interviewed and examined twice and the findings (of the two clinicians) compared. There was a wide range of variation among the observers. Variation in recording the presence or absence of axillary nodes was considerable (45%), as was that in sizing the primary lesion (55%). In 20% of cases the two assessments of primary lesion size differed by over 2 cm. In other respects the results were more encouraging; much variation could be eliminated by wording the proforma more clearly. Moreover, variation was not person-specific, so that these findings are probably reasonably representative. Any future trial of breast lesions should (a) design specific proformata, (b) define terminology, (c) make these definitions universally available, and (d) conduct observer variation studies before the start of the full trial.