Many health services research studies and cancer control evaluations will need to continue to rely on patient self-report of screening experience and it is important to know how accurate these self-reports are. Breast cancer screening is case in point. Studies depend on self-report because screening mammography may involve interaction with several providers at several different sites, making it difficult to review experience in the traditional manner (i.e. medical record audit). Not only do evaluators and researchers need to document screening participation (number, periodicity) but interest is mounting to document cost and utilization (e.g. follow-up, biopsy) which results from screening. This project has four objectives: a) to determine the validity of self-report of eleven mammography related measures by comparison to clinical records data; b) to document the extent to which respondent characteristics affect self-report accuracy; c) to compare two methods of self-report data collection: telephone and mail surveys; and d) to apply correction factors for three selected measures to an independent data set. Investigators in this study include representatives of the major radiology group and primary care physicians in a large metropolitan area in western Massachusetts. Female patients, ages 50-75, of primary physicians will be invited by mail into the study. Of the women who agree to participate, 1/2 will get a written survey and 1/2 will get a telephone interview. The accuracy of information provided by women will be compared to data taken from radiology reports and medical records. Regression techniques will be used to determine covariates of self-report accuracy, such as age, income, health history, medical care use and selected psychosocial measures as well as decay of accuracy with increasing time since the event. Correction factors will be calculated and subsequently compared to an existing data set from a survey of 1,000 Women.
