Are self-reported dates of mammograms accurate?

Background: This paper assesses the accuracy of self-reported dates of last mammograms from a postal survey compared to dates of mammograms in medical records. Methods: The subjects included women 50 years of age and older who were members of a local health maintenance organization, had reported ever having a mammogram, and had completed a postal questionnaire about breast cancer screening. The date of last mammogram from the postal surveys was compared to the date in medical records. Results: Of 78 women who self-reported both the month and year of last mammogram in a postal survey, agreement within 1 month with the medical record data was 62.8% and agreement within 3 months was 75.6%. A total of 32.1% of the subjects underestimated the time since their last mammogram, while only 5.1% overestimated. Based on self-reported dates, 85.9% of the subjects had a mammogram within the last year compared to 76.9% based on medical records. Using medical records as the 'gold standard', the sensitivity of self-reported mammogram within the last year versus more than 1 year ago was 98% and the specificity was 56%. Conclusions: The agreement between self-report and medical records was fairly good, although in women's self-reports that do not agree with medical records there is a tendency to underestimate the time since last mammogram. Inaccuracy of recall of the date of last mammogram will adversely affect adherence to the recommended screening guidelines by increasing the length of time between each mammogram.