Subgroup-specific effects of questionnaire wording on population-based estimates of mammography prevalence

Objectives. This study investigated whether an apparent downturn in prevale nce rates of mammography use reported in the 1992 Behavioral Risk Factor Su rveillance System (BRFSS) questionnaire resulted from a change in questionn aire wording. Methods. In a pretest-posttest design (1990-1991 vs 1992), piecewise linear regression analyses were based on monthly prevalence estimates of mammogra phy use among female BRFSS respondents 40 years or older. Results. Self-reported mammography use was lower by 3.5 percentage points ( 95% confidence interval [CI] = 1.5, 5.5) overall-and lower by 13.6 percenta ge points (95% CI = 2.6, 24.6) among Black women with less than a high scho ol education-when predicted from 1992 data than when predicted from 1990-19 91 data. Conclusions. A change in questionnaire wording in the BRFSS caused demograp hic-specific effects in population-based estimates of mammography use.