Perceived risk prior to mammography as an independent predictor of breast cancer

The predictive value of perceived risk of breast cancer has previously mainly been compared to statistical risk models. We aimed to compare the women's perceived risk with the observed risk of a breast cancer diagnosis.

Women referred to mammography at a Danish public hospital (N=1021) were asked to complete a questionnaire about 2 weeks before the examination. The women were asked to rate their perceived risk of having breast cancer on a six-point scale with the anchor points very low risk (1) and very high risk (6). The perceived risk was compared with the diagnosis from the mammography 1-3 weeks later. Multivariate logistic regression was used to adjust for age, category of referral, family history of breast cancer, and previous mammography.

The questionnaire was completed by 665 (65%) of the women. Compared to women without breast cancer, a larger percentage of women with breast cancer had estimated their risk as high. The risk difference was 16.5% (95% CI 2.5-30.6), using a cut-off between 4 and 5 on the six-point scale. The odds ratio was 3.6 (95% CI 1.4-8.9) after adjustment for age, category of referral, family history of breast cancer, and previous mammography. The sensitivity, specificity, and predictive value were 0.17, 0.96, and 23%, respectively.

Women who rated their risk as high had a statistically significant increased risk of having a breast cancer diagnosed at the subsequent mammography. The odds ratio based on the women's risk perceptions remained of the same magnitude after adjustment for category of referral, suggesting that information given by the women contains independent predictive value.