Perceived and actual breast cancer risk: optimistic and pessimistic biases.

Perceived risk can influence health behaviors. Studies using various populations and breast cancer risk bias assessment methods have identified both risk over- and underestimation. Among 1803 women in primary care settings, 47 percent were at average epidemiologic risk (Gail calculated relative risk ±50 percent of age-adjusted population average) and 55 percent perceived themselves to be at average risk (compared to same-age others) but there were mismatches or 'biases': 31 percent underestimated personal risk; 26 percent overestimated. Multiple logistic regression revealed that smokers were more likely to overestimate risk. Overestimation decreased with more education. Mammography use did not independently predict perception bias but, among never-screened women aged over 40 years, those contemplating mammograms were most likely to overestimate risk; precontemplators were most likely to underestimate. Implications for research and intervention are discussed.