Screening behavior was analyzed in multivariate models, controlling for the variables listed in Table 1 and area of residence for all four screening procedures (Table 4). We found that the strongest predictors of low levels of screening for all screening tests except BSE
in this model were having no insurance or only public insurance (i.e., Medicare/MediCal) compared to private health insurance, having infrequent use of medical services, not having a regular clinic, and not speaking English. For BSE the infrequent use of medical services
was a strong predictor. Consistent with the important role played by having regular medical care was the observation that the practice of taking hormone replacement therapy was a substantial and statistically significant indicator of adherence to preventive practices. In contrast, we found that race/ethnicity, in general, was not a good predictor of screening. Higher than
expected rates of BSE were reported by black and other (primarily Asian) women. Foreign-born Chinese women were much less likely than U.S.-born Chinese women to have had a Pap smear in the past 3 years. Measures commonly used to assess socioeconomic status like education, household income, and employment status were not predictive of screening behavior in this model, thus
highlighting the importance of factors more proximately related to access, specifically having insurance and a regular source of medical care.