Our analyses showed certain sociodemographic and psychosocial factors and health care access were significantly associated with receipt of a PSA test. To our knowledge, potential factors associated with PSA testing have not been well characterized for men aged 75 or older. In 2003, researchers found that higher education levels and more physician visits were significantly associated with having a PSA test (10). We confirmed these findings with 2006 BRFSS data. Additionally, our analyses suggested race/ethnicity was not associated with PSA testing. We also found that men who reported being satisfied with life, always having emotional support, being married or having a partner, or having health insurance were more likely to report having a PSA test. In a quality-of-life study, higher socioeconomic status, good health, and good social relationships were consistently associated with higher life satisfaction, and emotional support had the strongest association (13). Life satisfaction and emotional support are potent predictors of well-being, an integral part of health (13). People who perceived that they were in good health were more likely to believe that they would benefit from cancer screening (14). With a better understanding of patients' sociodemographic and psychosocial characteristics, physicians may be able to employ more appropriate and effective individualized strategies to frame discussions about PSA screening, such as formulating conversations with patients on the basis of their health behaviors or incorporating spouses or partners into these discussions.

Although no major professional associations have recommended PSA screening in elderly men, high prevalences of PSA screening have been reported (6-9,11,15). PSA screening and subsequent biopsy and treatment can lead to psychological and physical harm and additional medical cost (16,17). The difficulty in estimating patients' life expectancy and the convenience and low cost of PSA testing may contribute to the overuse of PSA tests in the elderly population (10,14). In addition, misaligned financial incentives for physician practices may also partially account for this problem; physicians with a laboratory on site are more likely to order a PSA test (18). Studies have suggested that a physician's advice is a major determinant in a man's decision to have a PSA test (19,20). The new USPSTF prostate cancer guidelines should prompt clinicians to discuss the potential implications of screening for prostate cancer before ordering a PSA test for men aged 75 or older.

Our study findings are generalizable because they were generated from a nationwide population-based sampling survey. However, this study is subject to several limitations. First, the BRFSS data were based on self-report and are thus subject to recall bias. Self-reports of screening behavior overestimate the extent of actual screening (6). However, the PSA testing rate reported in our study did not exceed the rate calculated from medical chart data in 46 community-based primary care settings (12). Second, BRFSS questionnaires did not ask if people had symptoms of prostate cancer when their PSA tests were done. We could not distinguish between tests for screening (testing an asymptomatic person) and tests that help physicians make diagnoses (testing a symptomatic person). Thus, some of the PSA tests reported in our study could have been used to assist in diagnosing disease. However, a cohort study of more than 500,000 veterans aged 70 or older who had no history of prostate cancer, elevated PSA, or prostate cancer symptoms found a PSA screening rate of 56% (11). The distinction between a screening test and a diagnostic test can be difficult to make. Studies suggest that men who have lower urinary tract symptoms derived from benign prostatic hyperplasia — a common prostate disease in older men — are not at higher risk for prostate cancer, aside from the risk conferred by their age (21). Screening for prostate cancer could be defined as testing in men who do and do not have lower urinary tract symptoms (21). To be conservative, we explained our rates as the use of PSA test, instead of screening test. In an analysis of data from NHIS, 86% and 77% of the PSA tests in 1999 were used for screening among men aged 65 to 79 and 80 or older, respectively (22). Therefore, most men in our study were likely to have been asymptomatic at the time the PSA test was done. Finally, because of limited sample size, we could not examine PSA testing at state or local levels. A geographic analysis study using multiple-year BRFSS data is warranted.

Jun Li, Centers for Disease Control and Prevention. 4770 Buford Hwy, Mailstop K-55, Atlanta, GA 30341, Phone: 770-488-3030, Email: ffa2/at/cdc.gov.