Dr. Curtis has received research and salary support from Allergan, GlaxoSmithKline, Lilly, Medtronic, Novartis, Ortho Biotech, OSI Eyetech, Pfizer, and Sanofi-Aventis. A detailed listing of Dr. Curtis' financial disclosures is available at http://www.dcri.duke.edu/research/coi.jsp
Kevin A. Schulman MD,
Center for Clinical Genetics and Economics, Duke Clinical Research Institute, Duke University School of Medicine, Durham, North Carolina
Division of General Internal Medicine, Department of Medicine, Duke University School of Medicine, Durham, North Carolina
Dr. Schulman has received research and/or salary support from Actelion, Allergan, Amgen, Arthritis Foundation, Astellas Pharma, Bristol-Meyers-Squibb, The Duke Endowment, Genentech, Inspire Pharmaceuticals, Johnson & Johnson, Kureha Corporation, LifeMasters Supported SelfCare, Medtronic, Nabi Pharmaceuticals, National Patient Advocate Foundation, North Carolina Biotechnology Center, Novartis, OSI Eyetech, Pfizer, Roche, Sanofi-Aventis, Schering-Plough, Scios, Tengion, Theravance, Thomson Healthcare, Vertex Pharmaceuticals, Wyeth, and Yamanouchi USA Foundation. Dr. Schulman has also received personal income for consulting from Avalere Health, LifeMasters Supported SelfCare, McKinsey & Company, and the National Pharmaceutical Council. He has equity in and serves on the board of directors of Cancer Consultants, has equity in and serves on the executive board of Faculty Connection LLC, and has equity in Alnylam Pharmaceuticals. A detailed listing of Dr. Schulman's financial disclosures is available at http://www.dcri.duke.edu/research/coi.jsp
Departmental funds were used to provide salary support for Dr. Scales. There were no other sources of funding for this study.
See editorial on pages 1278–81, this issue.
Disagreement exists on the use of prostate-specific antigen (PSA) tests for cancer-risk stratification in young men in the United States. Little is known about the use of PSA testing in these men. To understand policy implications of risk stratification, the authors sought to characterize PSA use among young men.
The authors used the 2002 Behavioral Risk Factor Surveillance System to study prostate-cancer screening in a representative sample of men aged 40 years and older (n = 58,511). The primary outcome was self-report of a PSA test in the previous year.
Among men aged 40 to 49 years, 22.5% (95% confidence interval [CI], 21.5-23.5) reported having had a PSA test in the previous year, compared with 53.7% (95% CI, 52.8-54.7; P < .001) of men aged ≥50 years. When sociodemographic characteristics were statistically controlled, young, black, non-Hispanic men were more likely than young, white, non-Hispanic men to report having had a PSA test in the previous year (odds ratio [OR], 2.42; 95% CI, 1.95-3.01; P < .001). In young men, annual household income ≥$35,000 (OR, 1.50; 95% CI, 1.26-1.78; P < .001) and an ongoing relationship with a physician (OR, 2.52; 95% CI, 2.06-3.07; P < .001) were associated with PSA testing.
A patient and his physician should discuss risks and benefits of screening and intervention for prostate cancer. For men at average risk who are aged 50 years or older and have a life expectancy of at least 10 years, the American Cancer Society and the American Urological Association recommend prostate cancer screening.1–3 For higher risk groups, such as African-American men, the recommendation is that screening begin at age 40 years.
Whether prostate cancer screening should be expanded to include average-risk younger men is a subject of debate in the medical literature.4, 5 The incidence of prostate cancer in this age group is low,6–9 and the clinical significance of detected prostate cancer in these men is unclear.10, 11 Recent data suggest that a prostate-specific antigen (PSA) value above 0.7 ng/mL in young men is associated with greater risk for development of prostate cancer.12 However, the men in this cohort had above-average risk, so the usefulness of generalizing this finding is not known.
PSA tests are common,13 may be overused in certain populations,14–16 and may be associated with misaligned financial incentives for physician practices.15 However, little is known about the extent of PSA screening among young men aged 40 to 49 years. Understanding the use of PSA screening in younger men will inform impact assessments and policy discussions of new risk stratification strategies for prostate cancer detection in this age group. Therefore, the objective of this study was to characterize the extent of and sociodemographic factors associated with PSA screening among men aged younger than 50 years.
MATERIALS AND METHODS
We obtained data from the 2002 Behavioral Risk Factor Surveillance System (BRFSS), an annual, population-based survey coordinated by the Centers for Disease Control and Prevention. BRFSS is a survey of civilian, noninstitutionalized adults who reside in households in the United States. The survey uses random-digit dialing to identify a probability sample of all households with telephones. One adult per household answers questions about his or her risk behaviors and healthcare use on 1 or more of the 10 leading causes of mortality.17
The study size was determined by the availability of response data to prostate cancer screening questions in the BRFSS. In 2002, 99,262 men completed the BRFSS survey. For this study, we excluded men aged younger than 40 years (n = 34,743) because they were not asked about prostate cancer screening. We also excluded men who reported a history of prostate cancer (n = 2524) and men who did not respond to questions about prostate cancer screening (n = 3484). The final sample consisted of 58,511 men aged 40 years or older.
The institutional review board of the Duke University Health System determined that the study was exempt from the requirement for approval.
Participants were asked a series of questions on prostate cancer screening (Fig. 1) and other healthcare issues. The questions did not ask whether a PSA test and digital rectal examination occurred together. Because the PSA test is the more sensitive test for prostate cancer, we selected it as the primary outcome. Participants were also asked about their general health and access to healthcare, including insurance status and whether they had a “personal physician or healthcare provider.” Finally, participants were asked about their sociodemographic characteristics.
To facilitate the analysis, we combined some response categories. For example, we grouped respondents who had less than a high school education into a single category. Similarly, we grouped respondents who were married or part of an unmarried couple into a single category; and we categorized respondents who were divorced, separated, never married, or widowed as “unmarried.” For employment status, we classified students, homemakers, retirees, and persons unable to work as not being in the labor force. We divided annual household income into 2 categories, <$35,000 and ≥$35,000. Race/ethnicity were reported by participants to the BRFSS survey administrators. In this analysis, we used the reported categories “white, non-Hispanic,” “black, non-Hispanic,” and “Hispanic” and combined all others as “other, non-Hispanic.”
We used Rao-Scott χ2 tests to compare weighted frequencies of prostate cancer screening.18 Results were considered statistically significant when 2-sided α < .05. We used a logistic regression model to evaluate PSA testing in each age group (aged 40 to 49 years and aged 50 years and older), controlling for race/ethnicity, access to care, education level, marital status, annual household income, employment status, general health, insurance status, and whether the participant had a personal physician. We retained participants with missing data for race/ethnicity, income, education level, employment status, marital status, general health, insurance status, having a personal physician, and having diabetes mellitus. We included dummy variables in the model to designate missing data. Missing responses for each variable ranged from 1% to 7%.
We used BRFSS sampling weights in all analyses, which account for the complex survey structure, demographic differences, and nonresponse. The sampling weights also account for the noncoverage error that results from the exclusion of persons living in nonresidential settings.17 We used SAS version 9.1 (SAS Institute, Cary, NC) for all statistical analyses.
Table 1 shows characteristics of the respondents. Among the 58,511 respondents, 20,388 (34.8%) were aged 40 to 49 years. Table 2 shows testing rates for the younger age group (aged 40 to 49 years). Older men were more likely than younger men to have ever had a PSA test (73.9% vs 36.7%; P < .001) and to have had a PSA test in the previous year (53.7% vs 22.5%; P < .001).
Table 1. Characteristics of the Respondents by Age Group
Values are expressed as weighted percentages unless otherwise indicated. The weighted percentages account for differences in demographic parameters and sampling weights and are representative of the US male population.
Age, mean (SE), y
Body mass index
Normal weight (<25.0 kg/m2)
Overweight (25.0-25.9 kg/m2)
Obese (≥30 kg/m2)
Annual household income
Less than high school
High school graduate
Not in the labor force
Health insurance coverage
Has at least 1 personal doctor
Table 2. Proportion of Respondents Who Reported Having Had a PSA Test*
PSA Test in Previous Year (95% CI)
PSA Test Ever (95% CI)
PSA indicates prostate-specific antigen.
Values are expressed as weighted percentage (95% confidence interval) unless otherwise indicated. The weighted percentages account for differences in demographic parameters and sampling bias and are representative of the US male population.
P<.001 versus black, non-Hispanic men aged 40 to 49 years.
Several sociodemographic characteristics were associated with PSA screening in younger men. As shown in Table 2, a higher proportion of black, non-Hispanic men reported having had a PSA test in the previous year than white, non-Hispanic men (33.6% vs 21.5%; P < .001). These differences persisted after we controlled for important sociodemographic characteristics. Table 3 shows the results of the multivariate analysis. Black, non-Hispanic men in the younger age group remained more likely than white, non-Hispanic men to report having had a PSA test in the previous year (odds ratio [OR], 2.43; 95% confidence interval [CI], 1.95-3.01; P < .001). We found similar results for digital rectal examination, as black, non-Hispanic young men had greater odds than white, non-Hispanic young men of undergoing digital rectal examination (OR, 1.45; 95% CI, 1.18-1.79; P < .001).
Table 3. Odds Ratios for PSA Screening in the Previous Year by Age Group
Age 40-49 Years OR (95% CI)
Age ≥50 Years OR (95% CI)
1.12 (1.10 to 1.15)
1.02 (1.02 to 1.03)
2.43 (1.95 to 3.02)
1.29 (1.09 to 1.52)
1.49 (0.85 to 2.63)
0.83 (0.59 to 1.15)
0.78 (0.56 to 1.11)
0.72 (0.55 to 0.94)
1.44 (1.11 to 1.87)
1.22 (1.00 to 1.48)
1.00 (0.60 to 1.65)
1.33 (0.95 to 1.87)
Body mass index
Normal (<25.0 kg/m2)
Overweight (25.0-29.9 kg/m2)
1.24 (1.08 to 1.44)
1.21 (1.11 to 1.33)
Obese (≥30.0 kg/m2)
1.48 (1.24 to 1.75)
1.29 (1.16 to 1.44)
Annual household income
1.50 (1.26 to 1.78)
1.45 (1.31 to 1.60)
1.55 (1.18 to 2.04)
1.47 (1.27 to 1.70)
1.29 (0.88 to 1.88)
0.83 (0.67 to 1.02)
Less than high school
High school graduate
1.24 (0.93 to 1.65)
1.34 (1.17 to 1.54)
1.29 (0.96 to 1.72)
1.53 (1.32 to 1.77)
1.42 (1.07 to 1.89)
1.73 (1.50 to 2.00)
0.04 (0.01 to 0.44)
1.25 (0.54 to 2.92)
1.07 (0.90 to 1.27)
0.94 (0.83 to 1.07)
0.98 (0.74 to 1.30)
0.85 (0.68 to 1.06)
Not in labor force
1.37 (1.03 to 1.82)
1.30 (1.17 to 1.45)
1.70 (0.44 to 6.63)
1.01 (0.47 to 2.18)
1.01 (0.88 to 1.17)
1.20 (1.10 to 1.30)
2.19 (0.53 to 8.99)
2.53 (0.72 to 8.91)
0.99 (0.85 to 1.16)
0.89 (0.80 to 1.00)
1.01 (0.85 to 1.20)
1.05 (0.94 to 1.18)
1.09 (0.82 to 1.44)
0.94 (0.82 to 1.08)
1.16 (0.78 to 1.73)
0.87 (0.72 to 1.03)
2.48 (0.90 to 6.87)
0.69 (0.41 to 1.16)
Health insurance coverage
1.78 (1.41 to 2.24)
1.48 (1.25 to 1.75)
0.27 (0.03 to 2.19)
0.83 (0.26 to 2.65)
Has at least 1 personal doctor
2.51 (2.05 to 3.07)
3.15 (2.75 to 3.61)
1.59 (0.15 to 16.4)
1.61 (0.58 to 8.92)
1.29 (1.00 to 1.67)
1.06 (0.94 to 1.20)
0.41 (0.12 to 1.47)
1.11 (0.61 to 2.04)
Other sociodemographic characteristics were also associated with PSA testing among young men (Table 3). Younger men with an annual household income of ≥$35,000 were more likely than younger men with annual household income of <$35,000 to have had a PSA test in the previous year (OR, 1.50; 95% CI, 1.26-1.78; P < .001). Access to care had a strong association with prostate cancer testing in young men. The odds of reporting a PSA test were greater among respondents who had an ongoing relationship with a physician (OR, 2.5; 95% CI, 2.05-3.07; P < .001). Insurance coverage was also positively associated with having had a PSA test (OR, 1.78; 95% CI, 1.41-2.24; P < .001). Finally, a higher education level was associated with having had a PSA test (OR, 1.42; 95% CI, 1.07-1.89; P < .001 for college graduates compared with respondents with less than a high school education).
Associations between having had a PSA test and sociodemographic characteristics differed between younger men and older men (Table 3). Increasing age had a stronger association with PSA testing among young men (OR, 1.12 per year; 95% CI, 1.10-1.15) when compared with older men (OR, 1.02 per year; 95% CI, 1.02-1.03). Although black, non-Hispanic men of all ages were more likely to report having had a PSA test, this effect was stronger among younger men. Marital status was not significantly associated with having had a PSA test among younger men but was positively associated with having had a PSA test among older men (OR, 1.20; 95% CI, 1.10-1.30; P < .001).
In this cross-sectional analysis of PSA testing among men aged 40 to 49 years, we found that PSA tests were not uncommon among young men. Approximately 1 in 5 younger men reported having had a PSA test in the previous year, and just over 1 in 3 reported ever having had a PSA test.
Sociodemographic characteristics were strongly associated with having a PSA test among younger men. Most notably, young black, non-Hispanic men were more likely than young white, non-Hispanic men to report having had a PSA test. This finding persisted after we controlled for important covariates such as income, education, and access to care. These results suggest that patient and provider practices now reflect the known risk differential for prostate cancer between black, non-Hispanic men and white, non-Hispanic men. However, PSA screening in this group remains potentially suboptimal; only one-third of these higher risk men reported having had a PSA test in the previous year.
Although we cannot exclude the possibility that this finding is the result of sampling bias in the survey design, the literature also suggests that the gap in PSA screening rates between black and white men has been closing. In the early 1990s, black men had lower odds than white men of having had a PSA test.19, 20 As PSA screening became more widespread in the late 1990s, screening among white men increased and stabilized.21 PSA screening among black men, although lower than in white men, increased but had not yet stabilized by 1998.21 In the 2000 National Health Interview Survey and the 2001 California Health Interview Survey, screening rates were not statistically different between black men and white men.22, 23 By using data from 2002 through 2004, the Southern Community Cohort Study documented higher PSA screening rates among young African-American men compared with white men, controlling for insurance status, marital status, household income, and education level.24 These results lend validity to our finding that among black, non-Hispanic men, the odds of having had a PSA screening are now greater.
Younger Hispanic men were more likely to undergo PSA testing than younger white, non-Hispanic men. PSA screening patterns in Hispanic men are not well documented (as compared with African-American men) in the medical literature. Although we can not exclude the possibility of sampling bias in the survey design, the proportion of Hispanic men undergoing PSA testing in this investigation is similar to findings from other data.25 Black men of Hispanic ethnicity may appropriately be screened at a younger age, although this group comprises only a small proportion of Hispanic men in our sample. Data from the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial suggest that Hispanic men have higher PSA levels than similar white men,26 a finding that may prompt providers to pursue PSA screening in Hispanic men more aggressively. Further investigation is required to understand the reasons for this observation.
Other sociodemographic characteristics were also associated with having had a PSA test. The probability of undergoing a PSA test increased with increasing obesity, an association that has been noted previously.27, 28 Higher household income and education level were associated with higher screening rates, as has been noted for other cancers.28–30 Health insurance coverage and an ongoing relationship with a physician were also strongly associated with having had a PSA test. Suboptimal processes of care for patients with poor access represent a systemic problem for cancer screening in the United States.
Although endorsed by several professional societies, prostate cancer screening remains controversial. There is much debate over both the sensitivity of PSA tests and the ultimate effect on morbidity and mortality of detecting prostate cancer at early, potentially clinically insignificant stages.31–34 Early detection is potentially more important in higher risk groups, such as African-American men and those with a family history of prostate cancer. For this reason, screening is recommended for younger men in these higher risk groups.1–3
In 2006, the National Comprehensive Cancer Network (NCCN) adopted by “nonuniform consensus” a risk-stratification strategy for young men that included a “baseline” PSA measurement at age 40 years and subsequent risk stratification based on the result of this initial test.10 The NCCN early detection protocol is supported by evidence that young men with PSA levels above the age-specific median are at greater risk for prostate cancer. In the Baltimore Longitudinal Study of Aging, the relative risk of prostate cancer was 3.75 for men aged 40 to 49 years with a PSA level above the age-specific median of 0.6 ng/mL, whereas the risk was similar among men in their 50s with PSA levels above and below the age-specific median.35 Similarly, among men aged 40 to 49 years who were at greater risk (either because of positive family history or African-American race), a PSA level between the age-specific median (0.7 ng/mL) and 2.5 ng/mL was associated with a higher risk of prostate cancer, with screening of 36 higher risk men required to detect 1 additional case of cancer. The external validity of this finding is unclear; it is not known how many young men at average risk would need a PSA test to detect 1 additional case of cancer.12
Overdiagnosis of prostate cancer using current detection strategies is not insignificant33, 34 and leads to potentially avoidable harm in the course of treatment.36–38 Etzioni et al34 estimated an overdiagnosis rate of 29% among white patients aged ≥60 years on the basis of model comparisons to the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) database. Similarly, by using data from the European Randomized Study of Screening for Prostate Cancer, Draisma et al33 found an overdiagnosis rate of 27% for men aged 55 years. High overdiagnosis rates make the benefit of prostate cancer screening unclear, but the adverse effects of treatment are clear and are similar for clinically detected and screening-related cancers.38 If all of the 2.1 million 40-year-old men in the United States39 underwent PSA testing, the result could be up to 48,000 additional biopsies performed for elevated PSA levels alone, with up to 12,000 men diagnosed with prostate cancer.40, 41 The incidence of treatment-related side effects (eg, impotence, incontinence) are well known, and given current life expectancy among men at age 40 years, the effects of treatment could result in a significant decrement in quality-adjusted life-years. These costs should be balanced against the potential benefits of prostate cancer screening in young men, which may include years of life saved and decreases in the costs of advanced disease management. Several cost-benefit analyses of prostate cancer screening exist in the literature.42–44 Coley et al44 estimated that the marginal cost of screening and treatment under favorable assumptions to be approximately $12,500 to $19,000 per year of life saved. However, the costs of screening and treatment became unfavorable as assumptions varied. Given the well recognized limitations of retrospectively identified screening strategies and cost-benefit models,36, 45 results of ongoing prospective PSA screening studies46–48 are needed to define the treatment benefit for prostate cancer diagnosed through screening.
In addition to the burden of treatment for potentially clinically insignificant prostate cancer, the policy implications of baseline PSA screening in men at age 40 years are potentially significant. Prostate cancer among men aged younger than 50 years has comprised 0.8% to 1.1% of cases,6, 8 although a more recent analysis estimated a frequency as high as 4%.7 According to data from the SEER database, the incidence of prostate cancer at the age of proposed baseline testing is 0.5%.9 Thus, the large majority of men in this age group who have a PSA test will not be diagnosed with prostate cancer, although some may develop it later in life. Additional baseline testing of young men will likely have substantial costs, both direct and indirect.
Healthcare resource use associated with PSA testing could increase substantially under the NCCN protocol. On the basis of our findings, 22.5% of men aged 40 to 49 years have a PSA test annually. In the United States, there are approximately 22.4 million men aged 40 to 49 years.39 This indicates that there are approximately 5 million annual PSA tests in this age group. If all men underwent PSA testing at age 40 years (2.1 million39 is the current US population of 40-year-old men), the sum would be 2 million annual tests. For men with PSA levels above the age-specific median, annual PSA testing with digital rectal examination is recommended (eg, up to 1 million annual PSA tests). Under this protocol, the number of PSA tests in men aged 40 to 49 years would more than double.
Our study has several limitations. First, results are based on patient self-report and may not correspond with actual screening behavior. PSA tests may not be used exclusively for prostate cancer screening, so the testing rates we observed may overestimate screening for prostate cancer alone. However, respondents were asked about PSA tests in the context of prostate cancer screening, and the prevalence of benign prostatic hyperplasia is low in young men, so we anticipate small effects from this limitation. Men younger than 40 years were not asked questions on prostate cancer screening. Thus, our results may underestimate the extent of PSA screening in men younger than 50 years. Odds ratios may exaggerate relative risks for PSA testing because of the relatively high proportion of men who were undergoing PSA testing in the study sample. As a cross-sectional survey, the BRFSS lacks data on outcomes after PSA screening. Thus, longer term outcomes, such as prostate biopsy, prostate cancer diagnosis and Gleason grade, and prostate cancer treatment, could not be addressed in this sample. Finally, the BRFSS did not include family history of prostate cancer, a covariate of particular importance in younger populations. A useful extension of this study would investigate whether men with a family history of prostate cancer are screened at a younger age.
These limitations notwithstanding, we believe the findings are robust. The BRFSS is designed to ascertain risk behaviors in a large, nationally representative sample. Our study is the first to specifically examine PSA screening in younger men, which provides an important assessment of quality of care, especially for high-risk groups. These findings have potential implications for PSA screening policies in young men. First, testing among black, non-Hispanic men remains suboptimal, as measured by current guidelines. Therefore, targeted efforts to increase screening in this population may be indicated. Second, characterizing the extent of PSA screening in young men informs resource use estimates in this population. Third, information about PSA screening rates in young men can be used to assess the impact of the new NCCN guidelines. Also, our analysis controls for factors that are known to influence cancer screening behavior, such as having an ongoing relationship with a physician, marital status, and income. Finally, we included dummy variables for missing responses (“don't know” or “refused”) to account for potential associations between item nonresponse and screening behavior.
In this cross-sectional analysis of a nationally representative sample of men aged 40 years or older in the United States, we found that approximately 1 in 5 men aged 40 to 49 years undergo PSA testing and that black, non-Hispanic men are tested at a higher rate, reflecting the current understanding of a higher prostate cancer risk in this group. Despite this advance, only one-third of young black, non-Hispanic men report having had an annual PSA test, so quality of care remains suboptimal in this population. Further investigation will be required to understand the impact of new risk-stratification strategies, with particular focus on policy implications of potentially large increases in healthcare resource use.
The authors thank Damon Seils of Duke University for editorial assistance and manuscript preparation. Mr. Seils did not receive compensation for his assistance apart from his employment at the institution where the study was conducted.