Use of the Prostate-Specific Antigen Test among U.S. Men: Findings from the 2005 National Health Interview Survey

Prostate cancer is the most common cancer, other than some types of skin cancer, and the second leading cause of cancer mortality among men in the United States (1). According to the Centers for Disease Control and Prevention, an estimated 189,075 American men were diagnosed with prostate cancer and 29,002 died from the disease in 2004 (1). Although prostate cancer incidence varies, at least to some extent, by demographic, social, and health characteristics, the overall incidence of prostate cancer has remained stable over the last decade and its mortality has fallen (2).

The U.S. Food and Drug Administration approved the prostate-specific antigen (PSA) test more than a decade ago for monitoring and early detection of prostate cancer (3). In 2002, the U.S. Preventive Services Task Force (USPSTF) concluded that the evidence was insufficient to advocate for or against routine prostate cancer screening using the PSA test and digital rectal examination (4). Specifically, the USPSTF determined that older men with comorbid conditions and a life expectancy of <10 years were unlikely to benefit from prostate screening (4). However, the USPSTF, under clinical considerations, indicated that if early detection improves health outcomes (which is not currently known), those most likely to benefit from screening are men at average risk of the disease ages 50 to 70 years and men ages ≥45 years at increased risk, such as African American men and those who have a first-degree relative with prostate cancer (4). The USPSTF and other organizations also recommended that physicians discuss potential benefits and risks associated with prostate cancer testing with their patients (4, 5).

Because African American men and men with a family history of prostate cancer are at increased risk of developing prostate cancer at a younger age than other men, the American Cancer Society recommends that physicians offer these patients the option of having a PSA test and digital rectal examination starting at age 45 (5). The American Cancer Society also recommends that men with several first-degree relatives who had prostate cancer at a young age begin testing at age 40 (5). Unlike the USPSTF, the American Cancer Society does not recommend that prostate cancer screening stop at age 70 (5). PSA testing for prostate cancer is common (4), and whether the test is done at the request of the physician or patient, >50% of men ages ≥50 years have had at least one PSA test in their lifetime (6-8). Previous research on PSA testing indicates that testing rates vary by age, race, and socioeconomic characteristics (6, 9).

Although PSA testing has become more common in the past decade, limited published data exist about the rates of PSA testing in subpopulations of men with demographic or social characteristics that could influence the use of PSA. We therefore conducted our analysis to compare PSA test use in men ages 40 to 49 years with use in older men using data from the 2005 National Health Interview Survey (NHIS; ref. 10). Specifically, we examined PSA test use along with demographic, socioeconomic, and healthcare access factors identified in the literature that might influence test use (6, 7, 11).

The NHIS is an annual health survey conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention (10). The survey represents the noninstitutionalized adult population in the United States. The NHIS has several core questionnaires with basic demographic and health items that it incorporates each year as well as one or more sets of questions (“supplements”) on current health topics that vary from year to year. The 2005 NHIS survey included a cancer supplement that contained items on family history of prostate cancer and prostate cancer screening using the PSA test. The survey information was collected through computer-assisted personal interviews conducted by trained U.S. Census Bureau staff. The adult sample included persons ages ≥18 years residing in the household. One adult per family was randomly selected for the interview. Black and Hispanic populations were oversampled to allow more precise estimates.

A total of 31,428 adults in 39,284 sampled families were interviewed. The final adult response rate was 69.0% (12). Of a sample of 7,885 men ages 40 to 79 years, the current analysis excluded 201 men who had a history of prostate cancer before their interview and 15 men for whom information about prostate cancer was missing. Of the remaining 7,669 men, 6,854 reported information about PSA testing, 164 men refused to answer whether they had a test, 242 did not know the answer to that question, and the information for 336 men was not ascertained. Additional 73 men were excluded because of missing information regarding the time of their most recent test.

Male respondents were asked several questions about prostate cancer and PSA testing, including whether they had ever had a PSA test and, if so, the time of their most recent test and the main reason for undergoing it. The 2005 NHIS did not collect information on digital rectal examination.

The National Center for Health Statistics recoded data on race or ethnicity into four categories: non-Hispanic White, non-Hispanic Black, Hispanic, and non-Hispanic other. Hispanics included Puerto Rican, Mexican, Mexican American, Cuban or Cuban American, Central or South American, and other Spanish. Poverty status was calculated as the income-to-poverty ratio and based on an index developed from U.S. Census Bureau information on 2004 poverty levels based on income (13).

We determined access to healthcare using three variables: (a) whether respondents had no health insurance, public health insurance, or private or military health insurance; (b) whether respondents had a usual source of care (excluding emergency room visits); and (c) whether respondents had seen a doctor or healthcare provider in the past year. The health status characteristics that might have influenced PSA test use according to the literature that we analyzed in this study included self-reported overall health, having a family history of prostate cancer (respondent's father, brother, or son was diagnosed previously with prostate cancer), number of self-reported chronic diseases (including hypertension, stroke, emphysema, diabetes, chronic bronchitis, failing kidneys, liver condition, and heart disease), and having ever been diagnosed with any cancer other than prostate cancer. We also recoded self-reported health status as excellent, very good, good, or fair or poor. These and other variables that could influence PSA test use are presented in Table 1.

The NHIS used a stratified, multistage-cluster sample (13). We used the statistical program SUDAAN (version 9.1) in the analysis to account for the complex sampling survey design and nonresponse (14). We obtained sample weights from the NHIS 2005 public use data file to calculate population estimates and their 95% confidence intervals.

We examined PSA test use for any reason in the previous 2 years. The decision to examine test use in the past 2 years rather than the past year intended to reduce the number of inaccurate responses about PSA testing dates that might have occurred if the testing dates were earlier than reported by participants. We decided to include PSA tests conducted for any purpose in this study after completing an additional analysis using only PSA tests for routine examination and obtaining very similar results to those of the current analysis.

We stratified this analysis into two age groups (40-49 and 50-79 years) to explore differences in demographic and health characteristics associated with PSA test use in these populations. We used χ² tests, which are analogous to the standard Pearson χ² tests for nonsurvey data, to examine the association between each covariate and PSA test use in the past 2 years. We used multivariate logistic regression models to estimate the percentages of men having had a PSA test in the past 2 years after adjusting for all other covariates in the models. We removed the covariate “seen physician or healthcare provider in the past year” from the multivariate analysis because the period mentioned did not match the period used in our analysis. We also removed from the analyses data with missing information due to a refusal to answer or a lack of knowledge about the time of the most recent PSA test.

We present the results as adjusted percentages or predicted margins. We calculated the predicted margins for each subgroup from the logistic regression model as the average of the predicted probabilities of PSA test use, assuming that all survey participants were in that subgroup (15). These adjusted percentages allow easy comparisons among categories within a covariate. We used general linear contrasts to assess the statistical differences of estimates compared with a reference level within a factor.

Of the 7,885 men ages 40 to 79 years in our sample, 7,669 had no history of prostate cancer, representing an estimated population of 55.8 million persons. Of these men, 2,777 were 40 to 49 years old and 4,892 were 50 to 79 years old (Fig. 1). Among men who underwent PSA testing, most had had their most recent test during a routine examination (86.5% for ages 40-49 years and 90.3% for ages 50-79 years). However, 9.0% of those ages 40 to 49 years and 7.2% ages 50 to 79 years were tested because of a problem.

The distributions of respondents' characteristics, which reflected the U.S. male population, are presented in Table 1. Men were mainly non-Hispanic White men, born in the United States, and married or living with a partner. Most had public or private health insurance and a usual source of care. However, compared with men ages 50 to 79 years, a larger percentage of those ages 40 to 49 years were Hispanic or never married, lacked health insurance or a usual source of medical care, reported being in excellent health and less often in fair or poor condition, and had no comorbid conditions and other types of cancer.

We present findings from the unadjusted analyses, in which we modeled the association between personal characteristics and PSA test use in the past 2 years (Table 2). Among men reporting information about PSA testing, an estimated 16% of men ages 40 to 49 years and 49% of men ages 50 to 79 years reported having had a PSA test in the past 2 years. Among the younger men, use of PSA test was highest in non-Hispanic Black men (23%), those from the northeast, college graduates, those with the highest poverty threshold, married men, those with private or military health insurance and a usual source of care, those with family history of prostate cancer, those who had a chronic disease, and those who had been diagnosed with another type of cancer. Men with less than high school education, those who had never married, and those who had no health insurance had the lowest percentages of PSA test use.

Men ages 50 to 79 years who had the highest percentages of PSA test use had similar characteristics to men ages 40 to 49 years with the highest percentages of testing. In addition, PSA test use in the older group was also highest among those ages 65 to 79 years, among non-Hispanic White men, those born in the United States, and those with one or more chronic diseases.

PSA test use within the past 2 years, adjusted for demographic and personal characteristics, is presented in Table 3. For most of the covariates, the patterns of PSA testing for both age groups were similar to those presented in Table 2. However, among men ages 40 to 49 years, non-Hispanic Black men and Hispanic men (with borderline significance) were more likely to be tested than non-Hispanic White men. Among men ages 50 to 79 years, Hispanic men had a higher percentage of test use than all other race or ethnicity groups. However, the difference was not significant.

Findings from this study revealed that among men reporting information about PSA testing 16% of men ages 40 to 49 years and 49% of those ages 50 to 79 years had a PSA test for any reason during the past 2 years. Most tests were done for routine exams, and the highest percentage of test use was among men ages 65 to 79 years. Our study showed that although many personal and health-related characteristics associated with PSA test use are similar in both younger and older men, test use varies by race and ethnicity in the two age groups.

In particular, we found that non-Hispanic Black men ages 40 to 49 years were more likely to have had a PSA test than men of other racial or ethnic backgrounds in the same age group. Because Black or African American men are at higher risk of developing prostate cancer, perhaps physicians and healthcare practitioners are implementing the USPSTF clinical recommendations (4). They may have discussed with African American patients their increased prostate cancer risk, and this knowledge may have translated into more testing, especially among men ages <50 years (16-18). Furthermore, increased media coverage and other community interventions (19-22) might have increased awareness of this disease.

By combining race and ethnicity into one variable with four distinct groups and examining it in stratified analyses of men ages 40 to 49 and 50 to 79 years, the current study offers additional insight into variations in PSA test use. Our multivariate analysis showed that the percentage of test use among non-Hispanic Black men ages 50 to 79 years was similar to the percentage among non-Hispanic White men. This study does not reflect the trend found in a study of the relationship between age, race, and prostate cancer screening, which found that PSA test use in elderly African American men has been increasing more rapidly than in White men (23).

The higher percentage of test use we found among Hispanic men than non-Hispanic White men in the multivariate analyses suggests the possibility that physicians are recommending testing to Hispanic men as part of their routine exam despite the lower incidence of prostate cancer in this population. The Hispanic population is heterogeneous and a more comprehensive analysis is required to explore the nature of variation within this population.

Our finding that men with higher socioeconomic status were more likely to be tested might be partially explained by the fact that these men tend to have greater access to healthcare and community resources and perhaps more knowledge of the disease, as shown in previous studies (6, 7, 11). Our finding that PSA test use was low among men who were divorced, widowed, and especially those who had never married might indicate that spouses and partners play an important role in encouraging men to be tested. This finding of the lowest testing percentages among single, never-married men was also presented in a previous study (7).

Perhaps because family history is a well-known risk factor for prostate cancer, PSA testing was greater among men who reported a family history of prostate cancer. The healthcare providers of these men might have encouraged these patients to consider testing. Previous studies show that when healthcare providers advise patients to undergo PSA testing, the overwhelming majority tend to comply (6), particularly if they have health insurance and a usual source of care (7).

An earlier study found that disparities in prostate cancer screening rates between patients who lacked a usual source of care and health insurance and those who had access to these resources are widening (11). Our study also found significant differences in PSA test use between men with private or military health insurance and a usual source of care and those who lacked access to these resources. Our finding that test use among men who had public insurance was similar to that of men with no insurance is noteworthy because variations in health insurance might partially explain the lack of access to preventive healthcare services, such as PSA testing. A recent systematic review of prostate cancer estimated the cost of a PSA test to range from $13.11 to $77.18, with an average of $37.23 (24). This average cost might be prohibitive for those without private or public health insurance, but it should be affordable for those with public insurance compared with other publicly funded cancer screening programs (25).

This study has some limitations. As in any interview- or survey-based study, participants might have misreported the date of their last PSA test because they had difficulty recalling the date accurately (26, 27). The rate of misreporting might have varied by personal and social characteristics, which could have led to bias (28). Also, because of small numbers in the sample, data on American Indians or Alaska natives, Asians, or other races were combined into one group, “non-Hispanic other,” in the race or ethnicity variable. Therefore, our results might not be representative of these populations.

Additionally, although a 69% response rate is quite substantial, an incomplete coverage of the population of men may limit our ability to draw inferences from the results if there are systematic differences between the survey respondents and nonrespondents. Nevertheless, we do not expect a distortion in the association between men's characteristics and PSA test use. Finally, because we did not collect data on type of PSA test used (such as complex PSA, free-to-total PSA, and other measures of PSA), this analysis is limited to any PSA testing. Although we excluded from our analysis men who reported having been diagnosed with prostate cancer, a small proportion of PSA tests reported in our sample had been conducted for diagnostic purposes rather than routine screening.

The strengths of this study stem from the fact that the data are from a large, well-established national survey. In addition, our examination of two age groups offers additional insight into the use of the PSA test.

Our findings could have important implications for medical and public health professionals. Our results imply that physicians and patients might be recognizing the fact that Black men are at increased risk of prostate cancer because, in spite of inconclusive evidence that PSA screening is beneficial, non-Hispanic Black men are tested more often at younger ages than non-Hispanic White men. Also, media and community campaigns that increase awareness of prostate cancer risk among Black men might have become more successful in promoting screening (20-22). However, the higher use of testing among Hispanic men than non-Hispanic White men in the multivariate analyses merits further examination to understand which factors are associated with increased PSA test use.