Prostate-specific antigen testing in older men in the USA: data from the behavioral risk factor surveillance system

Authors


Edward Messing, 601 Elmwood Avenue, Box 656, Rochester, NY 14642 USA. e-mail: Edward_Messing@urmc.rochester.edu

Abstract

Study Type – Diagnosis (cohort)

Level of Evidence 2a

What's known on the subject? and What does the study add?

Previous studies from the USA reported that, in recent years, prostate cancer screening with PSA was very common among the elderly men, including those whose life expectancy was substantially limited by advanced age and known comorbidities. In 2008, the US Preventive Services Task Force (USPSTF) recommended against PSA screening after age 75.

The purpose of this study was to examine the impact of the USPSTF 2008 recommendation on the frequency of PSA testing among elderly men in the USA. Our findings suggest that the USPSTF recommendation had no major impact on clinical practice.

OBJECTIVE

  • • To examine the frequency of PSA testing in men aged ≥75 years before and after the 2008 US Preventive Services Task Force (USPSTF) recommendation to stop prostate-specific antigen (PSA) screening at age 75.

MATERIALS AND METHODS

  • • Data were obtained from the Behavioral Risk Factor Surveillance System (BRFSS) surveys completed in 2006, 2008 and 2010.
  • • Men aged ≥ 76 years at the time of survey and without a prostate cancer diagnosis were included in the study.
  • • The percentage of men who had a PSA test in the year before the survey was computed separately for survey years 2006, 2008 and 2010.

RESULTS

  • • The estimated percentages of men with a PSA test in the year before the survey were 60% (95% CI: 58–62%), based on 9033 respondents interviewed in 2006, 63% (95% CI: 62–65%), based on 12 063 respondents interviewed in 2008, and 60% (95% CI: 59–61%), based on 14 782 respondents interviewed in 2010.

CONCLUSION

  • • No substantial reduction in the frequency of PSA testing was observed in the BRFSS 2010 survey data compared with the earlier years, suggesting that the USPSTF 2008 recommendation had no major impact on the frequency of PSA testing in older men in the USA.
Abbreviations
USPSTF

US Preventive Service Task Force

BRFSS

Behavioural Risk Factor Surveillance System

PC

prostate cancer

ERSPC

European Randomized Study of Screening for Prostate Cancer

NNS

number needed to screen

PLCO

Prostate Lung Colorectal and Ovarian

NHIS

National Health Interview Survey

CVD

cardiovascular disease

ACS

American Cancer Society

VHA

Veterans Health Administration.

INTRODUCTION

In the USA, prostate cancer (PC) is the second most commonly diagnosed malignancy in men (after skin cancer) and is the second most common cause of cancer death in men (after lung cancer), with an estimated 217 000 new cases and 32 000 deaths in 2010 [1]. Diagnosis and management of PC is often controversial because, in many patients, disease progression either does not occur or occurs very slowly even without treatment. As a result, many of these men die from unrelated causes before their PC becomes symptomatic or life-threatening. However, in some patients PC behaves very aggressively, producing significant local and systemic symptoms and eventually causing death.

Theoretically, PC mortality could be reduced by early detection. This question has been recently examined in three randomized trials. The European Randomized Study of Screening for Prostate Cancer (ERSPC), after a median follow-up of 9 years, reported a hazard ratio (HR) of 0.80 (95% CI: 0.65–0.98, P= 0.04) in favour of screening, with an absolute risk reduction of 0.071% and a number needed to screen (NNS) to prevent one death from PC of 1410 [2]. A Swedish trial of PC screening conducted in Norrköping, independently of the ERSPC study, found an absolute risk reduction of 0.30% (NNS = 333), favouring screening; however, this effect did not reach statistical significance after 20 years of follow-up (P= 0.065) [3]. The US Prostate Lung Colorectal and Ovarian (PLCO) cancer screening trial found no evidence of reduction in PC mortality in the screening arm relative to the control arm, when analysed using the intention-to-screen principle [4]. However, this trial reported a 52% contamination rate in the control arm owing to off-protocol screening and a 44% pre-screening rate (screening before randomization) in both intervention arms [4,5]. The PLCO trial has been re-analysed with stratification on comorbidities. After 10 years of follow-up, a significant decrease in PC mortality in the screening arm relative to control was observed in men with no or minimal comorbidities with a HR of 0.56 (95% CI: 0.33–0.95; P= 0.03), an absolute risk reduction of 0.14% and an NNS of 723 [5]. By contrast, among men with at least one significant comorbidity, screening was not associated with reduced PC mortality (HR 1.43; 95% CI: 0.96, 2.11, P= 0.08) [5].

All three trials of PC screening excluded men aged ≥75 years because it was believed that older men may not benefit from early detection of PC owing to their relatively short remaining life expectancy. In August 2008, the US Preventive Service Task Force (USPSTF) recommended against PSA screening after age 75 [6].

According to recent reports, PSA screening among US men aged ≥ 75 years was common before the 2008 USPSTF recommendation [7,8]. According to data from the 2005 National Health Interview Survey (NHIS), the estimated percentages of men with a screening PSA test in the year before the survey were 44% for men aged 75–79 years, 43% for men aged 80–84 years, and 26% for men aged ≥85 years [8]. It should be noted that the NHIS makes a distinction between a screening PSA test and a PSA test done for other reasons, and the reported percentages specifically reflect the screening frequency. In the NHIS analyses, PSA screening rates varied depending on age and estimated life expectancy, but PSA screening was fairly common even in older men with a relatively short life expectancy [7,8]. For example, in the NHIS 2005 analyses, 42% of ≥75-year-old men with a life expectancy of <5 years had a screening PSA in the 2 years before the survey [7].

Although PSA screening in older US men was common before the 2008 USPSTF recommendation, the impact of the recommendation on the screening practices after 2008 is uncertain because NHIS data on PSA screening are not available for the most recent years. An alternative approach to this question is to examine the frequency of PSA testing in older US men in years before and after year 2008 using data from the Behavioural Risk Factor Surveillance System (BRFSS) [9]. Unlike the NHIS, the BRFSS does not make a distinction between a screening PSA test and a PSA test done for symptoms; however, because PSA screening was common in older men before 2008, one would expect that if the 2008 USPSTF recommendation had a major impact on screening practices, the overall frequency of PSA testing in older men would substantially decrease after year 2008 compared with the earlier years (e.g. 2005–2007).

The primary objective of the present study was to examine the frequency of PSA testing in men ≥75 years old according to the BRFSS surveys completed in years 2006, 2008 and 2010. A secondary objective was to examine the association of various demographic variables and comorbidity indicators with PSA testing in men ≥75 years after 2008 (based on BRFSS 2010 survey data).

MATERIALS AND METHODS

Data for the present study were obtained from the BRFSS surveys completed in 2006, 2008 and 2010 [9]. These surveys were based on stratified random samples of the general US population. The surveys were administered by telephone using standard questionnaires. The survey response rate, defined as the ratio of the number of complete or partial interviews to the total number of eligible persons varied by state, with medians of 51% in year 2006, 53% in year 2008, and 55% in year 2010 [9]. For the purpose of the present study, all men aged ≥ 76 years at the time of survey who did not have a diagnosis of PC and provided information on PSA testing in the year before the survey were included in the analysis. Men aged <76 years were excluded because our interest was in PSA testing after age 75. The primary exposure variable was defined as the year of the BRFSS survey (2006, 2008 or 2010). The outcome variable was defined as history of PSA testing in the year before the survey (testing performed or not performed). Other variables included in the analysis were age at the time of survey (76–80 vs. ≥81 years), race (white, black or other), education (college, high school diploma or equivalent, no high school diploma or equivalent), marital status (married or living with a partner vs. widowed, divorced, separated, or never married), veteran status (yes or no), smoking (current, former, never), personal history of myocardial infarction or coronary heart disease (present or absent), personal history of stroke (present or absent), diagnosis of diabetes (present or absent), and body mass index (<25, 25–29, ≥30 kg/m2). Age was grouped into two categories (76–80 vs. ≥81 years) because more categories would result in a smaller sample size and low precision of estimation in all analyses stratified on age. The age-group threshold of 80 years was chosen to produce approximately equal sample sizes for the two age groups. For sensitivity analysis, age was also examined as a continuous variable.

To address the primary objective of the study, we estimated the percentage of men who had a PSA test in the year before the survey, separately for survey years 2006, 2008 and 2010. These analyses were also stratified according to age (76–80 vs. ≥81 years), and presence of cardiovascular disease (CVD) which was defined as at least one of the following: diabetes, myocardial infarction, coronary artery disease or stroke. To address the secondary objective of the study, we examined the association of the study variables with PSA testing in the past year according to the BRFSS 2010 survey in univariable and multivariable analysis using logistic regression.

All point estimates, CIs and P values reported in this paper were computed using the surveyfreq and the surveylogistic procedures in SAS, to take into account the complex sampling design of the BRFSS surveys [10,11]. Note that in these analyses, all estimates obtained from a given survey year are unbiased for the corresponding general population quantities. This assures in particular that the estimated differences in the frequency of PSA testing between the survey years 2006, 2008 and 2010 (using correct weighted estimators) are unbiased for the true differences in the population. All analyses reported in this paper were performed in SAS 9.2. All reported P values are two-sided.

RESULTS

The BRFSS datasets included a total of 12 172 men aged ≥76 years interviewed in 2006, 16 289 men aged ≥ 76 years interviewed in 2008, and 20 306 men aged ≥ 76 years interviewed in 2010. After excluding men who had a history of PC or whose PC status or PSA testing history were not known, the study sample included 9033 men interviewed in 2006, 12 063 men interviewed in 2008, and 14 782 men interviewed in 2010. Numbers and percentages of respondents remaining after application of each exclusion criterion are shown in Table 1. Information on CVD status was available for 98.5% of 9033 men interviewed in 2006, 98.4% of 12 063 men interviewed in 2008, and 98.5% of 14 782 men interviewed in 2010.

Table 1. Numbers of respondents remaining after application of exclusion criteria
 BRFSS 2006BRFSS 2008BRFSS 2010
  • *

    Percentages were computed by dividing the number in a given row by the number in the previous row (×100%).

All men aged ≥76 years12 17216 28920 306
Number remaining after excluding men with previous PC diagnosis (%*)10 273 (84)13 567 (83)16 779 (83)
Number remaining after excluding men with unknown PC diagnosis status (%*)9 779 (95)13 103 (97)15 981 (95)
Number remaining after excluding men whose PSA testing history in the past year was unknown (%*)9 033 (92)12 063 (92)14 782 (92)

The estimated percentages of men who had a PSA test in the year before the survey, stratified by year of survey, age and presence of CVD are shown in Table 2. Although these percentages varied between the survey years, the variation was small in magnitude. In particular, no substantial reduction in the frequency of PSA testing was observed in the BRFSS 2010 survey data compared with the earlier years. There was also no substantial variation in the frequency of PSA testing according to history of CVD (Table 2). Age had a noticeable association with PSA testing, with older men tested less frequently than younger men (Table 2).

Table 2. Percentage of men who had a PSA test in the past year according to the BRFSS survey, with data stratified by age, cardiovascular comorbidity status and year of survey
Age*, yearsCVDBRFSS 2006BRFSS 2008BRFSS 2010
N % tested95% CI N % tested95% CI N % tested95% CI
  • *

    Age at the time of survey.

76–80Yes23146258–663 0106764–703 5646663–68
No24826561–683 1576764–703 9206664–68
All48586361–666 2476765–697 5806664–67
≥81Yes19765753–612 7675956–623 4445552–58
No21245652–602 9355855–613 6355350–56
All41755653–595 8165856–607 2025452–56
AllAll90336058–6212 0636362–6514 7826059–61

The univariable and multivariable associations of the study variables with PSA testing in the past year, based on the BRFSS 2010 survey data, are shown in Table 3. In univariable analysis, men with a missing response on a particular variable were excluded from analyses involving this variable but were included in other analyses. Missing responses for each variable ranged from 0.1 to 1.9%. The multivariable model was based on respondents who had no missing values on any of the study variables (n= 13 835 or 94% of the 14 782 men who did not have a diagnosis of PC and provided information on PSA testing in the year before the survey, as described in Table 1). In multivariable analysis, age, education, smoking, marital status and body mass index were independent predictors of PSA testing (Table 3). These associations remained virtually the same in magnitude and statistical significance when age was entered in the model as a continuous variable (data not shown).

Table 3. Univariable and multivariable associations of the study variables with PSA testing in the past year according to the BRFSS 2010 survey
Variable N % testedCrude OR (95% CI) P Adjusted OR (95% CI) P
  • *

    Age at the time of survey. OR, odds ratio; HS,high school; Not single = married or living with a partner; single = divorced, widowed, separated or never married; MI, myocardial infarction; CHD, coronary heart disease; BMI, body mass index.

Age*, years      
 76–807 580661.61 (1.44–1.79)<0.0011.57 (1.40–1.76)<0.001
 ≥817 20254RefRefRefRef
Race      
 White13 485611.48 (1.08–2.03)0.011.23 (0.90–1.67)0.19
 Black563541.13 (0.76–1.70)0.551.14 (0.76–1.71)0.53
 Other52851RefRefRefRef
Education      
 College8 066641.94 (1.65–2.28)<0.0011.72 (1.44–2.05)<0.001
 HS diploma4 481591.58 (1.32–1.88)<0.0011.39 (1.15–1.67)<0.001
 No HS diploma2 19548RefRefRefRef
Marital status      
 Not single8 637641.61 (1.44–1.79)<0.0011.41 (1.26–1.59)<0.001
 Single6 10552RefRefRefRef
Veteran      
 Yes10 869611.17 (1.04–1.32)0.011.05 (0.92–1.19)0.50
 No3 81558RefRefRefRef
Smoking      
 Never5 340591.41 (1.07–1.87)0.011.29 (0.96–1.71)0.09
 Former8 641621.66 (1.26–2.18)<0.0011.54 (1.16–2.04)0.003
 Current70750RefRefRefRef
Ever had MI or CHD      
 Yes4 552590.92 (0.82–1.03)0.150.89 (0.78–1.00)0.06
 No9 95061RefRefRefRef
Ever had stroke      
 Yes1 629580.89 (0.76–1.05)0.160.90 (0.76–1.06)0.21
 No13 05360RefRefRefRef
Diabetes      
 Yes3 046631.15 (1.01–1.31)0.041.13 (0.98–1.30)0.10
 No11 71859RefRefRefRef
BMI, kg/m2      
 ≥302 453611.22 (1.04–1.43)0.011.08 (0.91–1.28)0.38
 25–296 775631.32 (1.18–1.49)<0.0011.20 (1.06–1.35)0.003
 <255 40256RefRefRefRef

DISCUSSION

The primary objective of the present study was to examine the frequency of PSA testing in men ≥75 years according to the BRFSS surveys completed in years 2006, 2008 and 2010. Because PSA screening was common in older men before 2008, one would expect that if the 2008 USPSTF recommendation had a major impact on screening practices, the overall frequency of PSA testing in older men would substantially decrease after 2008 compared with the earlier years. In fact, we found no substantial reduction in the frequency of PSA testing in the BRFSS 2010 data compared with the earlier years, suggesting that the USPSTF 2008 recommendation had no major impact on the frequency of PSA testing in older men in the USA.

These findings may potentially be explained by the following considerations. First, the USPSTF 2008 recommendation not to screen men after age 75 did not reflect a consensus opinion of all experts in the field and, in fact, there was a considerable disagreement on this issue. For example, according to the most recent edition of the National Comprehensive Cancer Network Guidelines on Prostate Cancer Early Detection, most experts believe that, in general, men aged >75 years have little to gain from a PSA test, which is in agreement with the USPSTF 2008 recommendation [12]. However, the AUA and the American Cancer Society (ACS) base their PSA screening recommendations on life expectancy rather than age alone [13,14]. Although the AUA and the ACS guidelines discourage PSA screening in men with a life expectancy of <10 years, many men aged ≥75 years have a life expectancy of >10 years. For example, the first quartile of life expectancy for 80-year-old men in the USA is 10.8 years, meaning that 25% of all 80-year-old men in the USA are expected to live 10.8 years or longer [15]. Given conflicting recommendations published by different professional organizations, it is possible that many primary care providers choose to order a screening PSA test for their elderly patients to ‘stay on the safe side’, particularly since PSA screening is covered by Medicare [16]. Adherence to clinical practice guidelines in general may also be influenced by a host of other factors, including physicians' lack of awareness of guideline recommendations, physicians' disagreement with guideline panel's interpretation of the evidence, patients' preferences and others [17].

Data from the NHIS surveys administered in 2000–2005 suggest that PSA testing in older men was influenced by their life expectancy, although PSA screening was fairly common even in older men with a relatively short life expectancy [7,8]. Life expectancy in NHIS data was computed by taking into account age, smoking status, body mass index, comorbidities, recent hospitalizations, self-perceived health, and functional status [8]. Because some of these variables were not available from the BRFSS database, we could not examine the association of PSA testing with life expectancy as defined in NHIS studies, although we were able to analyse the association of separate determinants of life expectancy with PSA testing. In our analysis, PSA testing was independently associated with age, smoking status and body mass index, but not with a personal history of CVD, including history of myocardial infarction/coronary heart disease, stroke or diabetes (Table 3).

The BRFSS data have been used previously to examine the frequency of PSA testing in the elderly. Specifically, Li et al. [18] used the BRFSS 2006 survey data to compute the PSA testing frequency in the year before the survey and estimated that 63% of men aged 76–80 years, 56% of men aged ≥81 years, and 60% of all men aged ≥76 years had a PSA test in the past year [18]. However, because these findings were based on data collected before 2008, they could not be used to assess the potential impact of the USPSTF 2008 recommendation on PSA testing in older men, which was the primary objective of the present study.

Recently, Zeliadt et al. [19] used data from the Veterans Health Administration (VHA) Pacific Northwest Network to examine the frequency of PSA testing in men without a PC diagnosis before and after the USPSTF 2008 recommendation. Similarly to the present study, the VHA network study did not make a distinction between a screening PSA test and a PSA test done for other reasons. However, if the screening frequency decreased substantially after 2008, the overall frequency of PSA testing would also noticeably decrease. In the VHA study, the frequency of PSA testing in men aged ≥75 years was 25% in the 6-month period just before the USPSTF recommendation, 24% in the 6-month period after the recommendation, and 22% in the 6-month period ending in March 2010. Although PSA testing frequency in older men in the VHA population decreased after the 2008 USPSTF recommendation, the magnitude of this change was relatively small. It should be noted that the VHA data analysis was based on the percentages of men who had a PSA test within a 6-month period, while we considered the 12-month period before the BRFSS survey in the present study, hence, percentages reported in our paper are higher. In addition, the VHA study was geographically limited to the Pacific Northwest and, within this area, substantial regional variation in the frequency of PSA testing was present [19]; therefore, it is possible that the VHA data for the Pacific Northwest may not be fully representative of the entire country. By contrast, the present analyses were based on the survey sample representing all geographical regions of the USA.

The present study was focused on men >75 years old at the time of survey because this age limit was used in the USPSTF 2008 recommendation. Patterns of PSA testing in younger men have been recently reported based on NHIS and BRFSS data [7,20]. For example, according to the BRFSS 2006 survey, the percentages of men with a PSA test in the past year before the survey were 12% at age 40 years, 33% at age 50 years, 60% at age 60 years, and 68% at age 70 years [20].

The strengths of the present study include a fairly large sample size representing all geographical regions of the USA, a large number of variables available for analysis, and a relatively small percentage of respondents with missing data on the study variables. However, the following limitations of the present study must be recognized. First, the BRFSS survey does not make a distinction between a screening PSA test and a PSA test done for other reasons, such as LUTS, so our analyses could not be specifically focused on screening. However, because PSA screening was common in older US men before 2008, one would expect that if the 2008 USPSTF recommendation had a major impact on screening practices, the overall frequency of PSA testing in older men would substantially decrease after year 2008 compared with the earlier years [7,8]. In fact, we observed no substantial reduction in the frequency of PSA testing in the BRFSS 2010 survey data compared with the earlier years, suggesting that the USPSTF 2008 recommendation had no major impact on the frequency of PSA testing in older men in the USA.

Another limitation of the present study is that all variables in the BRFSS surveys were documented by self-report. However, it is unlikely that any inaccuracies resulting from self-report differed systematically between the 2006, 2008 and 2010 surveys, particularly since the survey questions about PSA testing remained the same during these years. If any self-report bias was present, it would probably not vary in magnitude by year of the survey. For example, if the frequency of PSA testing was underestimated in BRFSS 2006, it would probably be underestimated by the same amount in 2008 and 2010, rendering the estimated differences between the survey years unbiased. Similarly, the non-response bias, if present, would probably not vary in magnitude between the survey years. The proportions of men excluded from the study owing to unknown history of PC diagnosis or unknown history of PSA testing in the past year before the survey were very similar or nearly identical in 2006, 2008 and 2010 (Table 1). Likewise, the proportions of men with missing data on CVD status remained virtually unchanged over time (as reported in the results section) so analyses related to the primary objective of the study (i.e. the comparison of PSA testing frequency between the survey years) would probably not be affected by self-report and non-response bias. These factors however could influence analyses related to the secondary objective of the study (Table 3) if the tendency to report PSA testing differed systematically between the levels of the study variables independently of the true testing history. For example, if among men who truly had a PSA test in the past year, the tendency to recall and report such testing was influenced by some of the variables in Table 2, then the associations reported in that table could potentially represent the influence of the study variables not only on the actual PSA testing but also on the ability/willingness to accurately recall and report such testing. Unfortunately presence or absence of this phenomenon could not be definitively determined with the available data, and this must be recognized as a limitation of the analyses. Despite these limitations, we believe the present study provides important information about the potential effect of the 2008 USPSTF recommendation on the frequency of PSA testing in older men.

In summary, the main objective of the present study was to examine the frequency of PSA testing in men aged ≥75 years in years before and after the 2008 USPSTF recommendation. Because PSA screening was common in older men before year 2008, one would expect that if the 2008 USPSTF recommendation had a major impact on screening practices, the overall frequency of PSA testing in older men would substantially decrease after year 2008 compared with the earlier years. In fact, we found no substantial reduction in the frequency of PSA testing in the BRFSS 2010 data compared with the earlier years, suggesting that the USPSTF 2008 recommendation had no major impact on the frequency of PSA testing in older men in the USA.

CONFLICT OF INTEREST

None declared. Funding for this work was provided by the Ashley Family Foundation.

Ancillary