Use of the prostate‐specific antigen (PSA) test in the United States for men age ≥65, 1999–2015: Implications for practice interventions

Abstract Background Various professional organizations have issued recommendations on use of the PSA test to screen for prostate cancer in different age groups. Aims Using Medicare claims databases, we aimed to determine rates of PSA testing in the context of screening recommendations during 1999–2015 for US men age ≥65, stratified by age group and census regions, after excluding claims relating to all prostate‐related conditions. Methods and Results Medicare claims databases encompassed 9.71–11.12 million men for the years under study. PSA testing rate was the proportion of men with ≥1 test(s) per 12 months of continuous enrollment. Men diagnosed with any prostate‐related condition were excluded. Annual percent change (APC) in PSA test use was estimated using joinpoint regression analysis. In 1999–2015, annual testing rate was 10.1%–23.1%, age ≥85; 16.6%–31.0%, age 80–84; 23.8%–35.8%, age 75–79; 28.3%–36.9%, age 70–74; and 26.4%–33.6%, age 65–69. From 1999 to 2015, PSA testing rate decreased 40.7%, 29.9%, 13.9%, and 2.9%, respectively, for men age ≥85, 80–84, 75–79, and 70–74. For men age 65–69, test use increased by 0.3%. Significant APC trends were: APC1999–2002 = +8.1%, P = .029 and APC2008–2015 = −9.0%, P < .001 for men age ≥85; APC2008–2015 = −7.1%, P = .001 for men age 80–84; APC2001–2015 = −2.5%, P < .001 for men age 75–79; APC2008–2015 = −3.3%, P = .007 for men age 70–74; and APC2010–2015 = −5.2%, P = .014 for men age 65–69. Coclusion Although decreased from 1999 to 2015, PSA testing rates remained high for men age ≥70. Further research could help understand why PSA testing continues inconsistent with recommendations.

Services Task Force (USPSTF) recommended against routine screening for prostate cancer by using the PSA test. 1 In 2002, the USPSTF concluded that there was insufficient evidence to make any recommendation to screen for prostate cancer using the PSA test. 2 In 2008, the USPSTF recommended against screening of men age ≥75 years, with no recommendation for younger men. 3 In 2012, the USPSTF recommended against PSA screening of men for prostate cancer regardless of age. 4 Subsequent to the 2012 recommendation, the USPSTF released a draft recommendation in April 2017 noting that men age ≥70 should not be screened, and that the potential benefits and harms of PSA-based screening were closely balanced in men age 55-69, and that the decision whether to be screened should be an individual one. 5 The USPSTF issued its most recent recommendations in May 2018, 6 that was identical to the 2017 draft recommendations. 5 The American Urological Association (AUA) 7 and American College of Physicians (ACP) 8 in 2013 both recommended against routine screening of men age ≥70. Like the USPSTF, 6 the AUA recommended that the decision for men age 50-69 to submit to screening be an individual one 7 and the ACP recommended individualized screening in this age group be limited to those with at least 10 years of life expectancy. 8 In 2016, the American Cancer Society (ACS) stipulated a starting age of 50 for individualized PSA-based screening of men with at least 10 years of life expectancy. 9 In 2001-2002, the AUA, ACP, and ACS all included informed and shared decision-making, with the AUA and ACP recommending a starting age of 50. 10 Using Medicare claims databases in the context of screening recommendations for prostate cancer, our objective was to determine the rates and trends for annual PSA testing in US men age ≥65 in 1999-2015, stratified by age group, after excluding claims relating to all prostaterelated conditions. Annual testing rates and trends were examined for different age groups overall and after stratifying by the four US Census regions. These analyses were undertaken to assess PSA screening uptake in older men of different age groups and particularly those age ≥70 for whom routine PSA screening is not recommended. Information gained from this study may be used to inform current practice interventions.

| Population
We used 100% Career files for Medicare claims data for men age ≥65 from January 1, 1999 through December 31, 2015, encompassing 9.71-11.12 million men with continuous enrollment for each calendar year. We excluded all men with claims containing one or more diagnoses or non-laboratory procedures relating to various prostate-related diseases or conditions, including those diagnostic codes related to elevated PSA or history of prostate cancer (see Appendix). Although men with nonmalignant prostate-related diseases or conditions may indeed undergo PSA testing for screening purposes, it is likely that some may have undergone PSA testing in the context of differential diagnosis in order to rule out prostate cancer. We therefore opted to conservatively exclude all populations with encounter claims for PSA testing that may have been performed for differential diagnosis, monitoring, or prognostic purposes. We included only claims with a Current Procedural Terminology (CPT, American Medical Association) Code 84152 for complexed PSA and Code 84153 for total PSA, and Healthcare Common Procedure Coding System (HCPCS, Centers for Medicare and Medicaid Services) code G0103 for total PSA testing. We did not use CPT code 84154 for free PSA because this test, unlike total or complexed PSA, is not initially used to screen for prostate cancer. If PSA test result is in the borderline range, usually considered 4.0-10.0 μg/L, free/total PSA ratio may be used to decide if a prostate biopsy is warranted.

| Annual PSA testing rates
Annual rates for PSA testing in each calendar year were the proportion of the study population with ≥1 PSA test(s) per 12 months continuous enrollment after excluding those with any prostate-related condition.

| Annual percent change and PSA testing trends
Annual percent change (APC) was used to estimate the change in PSA test use. APC was estimated using joinpoint regression analysis, fitting trend data to identify the log-linear model with the fewest number of inflection points. 11 Therefore, the software used fits trend data using a regression model that minimizes the number of inflection points, and therefore the number of APC line segments. APC was the loglinear slope of each trend line, and P values related to statistical significance of each APC estimate being different from zero.  WA, and WY). This stratification scheme was used to better understand the extent of overall and regional variation of PSA test use by age.

| Included populations
The final study sample included 6.46-7.79 million men in each year from 1999 through 2015 after excluding men having encounter claims associated with prostate-related diseases or conditions (see Methods).
As the result of this exclusion, 66.5%-70.1% of men remained for analysis of PSA testing rates and trends.  Figures 1, 2, and 3, respectively. Regional PSA testing trends are shown in Figure 4 for all male Medicare enrollees age ≥65.

ACKNOWLEDGEMENTS
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of CDC.

CONFLICT OF INTEREST
The authors have stated explicitly that there are no conflicts of interest in connection with this article.

AUTHOR CONTRIBUTIONS
Shahram Shahangian: concept and design of the study, data analysis, interpretation of data, drafting of the initial manuscript, review and revision of the manuscript, approval of the final manuscript as submitted, and accountability for all aspects of the work. Lin Fan and Krishna P. Sharma: concept of the study, data curation, interpretation of data, review and revision of the manuscript, approval of the final manuscript as submitted, and accountability for all aspects of the work. David A. Siegel: concept of the study, interpretation of data, review and revision of the manuscript, approval of the final manuscript as submitted, and accountability for all aspects of the work.

ETHICS STATEMENT
Institutional clearance and approval was obtained prior to submission.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.