Prostate-specific antigen testing after the U.S. Preventive Services Task Force recommendation: a population-based analysis of electronic health data

Abstract

Purpose:

This study describes longitudinal trends in the use of prostate-specific antigen (PSA)- based testing in two geographically-distinct healthcare systems following the 2011 US Preventive Services Task Force (USPSTF) recommendations against routine PSA screening.

Methods:

We analyzed population-based health claims data from 253,139 men aged 40-80 who were enrolled at two US healthcare systems. We assessed trends in the percentage of eligible men receiving ≥1 PSA test per year by time period (2000-8, 2009-11, 2012-14), age (40-54, 55-69, 70-80), and race (white, black, other, unknown), and conducted a joinpoint regression.

Results:

Men aged 55-69 and 70-80 years of all races had similar use of PSA testing between 2000-2011, ranging between 47-56% of eligible men by year, while only 22-26% of men aged 40-54 had a PSA test per year during this period. Overall, the percentage of men receiving at least one PSA test per year decreased by 26% between 2009-11 and 2012-14, with similar trends across race and age groups. PSA testing declined significantly after 2011 (annual percent change=−11.28).

Conclusions:

Following the 2011 USPSTF recommendations against routine PSA screening, declines in PSA testing were observed among men of all races and across all age groups in two large US healthcare systems.

Introduction

In October 2011, the U.S. Preventive Services Task Force (USPSTF) circulated a draft grade D recommendation against the routine use of prostate-specific antigen (PSA) testing for prostate cancer screening in all men.[1] This recommendation, finalized in May 2012, was of particular clinical relevance as the majority of primary care physicians cite the USPSTF guideline as the standard for determining their prostate cancer screening practices. [2,3] This recommendation was controversial as there is conflicting evidence suggesting a benefit to PSA screening in certain populations; in men ages 55-69, and in higher-risk populations which include black men and men with a family history of prostate cancer over age 40. [4–8] Recognizing this evidence, the American Urological Association (AUA) and National Comprehensive Care Network (NCCN) continued to recommend PSA screening for these men.[5,6] Given these conflicting guidelines, and the recent 2018 reversal of the USPSTF to recommend shared decision-making for men aged 55-69, [9,10] longer-term real-world trends in PSA testing, and especially in higher-risk populations, are of particular interest.

While some publications suggest that PSA testing rates have substantially decreased following the release of the 2011 USPSTF guidelines, other studies have not observed a notable decrease in testing.[11–14] Further, there are emerging data suggesting that following the 2011 USPSTF recommendations, declines in PSA screening may have also occurred among higher risk black men and men with a family history of prostate cancer.[14,15] To better characterize trends in PSA testing across prostate cancer risk groups, we conducted a population-based analysis of patients with longitudinal health claims data at two large, geographically diverse healthcare systems over a fifteen year period.

Materials and Methods

Data Sources

This analysis was conducted at two organizations participating in the National Cancer Institute-supported Cancer Research Network (CRN): the Meyers Primary Care Institute (MPCI) in Worcester, MA, and the Henry Ford Health System (HFHS) in Detroit, MI. The CRN is a nationwide consortium of research organizations affiliated with non-profit integrated healthcare delivery systems. [16,17] As member sites of the CRN, each site maintains a Virtual Data Warehouse (VDW), a system of compatible datasets constructed at each site from electronic medical records and healthcare claims, including Medicare claims, with additions from state or national registries. [18] Data in the VDW pass rigorous quality checks, and are aggregated for analysis. Data collected through the VDW and relevant to this analysis include sociodemographics, inpatient and outpatient utilization, procedures and diagnoses, and tumor data.

This analysis was approved by the Institutional Review Boards at the University of Massachusetts Medical School and at Henry Ford Health System. Data from the Massachusetts Cancer Registry was provided with approval from the Massachusetts Department of Public Health.

Study population

All men aged 40-80 years who were enrolled in the participating health plans with coverage during the study period (January 2000-December 2014), and who did not have a prior history of prostate cancer (defined by ICD-O-3 code C619 in linked tumor registry data), were eligible for inclusion in this analysis. Prostate cancer diagnoses among MPCI patients were identified through the Massachusetts Cancer Registry, which began collecting data in 1982, and captures all cancers diagnosed in the state, including those diagnosed out of network. For the HFHS population, prostate cancer cases were identified through the health system-based tumor registry, which identifies all cancers diagnosed since 1988. Patients diagnosed with prostate cancer during the study period were censored at date of diagnosis. Age was updated for each year of the study. The two study populations were combined for all analyses reported below.

Data collection

PSA tests (total PSA) were identified in health claims databases by CPT codes 84152, 84153, 84154, and G0103, which included both screening and diagnostic indications. To limit the inclusion of diagnostic monitoring tests, we excluded any patient once they were diagnosed with prostate cancer as detailed above. To estimate the frequency of annual PSA tests performed at each study site, we counted only one PSA test per person per year. Socio-demographic data (sex, date of birth, race [white, black, other, unknown]) were captured through health claims and electronic medical records. Enrollment time was calculated allowing for 30-day gaps in coverage.

Statistical analysis

We calculated the number of eligible participants who underwent at least one PSA test per calendar year from 2000-2014, and described testing frequencies across the three age groups (40-54, 55-69, 70-80 years) defined by current PSA screening recommendations from the American Urological Association. [5] Analyses of PSA testing across age groups were additionally stratified by race (black vs. other). The percentage of eligible men undergoing PSA testing per year was calculated by dividing the number of men with ≥1 claim for a PSA test by the total number of eligible men in the combined health systems during that year, for each year separately. For men who underwent ≥1 PSA test during the study period, we measured the time between PSA tests as the number of days between processed claims across the study period.

We conducted a joinpoint regression analysis to assess temporal trends in PSA testing across the study period. We selected a model incorporating two joinpoints and calculated the annual percent change (APC) for each segment of the regression line. We examined trends in the overall study population, and also in subgroups defined by age and race. Joinpoint analyses were conducted using software available from the National Cancer Institute. [19] All other analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC).

Results

In total, 253,139 men were included in this analysis (Table 1). Among the 129,090 (51%) men who had at least two PSA tests during the study period, the median time between a patient’s first two PSA tests averaged about 15 months (439 days), suggesting most PSA tests were performed at nearly annual intervals. Among men who had at least one PSA test, 14.5% of the population was black, 42% white, and 43% of other or unknown race. Men with at least one PSA test were enrolled in their health system for a median of 6.3 years (interquartile range [IQR]: 11.0) at HFHS and 5.0 years (IQR: 7.4) at MPCI. In comparison, all men aged 40-80, regardless of PSA testing status, were enrolled for a median of 5.5 years (IQR: 8.6) at HFHS and 3.3 years (IQR: 6.3) at MPCI.

Table 1.

Characteristics of the study population, men aged 40-80 years seeking care at two US healthcare systems, 2000-2014

Variable	Total populationa	Among men with ≥1 PSA test from 2000-2014
Total eligible men, N (%)	253,139	134,870
Meyers Primary Care Institute	152,112	70,055
Henry Ford Health System	101,027	64,815
Age in yearsb
40-54	166,067 (66%)	68,646 (51%)
55-69	66,860 (26%)	50,026 (37%)
70-80	20,212 (8%)	16,198 (12%)
Racec
Black	29,556 (12%)	19,556 (14%)
White	84,149 (33%)	56,603 (42%)
Other	7,213 (3%)	4,384 (3%)
Unknown	132,221 (52%)	54,327 (40%)
Family history of prostate cancerd
Yes	2,098 (1%)	1,978 (1.5%)
Mean time between first 2 PSA tests, dayse		439

^aAll men aged 40-80 enrolled in participating health systems for at least one day between January 2000 and December 2014, and who did not have a personal history of prostate cancer.

^bAge at beginning of study period or at enrollment in the health system during the study period.

^cSelf-reported race.

^dFamily history of prostate cancer defined as at least one occurrence of ICD-9 code V16.42 in claims data.

^eN = 129,090 men.

Among the entire study population during the 2009-2011 period, 37% of all men had at least one PSA test per year; however, by 2012-2014, just 27% of all men received at least one PSA test per year, representing a 26% decrease in the use of PSA testing among eligible men per year. Ongoing decreases in PSA testing were observed between 2012 and 2014, suggesting a continuing decline in the use of PSA testing not limited to the year immediately following the issuance of the USPSTF recommendation.

Men aged 55-69 and 70-80 years had similar use of PSA testing between 2000 and 2011, ranging between 47-56% of eligible men by year. Men aged 40-54 years were least likely to have a PSA test, with 22-26% of eligible men in this age group receiving at least one PSA test per year between 2000 and 2011. We observed notable declines in the use of PSA testing across all age groups between 2011 and 2014; there was a 39% decrease in testing among men aged 40-54, a 31% decrease among men aged 55-69, and a 28% decrease among men aged 70-80 years over this four-year time period.

Among younger men aged 40-54, black men were tested more often than men of other races across the study period (Fig 1a). However, the absolute percentage of eligible black men receiving PSA testing in a given year in this younger cohort was low overall, with just 20% tested in 2014. Older black men aged 55-80 years received PSA testing at similar frequencies to men of other races (Figure 1b–c). Notably, the percent of men receiving PSA testing in a given year declined similarly among black men and men of other races following the issuance of updated USPSTF guidelines (decrease of 31% among blacks and 30% among whites between 2009-11 and 2012-14).

Figure 1A-C.

Longitudinal trends in prostate-specific antigen testing by race^a from 2000 to 2014 for men aged: A) 40-54 years; B) 55-69 years; and C) 70-80 years at two US healthcare systems.

^a Race is self-reported; the category of all other races includes white, other, and unknown. The shaded bar represents the issuance of revised recommendations against routine PSA screening by the US Preventive Services Task Force in 2011.

Joinpoint analyses in the overall study population identified a non-significant decline in PSA testing between 2006 and 2011 (APC=−2.37), and a larger, statistically significant decline in testing between 2011 and 2014 (APC=−11.28; p<0.05). In models stratified jointly by age group and race, a significant decline in PSA testing was evident between 2011 and 2014 (APC range: −13.69 to −10.06; all p<0.05), with the exception of black men aged 40-54, where the decline began in 2010 (APC: −11.39).

Discussion

This population-based analysis of automated health data from two large US healthcare systems suggests that the issuance of the USPSTF grade D recommendation against routine PSA screening in late 2011 was associated with a subsequent decrease in the frequency of PSA testing across men of all races. Despite being at higher risk for prostate cancer, black men experienced similar declines in PSA testing compared to men of other races. We also observed similar decreases in PSA testing across all three studied age groups.

Our study provides objective, claims-based data in support of the decreases in PSA testing observed in the cross-sectional, patient-reported National Health Interview Survey and National Ambulatory Medical Care Survey, which showed an 18%-39% decrease in PSA testing between immediate years surrounding the issuance of the 2011 USPSTF guideline.[13,11] Furthermore, our results regarding trends in testing among black men mirror recently reported concurrent data from other large health systems,[15] suggesting that PSA screening rates have significantly decreased among black men in a variety of diverse geographical settings across the US following the 2011 USPSTF guideline. Additionally, despite a higher risk for developing significant prostate cancer, black men aged 55-69 years were tested at the same frequency as men of other races in our study. Our data further suggest that the decline in PSA testing was not limited to the immediate effects of the change in screening guidelines. Recent studies have already observed that declines in PSA screening may lead to delayed identification of significant disease, although the long-term impact of these changes is not yet evident.[11] However, the most recent update to USPSTF guidelines in May 2018 recommending individual decision-making for men aged 55-69 [10] may attenuate negative consequences of discontinued screening, although the full impact of these shifts in policy will not be evident for several years.

The earlier 2008 version of the USPSTF recommendation on PSA screening encouraged more discriminant screening in the late 2000s, but evidence suggests that the national impact on clinical practice was minimal.[20] In contrast, the 2011 revised USPSTF recommendations adopted a more direct recommendation against routinely screening all men. Evidence suggests that the 2011 recommendation may have achieved its goal, as several studies have observed decreases in overall PSA testing rates starting in 2012.[13,11,14,15,21] Our work adds to these observations by suggesting that men of all ages and of different racial groups appear to have experienced noticeable decreases in the use of PSA testing over this period. It is important to recognize that a consequence of the broad recommendation discouraging PSA screening among all men may have been also discouraging screening among high-risk populations who could benefit most from early prostate cancer detection. [5] It is also notable that despite the broad recommendation against screening, PSA testing rates decreased the least among older men, who may not experience as great a benefit from early disease detection given higher risks of comorbid conditions, and of dying from other, non-prostate cancer, causes.[22,23]

There are important caveats to consider in the interpretation of our results. We provide a 15-year longitudinal claims-based assessment of PSA testing that spans the period in which the two USPSTF guidelines were published. In addition, this study assesses temporal trends in PSA testing among a large population-based sample of men from two geographically diverse regions of the United States, with the ability to compare trends across age and race groups. Although our study sample includes a sizable cohort of black men, we were limited to incomplete race and ethnicity data present in health claims data. We were also unable to incorporate data on family history of prostate cancer as this variable was underutilized in claims data. As a result, our ability to analyze trends in PSA testing in high-risk populations was limited; however, our analyses by race suggest that high-risk populations are also being tested less frequently in recent years. Our estimates of PSA testing rates may have been underestimated as our denominators included all men enrolled in the health systems for at least one day. In our study populations, men with at least one PSA test appear to have been enrolled in their health plans for slightly longer than all males in their age group. However, across the CRN, health systems report high overall 1-year retention rates between 84-90%,[17] and among cancer patients, >88% are enrolled for at least 1 year prior to diagnosis. [24] In addition, the majority of men in our study population were enrolled for ≥5 years, suggesting any underestimation of testing rates is minimal. Furthermore, we could not distinguish screening PSA tests from diagnostic PSA tests in the claims data. Clinicians often use both categories of codes to indicate PSA testing in the screening setting despite the labeling differences, and as a result we included all codes for PSA testing in the analysis to best approximate the use of PSA in our health systems. Our analysis of one PSA test per person per year approximates annual screening, but has not been further validated. All observations reported here should be confirmed in other populations with access to detailed data on race and ethnicity.

Because our analysis utilized health claims data from two large US health systems, our findings may not be generalizable to an uninsured population, as insurance status may impact whether patients receive screening tests such as PSA. At the MPCI and HFHS sites, patients were primarily enrolled in private commercial payor plans (77% and 72%, respectively), with Medicare enrollees representing 24% and 11% of the cohorts, respectively. [24] In addition, we excluded any man with a prior history of prostate cancer using comprehensive state-level cancer registry data. Although it is possible that cases diagnosed out of state were missed, this number is likely small. Prior work assessing the stability of patient enrollment in the health plans captured in the CRN reveals 60% of HFHS and 67% of MPCI patients were enrolled in their health plan ≥10 years prior to cancer diagnosis,[24] suggesting the majority of past cancer diagnoses were identified and excluded.

Our findings are consistent with recent PSA screening trends reported from nationally representative surveys like the Centers for Disease Control’s Behavioral Risk Factors Surveillance System, National Health Interview and National Ambulatory Medical Care Survey, as well as data from other large health care systems. [13,11,15,21,25] Furthermore, our claims-based approach to assessing PSA testing provides objective evidence, and is not subject to recall bias, which is possible in survey-based assessments of testing practices. However, as with any claims-based approach, there is a risk for potential misclassification, which we were unable to quantify. Despite these limitations, our study provides important and objective insights into the ongoing evolution of PSA screening practices over the last 15 years. This study suggests that the notable decline in PSA testing observed since the updated 2011 USPSTF PSA screening recommendations has occurred in all men, regardless of age or race, and is not limited to the time period immediately following the guideline change.

Conclusions

There is evidence of a widespread decline in the use of PSA testing across all age groups following the 2011 USPSTF recommendations, with black men experiencing declines in testing similar to men of other races despite being at elevated risk of developing prostate cancer.