|Home | About | Journals | Submit | Contact Us | Français|
Controversy surrounds prostate-specific antigen (PSA) testing for prostate cancer screening especially among elderly men aged 75+. This study examines whether patient age results in differential use of PSA testing and if organizational attributes such as communication, stress, decision making and practice history of change predict PSA testing among men 75+.
Data come from chart audits of 1,149 men aged 50+ who were patients of 46 family medicine practices participating in 2 northeastern practice based research networks. Surveys administered to clinicians and staff in each practice provide practice level data. A stratified Cochran-Mantel-Haenszel (CHM) test was applied to examine whether PSA testing decreased with age. Hierarchical logistic regression analyses determined characteristics associated with PSA testing for men 75+.
Comparable rates for annual PSA testing of 77.2% for men aged 50-74 years and 74.6% for men 75+ were reported. The CMH test indicated no significant change in trend. Hierarchical models suggest that practice communication is the only organizational attribute that influences PSA testing for men 75+ (OR=5.04, p=0.022). Practices with higher communication scores (e.g., promoted constructive work relationships and a team atmosphere between staff and clinicians) screened men 75+ at lower rates than others.
Elderly men in community settings receive PSA testing at rates comparable to their younger counterparts even though major clinical practice guidelines discourage the practice for this population. Intra-office practice interventions that target PSA testing to the most appropriate populations and focus on communication (both within office and with patients) are needed.
Prostate cancer is the leading cause of new cancer cases and the second leading cause of cancer death among U.S. men. One in six American men will be diagnosed with prostate cancer during their lifetime.1 In 2007, there were an estimated 218,890 new cases and 27,050 deaths from prostate cancer.1 Controversy surrounds prostate cancer screening (i.e., prostate-specific antigen (PSA) testing in conjunction with digital rectal exam (DRE)) due to lack of definitive evidence about its effectiveness in improving clinical outcomes and reducing mortality.2, 3 The U.S. Preventive Services Task Force (USPSTF) neither recommends for nor against prostate cancer screening; the American College of Physicians and the American College of Preventive Medicine endorse individualizing the decision to screen, while the American Cancer Society (ACS) and the American Urological Association (AUA) recommend offering the PSA test in conjunction with DREs annually beginning at age 50 to men who have a life expectancy of at least 10 years.4-8
Surveys of primary care physicians suggest that PSA testing is routine practice for many.9-15 Patient surveys show prostate cancer screening to be fairly common,16 more common than colorectal cancer screening3, 17 for which there are evidence based screening recommendations. Published chart review studies conducted in primary care practices reported PSA testing rates of 25% to 40% in the 1990s.18, 19 PSA testing was associated with patient (age greater than 59 years, non-smoking status, having private insurance, race) and physician factors (older physicians were more likely to screen, being in a non-solo practice, readiness to change cancer screening behavior). 10, 18-20 A study using 2000 NHIS data found the majority of PSA test recipients were non-Hispanic white men aged 50-69 with relatively high levels of education, income, access to health care and general health.16
Older patient age is also associated with PSA testing.18 Yet, there are few data that support clinical decisions to test men over the age of 75.21-23 There are a number of concerns related to testing in this group.24 First, PSA tests cannot distinguish lethal versus non lethal prostate cancer;24, 25 therefore, it is unclear whom is likely to benefit from treatment. Second, while treatment with radical prostatectomy has recently been associated with moderate mortality reduction26-28 it is unclear whether treatment results in significant increases in lifespan7, 8 or quality of life.21, 23, 29 Therefore, there are a number of questions about the cost-benefit associated with testing and treatment which have led the USPSTF to issue a D “do not screen” recommendation for men 75 and older.24
None-the-less, several recent studies have found that men 75 and older routinely receive screening.30-32 Studies using physician reported data from the 1999-2002 National Ambulatory Medical Care Survey reports PSA testing to be approximately 28% in men older than 75 years.32 One retrospective study of PSA testing in 7 New England Veteran Health Administration hospitals found a rate of 18% for men 75 years or older.30 Another study of outpatient visits in the 104 hospitals overseen by the US Department of Veterans’ Affairs (VA) found decreasing rates of testing starting with 56% for men 75-79, 43% for men 80-84 and 36% for men 85 years or older.31
Studies documenting self-reported PSA testing report comparable rates. Lu-Yao et al 2003 21 report PSA testing rates of 39% for men aged 75-79 and 24% for men 80+ in 2000 based on the National Health Interview Survey. Further, the authors indicated that about 1/3 of PSA tests were initiated for diagnostic purposes and these tests were not included in the calculation of screening rates. Thus, it can be inferred that about 2/3 of PSA tests occurred for preventive prostate cancer screening. Several studies have suggested that PSA testing might be over-utilized among elderly men,21, 33 and resulted in overdiagnosis.34 For example, among men aged 85 years and older, 34% in best health had a PSA test compared with 36% in worst health in a VA study. 33
Much of the existing literature on this topic come from data generated in private health care settings that serve a fraction of the US public. There is no published data on PSA testing in community based primary care settings where the bulk of patients receive their care. In addition, there are no studies that examine how practice characteristics or organizational climate are associated with PSA testing. Yet, there is a substantial literature that suggests that office characteristics such as office practice structure35-38 and competing demands39-41 are associated with delivery of preventive services 35, 36, 42-46 and cancer screening.35-38The purpose of this study is to provide an updated examination of PSA testing in community-based family medicine practices and to assess whether organizational factors are associated with PSA testing in elderly men 75 years of age and older.
This study is based on a secondary analysis of data collected in a study of chronic disease management. We used cross-sectional data collected at baseline from April 2003 through December 2004 from a quality improvement intervention study, Using Learning Teams for Reflective Adaptation (ULTRA). The ULTRA study used a multi-method assessment process 47 to inform a facilitated team-building intervention 48 aimed at improving guideline adherence for multiple chronic diseases among 55 practices in New Jersey and Pennsylvania. Detailed information on sample recruitment has been previously published.49, 50 This study focuses on 46 non-residency community-based practices. The UMDNJ-Robert Wood Johnson Medical School Institutional Review Board approved this study.
Clinical outcome data were collected from medical records. Each participating practice generated lists of patients seen in their office during the previous twelve months based on billing codes for asthma, coronary artery disease, diabetes, and hypertension, as well as a list of living, current patients seen for any reason. Within each practice, twenty patients were randomly selected from each list of patients. In cases where there were fewer than 20 patients per diagnosis code, all patients were used. For all male patients, nurse chart auditors from the research team noted the dates of PSA testing as well as age and other patient information. This analysis focused only on men included in the chart review who were at least 50 years of age (n=1149).
In addition to the medical record review, practice managers and lead physicians completed a 46-item practice information form that solicited information on topics such as patient population (e.g., payer mix, race/ethnicity), staff turnover, use of clinical reminder and prevention systems, and implementation of electronic medical records. Practice employees (i.e., clinicians, nursing staff, office staff) also completed a self-administered practice climate survey that included the Survey of Organizational Attributes for Primary Care (SOAPC).51 This survey took approximately 15-20 minutes to complete and responses were kept confidential by the research team. Surveys were distributed once and a replacement survey was provided one month later to non-responders. Of 46 non-residency practices, 28 practices had at least a 65% response rate among the practice staff and were used in the following analyses examining relationships between PSA testing patterns and organizational characteristics.
A total of 755 practice members were surveyed across the 28 practices. We examined multiple characteristics of non-responders and responders for differences that may bias our findings. Non-responders were similar to responders in terms of gender, years employed in the practice and staff roles (i.e., physicians versus non physicians and clinical versus office staff). There was a difference, however, in number of hours worked per week (χ2= 7.10, p=.0077). Part-time employees were less likely to respond than full-time employees (57% vs. 68%, respectively). At the practice level, we examined practices excluded from the analysis because of the 65% cut off and found no significant differences between those and the ones included in the analysis on a number of traits including ownership (physicians vs. other), type of practice (solo, single specialty group, and multi-specialty group), and years under current ownership.
PSA testing was the primary outcome. Dates for the most recent PSA values and patients’ last visits to the practice were recorded to determine whether testing occurred within the past 1, 2 or 5 years.
Socio-demographic variables considered included patient age, patient age, number of visits within the last 2 years, number of major co-morbid conditions (including diabetes, asthma, hypertension or a heart condition) and whether the patient had received cholesterol screening in the past year. Cholesterol screening was used to determine whether it appeared the PSA testing was occurring as part of a standard panel of tests. Additionally, a number of practice level demographics including practice size, average number of years the practices had been in existence and practice estimates of payer mix, race/ethnicity of the patient population and office volume measured as the proportion of patient offices visits per physician clinician were assessed.
Organizational factors of interest in this study were assessed using the SOAPC,51 a validated instrument that examines practice organization in the primary care setting. Clinicians and staff members were asked to describe their level of agreement with a series of statements (1=Strongly Disagree, 5=Strongly Agree) designed to measure practice communication, decision-making models, stress and practice history of change.51 The communications scale included 4 items (α = 0.81) such as, “The staff and clinicians in this practice operate as a real team.” This scale describes whether all members of the practice are able to work through problems as a team through discussion and consultation with one another.51 The decision making scale included 8 items (α = 0.88) such as “The practice encourages staff input for making changes and improvements.” The stress scale was comprised of 6 items (α = 0.85) including “It’s hard to make any changes in this practice because we’re so busy seeing patients.” History of change was measured by 3 items (α = 0.73) including “Our practice has changed in how everyone relates.” High scores for these scales indicate better communication and decision making practices that encourage input from all employees, a stressful/overwhelming workload and numerous changes in the management and culture of the practice. 51 Measures were constructed for practices that achieved a response rate of a 65% or higher.
Descriptive statistics were calculated for subjects and practices. Frequencies for categorical variables and means and standard deviations for continuous variables were reported.
Each subject in the study was classified according to whether they had received PSA testing within the last 1, 2 or 5 years. To examine the relationship between age and testing, initial tables compared age group (5 year increments from 50 to 90 and at least 90) to whether PSA testing was conducted. A stratified Cochran-Mantel-Haenszel (CMH) test was applied to test for a decreasing trend in the rate of PSA testing for increasing ages, where the table was stratified with respect to practice membership.
Hierarchical logistic regression models examined whether particular practice or patient characteristics were associated with differential rates of PSA testing among those aged 75 years or older. Within these analyses, a binary response of whether each subject had received PSA testing within the last one year was modeled using a logit link function and a binomial distribution. Random effects representing practice accounted for similarities between subjects attending the same practice. For patients, we included fixed effects of patient age, number of visits within the last 2 years, number of major co-morbid conditions (including diabetes, asthma, hypertension or a heart condition) and whether the patient had received cholesterol screening in the past year as predictors. For the practices, potential fixed effect predictors included in all models were number years practice in existence, whether the practice was solo or group practice and average age of each practices’ patients. Additional practice-level descriptors were added individually to the model to examine significance in an effort not to dilute power due to the fact that they had missing values for at least one of the practices. These included office volume (# office visits/#clinicians), minority status of the practice and percent of payer mix with Medicare or Medicaid as well as organizational attributes as measured by aggregated responses to the SOAPC (communication, decision-making, stress and history of change).
All analyses implemented used the SAS/STAT software (SAS system for Windows, Version 9.1.3).52
The majority of practices were group practices (74%). Two thirds of the patient panel in each practice was estimated to be white (67%, see Table 1). Practices estimated that on average 23 percent of their patients received Medicare benefits in comparison with 8 percent who received Medicaid coverage. The average practice had been in existence 10.43 years (SD=9.06). The average age of men in the practice was 55.73 (SD=15.96) and the average age of men who were included in this chart audit was 65.05 (SD=10.30).
The average rate of PSA testing for men was 76.68% within the past year. Among men aged 50 to 74 and among those at least 75 years of age, these rates were 77.22% and 74.58%, respectively.
The CMH test indicated no significant trend in the rate of PSA testing depending on age (within past one year — chi-square=2.04, df=1, p=0.15; within past two years — chi-square=0.73, p=0.39; within past five years — chi-square=0.29, p=0.59). Table 2 provides the rates of PSA testing for each age category, summing over practices.
Table 3 provides results of the initial hierarchical model exploring predictors of PSA testing within the past year for those over 75. The only significant predictor is whether a patient had received cholesterol screening within the past year (p=0.012).
Table 4 contains significance of the additional organizational predictors, for which we had information on a subset of 28 practices. Only the organizational attribute labeled communication is significant (p=0.022). The effect size is large with an odds ratio of 0.20 when comparing practices at the 25th and 75th percentiles of communication. Thus, the odds of testing a patient within the last year are 80% smaller within a practice scoring at the 75th percentile on communication relative to a practice scoring at the 25th percentile.
Data from this study provide further support to the existing literature that shows PSA testing to be common in primary care practice9, 11-15 and provides an updated look at PSA testing in community based family medicine practices. The average practice testing rate of 77% for men aged 50-74 observed in this study is higher than rates reported in earlier chart audit studies18, 19 but comparable to self report data collected for NJ in the 2001 Behavioral Risk Factor Surveillance Study that put PSA testing rates between 61% and 77%.3 As well, our testing rate of 75% for men 75+ is higher than the rates of 18%30-56%31 that are published in other studies. However, this is not totally unexpected given a recent study that found that family physicians were more likely to screen men 75+ than other primary care providers.10 We hypothesize that increased awareness of prostate cancer from the early 1990s when the original chart audit studies were conducted and targeted public service announcements may account for the increase reported. In addition, the small number of men over 75 as well as continued testing among men 75+ in community settings may also account for the rates observed.
This study found an initial association between PSA testing and cholesterol screening. This association became insignificant, however, when other organizational factors such as communication were examined in the analysis. Specifically, we found that practices with higher communication scores (e.g., practices that promote constructive work relationships, a team atmosphere between staff and clinicians, and resolve conflict and tension) screened men 75+ at lower rates than others. We have seen in other studies that office systems and environment impact delivery of preventive screening. 35-37, 42-46 Our data suggest that these conditions may also impact on provision of PSA testing. It is important that future studies examine the larger context in which preventive services are offered to better understand and address the organizational dynamics that contribute to observed practice prostate cancer screening rates.
This study highlights a practice that has been largely ignored—PSA testing of men 75+ in community settings. Practice guidelines4-8 suggest that men over 75 with limited life expectancy (less than 10 years) have little to gain from prostate cancer screening and PSA testing unless they have a particularly aggressive tumor.53 Though there are a handful of studies that associate radical prostatectomy with moderate mortality reduction26-28 for localized prostate cancer, the larger body of literature suggests that treatment for elderly men neither substantially increases either lifespan7, 8 nor quality of life.21, 23, 29 Yet, we find in this study, not unlike others,10, 16, 31, 54 that a surprisingly high proportion (75%) of men over 75 are screened. Lu Yao et al 21 question the wisdom of screening men at such ages citing little benefit and comparable life expectancy for men over 70 who choose not to have treatment versus men who choose definitive treatment. The 2001 IOM report, Crossing the Quality Chasm: A New Health System for the 21st Century55 articulated several aims for health care including that it be effective, patient centered, and efficient. With the new guideline from the USPSTF that gives a D “do not screen” recommendation24 for men 75+ it is clear that the evidence is lacking that aggressive prostate cancer screening and subsequent treatment for elderly men is any of these.
While current guidelines recommend prostate cancer screening occur within the context of informed decision making,4-8 there are a growing number of studies documenting that informed decision making has yet to be incorporated as a routine part of primary care practice.16, 56, 57 Lu-Yao et al. report that 88% of elderly 75+ men indicated that their doctor suggested that they have PSA testing, yet, fewer than 2/3 reported discussing both pros and cons of PSA testing with their physicians.16, 21 Similarly, a study of community physicians found that physicians ordered PSA tests without discussing them with their patients citing medical and legal liability concerns.56 During the time period our data were collected, there was much discussion along the East Coast about a heavily publicized case of malpractice that involved PSA testing which may have affected testing decisions.58, 59 Data from our study and these others suggest that this is a research area in need of further study with major implications for potential design of targeted interventions focused on prostate cancer screening.
Study findings should be interpreted while considering several potential limitations. First, race of patients in this sample was largely unknown. Most medical records for patients in this study did not record patient race or ethnicity. Because we rely on chart, rather than self-report data, we are unable to examine the effects of patient race and ethnicity on PSA testing. Second, our outcome measure highlights PSA usage but does not provide a comprehensive look at prostate cancer screening. The measure does not include clinician use of DRE and does not distinguish between PSA testing that was initiated for diagnostic versus preventive screening purposes. The parent study did not include data on urologic problems or cancer status of patients. Therefore, we cannot distinguish between men who were being evaluated for lower urinary tract symptoms versus undergoing prostate cancer screening. While these are important limitations, guidelines from USPSTF24 and AUA60 recommend against PSA testing for either purpose in elderly populations calling into question its use value for men whose natural lifespan is less than 10 years. Therefore, we suggest that it is important to examine the practice of PSA testing regardless of the reason for initiation (i.e., preventive cancer screening, diagnosis, monitoring) in elderly populations.
Despite these limitations, data from this study are suggestive for interventions focused on PSA testing and prostate cancer screening in primary care settings. Much of the previous research on prostate cancer screening focuses on understanding patient and physician characteristics that influence screening decisions. It is unclear what is driving decision making processes on who receives PSA testing and prostate cancer screening in primary care practices. Is this a decision made unilaterally by physicians or nurse practitioners on an individual basis? Are there standing orders and if so how are they tailored? Are labs generating this data as part of routine blood work panel? How does the patient’s wish impact on these processes? We suggest that future research is needed to better understand the circumstances in which PSA testing and prostate cancer screening is initiated in primary care settings so that intervention studies focused on providing testing for the most appropriate populations can be better calibrated to address patient, physician and organization correlates of testing.
This research was conducted in conjunction with the New Jersey Family Medicine Research Network (NJFMRN), a shared resource of the Cancer Institute of New Jersey and the Eastern Pennsylvania Inquiry Collaborative Network (EPICnet). It was also supported by the Cancer Institute of New Jersey’s Survey Research and Qualitative Methods Shared Resource.
Support: This research was supported through grants from the National Cancer Institute (K01 CA131500-01A1), National Heart, Blood and Lung Institute (R01 HL70800), the New Jersey Commission on Cancer Research (03-40-CCR-S0), and the Department of Defense (DAMD 17-01-1-0755).
Conflict of Interest: None of the authors have any potential conflicts of interest to disclose.