Prostate Cancer Screening Decisions: Results From the National Survey of Medical Decisions (DECISIONS Study)

Background  Guidelines recommend informing patients about the risks and benefits of prostate cancer screening. We evaluated the medical decision-making process for prostate-specific antigen (PSA) testing.

Methods  We conducted a telephone survey of a randomly selected national sample of 3010 English-speaking US adults 40 years and older. Included in the survey were 375 men who had either undergone or discussed (with health care providers [HCPs]) PSA testing in the previous 2 years. We asked subjects about sociodemographic characteristics, prostate cancer screening discussion features, prostate cancer knowledge, and the importance of various decision factors and sources of information.

Results  Overall, 69.9% of subjects discussed screening before making a testing decision, including 14.4% who were not tested. Health care providers most often (64.6%) raised the idea of screening, and 73.4% recommended PSA testing. Health care providers emphasized the pros of testing in 71.4% of discussions but infrequently addressed the cons (32.0%). Although 58.0% of subjects felt well-informed about PSA testing, 47.8% failed to correctly answer any of the 3 knowledge questions. Only 54.8% of subjects reported being asked for their screening preferences. An HCP recommendation (odds ratio, 2.67; 95% confidence interval, 1.08-6.58) was the only discussion characteristic associated with testing. Valuing HCP information was also associated with testing (odds ratio, 1.26; 95% confidence interval, 1.04-1.54).

Conclusions  Recommendations and information from HCPs strongly influenced testing decisions. However, most prostate cancer screening decisions did not meet criteria for shared decision making because subjects did not receive balanced discussions of decision consequences, had limited knowledge, and were not routinely asked for their preferences.

The majority of American men older than 50 years have been screened for prostate cancer with the prostate-specific antigen (PSA) blood test.^1,2 However, prostate cancer screening is controversial because there is no convincing evidence that screening reduces disease-specific morbidity and mortality. The European Randomized Study of Screening for Prostate Cancer recently published results suggesting a survival benefit for screening, but the absolute benefit was negligible (decrease in mortality, favoring screening, of 0.7 death per 1000 men after 9 years of follow-up) and could not be easily disentangled from treatment effects.³ Meanwhile, the simultaneously published results from the American Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial found no survival benefit for screening.⁴ There are considerable data suggesting that treatments for early stage cancers—the targets for screening—may lead to important complications⁵ and are only marginally beneficial, especially for men 65 years and older.⁶

Given the uncertain benefit for screening and known treatment risks, prostate cancer screening decisions should be guided by patient preferences. Indeed, most professional organizations recommend that the first step in screening should be a discussion between health care providers (HCPs) and patients about the risks and benefits of early detection and treatment so that patients can make informed decisions about whether to be screened.^7-9 Although previous studies have addressed the occurrence of prostate cancer screening discussions and the quality of screening decisions, most have been conducted in a single health care system or geographic area.^10-13

The National Survey of Medical Decisions (DECISIONS study), a survey of US adults 40 years and older, provides a unique opportunity to characterize patient decision making about prostate cancer screening. The DECISIONS study is the first to systematically assess multiple common medical decisions using a consistent survey methodology and nationally representative sample. In this article, we report on men 40 years and older who faced decisions about prostate cancer screening in the previous 2 years and completed a prostate cancer–specific module in the survey. Our objectives were to characterize the decision-making process and evaluate factors associated with discussing screening before making a PSA testing decision and undergoing PSA testing following a discussion.

The DECISIONS study consisted of a random-digit-dial telephone survey of a national probability sample of English-speaking US adults 40 years and older. Participants completed a set of screening questions and were then eligible for decision-specific question modules if they had taken a medical action or discussed taking that action with HCPs for 1 of 9 common medical decisions within the past 2 years. Modules covered decisions related to (1) initiating prescription medications for hypertension, hypercholesterolemia, or depression; (2) elective surgical procedures for knee or hip replacement, cataracts, or low back pain; or (3) cancer screening tests for prostate, colorectal, or breast cancer. Full details of the sampling, instrument development, and data collection methodology are described elsewhere.¹⁴ The institutional review boards of the University of Michigan and Ann Arbor Department of Veterans Affairs Medical Center approved the study.

The present analyses focus on subjects who completed the DECISIONS prostate cancer screening module. Subjects eligible for this module had, in the past 2 years, either undergone PSA testing or discussed such screening with HCP but were not tested.

The 46 questions in the prostate cancer screening module (available from the corresponding author) addressed subject characteristics (demographic, socioeconomic, and health status); descriptive reports of the decision-making process, including the content and structure of patient-HCP discussions; the types of decision attributes subjects rated as most important to their decisions; the importance of various sources of information that subjects relied on in their decision making; and knowledge of key facts about prostate cancer that a panel of experts deemed that an informed patient should reasonably be expected to know (see “Results” section). Items evaluating the importance of decision factors and sources of information were scaled from 0 to 10, where 0 was least important and 10 was most important.

Telephone interviews were conducted from November 2006 through May 2007. Subjects were compensated for their participation. From the original list of 10 687 unique telephone numbers, initial screening identified 3420 as having an eligible respondent and 3299 as ineligible, with 2960 numbers of unknown eligibility. Overall, 3010 subjects completed interviews for a weighted cooperation rate of 86.5% and a weighted response rate of 51.6%. Eligible subjects were randomized to complete up to 2 decision modules (to limit respondent burden), with the probability of assignment inversely related to the expected prevalence of each condition. This strategy meant that we oversampled subjects who faced the less frequently encountered decisions. Among the 1302 male survey respondents, 674 (51.7%) were eligible for the prostate module (88.0% reported undergoing PSA testing in the past 2 years) and 375 were randomly selected to complete the survey.

Before beginning the primary survey analyses, we constructed selection, nonresponse, and poststratification weights to permit generalization from this sample to the full population of persons 40 years or older in the United States. We estimated population means with 95% confidence intervals to describe subject characteristics, the testing discussion, decision factors, and information sources using the survey procedures in SAS version 9.13 statistical software (SAS Institute Inc, Cary, North Carolina) to adjust for the weighting and stratification of the sample. For descriptive statistics, we collapsed the 11-point response scales into 4 categories (0-3, 4-7, 8-9, and 10) because data were highly skewed and these categorizations captured most of the underlying variation. However, these variables were modeled as continuous in the multivariate regression analyses.

We used multivariate logistic regression analysis (SAS version 9.13) to assess the associations of subject characteristics with having discussed screening before making a PSA testing decision and to assess the associations of subject characteristics, discussion processes, decision factors, information sources, and knowledge with undergoing PSA testing follow a discussion.

Subject characteristics (weighted estimates) are given in Table 1. Generally, the majority of the subjects were married, in good to excellent health, with relatively high socioeconomic status and access to health care. Nearly 85% of the subjects reported themselves to be at average or low risk for prostate cancer and more than 90% believed that screening would reduce prostate cancer mortality.

Subjects most frequently cited HCPs as an extremely important source of information about prostate cancer screening (response of 10 on the rating scale), followed by family and friends. The majority of subjects reported feeling very well informed about prostate cancer. However, 92.8% of the subjects could not correctly answer more than 1 of the following 3 knowledge questions:

• Of every 100 men, about how many do you think will die of prostate cancer?

• Of 100 men, about how many will be diagnosed as having prostate cancer at some time in their lives?

• For every 100 times a PSA test result suggests the need for further testing, about how many times does it turn out to be cancer?

Most incorrect responses were in the direction of overestimating risks for diagnosis and mortality and overestimating the accuracy of the PSA test (Table 1). Men believing themselves to be at low or average lifetime risk for prostate cancer diagnosis substantially overestimated this risk (33.9% [SD, 25.3%] and 34.7% [SD, 23.2%], respectively). Men who believed themselves to be at high risk for prostate cancer estimated the lifetime risk to be 44.6% (SD, 26.4%), which is significantly higher than the estimates for men perceiving themselves to be at low or average risk men (P = .001). We also found a significant inverse correlation between feeling informed and correctly answering the knowledge question about lifetime incidence risk (r = −0.17; P < .001); correlations with knowledge of lifetime mortality risk and knowledge of the predictive value of PSA were not significant.

Overall, only 69.9% of subjects reported having discussed screening with an HCP before making a decision about PSA testing, including 14.4% who did not subsequently get tested. Table 2 gives the characteristics of the screening discussions. The HCP was more likely than the subject to first raise the idea of getting tested; most of the time testing was presented as part of a plan for regular testing, though discontinuing testing was rarely addressed (4.4%). Few subjects sought second opinions from other HCPs. Almost all subjects reported discussing the pros of PSA testing, though substantially fewer reported discussing the cons of testing. Only 20.6% reported discussing pros and cons of PSA testing and being asked their preference for testing.

Respondents reported that most HCPs had opinions on PSA testing and usually recommended screening. Subjects generally reported that testing was explained well, that the final decision was made together with the HCP, and that their level of involvement in the decision was appropriate, although only 54.8% reported being asked their preference about testing. Subjects who discussed only the pros (86.6%) of PSA testing were almost as likely as those who discussed both the pros and cons (94.9%) to believe that testing was explained well. Nearly 60% of the subjects were extremely confident (response of 10 on the rating scale) in their testing decision.

The factors associated with discussing screening before making a testing decision (either to test or not to test) are displayed in Table 3. Self-reported health status (good/fair/poor vs excellent/very good) was the only factor in multivariate analysis significantly associated with having a screening discussion.

Overall, 79.3% of those who discussed screening went on to have a PSA test. We looked at the multivariate associations between subject characteristics, discussion processes, decision factors, sources of information, and knowledge on having a PSA test with each domain modeled separately (Table 4). The only patient characteristic associated with testing was increasing age. An HCP recommendation for testing was the only discussion process significantly associated with testing. Although not statistically significant, the odds ratio for the contrast between discussing the pros of testing a lot vs no discussion at all was 6.58, with an upper 95% confidence interval of 53.4. Several decision factors were associated with PSA testing, including higher perceived risk for cancer, believing that PSA testing reduced the risk of dying from cancer, and valuing the importance of HCP advice about testing. Valuing the importance of early detection was inversely associated with testing. Valuing information provided by the HCP was associated with testing, while valuing the importance of media information was associated with being less likely to get tested. No knowledge measures were associated with PSA testing.

The DECISIONS study systematically assessed medical decision making in a nationally representative sample of adults 40 years and older. The prostate module characterized the occurrence and content of prostate cancer screening discussions and the characteristics and outcomes of the decision-making process. Among the men in the national sample without prostate cancer, 42.2% reported having undergone PSA testing in the previous 2 years. Nearly 90% of those selected for the prostate module reported PSA testing. However, only 69.9% of these subjects discussed PSA testing before making a testing decision, including 14.4% who did not subsequently get tested.

The finding that 30.1% of subjects underwent PSA testing without first discussing screening, although consistent with other reports from the literature,^12,15,16 is a disconcerting finding. Only 20.6% of discussions presented both the pros and cons of screening and elicited the subject's preferences for testing. Given that some men may never be aware that a PSA test was ordered, we may have overestimated the proportion of men having any screening discussion, let alone making an informed or shared decision. Prostate cancer screening is considered to be an ideal opportunity for informed medical decision making because there is a lack of randomized controlled trial data supporting efficacy.^17,18 In addition, decisions for screening and treatment are sensitive to a patient's preferences regarding the downstream consequences of PSA testing, including persistent PSA test–induced anxiety following a negative biopsy result¹⁹ and treatment-related urinary, sexual, and bowel dysfunction.⁵ Guidelines from major professional organizations emphasize the need for educating patients about the risks and benefits of screening and treatment and helping them make informed decisions.^7-9 Ordering tests without accounting for patient preferences obviously undermines the goal of achieving an informed decision.

Being in poorer health was the only patient characteristic significantly associated with discussing PSA testing. Subjects reporting less than very good health were as likely to raise the issue of screening as those who reported very good or excellent health (33.0% vs 31.4%; P = .84). However, HCPs were significantly less likely to recommend screening (70.4% vs 78.1%) and more likely to offer no recommendation (29.1% vs 15.3%) for patients reporting less than very good health (P = .005). This finding suggests that HCPs are aware that guidelines advise against screening when comorbidity and/or age limit life expectancy to less than 10 years.^7-9 This also suggests that healthier men are being tested routinely without any discussion. However, we still found that more than 80% of subjects describing themselves as being in fair or poor health reported recent PSA testing. Analyses of a linked Medicare–Department of Veterans Affairs database and responses to the 2000 National Health Interview Survey (NHIS) have also documented substantial screening in sicker patients.^1,20 Apparently, even when screening is discussed, patients may not receive sufficient information to make an informed decision. Studies of prostate cancer screening decision aids, defined as written or audiovisual tools designed to facilitate decision making by informing patients about options and helping them clarify their values regarding decisions for which there is more than 1 reasonable strategy for clinical management,^18,21 have consistently shown that enhanced knowledge is associated with decreased interest in testing.¹⁰

Braddock and colleagues²² have explored the theme that a discussion is necessary but not sufficient for achieving informed decision making. They identified 7 key content elements for informed decision making, including discussing the patient's role in decision making; discussing the clinical issue or nature of the decision; discussing alternatives; discussing the pros and cons of the alternatives; discussing uncertainty; assessing the patient's understanding; and exploring patient preferences. While our survey questions did not address all of these elements, we found that a minority of subjects (32.0%) discussed the cons of testing to any degree and only 54.8% of the subjects were asked for their testing preferences—even though 60.1% described a shared decision-making process. Nearly 90% of subjects believed that PSA testing was well explained, and 58.0% felt very knowledgeable (score of 8 to 10 on the rating scale) about PSA testing. However, the poor performance on the knowledge questions at the time of completing the module, characterized by substantially overestimating the lifetime risks for prostate cancer diagnosis and mortality, suggested a potentially inadequate understanding of the clinical issues, or at least a diminution of knowledge retention. We also found that the performance on knowledge questions was inversely related to the perception of being informed. This suggests a barrier to informed decision making, given that subjects may be unaware of their knowledge deficits.

Health care providers initiated the majority of the screening discussions and were highly rated as important sources of information. Indeed, few subjects sought second opinions. Participants in the 2000 NHIS similarly reported that HCPs initiated most screening discussions.¹⁵ Our subjects also reported that HCPs often offered their personal opinions about screening, with the majority recommending PSA testing. A recent physician survey reported that 87% of male physicians 50 years and older endorsed personal screening, which could be an important factor behind screening discussions and recommendations.²³

Among subjects participating in screening discussions, most were satisfied with the decision-making process. More than 80% believed that their level of involvement in the decision was appropriate. Overall, 57.8% were highly confident in their decision, though 64.1% of those opting for PSA testing reported high confidence in their decision compared with only 30.4% of those who did not get tested. The relative lack of confidence in no-testing decisions suggests that these subjects might have been conflicted by their decision, though we did not explicitly measure this construct. Even though prostate cancer screening decision aids have been shown to reduce decisional conflict,^10,21 fully informing a patient about the complexity of the PSA testing decision could potentially increase uncertainty.

Health care provider recommendations were strongly associated with having PSA testing. The 2003 Health Information National Trends Survey (HINTS) similarly identified HCP recommendations as a highly significant predictor of PSA testing (odds ratio, 236.3; 95% confidence interval, 70.5-791.4).²⁴ The HINTS analysis also found that involving patients in decisions was associated with testing (odds ratio, 1.76; 95% confidence interval, 1.02-3.03). We did not find that shared decisions or eliciting preferences were associated with PSA testing within the past 2 years; however, we restricted our analysis to subjects who reported discussing screening while the HINTS analysis was based on the entire cohort. We found that higher perceived risk for cancer, belief in the efficacy of screening, and valuing the importance of HCP advice were associated with PSA testing. Curiously, placing more importance on early detection was inversely associated with PSA testing. However, this finding appeared to be mediated by belief in the efficacy of screening. Among those who placed most importance on early detection (response = 10), the proportion being tested was 81.1% for those who believed in the efficacy of screening compared with only 44.2% for those who did not. This suggests that respondents who valued early detection did not necessarily believe in the value of PSA testing. Valuing media information was inversely associated with testing, an unexpected finding given that media coverage of prostate cancer screening is often favorable and uncritical.²⁵ However, only 10.5% of respondents rated media information as extremely important, while 14.5% rated it as not at all important.

The study had several important limitations. The results were susceptible to recall bias because we relied on patient self-report to characterize the testing process—whether and how discussions occurred and their content—and there could be up to a 2-year lag time from the discussing screening to being surveyed. We did not survey HCPs to confirm subjects' reports of screening discussions. However, the proportion of DECISIONS study subjects 40 years and older with PSA testing in the past 2 years (49.2%) is consistent with results of other national benchmark surveys that used shorter recall periods. Among men 50 years and older participating in the 2003 HINTS, approximately 44% reported PSA testing in the previous year,²⁶ and 41.0% of men 50 years and older in the 2000 NHIS reported PSA testing in the previous year.²⁴

Even though screening discussion might have been less memorable than discussions surrounding medical or surgical therapies, we did find substantial differences in describing the screening discussions among the 129 men who completed both the colorectal cancer and prostate cancer screening modules. For example, 82.1% recalled discussing colorectal cancer screening before testing but only 68.5% reported first discussing prostate cancer screening. In addition, 48.7% reported being asked for their testing preferences in prostate screening discussions compared with only 25.4% in colorectal cancer discussions.

The study sample was largely composed of white, married men with relatively high socioeconomic status. We also restricted the survey to English-speaking subjects who had telephones. Our findings may be less generalizable to poor and minority populations.

The knowledge questions, though based on validated measures, addressed only a limited spectrum of prostate cancer knowledge. We did not ask subjects about the potential downstream consequences of screening, including issues related to biopsies and treatments, that are also considered important for making informed decisions for prostate cancer screening.²⁷

Another limitation of our survey was that we did not formally assess health literacy, defined as “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions.”²⁸ Although our subjects were relatively well educated, this does not preclude problems with health literacy. A systematic review on the prevalence of limited health literacy suggested that 26% of patients have inadequate health literacy skills and another 20% perform at marginal levels.²⁹ While literacy was associated with level of education, the prevalence of low literacy was still 11% in studies in the top quartile of high school graduation rates. Patients with health literacy deficits have greater difficulty understanding and recalling complex medical information³⁰ and are less likely to actively participate in the decision-making process.³¹ Investigators have also found an inverse relationship between the overall amount of information given and the proportion that a patient can recall.^32,33 Therefore, the adequacy of health literacy skills may affect recall and comprehension of information provided in patient-HCP interactions, which has implications for achieving informed decision making; satisfaction, conflict and confidence with decisions; adherence with medical recommendations; and use of screening and preventive services.

Prostate-specific antigen testing in a randomly selected national sample of men 40 years and older was highly influenced by HCPs. Although respondents generally endorsed shared decision-making process and felt informed, only 69.9% actually discussed screening before making a testing decision, few subjects reported having discussed the cons of screening, 45.2% said they were not asked for their preference about PSA testing, and performance on knowledge testing was poor. Therefore, these discussions—when held—did not meet criteria for shared decision making. Our findings suggest that patients need a greater level of involvement in screening discussions and to be better informed about prostate cancer screening issues. Given the challenges of meeting these expectations in routine clinical encounters, alternative strategies, such as decision aids, need to be considered to ensure a process that engages patients in decision making, provides them with information about alternative strategies, and facilitates the incorporation of their preferences and values into the medical plan.³⁴

Correspondence: Richard M. Hoffman, MD, MPH, 1501 San Pedro Dr SE, Mail Stop 111, Albuquerque, NM 87108 (rhoffman@unm.edu).

Accepted for Publication: June 7, 2009.

Author Contributions: Drs Zikmund-Fisher, Couper, and VanHoewyk had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Couper, Zikmund-Fisher, Levin, and Barry. Acquisition of data: Couper and Zikmund-Fisher. Analysis and interpretation of data: Hoffman, Couper, Zikmund-Fisher, Levin, McNaughton-Collins, Helitzer, VanHoewyk, and Barry. Drafting of the manuscript: Hoffman, Couper, and Helitzer. Critical revision of the manuscript for important intellectual content: Hoffman, Couper, Zikmund-Fisher, Levin, McNaughton-Collins, VanHoewyk, and Barry. Statistical analysis: Couper, Levin, and VanHoewyk. Obtained funding: Couper and Zikmund-Fisher. Administrative, technical, and material support: Couper, Helitzer, and Barry.

Financial Disclosure: Dr Levin is Director of Research at the Foundation for Informed Medical Decision (FIMDM), and she contributed to the design of the study, interpretation of the data, and the review of the manuscript. Drs Barry, Hoffman, and McNaughton-Collins receive partial salary support from FIMDM.

Funding/Support: This study was supported by the not-for-profit FIMDM, Boston, Massachusetts, and by the New Mexico VA Health Care System, Albuquerque. Dr Zikmund-Fisher is supported by a career development award from the American Cancer Society (MRSG-06-130-01-CPPB).

Role of the Sponsor: The funding source had no other role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation, review, or approval of the manuscript.

Disclaimer: The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the Foundation for Informed Medical Decision Making.

Previous Presentation: This study was presented in part in a poster session at the 30th annual meeting of the Society for Medical Decision Making; October 20, 2008; Philadelphia, Pennsylvania.

Additional Information: The prostate cancer screening module is available from the corresponding author.