Psychological benefits of prostate cancer screening: the role of reassurance

Abstract

Objectives  The role of reassurance in decision‐making about screening for health problems is largely unknown. We examined the reassurance value of prostate cancer screening in primary care patients.

Setting and participants  One hundred and sixty eight men, aged 45–70 years, who had no history of prostate cancer, from three family medicine practices in southeast Texas.

Methods  A hypothetical scenario was developed where men were asked to assume they did not have prostate cancer, and then to rank three predefined screening states with regard to reassurance value: (A) unknown (no screening), (B) normal by screening [a prostate‐specific antigen (PSA) test and a digital rectal examination (DRE)] and (C) normal by biopsy (abnormal PSA test and DRE results but a negative ultrasound‐guided prostate biopsy).

Results  Most of the men (96.8%) associated some reassurance value with screening, considering health state A to be the worst possible health state. Results from a multivariate analysis showed that preference for screening state C was associated with a family history of prostate cancer and perceived greater risk for prostate cancer compared with other men.

Conclusion  These findings suggest that prostate cancer screening may have some reassurance value for men, and that increased risk status may explain why some men prefer the added reassurance afforded by screening.

Introduction

Prostate cancer is the second leading cause of cancer deaths among men in the United States. In 1999, approximately 179 300 men were diagnosed with the disease; another 37 000 died of it. ¹ Although screening tests for prostate cancer, such as the digital rectal examination (DRE) and the prostate‐specific antigen (PSA) test, are now used often to identify its early stages, the benefits of such tests remain unclear. No randomized clinical trial to date has demonstrated any correlation between prostate cancer‐screening results and length or quality of life.

Other concerns have also called the use of prostate cancer screening into question. Firstly, prostate cancer is a slow‐growing disease, which makes `watchful waiting' a reasonable management approach even if early stage disease is found. Secondly, treatments for prostate cancer, including radical prostatectomy and pelvic radiotherapy, often lead to adverse outcomes such as impotence, incontinence and even death. Thirdly, diagnostic tests such as the DRE and the PSA test are not sensitive enough to detect all clinically significant prostate cancers, and they also identify many cancers that are not clinically significant. ² Fourthly, conclusive proof does not exist to demonstrate that early detection and treatment of prostate cancer improves survival. Nevertheless, organizations such as the American Urological Association and the American Cancer Society have recommended prostate cancer screening for all men after age 50 and for younger men in high‐risk groups (e.g. African Americans and men with a family history of prostate cancer).

Despite the uncertainties of prostate cancer screening, many asymptomatic men visit primary care clinics specifically to be screened for the disease. Whether their visits are prompted by advertisements that feature well‐known personalities such as Senator Robert Dole and General Norman Schwartzkopf or by the deaths of celebrities such as Telly Savalas and Frank Zappa, having prostate cancer is of great concern to them. Any favourable outcome of prostate cancer screening, then, would undoubtedly be reassuring.

Psychological attributes, including regret and reassurance, have been recognized as important but not well‐understood aspects of health‐care decision‐making. Certainly, people make decisions based not only on the potential health outcomes, but also on potential psychological outcomes. However, studies have not examined the role of these psychological factors as patients and physicians make clinical decisions. Most prior work on psychological attributes has focused on regret, either from the patient's ³ or physician's ⁴ perspective. Few studies, however, have examined the relationship between health‐screening behaviour and the need for patient reassurance, especially with regard to cancer screening. Guided by previous research on factors related to cancer screening behaviour in general, ⁵ , 6, 7, 8, 9, ^–10 we undertook the study we describe herein to determine the reassurance value of prostate cancer screening by DRE and the PSA test in a primary care setting. To discern possible psychological benefits of prostate cancer screening, we evaluated the relationship of sociodemographic factors, past screening behaviour, perceived cancer risk and beliefs and misconceptions about prostate cancer screening, with the reassurance value of prostate cancer screening tests available.

Methods

Recruitment of study group

Study participants were those patients enrolled in the Couples Prostate Cancer Screening Project, a study of the preferences of male primary care patients and their spouses for the outcomes of prostate cancer screening and treatment. The men were English‐ or Spanish‐speaking, were 45–70 years of age, and did not have a history of prostate cancer.

All 168 men enrolled in the study were recruited from three family medicine clinics that serve an ethnically and socioeconomically diverse patient population in southeast Texas. The men were recruited by direct solicitation in clinic waiting areas, by phone if they had scheduled non‐acute office visits and by self‐referral through study notices posted at the three clinics. Approval for the use of human subjects was granted by the Institutional Review Board of The University of Texas Medical Branch at Galveston. Written informed consent was obtained from every participant.

Data collection

The subjects were first asked to complete a self‐report questionnaire that contained the study measures, including sociodemographic data, previous screening behaviour, prostate cancer risk and knowledge/beliefs about prostate cancer. Research staff were available to assist subjects in completing the questionnaire as needed. The subjects were then interviewed face‐to‐face, and their preferences for prostate cancer screening outcomes were assessed using a scenario‐based (hypothetical) approach combined with a simple ranking procedure. Because the screening states the subjects were asked to consider were short‐term and their impact was primarily psychological rather than physical, the ranking procedure provided a simple and practical method for evaluating the subjects' preferences compared with other utility assessment procedures such as standard reference gamble, time trade‐off, category scaling and willingness‐to‐pay. The procedure did not directly yield utilities, but rather provided a rank‐ordered evaluation of preferences.

The assessment began with an extensive patient education phase. All subjects were given descriptions of prostate cancer, prostate cancer screening methods and diagnostic tests, including biopsy, and possible adverse effects of prostate cancer treatment. Interviewers explained details regarding the screening and diagnostic procedures. Graphical materials were used, along with text‐based informational summaries for patients printed on laminated cards and provided for their reference throughout the assessments. The purpose of the education sessions on prostate cancer, screening and treatment was to better inform the men for the health state rank‐ordering process which followed. This sequence of education and interviewing is standard methodology for utility assessment, i.e. to make the subjects more familiar with the health states in question. ¹¹ Patients were then asked to consider a scenario where it was assumed they did not have prostate cancer. They were then asked to rank, using a forced‐choice format, three possible prostate cancer screening scenarios, again assuming that the underlying health state was not having prostate cancer. The three prostate cancer screening scenarios were described as follows (note that in this context, the word `confirm' should be interpreted as `corroborate', not as `guarantee with certainty'):

(A) You do not have a PSA test or a digital rectal exam. Therefore, you do not have test results to confirm you do not have prostate cancer.

(B) You have a PSA test and a digital rectal exam. Both tests are normal, confirming that you do not have prostate cancer.

(C) You have a PSA test and a digital rectal exam. These tests are abnormal, so you have an ultrasound with biopsy to determine whether there is cancer. The ultrasound with biopsy is normal, indicating you do not have prostate cancer.

The subjects were required to rank the three prostate cancer‐screening scenarios numerically as `1', `2' or `3', indicating, respectively, which outcome was `best', `intermediate' and `worst.' Interviewers helped to clarify the subjects' responses by saying statements such as, `This means you preferred health state B the most, followed by health state C, with health state A being the least preferred'. No equivalent responses were allowed.

Because of its simplicity, we felt that rank ordering would be more likely to be reproducible compared with methods that evaluated strength of preference. Rank ordering is a `step up' from visual analog scale method, which outranked other preference methods (e.g. time trade‐off and standard gamble) in differentiation and consistency measures. ¹²

By using this forced‐choice comparison of screening states, our intention was to examine the reassurance value derived from either favourable (i.e. negative) PSA test and DRE or biopsy results over that derived from not being screened. In this context, we defined reassurance as the psychological benefit of a favourable (i.e. no cancer) test result. In other words, health state A would provide no reassurance, health state B would provide some reassurance and health state C would provide the greatest reassurance. Our premise was that for some patients, such reassurance would outweigh the physical discomfort of a blood test, a physical examination or an invasive diagnostic procedure (i.e. biopsy).

Independent variables

Sociodemographic factors

Indicators of patient age, ethnicity, highest education level achieved and annual household income were included in the patient questionnaire.

Past screening behaviour

The subjects were asked whether they had had a previous PSA test, with response options of `yes', `no', or `I don't know'.

Prostate cancer risk

Two measures of prostate cancer risk were included in the questionnaire. As an objective measure of personal risk, the subjects were asked about their family history of prostate cancer – whether their father, a brother, an uncle, a grandfather or a son had had the disease. As a subjective measure of personal risk, the subjects were asked how likely it is that they will develop prostate cancer when compared with other men their age.

Beliefs and misconceptions about prostate cancer and PSA testing

Responses to three prostate cancer knowledge and beliefs questions were used as indicators of the subjects' prostate cancer beliefs and misconceptions. Each item and the response option representing a misconception about prostate cancer are given here:

(1) How many untreated men with early stage prostate cancer will die from it? (Misconception: `Most or all will die'. Correct response: `Most of them will not die').

(2) A high PSA level usually means. (Misconception: `Cancer of the prostate'. Correct response: `An enlarged prostate gland associated with aging').

(3) How many men with a high PSA level will have prostate cancer? (Misconception: `Most or all will have prostate cancer'. Correct response: `Most will NOT have prostate cancer').

These misconceptions about prostate cancer and prostate cancer screening, if believed by men, may lead to inappropriate or suboptimal decision‐making regarding their health‐care choices.

Data analysis

The percentage of subjects favouring each of the screening states was computed. Because few patients preferred screening state A of no screening, we focused our analysis on identifying those attributes that were predictive of a preference for screening state C (reassurance through biopsy, a more definitive form of reassurance) as opposed to screening state B (reassurance through PSA test and DRE, a less certain form of reassurance). We then conducted a series of bivariate analyses, examining the association between each independent variable and a preference for screening state C over state B, using contingency tables with χ ² as the test statistic. A logistic regression model was then specified, wherein all independent variables were initially included in the analysis and backward stepwise elimination was used to identify an optimal model. Again, preferences for health state C over health state B served as the dependent variable. The likelihood ratio statistic was used for removal of independent variables. All analyses were performed by using SPSS for Windows Version 10 (SPSS Inc., Chicago, IL).

Results

Sociodemographic characteristics

The sociodemographic characteristics of the study subjects are presented in Table 1. The study sample was ethnically diverse: 34.5% of the subjects were minorities. Most (71.4%) of the subjects were over the age of 50. The level of education among the subjects varied widely – 38.0% of the subjects had a high school diploma or less, and 27.4% had a college diploma or some form of graduate study. The annual household income also varied widely – 37.5% had an income of less than $20 000 and 40.5% had an income of greater than $40 000.

Table 1.

  Characteristics of study subjects (n = 168)

Preferences for prostate cancer screening states

The subjects' preference rankings for the three prostate cancer screening states described above are detailed and summarized in Table 2. Of the 168 men surveyed, 161 (96.8%) associated at least some reassurance with being screened. Although screening state A was considered to be the worst state by many (86.9%) of the men, a few (4.2%) considered it to be the best. More than half (57.1%) of the men thought screening state B to be the best state; only a few (1.8%) considered it to be the worst. With regard to screening state C, 65 (38.7%) of the men considered it to be the best, whereas 19 (11.3%) of the men considered it to be the worst.

Table 2.

  Rank order for prostate cancers‐screening health states*

Relationship between independent variables and preference for health state C over health state B

3 4Table 3 gives the results of bivariate analyses, comparing subjects with a preference for screening state C with those with preferences for screening state B. Age, ethnicity, education and income were not significantly related to a preference for screening state C over state B. The subjects who reported having had a PSA test in the past were no more likely to prefer screening state C than were the other subjects.

Table 3.

  Characteristics of subjects preferring screening state C (greater reassurance) over other screening state B (lesser reassurance)

Table 4.

Table 3   Continued

A family history of prostate cancer was strongly associated with a preference for health state C (3, 4 P = 0.001; Table 3). Nineteen (29.2%) of the 65 men who preferred health state C reported having a relative with prostate cancer, as compared with nine (9.4%) of the 94 men who preferred health state B. Nine (13.8%) of the 65 men who preferred health state C indicated that they had a greater risk of prostate cancer compared with other men, while two (2.1%) of the 94 men preferring health state B considered themselves at greater risk (P = 0.02). No relationship was observed between prostate cancer beliefs and misconceptions and a preference for screening state C.

Results from the logistic regression model predicting preferences for screening state C over state B are given in Table 4. All independent variables were initially included in the model. Subjects reporting a family history of prostate cancer were more likely to report a preference for state C over state B than those subjects reporting no family history (OR = 3.10, P = 0.02). Interestingly, subjects who did not know about their family history were less likely than subjects reporting no family history to favour state C over state B. Finally, subjects who considered themselves at greater risk of prostate cancer than other men were more likely to prefer state C than subjects reporting a lesser chance (OR = 7.13, P = 0.04).

Table 4.

  Results from logistic regression analysis predicting preference for screening state C over state B

Discussion

While the health benefits of early detection of prostate cancer in asymptomatic men remain uncertain, our sample of primary care patients indicated that negative results from prostate cancer screening had reassurance value. The fact that the patients overwhelmingly preferred the health states that included the PSA test supports the clinical observation that as men get older they want to be screened for prostate cancer. ⁵ A significant number of our patients were even willing to endure the anxiety of a false‐positive PSA test result and the pain of biopsy to more definitively establish the absence of cancer.

The importance of reassurance is underscored by Kattan et al. ¹³ in their decision analysis of treatment for clinically localized prostate cancer. They found that the health state of `watchful waiting' (i.e. a positive diagnostic test but no active treatment) produced such a significant psychological burden that radical prostatectomy was preferred. The latter clinical strategy was preferred because it maximized quality‐adjusted life expectancy.

Although we used the literature on cancer screening behaviour to guide our analysis, on the whole we did not find that factors related to screening were also related to reassurance. For example, Close et al. ⁵ found that the percentages of men undergoing a PSA test dramatically increased with age, but we did not find such a gradient in our study sample. And although Simoes et al. ⁷ found that lack of financial barriers to medical care and a high school‐level education were associated with increased compliance with cervical cancer screening, we found no association between education and income levels and reassurance.

We also were interested in the relationship between ethnicity and screening behaviour. Because African‐American men have been found to have a greater lifetime risk of developing prostate cancer and of dying of the disease, ^{14 , 15} the American Cancer Society now recommends that they begin screening earlier than the prescribed age of 50 years. Ethnicity was not significantly associated with preference in this study. Yet perceived risk was related to preference, suggesting that men who are aware of their increased risk status want the reassurance of screening.

Some studies have suggested that engaging in cancer screening is akin to engaging in other healthy behaviours, such as screening for other types of cancer. For example, Simon et al. ⁸ found that past participation in breast cancer screening (i.e. mammography) was an important predictor of current mammography use among a sample of women who were members of a large health maintenance organization in Detroit. In addition, Close et al. ⁵ found that behaviours now associated with good health, such as regular exercise and a healthy diet, were significantly associated with prostate cancer screening in a population‐based sample of men. In our study sample, however, we found no such associations between previous PSA testing and reassurance.

The only variables we found to be statistically significant for a greater likelihood of wanting `ultimate reassurance' for the absence of prostate cancer were family history and perceived comparative risk. This finding is consistent with the review by Vernon ⁹ and the research of Clavel‐Chapelon et al. ¹⁶ who found that individuals with a family history of colorectal cancer are more likely to be screened for the disease than those without affected family members. Clearly, there is a correlation between family history and comparative risk. If a patient who knows that members of his family have had prostate cancer, then he should also know that he has a greater chance of developing the disease than do other men in his age group. Furthermore, perceived comparative risk appears to impact preferences for reassurance somewhat independently of family history. This finding suggests that some subjects without a family history of prostate cancer may still consider themselves at greater risk of prostate cancer. The reasons for this perception are not entirely clear from this study.

Reassurance, i.e. knowledge about a health state, may not always be of value. In a study of the prenatal diagnostic value of ultrasound, Berwick and Weinstein ¹⁷ found that some parents preferred to suppress specific information elicited from the test, that is, they did not want to know the sex of their baby until his or her delivery. The reassurance value came instead from the parents knowing that `everything was all right', meaning their child bore no major physical abnormalities. In contrast, in a decision analysis of prenatal diagnostic tests for foetal chromosomal abnormalities, Heckerling and Verp ¹⁸ compared the reassurance value of chorionic villus sampling as opposed to amniocentesis. They found that if the reduction in parental anxiety afforded by these tests had any measurable value, then chorionic villus sampling would be recommended, even though it must be performed at an earlier stage of foetal development and carries a greater risk of miscarriage than does amniocentesis.

Perhaps these examples of reassurance in medical decision‐making seem far removed from the question of prostate cancer screening, but there are few published examples where psychological attributes are explicitly incorporated in the decision‐making. Prenatal testing is an important one, especially because usually everything is `all right', and no clinical action is required. However, the reassurance is extremely valuable for prospective parents.

No matter the sophistication of the diagnostic technique or the sensitivity of the screening test, the possibility of uncertainty in medical diagnoses will always be a factor. In his theoretical paper on the temporal resolution of uncertainty, Wu ¹⁹ points out the reassurance value of resolving uncertain diagnoses as quickly as possible. In the clinical examples that he presents, uncertainty typically can be resolved. But because there is no diagnostic test that is completely accurate for any disease or medical condition, there is always the slightest possibility that a patient's health state could unexpectedly change for the worse.

The results of this study might not be generalizable to other patient populations for several reasons: (1) the study subjects were recruited from only three primary care clinics in Texas; (2) only married men were enrolled; (3) the number of subjects was limited and (4) preferences were given for hypothetical clinical scenarios. Whether the results would transfer to a larger or more diverse sample remains to be tested. The sample of men was constructed by methods of convenience from three family medicine clinics. Because the men who participated were volunteers recruited from waiting rooms and through posters in the clinic facilities, the men selected are self‐referred. Thus, there can be no `refusal' or `drop out' rate, and the men in the study may not be entirely representative of married men from a family medicine population.

Another limitation is that we did not assess preference scores using formal methods of utility elicitation for the various health states. It is especially challenging to formally perform utility assessment for health states involving psychological attributes. Rather, by focusing on the rank order of the prostate cancer screening health states, we were able to get a sense of how much knowledge would affect preference and who it would affect.

Although we asked patients for their preferences for these sequences of health states, we did not `tease out' the reasons why the patients had particular preferences. Most patients preferred having the hypothetical sequence described as health state B compared with health state C. We hypothesize that the reason health state B is preferred is because it avoids the false‐positive result as described in health state C. There may be more to these health states than just the psychological attribute of reassurance; however, the psychological aspects seem to be much more salient than the physical aspects for these health state descriptions.

It should be noted that we dealt with only one side of the reassurance issue, i.e. what happens after a negative diagnostic test. It may be the case that patients are reassured simply by having more definitive `knowledge' about their true prostate cancer status even if they learn they have cancer. We decided to focus on only one side of the reassurance issue because it is far more common, based on prostate cancer prevalence, that the tests for prostate cancer will rule out disease rather than rule it in.

Eddy ²⁰ has suggested that the increased cost of medical care has resulted because of physicians' need to lessen uncertainty of diagnoses. If more and more patients require that their uncertainty be lessened and insist that their preferences for prostate or other cancer‐screening health states be met, then they will seek those diagnostic tests or procedures that will most greatly decrease the probability of disease and provide `ultimate reassurance.' As we have seen from our preliminary study, the tests and procedures chosen for their reassurance value most likely will be more invasive and therefore more costly. Justifying the greater costs to achieve these psychological benefits for patients has implications for public health policy.

Acknowledgements

The authors thank Michael J. Barry, MD, for his contributions as a consultant to this study. The authors also thank Walter J. Pagel and Pamela P. Tice for their editorial comments. The authors appreciated the comments of Margaret Holmes‐Rovner, PhD, and Jonathan M. Baron, PhD, on an earlier version of the manuscript.

Financial support for this study was provided, in part, by a grant from the Agency for Healthcare Research and Quality (Grant No. R01‐HS08992). The funding agreement ensured the authors' independence in designing the study, interpreting the data, writing and publishing the report.