- Email: [email protected]
High Rates of Prostate-specific Antigen Testing in Men with Evidence of Benign Prostatic Hyperplasia
CLINICAL STUDIES
High Rates of Prostate-specific Antigen Testing in Men with Evidence of Benign Prostatic Hyperplasia James B. Meigs, MD, MPH, Michael J. Barry, MD, Edward Giovannucci, MD, Eric B. Rimm, ScD, Meir J. Stampfer, MD, Ichiro Kawachi, MD, PhD PURPOSE: Benign prostatic hyperplasia is common among men who may be candidates for prostate cancer screening using prostate-specific antigen (PSA) testing. Patterns of PSA testing among men with evidence of benign prostatic hyperplasia have not been studied. METHODS: We examined the prevalence and correlates of a self-reported history of PSA testing. In 1994, 33,028 US health professionals without prostate cancer aged 47 to 85 years provided information on prior PSA testing, lower urinary tract symptoms characteristic of benign prostatic hyperplasia, history of prostatectomy, and prostate cancer risk factors. In 1995, a subset of 7,070 men provided additional information on diagnosis and treatment of benign prostatic hyperplasia. RESULTS: From 39% of men in their 50s to 53% of men in their 80s reported PSA testing in the prior year (P ,0.0001 for trend with age). Men were more likely to report PSA testing if
they had lower urinary tract symptoms characteristic of benign prostatic hyperplasia (age-adjusted odds ratio for severe symptoms 2.2, 95% confidence interval 1.8 to 2.6), a prior history of prostatectomy (age-adjusted odds ratio 1.1, 95% confidence interval 1.02 to 1.2), or a physician diagnosis of benign prostatic hyperplasia (odds ratio 1.9, 95% confidence interval 1.7 to 2.2; adjusted for age, signs or symptoms of benign prostatic hyperplasia, and prostate cancer risk factors). CONCLUSIONS: These US health professionals reported preferential use of PSA testing among men least likely to benefit from early cancer detection (older men) and among men most likely to have a false-positive PSA result (men with benign prostatic hyperplasia). Physician and patient education are needed to promote more rational and selective use of this screening test. Am J Med. 1998;104:517–525. q1998 by Excerpta Medica, Inc.
P
cancer does more good than harm (4), even advocates of early detection question its value in men older than 70 to 75 years, in whom benign prostatic hyperplasia is especially prevalent (5). Prostate-specific antigen (PSA) testing has become increasingly popular for early detection of prostate cancer (6,7). Unfortunately, among men with benign prostatic hyperplasia, PSA testing appears to provide little useful additional information about the presence of prostate cancer (8 –10), and is likely to result in false-positive results. For example, the false-positive rate for a PSA level greater than 4.0 ng/mL in men not known to have benign prostatic hyperplasia is 8% to 13%, but increases to 32% to 60% in men with this condition (11,12). Furthermore, neither a history of benign prostatic hyperplasia nor the presence of lower urinary tract symptoms characteristic of benign prostatic hyperplasia appear to increase the risk of prostate cancer (3,13–15). Despite uncertain benefits, no data are available on rates of PSA testing in men with benign prostatic hyperplasia. We examined the prevalence and correlates of PSA testing during 1993 and 1994 among men with evidence of benign prostatic hyperplasia in a large cohort of US health professionals.
rostate cancer and benign prostatic hyperplasia are major health problems among men in the United States (1,2). As older age is a risk factor for both of these diseases (1,3), the potential target population for early detection of prostate cancer is enriched with men who have benign prostatic hyperplasia. Although it is unknown whether early detection and treatment of prostate
From the General Medicine Division of the Medical Services (JBM, MJB), Massachusetts General Hospital, and the Department of Medicine, Harvard Medical School; Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women’s Hospital (EG, EBR, MJS, IK); Departments of Epidemiology and Nutrition, Harvard School of Public Health (EG, EBR, MJS), and Departments of Health and Social Behavior (IK), Boston, Massachusetts. Supported by Research Grants DK45779, CA55075, and HL35464 from The National Institutes of Health and by PORT II Grant No. HS 08397 from the Agency for Health Care Policy and Research. Requests for reprints should be addressed to James B. Meigs, MD, MPH, General Internal Medicine Unit S50-9, Massachusetts General Hospital, Boston, Massachusetts 02114. Presented in abstract form at the Society of General Internal Medicine Annual Meeting, Washington, DC, May 1996. Manuscript submitted May 14, 1997 and accepted in revised form March 17, 1998. q1998 by Excerpta Medica, Inc. All rights reserved.
0002-9343/98/$19.00 517 PII S0002-9343(98)00113-2
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
METHODS Study Population The Health Professionals Follow-up Study is a prospective cohort of 51,529 US male dentists, optometrists, osteopaths, podiatrists, pharmacists, and veterinarians born between January 1, 1910, and December 31, 1946, who responded to a mailed questionnaire in 1986. We collected information on age, ancestry, a family history of prostate cancer, smoking behavior, and the presence and year of occurrence of a variety of medical conditions, including prostate cancer, vasectomy, and prostatectomy for benign prostatic hyperplasia. Further details about the cohort study and its participants have been reported (16). We mailed follow-up questionnaires to cohort members biennially to ascertain new cases of disease and to update exposure information. In 1992 and 1994 we added a section to assess prevalent lower urinary tract symptoms, and in 1994, a question regarding prior PSA testing. Lower urinary tract symptoms were measured using the American Urological Association (AUA) symptom index, a validated 7-item questionnaire that assesses the frequency over the preceding month of lower urinary tract symptoms characteristic of benign prostatic hyperplasia (symptoms include nocturia, weak urinary stream, hesitancy, intermittence, incomplete bladder emptying, urinary frequency, and urgency; the score range from 0 to 35, with higher scores reflecting more severe symptoms) (17). In 1995, we mailed a supplemental questionnaire to all men who reported high-moderate to severe symptoms in 1994 (AUA symptom score $15, n 5 1,452) and an age-stratified random sample of men with none to lowmoderate symptoms (AUA symptom score #14, n 5 8,278) to obtain more detailed information about benign prostatic hyperplasia-related symptoms, diagnoses, and treatments. From the 38,706 respondents to the 1994 questionnaire we excluded men for whom we had incomplete information on PSA testing and lower urinary tract symptoms (n 5 4,129) or who reported prostate cancer as of 1994 (n 5 1,549), leaving 33,028 men for analysis of PSA testing rates. The validity of a self-reported diagnosis of prostate cancer in this cohort has been established (18). Of 8,208 men who responded to the 1995 benign prostatic hyperplasia questionnaire (response rate 85% for men with high-moderate to severe symptoms and 84% for men with none to low-moderate symptoms), 7,070 men free of prostate cancer in 1994 and for whom we had complete information on PSA testing and lower urinary tract symptoms were available for more detailed analysis.
Assessment of PSA Testing We ascertained prior PSA testing in the entire cohort as of 1994 with the question: “Have you ever had a PSA blood test for prostate cancer screening?” For affirmative responses, we recorded the answers to: “When was your 518
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104
most recent test: before 1988, ’88 –’89, ’90 –’91, ’92, ’93, or ’94?” If they reported a PSA test, we asked whether the result was elevated. We also asked whether they had ever had a prostate biopsy or a transrectal ultrasound examination of the prostate. On the 1995 benign prostatic hyperplasia questionnaire, the question was phrased, “Since January 1, 1994, have you had screening for prostate cancer by using PSA (prostate specific antigen)?” Men in the entire cohort were classified as having PSA testing in the prior year if the most recent PSA test was performed in 1993 or 1994, and for men in the subgroup, if a PSA test had been performed since January 1, 1994.
Assessment of Benign Prostatic Hyperplasia We used a variety of definitions to classify subjects as having evidence of benign prostatic hyperplasia because there are no standard criteria for this condition for use in epidemiologic studies (19). Using the AUA symptom index, we ascertained the frequency, over the preceding month, of lower urinary tract symptoms characteristic of benign prostatic hyperplasia (17,20,21). Symptom scores allowed classification of men with none to mild (AUA symptom score #7), moderate (score 8 to 19), and severe (score $20) lower urinary tract symptoms. Because there is no generally accepted threshold for the diagnosis of benign prostatic hyperplasia, data were analyzed by AUA symptom score class to assess the “dose-response” relation between increasingly severe lower urinary tract symptoms and PSA testing. The AUA symptom index was administered to the cohort in 1992, 1994, and 1995; subjects were classified based on 1994 responses for the entire cohort and 1995 responses for the more detailed subgroup to correlate symptoms with concurrent PSA testing. We also calculated the change in lower urinary tract symptoms between 1992 and 1994 (the period concurrent with or immediately preceding reported PSA testing). We classified men as having stable to improved lower urinary tract symptoms, as indicated by remaining in the same or moving to a less symptomatic class (ie, from moderate symptoms in 1992 to none to mild symptoms in 1994), or as having worsening lower urinary tract symptoms, as indicated by moving to a more symptomatic class. We also assessed symptom change by classifying men as having worsening symptoms if the AUA symptom score increased by 3 or more points between 1992 and 1994 (21). Results using AUA symptom class or score increase were similar; only those using class increase are reported here. Symptom change analysis was limited to the 26,052 men who provided complete AUA symptom index data in both 1992 and 1994. Alternatively, we considered subjects to have benign prostatic hyperplasia if they reported a history of prostatectomy for this condition. Subjects were classified as having had a prostatectomy during 1993 and 1994, prior to 1993, or never. The validity of a self-reported prosta-
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
tectomy in this cohort has been established (16). In 1988 we requested permission to review the medical records of 99 randomly selected men who reported having a prostatectomy. Of 77 men who granted permission, 74 records were obtained, all of which confirmed the surgery. The 1995 supplemental questionnaire provided additional information on benign prostatic hyperplasia diagnoses and drug therapy. Men in this subgroup were considered to have benign prostatic hyperplasia if they responded affirmatively to either question: “Since January 1, 1990, have you had an enlarged prostate detected by rectal exam?” or “Since January 1, 1990, has a physician told you that you have BPH (benign prostatic hyperplasia)?” Subjects were classified as receiving drug treatment for benign prostatic hyperplasia if they reported current, regular use of finasteride or the alpha-blocker drugs terazosin, prazosin, or doxazosin specifically for prostate problems.
Assessment of Prostate Cancer Risk Factors To provide a contrast to men with benign prostatic hyperplasia, we analyzed PSA testing patterns among men with risk factors for prostate cancer, who might be expected to have higher rates of PSA testing. Risk factors were selected based on previously reported associations with prostate cancer (3,15) and on assumptions about which factors might commonly influence the clinical decision to order a PSA test. We classified men as African American or other (Southern European or Mediterranean, Scandinavian, other Caucasian, Asian or Oriental, or other origin). We classified men as cigarette smokers if they reported current cigarette use in 1994, as having had a vasectomy if they reported this procedure during or prior to 1994, and as having a family history of prostate cancer if they reported a father or brother with this condition. Dietary risk factors for prostate cancer, although available, were not included based on the assumption that information about dietary risk is rarely considered in routine PSA testing decisions.
Data Analysis Two sets of analyses were performed—the first on the entire study cohort as of 1994, and the second on the subgroup for whom more detailed information was available on the diagnosis and treatment of benign prostatic hyperplasia. We estimated the age-stratified prevalence of PSA testing, lower urinary tract symptoms, and prostatectomy for the entire cohort. We then compared the age-stratified proportions of men with or without benign prostatic hyperplasia and with or without prostate cancer risk factors who reported PSA testing during the prior year. Trend in the proportion reporting PSA testing across age decades was assessed with the Mantel-Haenszel test for trend (22). Differences in proportions reporting PSA testing were summarized over age decades with the Mantel-Haenszel estimators (22). Odds ratios and 95%
confidence intervals (CI) for PSA testing adjusted for age were calculated using logistic regression (23). Models assessing the effect of a change in AUA symptom score class between 1992 and 1994 on PSA testing included a covariate for AUA symptom score class in 1992. Although the majority of men (89.1%) reported having a periodic health examination during 1990 through 1994, we assessed the possibility that surveillance bias might affect our results by repeating the analysis among this subset of men. Results obtained in this subset were essentially identical to those obtained among the full set of eligible men; only the latter results are shown. Multivariate odds ratios and 95% CI for PSA testing were calculated using separate models with benign prostatic hyperplasia diagnoses and prostate cancer risk factors each entered separately or in combination. Additional covariates in the multivariate models included age, profession, report of a periodic health examination between 1990 and 1994, and use of finasteride or alphablocker medications. Adequacy of each multivariate model was assessed with the Hosmer and Lemeshow goodness-of-fit test; additional regression diagnostics were performed by plotting changes in the deviance, the chi-square goodness-of-fit statistic, and the confidence interval displacement diagnostic against the predicted probabilities for each observation (23). These procedures indicated adequate fit for the multivariate logistic models overall and for the majority of covariate patterns. The area under the receiver-operating-curves for the multivariate logistic regression models ranged from 0.68 to 0.71. SAS statistical software was used for the analysis (24). Statistical significance was defined as a two-tailed P value ,0.05.
RESULTS The majority of men in the cohort were in the age range when PSA testing for early detection of prostate cancer might be recommended: 87.4% were between the ages of 50 and 79 years (Table 1). PSA testing was common and increased significantly with age. Although there was a slight fall-off among men in their 80s, prevalence rates remained high even among the oldest men: 74.4% of men in their 70s and 66.6% of men in their 80s reported PSA testing at least once during or prior to 1994. The majority of men in every age decade reported their most recent PSA test during the preceding year (1993 to 1994), suggesting either that PSA testing became increasingly common in this cohort between 1990 and 1994 or that many of these men had begun to receive annual PSA testing. Benign prostatic hyperplasia, defined either on the basis of lower urinary tract symptoms or a history of prostatectomy, also was common and increased significantly with age. The majority of men who reported a prostatectomy June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 519
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
Table 1. Age-stratified Prevalence of Prostate-specific Antigen Testing (PSA), Symptoms of Benign Prostatic Hyperplasia, and Prostatectomy for Benign Prostatic Hyperplasia among 33,028 Health Professionals* Age (Years)
Number (%) Most recent PSA test Never Prior to 1990 1990–1992 1993–1994 AUA Symptom Index class (1994) None–mild (score #7) Moderate (score 8–19) Severe (score $20) History of prostatectomy for benign prostatic hyperplasia Never Prior to 1993 1993–1994
47–49
50–59
60–69
70–79
80–85
3216 (9.7)
11633 (35.2)
10359 (31.4)
6911 (20.9)
909 (2.8)
2374 (73.8) 13 (0.4) 208 (6.5) 621 (19.3)
5808 (49.9) 53 (0.5) 1205 (10.4) 4567 (39.3)
3193 (30.8) 63 (0.6) 1373 (13.3) 5730 (55.3)
1766 (25.6) 55 (0.8) 842 (12.2) 4248 (61.5)
304 (33.4)† 19 (2.1) 108 (11.9) 478 (52.6)
2959 (92.0) 244 (7.6) 13 (0.4)
10022 (86.2) 1496 (12.9) 115 (1.0)
7897 (76.2) 2265 (21.9) 197 (1.9)
4696 (67.9) 2003 (29.0) 212 (3.1)
602 (66.2)† 271 (29.8) 36 (4.0)
3205 (99.7) 9 (0.3) 2 (0.1)
11499 (98.8) 105 (0.9) 29 (0.2)
9729 (93.9) 557 (5.4) 73 (0.7)
5563 (80.5) 1257 (18.2) 91 (1.3)
568 (62.5)† 328 (36.1) 13 (1.4)
* Data are n (%). † P ,0.0001 for trend across age decades. PSA 5 prostate-specific antigen; AUA 5 American Urological Association.
had their surgery prior to the time interval during which PSA testing was assessed. Of men who reported ever having a PSA test, 1.6% of men in their 40s, 4.0% of men in their 50s, 9.1% of men in their 60s, 13.8% of men in their 70s, and 14.8% of men in their 80s reported that the PSA level was elevated (P ,0.001 for trend). In addition, 1.7% of men in their 40s, 4.6% of men in their 50s, 14.3% of men in their 60s, 28.2% of men in their 70s, and 33.8% of men in their 80s reported having ever had a prostate biopsy or rectal ultrasound for prostate examination (P ,0.001 for trend).
Prevalence of PSA Testing in Men with Benign Prostatic Hyperplasia In every age decade, PSA testing was significantly more common among men with benign prostatic hyperplasia than among men without benign prostatic hyperplasia (Table 2). Increasing severity of lower urinary tract symptoms, a worsening of symptoms in the 1 or 2 years preceding PSA testing, and a history of prostatectomy prior to the year of PSA testing were each associated with a markedly increased prevalence of PSA testing. Only among 80-year-old men did PSA testing rates decline slightly. Nonetheless, over half of 80-year-old men with benign prostatic hyperplasia reported PSA testing during the preceding year. Older men with moderate or severe lower urinary tract symptoms had especially high PSA testing rates, including 63.5% of men in their 60s with moderate symptoms and 72.2% of men in their 80s with severe symptoms. Among men with moderate to severe lower urinary tract symptoms (AUA symptom score 8 to 35) who re520
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104
ported ever having a PSA test, 2.8% of men in their 40s, 8.6% of men in their 50s, 15.2% of men in their 60s, 18.6% of men in their 70s, and 17.6% of men in their 80s reported that the PSA level was elevated (P ,0.001 for trend). Among men reporting an elevated PSA level, 30.8% of men in their 40s, 55.5% of men in their 50s, 66.6% of men in their 60s, 73.1% of men in their 70s, and 79.1% of men in their 80s reported having ever had a prostate biopsy or rectal ultrasound for prostate examination (P ,0.001 for trend). Men with risk factors for prostate cancer, on the other hand, did not uniformly report higher PSA testing rates than men without risk factors (Table 2). PSA testing was more common among men with a family history of prostate cancer or a history of vasectomy, whereas AfricanAmerican men were slightly, but not significantly, less likely to receive such testing. The limited number of African-American men in this cohort (207, 0.6%) limits our power to be certain about the effect of race on PSA testing. Current cigarette smokers were actually less likely than nonsmokers to have had PSA testing.
Odds of PSA Testing After Adjustment for Age and Potential Confounding Age-adjusted odds ratios for PSA testing, according to the presence of benign prostatic hyperplasia or risk factors for prostate cancer, are presented in Table 3 for both the entire cohort and for the more detailed subgroup. Even after age adjustment, an increasing burden of lower urinary tract symptoms, worsening symptoms, or a history of prostatectomy all significantly increased the odds of PSA testing. In the entire cohort men with moderate
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
Table 2. Age-stratified Prevalence of Prostate-specific Antigen (PSA) Testing during 1993–1994 among 33,028 Health Professionals with or without Benign Prostatic Hyperplasia or Risk Factors for Prostate Cancer* Percent Reporting PSA Testing by Age (Years)
AUA Symptom Index class (1994) None–mild (score #7) Moderate (score 8–19) Severe (score $20) Change in AUA Symptom Index class 1992–1994 Same or improved Worsened History of prostatectomy for benign prostatic hyperplasia Never Prior to 1993 1993–1994 Family history of prostate cancer No family history of prostate cancer Vasectomy No history of vasectomy African American Non-African American Current smoker Nonsmoker
47–49
50–59
60–69
70–79
80–85
18.3 30.3 38.5
37.1 51.8 63.5
52.6 63.5 69.5
58.3 67.7 73.1
50.5†‡ 54.6 72.2
18.7 32.8
38.2 49.2
54.7 60.3
61.2 65.4
52.9†‡ 51.4
19.2 33.3 100.0 29.3 18.8 23.0 17.7 27.3 19.1 17.9 19.4
38.9 66.7 86.2 46.6 38.8 42.4 37.8 31.8 39.4 29.8 39.9
54.3 68.2 87.7 60.3 55.0 59.2 53.9 51.7 55.4 43.8 56.1
60.5 64.1 85.7 69.0 61.0 63.3 61.1 56.8 61.6 46.5 62.2
52.8†‡ 50.6 92.3 59.7†‡ 52.1 50.0†‡ 52.9 0.0† 52.7 34.6†‡ 53.1
* Data are % reporting PSA testing in each category. † P ,0.0001 for trend across age decades. ‡ P ,0.0001 for Mantel-Haenszel summary estimate for differences within age groups. PSA 5 prostate-specific antigen; AUA 5 American Urological Association.
symptoms had 1.6-fold greater odds of having had a PSA test, whereas men with severe symptoms had 2.2-fold greater odds. Among men in the subgroup, those who had been told they had benign prostatic hyperplasia, or who reported an enlarged prostate detected by rectal examination, or who were under benign prostatic hyperplasia-specific drug treatment all were more likely to report PSA testing during the preceding year. Compared with men in the entire cohort with no or mild lower urinary tract symptoms, those with moderate symptoms were more likely to report that the PSA level was elevated (age adjusted, odds ratio 5 2.1, 95% CI 1.9 to 2.4); those with severe lower urinary tract symptoms had 2.7-fold greater odds (95% CI 2.1 to 3.4). Men with an elevated PSA level had 9.7-fold greater odds to report having ever had a prostate biopsy or rectal ultrasound for prostate examination (age and symptom adjusted; 95% CI 8.6 to 10.9). A family history of prostate cancer or a history of vasectomy also significantly increased the odds of PSA testing after adjustment for age in both groups, although in the smaller subgroup, the effect of family history was attenuated. African-American men were no more likely to have received PSA testing. Current cigarette smokers were significantly less likely than nonsmokers to have received PSA testing during the previous year. In addition to age, the subjects’ professions, having
seen a physician for a physical examination (thereby providing the opportunity for PSA testing), being on treatment with benign prostatic hyperplasia-specific drugs (a possible indication for PSA testing) (25), or the presence of prostate cancer risk factors, could have confounded the association between benign prostatic hyperplasia and PSA testing and PSA testing. Table 4 displays a series of multiple logistic regression models designed to estimate the effect of benign prostatic hyperplasia on PSA testing while controlling for these covariates. Entering all four prostate cancer risk factors simultaneously into a logistic regression model did not qualitatively change their bivariate relationships to PSA testing. A positive family history or history of vasectomy were still associated, although not significantly, with PSA testing. Next, simultaneous adjustment for risk factors, possible confounders, and several benign prostatic hyperplasia definitions demonstrated that an increasing burden of lower urinary tract symptoms, an enlarged prostate detected by rectal examination, and a history of prostatectomy all independently increased the odds that a man received PSA testing. Finally, addition of a physician diagnosis of benign prostatic hyperplasia to a logistic regression model containing prostate cancer risk factors and signs and symptoms consistent with benign prostatic hyperplasia weakened the association of lower urinary tract symptoms or prior prostatectomy with PSA testing, probably because these June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 521
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
Table 3. Age-adjusted Odds Ratios for Prostate-specific Antigen (PSA) Testing during the Previous Year in Male Health Professionals with Benign Prostatic Hyperplasia (BPH) or Risk Factors for Prostate Cancer* 33,028 Men in Entire Cohort PSA/No PSA N with PSA testing/N without PSA testing AUA Symptom Index class† None-mild (score #7) Moderate (score 8–19) Severe (score $20) Worsened AUA Symptom Index class 1992–1994‡ History of prostatectomy for benign prostatic hyperplasia Never Prior to 1993 1993–1994 Family history of prostate cancer Vasectomy African American Current Smoker Physician-diagnosed benign prostatic hyperplasia§ Enlarged prostate detected by rectal examination Finasteride or alpha-blocker therapy
OR
95% CI
15,644/17,384
7,070 Men in Subgroup PSA/No PSA
OR
95% CI
3,654/3,416
11,457/14,719 3,791/2,488 396/177 1,329/1,008
1.0 1.6 2.2 1.4
Referent 1.5, 1.7 1.8, 2.6 1.3, 1.5
1,917/2,296 1,437/974 300/146 542/407
1.0 1.4 1.9 1.2
Referent 1.3, 1.6 1.5, 2.4 1.02, 1.4
14,038/16,526 1,425/831 181/27 1,130/937 4,128/4,413 94/113 701/1,244
1.0 1.1 5.6 1.4 1.3 0.9 0.6
Referent 1.02, 1.2 3.7, 8.4 1.3, 1.5 1.2, 1.3 0.7, 1.2 0.6, 0.7
3,287/3,227 316/175 50/4 263/213 987/912 13/13 156/233 1,270/459 679/322 462/99
1.0 1.0 9.2 1.2 1.2 0.9 0.7 2.7 2.0 3.9
Referent 0.8, 1.2 3.3, 26 0.97, 1.4 1.1, 1.3 0.4, 1.9 0.5, 0.8 2.4, 3.1 1.7, 2.3 3.1, 4.9
* Data are odds ratios (OR) and 95% confidence intervals (95% CI) for PSA testing during 1993–1994 (n 5 33,028) or 1994 –1995 (n 5 7,070) age-adjusted using age as a continuous variable. † American Urological Association (AUA) Symptom Index class in 1994 (n 5 33,028) or 1995 (n 5 7,070). ‡ Change in AUA Symptom Index class adjusted for age and baseline class; referent group was men with stable or improved class. § Data on physician-diagnosed benign prostatic hyperplasia, enlarged prostate detected by rectal examination, and drug therapy were not available for the entire cohort. PSA 5 prostate-specific antigen.
conditions lead directly to the diagnosis of benign prostatic hyperplasia by the physician. Nonetheless, a physician diagnosis of benign prostatic hyperplasia and the presence of an enlarged prostate detected by rectal examination both independently increased the odds that a man reported PSA testing even after adjustment for lower urinary tract symptoms, a history of prostatectomy, prostate cancer risk factors, age, and other potential confounders. Men whose physicians told them they had benign prostatic hyperplasia had nearly twice the odds of PSA testing for early detection of prostate cancer as men without such a diagnosis.
DISCUSSION In this study we found that men with evidence of benign prostatic hyperplasia were significantly more likely than men without such evidence to receive PSA testing for early detection of prostate cancer. We also confirmed previous reports of high PSA testing rates among older men in the US (6,7). In this large, well-characterized cohort, PSA testing for prostate cancer was extremely common, especially among older men, in whom the benefits
522
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104
of early cancer detection are the most questionable. Three fourths of men in their 70s and two thirds of men in their 80s reported having a PSA test at least once as of 1994. The majority of tests were performed during 1993 and 1994. Given the controversy surrounding the value of early detection of prostate cancer (4), these PSA testing rates seem remarkably high. Although men in the Health Professionals Follow-up Study are generally health conscious, and as such may not be completely representative of all US men, the prevalence of PSA testing in this cohort is consistent with testing rates projected from data from a community study (7) or from four National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) sites (6). Among men in Olmsted County, for example, PSA testing rates rose steadily beginning in 1987, so that by 1992, 6% of men in their 40s, 23% of men in their 50s, 43% of men in their 60s, and 64% of men in their 70s had at least one PSA test during their lifetime (7). These rates suggest that physicians were not following 1989 guidelines published by the US Preventive Services Task Force, who recommended against any routine screening for prostate cancer (26). They are more consistent with 1992–1993 recommendations by the American
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
Table 4. Multivariate Odds Ratios for Prostate-specific Antigen (PSA) Testing during the Previous Year in 7,070 Male Health Professionals with Benign Prostatic Hyperplasia or Risk Factors for Prostate Cancer*
†
Family history of prostate cancer Vasectomy African American Current smoker AUA Symptom Index class (1995) None–mild (score #7) Moderate (score 8–19) Severe (score $20) Enlarged prostate detected by rectal examination History of prostatectomy for benign prostatic hyperplasia Never Prior to 1993 1993–1994 Physician-diagnosed benign prostatic hyperplasia
Prostate Cancer Risk Factors, Signs or Symptoms of Benign Prostatic Hyperplasia, and a Physician Diagnosis of Benign Prostatic Hyperplasia
Prostate Cancer Risk Factors
Prostate Cancer Risk Factors and Signs or Symptoms of Benign Prostatic Hyperplasia
OR
95% CI
OR
95% CI
OR
95% CI
1.2 1.1 0.7 0.7
0.96, 1.5 0.99, 1.2 0.3, 1.6 0.5, 0.9
1.2 1.1 0.7 0.7
0.97, 1.5 0.99, 1.2 0.3, 1.7 0.6, 0.9
1.2 1.1 0.8 0.7
0.96, 1.5 0.98, 1.2 0.3, 1.8 0.6, 0.9
1.0 1.2 1.5 1.7
Referent 1.1, 1.4 1.2, 1.8 1.5, 2.0
1.0 1.1 1.2 1.4
Referent 0.97, 1.2 0.96, 1.5 1.2, 1.7
1.0 1.3 8.4
Referent 1.01, 1.6 3.0, 24
1.0 1.04 5.7 1.9
Referent 0.8, 1.3 2.0, 16 1.7, 2.2
* Data are odds ratios (OR) and 95% confidence intervals (95% CI) for PSA testing during 1994 –1995 adjusted for age, profession, having a physical examination 1990 –1994, and use of finasteride or alpha-blocker medication. † The referent groups for prostate cancer risk factor odds ratios are those men without the specified characteristic. PSA 5 prostate-specific antigen; AUA 5 American Urological Association.
Cancer Society and the American Urological Association, which recommended annual PSA testing in combination with a digital rectal examination for early detection of prostate cancer in men older than 50 years (5,27). However, the American Cancer Society acknowledged that early detection may only benefit men with at least a 10year life expectancy (5), equivalent to about age 75 in men with average comorbidities (28). The high rate of testing among men in their 70s and 80s in this cohort is of particular concern, as men in this age group may be the least likely to benefit from early aggressive treatment of a prostate cancer (29 –31). PSA testing rates were significantly higher in men of all ages with evidence of benign prostatic hyperplasia, whether defined by symptoms, a history of prostatectomy, an enlarged prostate on rectal examination, or by a physician diagnosis of benign prostatic hyperplasia. Even after accounting for age, prostate cancer risk factors, and signs, symptoms, and treatments for benign prostatic hyperplasia, men who had been told by their physicians that they had benign prostatic hyperplasia were more likely to report recent PSA testing. Not unexpectedly, we found
that men with benign prostatic hyperplasia who had a PSA test were more likely than men without benign prostatic hyperplasia to report that the PSA level was elevated, and that men reporting elevated PSA levels were very likely to report a prostate biopsy or transrectal ultrasound. Well over half of men older than 50 years of age, and four of five men older than 80 years of age, who had an elevated PSA level reported having such additional evaluations. SEER data also suggest that elevated PSA levels lead to increased rates of prostate biopsy and transrectal ultrasound examination (6). Given the poor specificity of PSA testing in patients with benign prostatic hyperplasia, our data suggest that in this cohort PSA testing has resulted in a high likelihood of false-positive results. Such inefficiency may lead to unnecessary additional evaluation with attendant risk, cost, and patient anxiety. We have previously estimated that the evaluation of a suspicious PSA or digital rectal examination in a man older than 65 years of age costs at least $444 (32), equivalent to a per-member per-month payment under many capitated insurance plans for Medicare beneficiaries. Evidence of inefficient, and possibly ineffective, use of med-
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 523
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
ical resources should be of concern to patients, physicians, and health policy-makers. A limitation of this analysis is uncertainty about the real reason why PSA tests were ordered. Our ascertainment of PSA testing relied on the self-reported response to a question about “prostate cancer screening,” but there may have been many reasons why men received PSA testing. Screening refers to testing men without any symptoms that raise the probability of prostate cancer, whereas early detection refers to testing men with symptoms that could suggest the presence of an early, occult malignancy. Although the presence of lower urinary tract symptoms does not appear to raise the probability of prostate cancer (13–15), PSA tests may still have been ordered as part of a work-up strategy in men with potential signs or symptoms of prostate cancer. Increased use of PSA testing associated with increasingly severe lower urinary tract symptoms is consistent with the concern that these symptoms represented prostate cancer. Other reasons for PSA testing may have included initiation and management of finasteride therapy in men with benign prostatic hyperplasia, as an automatic component of a routine chemistry panel, because the patient requested that a PSA test be performed, or because of concern on the part of the ordering physician about malpractice litigation. An additional limitation is lack of information about whether primary care physicians, urologists, or medical specialists were responsible for ordering PSA tests. Evidence from a national survey demonstrated that primary care physicians were much more likely to perform routine PSA testing for early cancer detection in older men than were urologists to recommend such testing. For example, 53% of primary care physicians reported routine PSA testing in men older than 79 years of age, whereas only 25% of urologists would have recommended testing for men this age (33). Primary care physicians were likely to be responsible for the majority of PSA testing in this cohort, and represent a group that might benefit from educational efforts to promote a more rational approach to prostate cancer screening. In summary, we have demonstrated high rates of PSA testing during 1993 and 1994, especially among older men and men with benign prostatic hyperplasia, in a large well-characterized cohort of US health professionals. These data indicate a need for physician education about the apparent lack of correlation among lower urinary tract symptoms and risk for prostate cancer, the high false-positive rate of PSA in the setting of such symptoms, and the uncertainty surrounding the balance of harms and benefits of early prostate cancer detection and treatment, especially among older men. Further research is needed on the relationship among lower urinary tract symptoms, benign prostatic hyperplasia, and prostate cancer in order to define the appropriate target population, if any, for early cancer detection. Until randomized 524
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104
trials indicate that prostate cancer screening does more good than harm (34), a more selective approach to PSA testing is appropriate and should be coupled with informed consent procedures for patients considering screening (35,36).
REFERENCES 1. Guess H. Benign prostatic hyperplasia: antecedents and natural history. Epidemiol Rev. 1992;14:131–153. 2. Parker SL, Tong T, Bolden S, Wingo PA. Cancer Statistics, 1996. CA Cancer J Clin. 1996;46:5–27. 3. Pienta KJ, Esper PS. Risk factors for prostate cancer. Ann Intern Med. 1993;118:793– 803. 4. Woolf SH. Screening for prostate cancer with prostate-specific antigen. NEJM. 1995;333:1401–1405. 5. Mettlin C, Jones G, Averette H, et al. Defining and updating the American Cancer Society guidlines for the cancer-related checkup: prostate and endometrial cancers. CA Cancer J Clin. 1993;43:42– 47. 6. Potosky AL, Miller BA, Albertson PC, Kramer BS. The role of increasing detection in the rising incidence of prostate cancer. JAMA. 1995;273:548 –552. 7. Jacobsen SJ, Katusic SK, Bergstralh EJ, et al. Incidence of prostate cancer diagnosis in the eras before and after serum prostate-specific antigen testing. JAMA. 1995;274:1445–1449. 8. Oesterling JE. Prostate specific antigen. A critical assessment of the most useful tumor marker for adenocarcinoma of the prostate. J Urol. 1991;145:907–923. 9. Monda JM, Barry MJ, Oesterling JE. Prostate specific antigen cannot distinguish stage T1a (A1) prostate cancer from benign prostatic hyperplasia. J Urol. 1994;151:1291–1295. 10. Sershon PD, Barry MJ, Oesterling JE. Serum prostate-specific antigen discriminates weakly between men with benign prostatic hyperplasia and patients with organ-confined prostate cancer. Euro Urol. 1994;25:281–287. 11. Gillatt D, Reynard JM. What is the “normal range” for prostatespecific antigen? Use of a receiver operating characteristic curve to evaluate a serum marker. Br J Urol. 1995;75:341–346. 12. Meigs JB, Barry MJ, Oesterling JE, Jacobsen SJ. Interpreting results of prostate-specific antigen testing for early detection of prostate cancer. J Gen Intern Med. 1996;11:505–512. 13. Mettlin C, Lee F, Drago J, Murphy GP. The American Cancer Society National Prostate Cancer Detection Project. Findings on the detection of early prostate cancer in 2425 men. Cancer. 1991;67: 2949 –2958. 14. Catalona WJ, Richie JP, Ahmann FR, et al. Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. J Urol. 1994;151:1283–1290. 15. Hiatt RA, Armstrong MA, Klatsky AL, Sidney S. Alcohol consumption, smoking, and other risk factors and prostate cancer in a large health plan cohort in California (United States). Cancer Cause Control. 1994;5:66 –72. 16. Giovannucci E, Rimm EB, Chute CG, et al. Obesity and benign prostatic hyperplasia. Am J Epidemiol. 1994;140:989 –1002. 17. Barry MJ Jr, Fowler FJ, O’Leary MP, et al. The American Urological Association Symptom Index for benign prostatic hyperplasia. J Urol. 1992;148:1549 –1557. 18. Giovannucci E, Ascherio A, Rimm EB, et al. A prospective cohort study of vasectomy and prostate cancer in U.S. men. JAMA. 1992; 269:873– 877. 19. Meigs JB, Barry MJ. The natural history of benign prostatic hyperplasia. In: Kirby RS, McConnell JD, Fitzpatrick JM, et al, eds. Text-
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
20.
21.
22. 23. 24. 25. 26. 27. 28.
29.
book of Benign Prostatic Hyperplasia. Oxford: Isis Medical Media; 1996:125–135. Barry MJ, Fowler FJ, O’Leary MP, et al. Measuring disease-specific health status in men with benign prostatic hyperplasia. Med Care. 1995;33(suppl):AS145–AS155. Barry MJ, Willerford WO, Chang Y, et al. Benign prostatic hyperplasia specific health status measurements in clinical research: how much change in the American Urological Association symptom index and the benign prostatic hyperplasia impact index is perceptible to patients? J Urol. 1995;154:1770 –1774. Rothman KJ. Modern Epidemiology. Boston, Mass: Little, Brown, and Co; 1986. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: Wiley-Interscience; 1989. SAS/STAT User’s Guide. Version 6, 4th ed. Cary, NC: SAS Institute, Inc; 1989. Physicians’ Desk Reference. Montvale, NJ: Medical Economics; 1996. U.S. Preventive Services Task Force. Guide to Clinical Preventive Services. Baltimore, Md: William and Wilkins; 1989. American Urological Association Executive Committee Report, 1992. Baltimore, Md: American Urological Association; 1992. Albertsen PC, Fryback DG, Storer BE, et al. Long-term survival among men with conservatively treated localized prostate cancer. JAMA. 1995;274:626 – 631. Lu-Yao GL, McLerran D, Wasson J, Wennberg JE. An assessment of
30.
31.
32.
33.
34.
35.
36.
radical prostatectomy: time trends, geographic variation, and outcomes. JAMA. 1993;269:2633–2636. Fleming C, Wasson JH, Albertsen PC, et al. A decision analysis of alternative treatment strategies for clinically localized prostate cancer. JAMA. 1993;269:2650 –2658. Johansson J-E, Holmberg L, Johansson S, et al. Fifteen-year survival in prostate cancer: a prospective, population-based study in Sweden. JAMA. 1997;277:467– 471. Barry MJ, Fleming C, Coley CM, et al. Should Medicare provide reimbursement for prostate-specific antigen testing for early detection of prostate cancer? Part IV: estimating the risks and benefits of an early detection program. Urology. 1995;46:445– 461. Fowler FJ, Bin L, McNaughton Collins M, et al. Prostate cancer screening and beliefs about treatment efficacy: a national survey of primary care physicians and urologists. Am J Med. 1998;104:526 – 532. Gohogan JK, Prorok PC, Kramer BS, Cornett JE. Prostate cancer screening in the prostate, lung, colorectal, and ovarian cancer screening trial of the National Cancer Institute. J Urol. 1994;152: 1905–1909. Flood AB, Wennberg JE, Nease RF, et al. The importance of patient preference in the decision to screen for prostate cancer. J Gen Intern Med. 1996;11:342–349. Wolf AMD, Nasser JF, Wolf AM, Schorling JB. The impact of informed consent on patient interest in prostate-specific antigen screening. Arch Intern Med. 1996;156:1333–1336.
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 525
High Rates of Prostate-specific Antigen Testing in Men with Evidence of Benign Prostatic Hyperplasia James B. Meigs, MD, MPH, Michael J. Barry, MD, Edward Giovannucci, MD, Eric B. Rimm, ScD, Meir J. Stampfer, MD, Ichiro Kawachi, MD, PhD PURPOSE: Benign prostatic hyperplasia is common among men who may be candidates for prostate cancer screening using prostate-specific antigen (PSA) testing. Patterns of PSA testing among men with evidence of benign prostatic hyperplasia have not been studied. METHODS: We examined the prevalence and correlates of a self-reported history of PSA testing. In 1994, 33,028 US health professionals without prostate cancer aged 47 to 85 years provided information on prior PSA testing, lower urinary tract symptoms characteristic of benign prostatic hyperplasia, history of prostatectomy, and prostate cancer risk factors. In 1995, a subset of 7,070 men provided additional information on diagnosis and treatment of benign prostatic hyperplasia. RESULTS: From 39% of men in their 50s to 53% of men in their 80s reported PSA testing in the prior year (P ,0.0001 for trend with age). Men were more likely to report PSA testing if
they had lower urinary tract symptoms characteristic of benign prostatic hyperplasia (age-adjusted odds ratio for severe symptoms 2.2, 95% confidence interval 1.8 to 2.6), a prior history of prostatectomy (age-adjusted odds ratio 1.1, 95% confidence interval 1.02 to 1.2), or a physician diagnosis of benign prostatic hyperplasia (odds ratio 1.9, 95% confidence interval 1.7 to 2.2; adjusted for age, signs or symptoms of benign prostatic hyperplasia, and prostate cancer risk factors). CONCLUSIONS: These US health professionals reported preferential use of PSA testing among men least likely to benefit from early cancer detection (older men) and among men most likely to have a false-positive PSA result (men with benign prostatic hyperplasia). Physician and patient education are needed to promote more rational and selective use of this screening test. Am J Med. 1998;104:517–525. q1998 by Excerpta Medica, Inc.
P
cancer does more good than harm (4), even advocates of early detection question its value in men older than 70 to 75 years, in whom benign prostatic hyperplasia is especially prevalent (5). Prostate-specific antigen (PSA) testing has become increasingly popular for early detection of prostate cancer (6,7). Unfortunately, among men with benign prostatic hyperplasia, PSA testing appears to provide little useful additional information about the presence of prostate cancer (8 –10), and is likely to result in false-positive results. For example, the false-positive rate for a PSA level greater than 4.0 ng/mL in men not known to have benign prostatic hyperplasia is 8% to 13%, but increases to 32% to 60% in men with this condition (11,12). Furthermore, neither a history of benign prostatic hyperplasia nor the presence of lower urinary tract symptoms characteristic of benign prostatic hyperplasia appear to increase the risk of prostate cancer (3,13–15). Despite uncertain benefits, no data are available on rates of PSA testing in men with benign prostatic hyperplasia. We examined the prevalence and correlates of PSA testing during 1993 and 1994 among men with evidence of benign prostatic hyperplasia in a large cohort of US health professionals.
rostate cancer and benign prostatic hyperplasia are major health problems among men in the United States (1,2). As older age is a risk factor for both of these diseases (1,3), the potential target population for early detection of prostate cancer is enriched with men who have benign prostatic hyperplasia. Although it is unknown whether early detection and treatment of prostate
From the General Medicine Division of the Medical Services (JBM, MJB), Massachusetts General Hospital, and the Department of Medicine, Harvard Medical School; Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women’s Hospital (EG, EBR, MJS, IK); Departments of Epidemiology and Nutrition, Harvard School of Public Health (EG, EBR, MJS), and Departments of Health and Social Behavior (IK), Boston, Massachusetts. Supported by Research Grants DK45779, CA55075, and HL35464 from The National Institutes of Health and by PORT II Grant No. HS 08397 from the Agency for Health Care Policy and Research. Requests for reprints should be addressed to James B. Meigs, MD, MPH, General Internal Medicine Unit S50-9, Massachusetts General Hospital, Boston, Massachusetts 02114. Presented in abstract form at the Society of General Internal Medicine Annual Meeting, Washington, DC, May 1996. Manuscript submitted May 14, 1997 and accepted in revised form March 17, 1998. q1998 by Excerpta Medica, Inc. All rights reserved.
0002-9343/98/$19.00 517 PII S0002-9343(98)00113-2
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
METHODS Study Population The Health Professionals Follow-up Study is a prospective cohort of 51,529 US male dentists, optometrists, osteopaths, podiatrists, pharmacists, and veterinarians born between January 1, 1910, and December 31, 1946, who responded to a mailed questionnaire in 1986. We collected information on age, ancestry, a family history of prostate cancer, smoking behavior, and the presence and year of occurrence of a variety of medical conditions, including prostate cancer, vasectomy, and prostatectomy for benign prostatic hyperplasia. Further details about the cohort study and its participants have been reported (16). We mailed follow-up questionnaires to cohort members biennially to ascertain new cases of disease and to update exposure information. In 1992 and 1994 we added a section to assess prevalent lower urinary tract symptoms, and in 1994, a question regarding prior PSA testing. Lower urinary tract symptoms were measured using the American Urological Association (AUA) symptom index, a validated 7-item questionnaire that assesses the frequency over the preceding month of lower urinary tract symptoms characteristic of benign prostatic hyperplasia (symptoms include nocturia, weak urinary stream, hesitancy, intermittence, incomplete bladder emptying, urinary frequency, and urgency; the score range from 0 to 35, with higher scores reflecting more severe symptoms) (17). In 1995, we mailed a supplemental questionnaire to all men who reported high-moderate to severe symptoms in 1994 (AUA symptom score $15, n 5 1,452) and an age-stratified random sample of men with none to lowmoderate symptoms (AUA symptom score #14, n 5 8,278) to obtain more detailed information about benign prostatic hyperplasia-related symptoms, diagnoses, and treatments. From the 38,706 respondents to the 1994 questionnaire we excluded men for whom we had incomplete information on PSA testing and lower urinary tract symptoms (n 5 4,129) or who reported prostate cancer as of 1994 (n 5 1,549), leaving 33,028 men for analysis of PSA testing rates. The validity of a self-reported diagnosis of prostate cancer in this cohort has been established (18). Of 8,208 men who responded to the 1995 benign prostatic hyperplasia questionnaire (response rate 85% for men with high-moderate to severe symptoms and 84% for men with none to low-moderate symptoms), 7,070 men free of prostate cancer in 1994 and for whom we had complete information on PSA testing and lower urinary tract symptoms were available for more detailed analysis.
Assessment of PSA Testing We ascertained prior PSA testing in the entire cohort as of 1994 with the question: “Have you ever had a PSA blood test for prostate cancer screening?” For affirmative responses, we recorded the answers to: “When was your 518
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104
most recent test: before 1988, ’88 –’89, ’90 –’91, ’92, ’93, or ’94?” If they reported a PSA test, we asked whether the result was elevated. We also asked whether they had ever had a prostate biopsy or a transrectal ultrasound examination of the prostate. On the 1995 benign prostatic hyperplasia questionnaire, the question was phrased, “Since January 1, 1994, have you had screening for prostate cancer by using PSA (prostate specific antigen)?” Men in the entire cohort were classified as having PSA testing in the prior year if the most recent PSA test was performed in 1993 or 1994, and for men in the subgroup, if a PSA test had been performed since January 1, 1994.
Assessment of Benign Prostatic Hyperplasia We used a variety of definitions to classify subjects as having evidence of benign prostatic hyperplasia because there are no standard criteria for this condition for use in epidemiologic studies (19). Using the AUA symptom index, we ascertained the frequency, over the preceding month, of lower urinary tract symptoms characteristic of benign prostatic hyperplasia (17,20,21). Symptom scores allowed classification of men with none to mild (AUA symptom score #7), moderate (score 8 to 19), and severe (score $20) lower urinary tract symptoms. Because there is no generally accepted threshold for the diagnosis of benign prostatic hyperplasia, data were analyzed by AUA symptom score class to assess the “dose-response” relation between increasingly severe lower urinary tract symptoms and PSA testing. The AUA symptom index was administered to the cohort in 1992, 1994, and 1995; subjects were classified based on 1994 responses for the entire cohort and 1995 responses for the more detailed subgroup to correlate symptoms with concurrent PSA testing. We also calculated the change in lower urinary tract symptoms between 1992 and 1994 (the period concurrent with or immediately preceding reported PSA testing). We classified men as having stable to improved lower urinary tract symptoms, as indicated by remaining in the same or moving to a less symptomatic class (ie, from moderate symptoms in 1992 to none to mild symptoms in 1994), or as having worsening lower urinary tract symptoms, as indicated by moving to a more symptomatic class. We also assessed symptom change by classifying men as having worsening symptoms if the AUA symptom score increased by 3 or more points between 1992 and 1994 (21). Results using AUA symptom class or score increase were similar; only those using class increase are reported here. Symptom change analysis was limited to the 26,052 men who provided complete AUA symptom index data in both 1992 and 1994. Alternatively, we considered subjects to have benign prostatic hyperplasia if they reported a history of prostatectomy for this condition. Subjects were classified as having had a prostatectomy during 1993 and 1994, prior to 1993, or never. The validity of a self-reported prosta-
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
tectomy in this cohort has been established (16). In 1988 we requested permission to review the medical records of 99 randomly selected men who reported having a prostatectomy. Of 77 men who granted permission, 74 records were obtained, all of which confirmed the surgery. The 1995 supplemental questionnaire provided additional information on benign prostatic hyperplasia diagnoses and drug therapy. Men in this subgroup were considered to have benign prostatic hyperplasia if they responded affirmatively to either question: “Since January 1, 1990, have you had an enlarged prostate detected by rectal exam?” or “Since January 1, 1990, has a physician told you that you have BPH (benign prostatic hyperplasia)?” Subjects were classified as receiving drug treatment for benign prostatic hyperplasia if they reported current, regular use of finasteride or the alpha-blocker drugs terazosin, prazosin, or doxazosin specifically for prostate problems.
Assessment of Prostate Cancer Risk Factors To provide a contrast to men with benign prostatic hyperplasia, we analyzed PSA testing patterns among men with risk factors for prostate cancer, who might be expected to have higher rates of PSA testing. Risk factors were selected based on previously reported associations with prostate cancer (3,15) and on assumptions about which factors might commonly influence the clinical decision to order a PSA test. We classified men as African American or other (Southern European or Mediterranean, Scandinavian, other Caucasian, Asian or Oriental, or other origin). We classified men as cigarette smokers if they reported current cigarette use in 1994, as having had a vasectomy if they reported this procedure during or prior to 1994, and as having a family history of prostate cancer if they reported a father or brother with this condition. Dietary risk factors for prostate cancer, although available, were not included based on the assumption that information about dietary risk is rarely considered in routine PSA testing decisions.
Data Analysis Two sets of analyses were performed—the first on the entire study cohort as of 1994, and the second on the subgroup for whom more detailed information was available on the diagnosis and treatment of benign prostatic hyperplasia. We estimated the age-stratified prevalence of PSA testing, lower urinary tract symptoms, and prostatectomy for the entire cohort. We then compared the age-stratified proportions of men with or without benign prostatic hyperplasia and with or without prostate cancer risk factors who reported PSA testing during the prior year. Trend in the proportion reporting PSA testing across age decades was assessed with the Mantel-Haenszel test for trend (22). Differences in proportions reporting PSA testing were summarized over age decades with the Mantel-Haenszel estimators (22). Odds ratios and 95%
confidence intervals (CI) for PSA testing adjusted for age were calculated using logistic regression (23). Models assessing the effect of a change in AUA symptom score class between 1992 and 1994 on PSA testing included a covariate for AUA symptom score class in 1992. Although the majority of men (89.1%) reported having a periodic health examination during 1990 through 1994, we assessed the possibility that surveillance bias might affect our results by repeating the analysis among this subset of men. Results obtained in this subset were essentially identical to those obtained among the full set of eligible men; only the latter results are shown. Multivariate odds ratios and 95% CI for PSA testing were calculated using separate models with benign prostatic hyperplasia diagnoses and prostate cancer risk factors each entered separately or in combination. Additional covariates in the multivariate models included age, profession, report of a periodic health examination between 1990 and 1994, and use of finasteride or alphablocker medications. Adequacy of each multivariate model was assessed with the Hosmer and Lemeshow goodness-of-fit test; additional regression diagnostics were performed by plotting changes in the deviance, the chi-square goodness-of-fit statistic, and the confidence interval displacement diagnostic against the predicted probabilities for each observation (23). These procedures indicated adequate fit for the multivariate logistic models overall and for the majority of covariate patterns. The area under the receiver-operating-curves for the multivariate logistic regression models ranged from 0.68 to 0.71. SAS statistical software was used for the analysis (24). Statistical significance was defined as a two-tailed P value ,0.05.
RESULTS The majority of men in the cohort were in the age range when PSA testing for early detection of prostate cancer might be recommended: 87.4% were between the ages of 50 and 79 years (Table 1). PSA testing was common and increased significantly with age. Although there was a slight fall-off among men in their 80s, prevalence rates remained high even among the oldest men: 74.4% of men in their 70s and 66.6% of men in their 80s reported PSA testing at least once during or prior to 1994. The majority of men in every age decade reported their most recent PSA test during the preceding year (1993 to 1994), suggesting either that PSA testing became increasingly common in this cohort between 1990 and 1994 or that many of these men had begun to receive annual PSA testing. Benign prostatic hyperplasia, defined either on the basis of lower urinary tract symptoms or a history of prostatectomy, also was common and increased significantly with age. The majority of men who reported a prostatectomy June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 519
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
Table 1. Age-stratified Prevalence of Prostate-specific Antigen Testing (PSA), Symptoms of Benign Prostatic Hyperplasia, and Prostatectomy for Benign Prostatic Hyperplasia among 33,028 Health Professionals* Age (Years)
Number (%) Most recent PSA test Never Prior to 1990 1990–1992 1993–1994 AUA Symptom Index class (1994) None–mild (score #7) Moderate (score 8–19) Severe (score $20) History of prostatectomy for benign prostatic hyperplasia Never Prior to 1993 1993–1994
47–49
50–59
60–69
70–79
80–85
3216 (9.7)
11633 (35.2)
10359 (31.4)
6911 (20.9)
909 (2.8)
2374 (73.8) 13 (0.4) 208 (6.5) 621 (19.3)
5808 (49.9) 53 (0.5) 1205 (10.4) 4567 (39.3)
3193 (30.8) 63 (0.6) 1373 (13.3) 5730 (55.3)
1766 (25.6) 55 (0.8) 842 (12.2) 4248 (61.5)
304 (33.4)† 19 (2.1) 108 (11.9) 478 (52.6)
2959 (92.0) 244 (7.6) 13 (0.4)
10022 (86.2) 1496 (12.9) 115 (1.0)
7897 (76.2) 2265 (21.9) 197 (1.9)
4696 (67.9) 2003 (29.0) 212 (3.1)
602 (66.2)† 271 (29.8) 36 (4.0)
3205 (99.7) 9 (0.3) 2 (0.1)
11499 (98.8) 105 (0.9) 29 (0.2)
9729 (93.9) 557 (5.4) 73 (0.7)
5563 (80.5) 1257 (18.2) 91 (1.3)
568 (62.5)† 328 (36.1) 13 (1.4)
* Data are n (%). † P ,0.0001 for trend across age decades. PSA 5 prostate-specific antigen; AUA 5 American Urological Association.
had their surgery prior to the time interval during which PSA testing was assessed. Of men who reported ever having a PSA test, 1.6% of men in their 40s, 4.0% of men in their 50s, 9.1% of men in their 60s, 13.8% of men in their 70s, and 14.8% of men in their 80s reported that the PSA level was elevated (P ,0.001 for trend). In addition, 1.7% of men in their 40s, 4.6% of men in their 50s, 14.3% of men in their 60s, 28.2% of men in their 70s, and 33.8% of men in their 80s reported having ever had a prostate biopsy or rectal ultrasound for prostate examination (P ,0.001 for trend).
Prevalence of PSA Testing in Men with Benign Prostatic Hyperplasia In every age decade, PSA testing was significantly more common among men with benign prostatic hyperplasia than among men without benign prostatic hyperplasia (Table 2). Increasing severity of lower urinary tract symptoms, a worsening of symptoms in the 1 or 2 years preceding PSA testing, and a history of prostatectomy prior to the year of PSA testing were each associated with a markedly increased prevalence of PSA testing. Only among 80-year-old men did PSA testing rates decline slightly. Nonetheless, over half of 80-year-old men with benign prostatic hyperplasia reported PSA testing during the preceding year. Older men with moderate or severe lower urinary tract symptoms had especially high PSA testing rates, including 63.5% of men in their 60s with moderate symptoms and 72.2% of men in their 80s with severe symptoms. Among men with moderate to severe lower urinary tract symptoms (AUA symptom score 8 to 35) who re520
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104
ported ever having a PSA test, 2.8% of men in their 40s, 8.6% of men in their 50s, 15.2% of men in their 60s, 18.6% of men in their 70s, and 17.6% of men in their 80s reported that the PSA level was elevated (P ,0.001 for trend). Among men reporting an elevated PSA level, 30.8% of men in their 40s, 55.5% of men in their 50s, 66.6% of men in their 60s, 73.1% of men in their 70s, and 79.1% of men in their 80s reported having ever had a prostate biopsy or rectal ultrasound for prostate examination (P ,0.001 for trend). Men with risk factors for prostate cancer, on the other hand, did not uniformly report higher PSA testing rates than men without risk factors (Table 2). PSA testing was more common among men with a family history of prostate cancer or a history of vasectomy, whereas AfricanAmerican men were slightly, but not significantly, less likely to receive such testing. The limited number of African-American men in this cohort (207, 0.6%) limits our power to be certain about the effect of race on PSA testing. Current cigarette smokers were actually less likely than nonsmokers to have had PSA testing.
Odds of PSA Testing After Adjustment for Age and Potential Confounding Age-adjusted odds ratios for PSA testing, according to the presence of benign prostatic hyperplasia or risk factors for prostate cancer, are presented in Table 3 for both the entire cohort and for the more detailed subgroup. Even after age adjustment, an increasing burden of lower urinary tract symptoms, worsening symptoms, or a history of prostatectomy all significantly increased the odds of PSA testing. In the entire cohort men with moderate
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
Table 2. Age-stratified Prevalence of Prostate-specific Antigen (PSA) Testing during 1993–1994 among 33,028 Health Professionals with or without Benign Prostatic Hyperplasia or Risk Factors for Prostate Cancer* Percent Reporting PSA Testing by Age (Years)
AUA Symptom Index class (1994) None–mild (score #7) Moderate (score 8–19) Severe (score $20) Change in AUA Symptom Index class 1992–1994 Same or improved Worsened History of prostatectomy for benign prostatic hyperplasia Never Prior to 1993 1993–1994 Family history of prostate cancer No family history of prostate cancer Vasectomy No history of vasectomy African American Non-African American Current smoker Nonsmoker
47–49
50–59
60–69
70–79
80–85
18.3 30.3 38.5
37.1 51.8 63.5
52.6 63.5 69.5
58.3 67.7 73.1
50.5†‡ 54.6 72.2
18.7 32.8
38.2 49.2
54.7 60.3
61.2 65.4
52.9†‡ 51.4
19.2 33.3 100.0 29.3 18.8 23.0 17.7 27.3 19.1 17.9 19.4
38.9 66.7 86.2 46.6 38.8 42.4 37.8 31.8 39.4 29.8 39.9
54.3 68.2 87.7 60.3 55.0 59.2 53.9 51.7 55.4 43.8 56.1
60.5 64.1 85.7 69.0 61.0 63.3 61.1 56.8 61.6 46.5 62.2
52.8†‡ 50.6 92.3 59.7†‡ 52.1 50.0†‡ 52.9 0.0† 52.7 34.6†‡ 53.1
* Data are % reporting PSA testing in each category. † P ,0.0001 for trend across age decades. ‡ P ,0.0001 for Mantel-Haenszel summary estimate for differences within age groups. PSA 5 prostate-specific antigen; AUA 5 American Urological Association.
symptoms had 1.6-fold greater odds of having had a PSA test, whereas men with severe symptoms had 2.2-fold greater odds. Among men in the subgroup, those who had been told they had benign prostatic hyperplasia, or who reported an enlarged prostate detected by rectal examination, or who were under benign prostatic hyperplasia-specific drug treatment all were more likely to report PSA testing during the preceding year. Compared with men in the entire cohort with no or mild lower urinary tract symptoms, those with moderate symptoms were more likely to report that the PSA level was elevated (age adjusted, odds ratio 5 2.1, 95% CI 1.9 to 2.4); those with severe lower urinary tract symptoms had 2.7-fold greater odds (95% CI 2.1 to 3.4). Men with an elevated PSA level had 9.7-fold greater odds to report having ever had a prostate biopsy or rectal ultrasound for prostate examination (age and symptom adjusted; 95% CI 8.6 to 10.9). A family history of prostate cancer or a history of vasectomy also significantly increased the odds of PSA testing after adjustment for age in both groups, although in the smaller subgroup, the effect of family history was attenuated. African-American men were no more likely to have received PSA testing. Current cigarette smokers were significantly less likely than nonsmokers to have received PSA testing during the previous year. In addition to age, the subjects’ professions, having
seen a physician for a physical examination (thereby providing the opportunity for PSA testing), being on treatment with benign prostatic hyperplasia-specific drugs (a possible indication for PSA testing) (25), or the presence of prostate cancer risk factors, could have confounded the association between benign prostatic hyperplasia and PSA testing and PSA testing. Table 4 displays a series of multiple logistic regression models designed to estimate the effect of benign prostatic hyperplasia on PSA testing while controlling for these covariates. Entering all four prostate cancer risk factors simultaneously into a logistic regression model did not qualitatively change their bivariate relationships to PSA testing. A positive family history or history of vasectomy were still associated, although not significantly, with PSA testing. Next, simultaneous adjustment for risk factors, possible confounders, and several benign prostatic hyperplasia definitions demonstrated that an increasing burden of lower urinary tract symptoms, an enlarged prostate detected by rectal examination, and a history of prostatectomy all independently increased the odds that a man received PSA testing. Finally, addition of a physician diagnosis of benign prostatic hyperplasia to a logistic regression model containing prostate cancer risk factors and signs and symptoms consistent with benign prostatic hyperplasia weakened the association of lower urinary tract symptoms or prior prostatectomy with PSA testing, probably because these June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 521
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
Table 3. Age-adjusted Odds Ratios for Prostate-specific Antigen (PSA) Testing during the Previous Year in Male Health Professionals with Benign Prostatic Hyperplasia (BPH) or Risk Factors for Prostate Cancer* 33,028 Men in Entire Cohort PSA/No PSA N with PSA testing/N without PSA testing AUA Symptom Index class† None-mild (score #7) Moderate (score 8–19) Severe (score $20) Worsened AUA Symptom Index class 1992–1994‡ History of prostatectomy for benign prostatic hyperplasia Never Prior to 1993 1993–1994 Family history of prostate cancer Vasectomy African American Current Smoker Physician-diagnosed benign prostatic hyperplasia§ Enlarged prostate detected by rectal examination Finasteride or alpha-blocker therapy
OR
95% CI
15,644/17,384
7,070 Men in Subgroup PSA/No PSA
OR
95% CI
3,654/3,416
11,457/14,719 3,791/2,488 396/177 1,329/1,008
1.0 1.6 2.2 1.4
Referent 1.5, 1.7 1.8, 2.6 1.3, 1.5
1,917/2,296 1,437/974 300/146 542/407
1.0 1.4 1.9 1.2
Referent 1.3, 1.6 1.5, 2.4 1.02, 1.4
14,038/16,526 1,425/831 181/27 1,130/937 4,128/4,413 94/113 701/1,244
1.0 1.1 5.6 1.4 1.3 0.9 0.6
Referent 1.02, 1.2 3.7, 8.4 1.3, 1.5 1.2, 1.3 0.7, 1.2 0.6, 0.7
3,287/3,227 316/175 50/4 263/213 987/912 13/13 156/233 1,270/459 679/322 462/99
1.0 1.0 9.2 1.2 1.2 0.9 0.7 2.7 2.0 3.9
Referent 0.8, 1.2 3.3, 26 0.97, 1.4 1.1, 1.3 0.4, 1.9 0.5, 0.8 2.4, 3.1 1.7, 2.3 3.1, 4.9
* Data are odds ratios (OR) and 95% confidence intervals (95% CI) for PSA testing during 1993–1994 (n 5 33,028) or 1994 –1995 (n 5 7,070) age-adjusted using age as a continuous variable. † American Urological Association (AUA) Symptom Index class in 1994 (n 5 33,028) or 1995 (n 5 7,070). ‡ Change in AUA Symptom Index class adjusted for age and baseline class; referent group was men with stable or improved class. § Data on physician-diagnosed benign prostatic hyperplasia, enlarged prostate detected by rectal examination, and drug therapy were not available for the entire cohort. PSA 5 prostate-specific antigen.
conditions lead directly to the diagnosis of benign prostatic hyperplasia by the physician. Nonetheless, a physician diagnosis of benign prostatic hyperplasia and the presence of an enlarged prostate detected by rectal examination both independently increased the odds that a man reported PSA testing even after adjustment for lower urinary tract symptoms, a history of prostatectomy, prostate cancer risk factors, age, and other potential confounders. Men whose physicians told them they had benign prostatic hyperplasia had nearly twice the odds of PSA testing for early detection of prostate cancer as men without such a diagnosis.
DISCUSSION In this study we found that men with evidence of benign prostatic hyperplasia were significantly more likely than men without such evidence to receive PSA testing for early detection of prostate cancer. We also confirmed previous reports of high PSA testing rates among older men in the US (6,7). In this large, well-characterized cohort, PSA testing for prostate cancer was extremely common, especially among older men, in whom the benefits
522
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104
of early cancer detection are the most questionable. Three fourths of men in their 70s and two thirds of men in their 80s reported having a PSA test at least once as of 1994. The majority of tests were performed during 1993 and 1994. Given the controversy surrounding the value of early detection of prostate cancer (4), these PSA testing rates seem remarkably high. Although men in the Health Professionals Follow-up Study are generally health conscious, and as such may not be completely representative of all US men, the prevalence of PSA testing in this cohort is consistent with testing rates projected from data from a community study (7) or from four National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) sites (6). Among men in Olmsted County, for example, PSA testing rates rose steadily beginning in 1987, so that by 1992, 6% of men in their 40s, 23% of men in their 50s, 43% of men in their 60s, and 64% of men in their 70s had at least one PSA test during their lifetime (7). These rates suggest that physicians were not following 1989 guidelines published by the US Preventive Services Task Force, who recommended against any routine screening for prostate cancer (26). They are more consistent with 1992–1993 recommendations by the American
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
Table 4. Multivariate Odds Ratios for Prostate-specific Antigen (PSA) Testing during the Previous Year in 7,070 Male Health Professionals with Benign Prostatic Hyperplasia or Risk Factors for Prostate Cancer*
†
Family history of prostate cancer Vasectomy African American Current smoker AUA Symptom Index class (1995) None–mild (score #7) Moderate (score 8–19) Severe (score $20) Enlarged prostate detected by rectal examination History of prostatectomy for benign prostatic hyperplasia Never Prior to 1993 1993–1994 Physician-diagnosed benign prostatic hyperplasia
Prostate Cancer Risk Factors, Signs or Symptoms of Benign Prostatic Hyperplasia, and a Physician Diagnosis of Benign Prostatic Hyperplasia
Prostate Cancer Risk Factors
Prostate Cancer Risk Factors and Signs or Symptoms of Benign Prostatic Hyperplasia
OR
95% CI
OR
95% CI
OR
95% CI
1.2 1.1 0.7 0.7
0.96, 1.5 0.99, 1.2 0.3, 1.6 0.5, 0.9
1.2 1.1 0.7 0.7
0.97, 1.5 0.99, 1.2 0.3, 1.7 0.6, 0.9
1.2 1.1 0.8 0.7
0.96, 1.5 0.98, 1.2 0.3, 1.8 0.6, 0.9
1.0 1.2 1.5 1.7
Referent 1.1, 1.4 1.2, 1.8 1.5, 2.0
1.0 1.1 1.2 1.4
Referent 0.97, 1.2 0.96, 1.5 1.2, 1.7
1.0 1.3 8.4
Referent 1.01, 1.6 3.0, 24
1.0 1.04 5.7 1.9
Referent 0.8, 1.3 2.0, 16 1.7, 2.2
* Data are odds ratios (OR) and 95% confidence intervals (95% CI) for PSA testing during 1994 –1995 adjusted for age, profession, having a physical examination 1990 –1994, and use of finasteride or alpha-blocker medication. † The referent groups for prostate cancer risk factor odds ratios are those men without the specified characteristic. PSA 5 prostate-specific antigen; AUA 5 American Urological Association.
Cancer Society and the American Urological Association, which recommended annual PSA testing in combination with a digital rectal examination for early detection of prostate cancer in men older than 50 years (5,27). However, the American Cancer Society acknowledged that early detection may only benefit men with at least a 10year life expectancy (5), equivalent to about age 75 in men with average comorbidities (28). The high rate of testing among men in their 70s and 80s in this cohort is of particular concern, as men in this age group may be the least likely to benefit from early aggressive treatment of a prostate cancer (29 –31). PSA testing rates were significantly higher in men of all ages with evidence of benign prostatic hyperplasia, whether defined by symptoms, a history of prostatectomy, an enlarged prostate on rectal examination, or by a physician diagnosis of benign prostatic hyperplasia. Even after accounting for age, prostate cancer risk factors, and signs, symptoms, and treatments for benign prostatic hyperplasia, men who had been told by their physicians that they had benign prostatic hyperplasia were more likely to report recent PSA testing. Not unexpectedly, we found
that men with benign prostatic hyperplasia who had a PSA test were more likely than men without benign prostatic hyperplasia to report that the PSA level was elevated, and that men reporting elevated PSA levels were very likely to report a prostate biopsy or transrectal ultrasound. Well over half of men older than 50 years of age, and four of five men older than 80 years of age, who had an elevated PSA level reported having such additional evaluations. SEER data also suggest that elevated PSA levels lead to increased rates of prostate biopsy and transrectal ultrasound examination (6). Given the poor specificity of PSA testing in patients with benign prostatic hyperplasia, our data suggest that in this cohort PSA testing has resulted in a high likelihood of false-positive results. Such inefficiency may lead to unnecessary additional evaluation with attendant risk, cost, and patient anxiety. We have previously estimated that the evaluation of a suspicious PSA or digital rectal examination in a man older than 65 years of age costs at least $444 (32), equivalent to a per-member per-month payment under many capitated insurance plans for Medicare beneficiaries. Evidence of inefficient, and possibly ineffective, use of med-
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 523
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
ical resources should be of concern to patients, physicians, and health policy-makers. A limitation of this analysis is uncertainty about the real reason why PSA tests were ordered. Our ascertainment of PSA testing relied on the self-reported response to a question about “prostate cancer screening,” but there may have been many reasons why men received PSA testing. Screening refers to testing men without any symptoms that raise the probability of prostate cancer, whereas early detection refers to testing men with symptoms that could suggest the presence of an early, occult malignancy. Although the presence of lower urinary tract symptoms does not appear to raise the probability of prostate cancer (13–15), PSA tests may still have been ordered as part of a work-up strategy in men with potential signs or symptoms of prostate cancer. Increased use of PSA testing associated with increasingly severe lower urinary tract symptoms is consistent with the concern that these symptoms represented prostate cancer. Other reasons for PSA testing may have included initiation and management of finasteride therapy in men with benign prostatic hyperplasia, as an automatic component of a routine chemistry panel, because the patient requested that a PSA test be performed, or because of concern on the part of the ordering physician about malpractice litigation. An additional limitation is lack of information about whether primary care physicians, urologists, or medical specialists were responsible for ordering PSA tests. Evidence from a national survey demonstrated that primary care physicians were much more likely to perform routine PSA testing for early cancer detection in older men than were urologists to recommend such testing. For example, 53% of primary care physicians reported routine PSA testing in men older than 79 years of age, whereas only 25% of urologists would have recommended testing for men this age (33). Primary care physicians were likely to be responsible for the majority of PSA testing in this cohort, and represent a group that might benefit from educational efforts to promote a more rational approach to prostate cancer screening. In summary, we have demonstrated high rates of PSA testing during 1993 and 1994, especially among older men and men with benign prostatic hyperplasia, in a large well-characterized cohort of US health professionals. These data indicate a need for physician education about the apparent lack of correlation among lower urinary tract symptoms and risk for prostate cancer, the high false-positive rate of PSA in the setting of such symptoms, and the uncertainty surrounding the balance of harms and benefits of early prostate cancer detection and treatment, especially among older men. Further research is needed on the relationship among lower urinary tract symptoms, benign prostatic hyperplasia, and prostate cancer in order to define the appropriate target population, if any, for early cancer detection. Until randomized 524
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104
trials indicate that prostate cancer screening does more good than harm (34), a more selective approach to PSA testing is appropriate and should be coupled with informed consent procedures for patients considering screening (35,36).
REFERENCES 1. Guess H. Benign prostatic hyperplasia: antecedents and natural history. Epidemiol Rev. 1992;14:131–153. 2. Parker SL, Tong T, Bolden S, Wingo PA. Cancer Statistics, 1996. CA Cancer J Clin. 1996;46:5–27. 3. Pienta KJ, Esper PS. Risk factors for prostate cancer. Ann Intern Med. 1993;118:793– 803. 4. Woolf SH. Screening for prostate cancer with prostate-specific antigen. NEJM. 1995;333:1401–1405. 5. Mettlin C, Jones G, Averette H, et al. Defining and updating the American Cancer Society guidlines for the cancer-related checkup: prostate and endometrial cancers. CA Cancer J Clin. 1993;43:42– 47. 6. Potosky AL, Miller BA, Albertson PC, Kramer BS. The role of increasing detection in the rising incidence of prostate cancer. JAMA. 1995;273:548 –552. 7. Jacobsen SJ, Katusic SK, Bergstralh EJ, et al. Incidence of prostate cancer diagnosis in the eras before and after serum prostate-specific antigen testing. JAMA. 1995;274:1445–1449. 8. Oesterling JE. Prostate specific antigen. A critical assessment of the most useful tumor marker for adenocarcinoma of the prostate. J Urol. 1991;145:907–923. 9. Monda JM, Barry MJ, Oesterling JE. Prostate specific antigen cannot distinguish stage T1a (A1) prostate cancer from benign prostatic hyperplasia. J Urol. 1994;151:1291–1295. 10. Sershon PD, Barry MJ, Oesterling JE. Serum prostate-specific antigen discriminates weakly between men with benign prostatic hyperplasia and patients with organ-confined prostate cancer. Euro Urol. 1994;25:281–287. 11. Gillatt D, Reynard JM. What is the “normal range” for prostatespecific antigen? Use of a receiver operating characteristic curve to evaluate a serum marker. Br J Urol. 1995;75:341–346. 12. Meigs JB, Barry MJ, Oesterling JE, Jacobsen SJ. Interpreting results of prostate-specific antigen testing for early detection of prostate cancer. J Gen Intern Med. 1996;11:505–512. 13. Mettlin C, Lee F, Drago J, Murphy GP. The American Cancer Society National Prostate Cancer Detection Project. Findings on the detection of early prostate cancer in 2425 men. Cancer. 1991;67: 2949 –2958. 14. Catalona WJ, Richie JP, Ahmann FR, et al. Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. J Urol. 1994;151:1283–1290. 15. Hiatt RA, Armstrong MA, Klatsky AL, Sidney S. Alcohol consumption, smoking, and other risk factors and prostate cancer in a large health plan cohort in California (United States). Cancer Cause Control. 1994;5:66 –72. 16. Giovannucci E, Rimm EB, Chute CG, et al. Obesity and benign prostatic hyperplasia. Am J Epidemiol. 1994;140:989 –1002. 17. Barry MJ Jr, Fowler FJ, O’Leary MP, et al. The American Urological Association Symptom Index for benign prostatic hyperplasia. J Urol. 1992;148:1549 –1557. 18. Giovannucci E, Ascherio A, Rimm EB, et al. A prospective cohort study of vasectomy and prostate cancer in U.S. men. JAMA. 1992; 269:873– 877. 19. Meigs JB, Barry MJ. The natural history of benign prostatic hyperplasia. In: Kirby RS, McConnell JD, Fitzpatrick JM, et al, eds. Text-
PSA Testing in Benign Prostatic Hyperplasia/Meigs et al
20.
21.
22. 23. 24. 25. 26. 27. 28.
29.
book of Benign Prostatic Hyperplasia. Oxford: Isis Medical Media; 1996:125–135. Barry MJ, Fowler FJ, O’Leary MP, et al. Measuring disease-specific health status in men with benign prostatic hyperplasia. Med Care. 1995;33(suppl):AS145–AS155. Barry MJ, Willerford WO, Chang Y, et al. Benign prostatic hyperplasia specific health status measurements in clinical research: how much change in the American Urological Association symptom index and the benign prostatic hyperplasia impact index is perceptible to patients? J Urol. 1995;154:1770 –1774. Rothman KJ. Modern Epidemiology. Boston, Mass: Little, Brown, and Co; 1986. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: Wiley-Interscience; 1989. SAS/STAT User’s Guide. Version 6, 4th ed. Cary, NC: SAS Institute, Inc; 1989. Physicians’ Desk Reference. Montvale, NJ: Medical Economics; 1996. U.S. Preventive Services Task Force. Guide to Clinical Preventive Services. Baltimore, Md: William and Wilkins; 1989. American Urological Association Executive Committee Report, 1992. Baltimore, Md: American Urological Association; 1992. Albertsen PC, Fryback DG, Storer BE, et al. Long-term survival among men with conservatively treated localized prostate cancer. JAMA. 1995;274:626 – 631. Lu-Yao GL, McLerran D, Wasson J, Wennberg JE. An assessment of
30.
31.
32.
33.
34.
35.
36.
radical prostatectomy: time trends, geographic variation, and outcomes. JAMA. 1993;269:2633–2636. Fleming C, Wasson JH, Albertsen PC, et al. A decision analysis of alternative treatment strategies for clinically localized prostate cancer. JAMA. 1993;269:2650 –2658. Johansson J-E, Holmberg L, Johansson S, et al. Fifteen-year survival in prostate cancer: a prospective, population-based study in Sweden. JAMA. 1997;277:467– 471. Barry MJ, Fleming C, Coley CM, et al. Should Medicare provide reimbursement for prostate-specific antigen testing for early detection of prostate cancer? Part IV: estimating the risks and benefits of an early detection program. Urology. 1995;46:445– 461. Fowler FJ, Bin L, McNaughton Collins M, et al. Prostate cancer screening and beliefs about treatment efficacy: a national survey of primary care physicians and urologists. Am J Med. 1998;104:526 – 532. Gohogan JK, Prorok PC, Kramer BS, Cornett JE. Prostate cancer screening in the prostate, lung, colorectal, and ovarian cancer screening trial of the National Cancer Institute. J Urol. 1994;152: 1905–1909. Flood AB, Wennberg JE, Nease RF, et al. The importance of patient preference in the decision to screen for prostate cancer. J Gen Intern Med. 1996;11:342–349. Wolf AMD, Nasser JF, Wolf AM, Schorling JB. The impact of informed consent on patient interest in prostate-specific antigen screening. Arch Intern Med. 1996;156:1333–1336.
June 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 525