Abstract
This study determined the accuracy of self-reports of prostatespecific antigen (PSA) testing. Men (N = 402) attending 2 outpatient clinics were asked: “Did you have a PSA test today?” and their medical records were checked. Concordance, sensitivity, and false-negative values were 65%, 67%, and 33%, respectively, at 1 clinic site and 88%, 64%, and 36% at the other. The accuracy of self-reports of PSA testing should be interpreted with caution.
Screening for prostate cancer with prostate-specific antigen (PSA) is controversial because it is not clear whether regular testing reduces mortality.1–5 Therefore, professional organizations recommend informed decisionmaking for PSA testing.6–10 To promote educational efforts for informed decisionmaking and to determine screening prevalence, which often relies on selfreported data, investigators need data on the accuracy of self-reports of PSA testing. We hypothesized that the accuracy of self-reports of PSA testing would be high if patients were asked about testing on the same day as their clinic visit. We determined whether any demographic, knowledge, or experience variables predicted accurate self-reports.
Study participants were enrolled in another study11 and attended the general internal medicine outpatient clinics at Kelsey-Seybold Clinic in Houston, Tex, and the University of Texas–Houston. Men attending KelseySeybold Clinic were scheduled for an annual health maintenance examination. Men attending the University of Texas–Houston were scheduled for nonurgent care visits.
From April to July 2001, we approached 677 men aged 50 years or older on-site after they had visited with their physician. To be eligible, men had to have no history of prostate cancer and at least a sixth-grade education. Ninety men were ineligible, and 157 refused to participate, resulting in a sample of 430 men. Each participant was paid $10 to complete a self-administered survey. A medical record review between January and March 2002 by an internal medicine physician was used as the gold standard to determine whether men had received a PSA test with their visit. After participants with unavailable records were excluded, 265 men from Kelsey-Seybold Clinic and 137 men from the University of Texas–Houston remained.
We used sections of a survey from another study.11 The main dependent variable was concordance12 of patient self-reports of PSA testing to the following question: “Did you have a PSA test today?” Independent variables were demographic, knowledge, and experience variables.
We excluded from our analysis 46 men at Kelsey-Seybold Clinic (17%) and 7 men from the University of Texas–Houston (5%) who responded “don’t know” to “Did you have a PSA test today?”
We conducted χ2 tests to describe the demographic characteristics of the respondents and to assess the accuracy of PSA self-reports for independent variables. Variables significant at P ≤ .25 were entered into a stepwise logistic regression model to identify the predictors of accurate self-reports at each site.11,12
Respondents at the Kelsey-Seybold Clinic had a higher annual household income than did respondents at the University of Texas–Houston but otherwise had similar demographic characteristics (Table 1).
In response to the question “Did you have a PSA test today?” at Kelsey-Seybold Clinic, 145 men responded “yes” and 74 men responded “no.” At the University of Texas–Houston, 25 men responded “yes” and 105 men responded “no.” The concordance was 65% at Kelsey-Seybold Clinic and 88% at the University of Texas–Houston. The sensitivity and false-negative values were 67% and 33%, respectively, at Kelsey-Seybold Clinic and 64% and 36%, respectively, at the University of Texas–Houston. The specificity and false-positive values were 43% and 57%, respectively, at KelseySeybold Clinic and 98% and 2%, respectively, at the University of Texas–Houston.
At Kelsey-Seybold Clinic, 5 variables were significantly associated with accurate self-reports of PSA testing: (1) more education, (2) “ever heard of a PSA,” (3) “ever been told to have a PSA,” (4) “ever had a PSA,” and (5) “doctor recommended a PSA today” (Table 2).
The prevalence of PSA testing was 90% at Kelsey-Seybold Clinic and 28% at the University of Texas–Houston by medical record review. Among the 53 men who responded “don’t know” to the question “Did you have a PSA test today?”, the prevalence of PSA testing was 91% at Kelsey-Seybold Clinic and 28% at the University of Texas–Houston. Among the 220 men who responded “yes” to the question “Did your doctor recommend a PSA test today?”, the prevalence of PSA testing was 93% at Kelsey-Seybold Clinic and 64% at the University of Texas–Houston. Among the 146 men who responded “yes” to the question “Did your doctor discuss the advantages and disadvantages of the PSA test with you?”, the prevalence of PSA testing was 88% at Kelsey-Seybold Clinic and 38% at the University of Texas–Houston.
Our findings and those of others13,14 suggest that the sensitivity of asking a man whether he had completed a PSA test is relatively low, regardless of the time frame for self-reporting a test. We found a false-negative response rate of one third, which is of concern because of the potential harm that may arise when men must cope with the consequences of a test that they did not even realize they had taken. Because professional organizations recommend informed decisionmaking before PSA testing, future studies should focus on improving physician–patient communication. We recommend caution when interpreting the accuracy of self-reports of PSA testing.
aThere were 402 study participants: 265 at Kelsey-Seybold Clinic and 137 at the University of Texas–Houston. *P < .001 by χ2 test across sites. ** P < .001 by χ2 test across sites. This is the only variable that was based on the medical record review. aRespondents could respond “yes,” “no,” or “don’t know” to the experience variables shown. “Don’t know” responses were omitted. *P ≤ .001 by χ2 test between “yes” and “no” responses for a variable at that site only.
Kelsey-Seybold Clinica The University of Texas–Houstona n % n % Age, y 50–59 150 57 65 47 60–69 80 30 50 37 ≤ 70 35 13 22 16 Race/Ethnicity White 145 55 88 64 Hispanic 24 9 9 7 African American 81 31 37 27 Other 13 5 2 1 No response 2 1 1 1 Marital status Married 214 81 105 77 Not married 49 18 31 23 No response 2 1 1 1 Education < High school 18 7 15 11 Completed high school/general equivalency diploma 27 10 19 14 Some college or trade school 60 23 35 26 Bachelor’s degree 75 28 28 20 Master’s degree or beyond 84 32 39 28 No response 1 0.4 1 1 Annual household income,* $ < 15 000 5 2 13 9 15 000–99 999 180 68 73 53 ≥ 100 000 65 25 45 33 No response 15 6 6 4 Family history of prostate cancer Yes 39 15 17 12 No 217 82 117 85 No response 9 3 3 2 PSA test with clinic visit as documented in medical record** Yes 238 90 38 28 No 27 10 99 72 Kelsey-Seybold Clinic University of Texas–Houston Experience Variables (Response Categories = Yes or No)a Self-Report of PSA Matched Chart, % Total Group N Self-Report of PSA Matched Chart, % Total Group N Have you ever heard of a PSA test? Yes 71* 182 88 104 No 32 28 100 1 Have you ever been told by a doctor that you should have a PSA blood test? Yes 75* 174 87 94 No 27 33 93 29 Have you ever had a PSA test? Yes 77* 163 88 86 No 28 40 88 34 Did your doctor recommend a PSA test to you today? Yes 76* 162 89 87 No 33 48 85 34 Did your doctor discuss the advantages and disadvantages of the PSA test with you? Yes 72 88 82 44 No 62 117 93 83
This study was funded by National Cancer Institute grant K08-CA78615, awarded to Dr Chan as a clinical scientist award, and by Centers for Disease Control and Prevention/Association of Schools of Public Health grant S1171-19/20. Additional technical support was provided by National Institutes of Health grant M01-RR02558 to the Clinical Research Center at The University of Texas–Houston.
This brief was presented at the annual meeting of the American Public Health Association, San Francisco, Calif, November 17, 2003.
We appreciate the assistance of Dr Donnie Aga of Kelsey-Seybold Clinic, Houston, Tex.
Human Participant Protection The study was approved by the institutional review board of The University of Texas–Houston Health Sciences Center.
