Prostate carcinoma knowledge, attitudes, and screening behavior among African-American men in Central Harlem, New York City †
Article first published online: 4 JAN 2001
Copyright © 2001 American Cancer Society
Volume 91, Issue 1, pages 164–172, 1 January 2001
How to Cite
Ashford, A. R., Albert, S. M., Hoke, G., Cushman, L. F., Miller, D. S. and Bassett, M. (2001), Prostate carcinoma knowledge, attitudes, and screening behavior among African-American men in Central Harlem, New York City . Cancer, 91: 164–172. doi: 10.1002/1097-0142(20010101)91:1<164::AID-CNCR21>3.0.CO;2-A
This article is a US Government work and, as such, is in the public domain in the United States of America.
- Issue published online: 4 JAN 2001
- Article first published online: 4 JAN 2001
- Manuscript Accepted: 25 SEP 2000
- Manuscript Revised: 26 JUN 2000
- Manuscript Received: 5 APR 1999
- State of New York
- Department of Health
- Centers for Disease Control and Prevention. Grant Number: 210704
- prostate carcinoma;
- African Americans;
- prostate specific antigen (PSA);
- health behaviors
Although the benefits of prostate carcinoma screening in reducing mortality rates have not been proven or shown to be cost-effective, screening, particularly using prostate specific antigen (PSA) tests, is widespread. A better understanding of screening behavior, knowledge of prostate carcinoma risk, and attitudes toward screening among men at high risk, such as African-American men, would be valuable.
A prevalence survey was conducted using 2 samples of African-American men, aged 50–74 years: a clinic sample drawn from all clinics in Central Harlem (n = 404) and a random-digit dial sample from the same geographic region (n = 319). The prevalence of self-reported PSA screening was estimated using a cognitive survey methodology based on the internal consistency of answers to four different questions. Prevalence estimates were adjusted to take into account the high proportion of nontelephone residences.
The clinic sample, representing a poorer, more ill population (as determined by MOS Physical Function Scores, was less likely to report PSA screening than the community sample (11.1% in clinic sample vs. 25.5% in community). The prevalence of PSA testing in Central Harlem overall in this age group by using two different techniques was estimated to be 24%. In multiple logistic models, self-reported PSA screening was associated with age, education, favorable attitudes toward screening, and knowing someone who had prostate carcinoma. However, the association between these factors and the likelihood of self-reported PSA screening differed between clinic and community samples.
The prevalence of self-reported PSA screening in Central Harlem was lower than that reported for other populations. These findings may be useful in the design of health education campaigns and for counseling innercity, low-income African-American patients appropriately about the disease. Cancer 2001;91:164–72. Published 2001 American Cancer Society.
Prostate carcinoma is the second leading cause of cancer deaths among American men.1 The American Cancer Society estimated that 184,500 new cases would be diagnosed in 1998.2 Prostate carcinoma morbidity and mortality rates are significantly elevated among African-American men. In 1994, the incidence rate for whites was 185.3 per 100,000, whereas for African-American men it was 234.4; also, the lifetime risk of prostate carcinoma mortality is twice as high among African-American men as compared with whites.3 African-American men are less likely to be screened for prostate carcinoma than whites and more likely to have advanced disease at the time of diagnosis.4–9
Prostate carcinoma screening is controversial. Scientific and medical organizations differ regarding recommendations for screening and detection. Although the benefits of prostate carcinoma screening in reducing mortality rates have not been proven or shown to be cost-effective,10–17 screening, particularly using prostate specific antigen (PSA), is widespread. A better understanding of screening behavior, knowledge of prostate carcinoma risk, and attitudes toward screening among men at high risk, such as African-American men, would be valuable.
Because African-American men are at such high risk for prostate carcinoma, early efforts must be made to identify the knowledge, attitudes, and beliefs of African-American men now to develop and target appropriate health messages. This is especially important in preparation for any additional developments in the control of prostate carcinoma. Obtaining information about the knowledge, attitudes, and beliefs of African-American men regarding prostate carcinoma also may help in understanding the low rates of participation of African-American men in prevention trials. However, few population-based surveys of PSA screening in minority populations are available. Most studies have relied on convenience samples and hence do not offer unbiased estimates of the prevalence of screening.
This research examines knowledge of prostate carcinoma risk, attitudes toward prostate carcinoma screening, and screening behavior in a cohort of African-American men residing in Central Harlem, New York City. We examine the prevalence of self-reported PSA screening, attitudes toward prostate carcinoma screening, and accuracy of knowledge regarding prostate carcinoma risk. We assess the reliability of self-reported PSA screening by examining the consistency of self-reports in light of answers to a series of related questions built into the Harlem Men's Health Cohort survey. We concentrate on PSA screening because virtually all men reporting having a PSA test also reported having a digital rectal examination for prostate carcinoma screening (see below).
METHODS AND SAMPLE
Eligibility criteria for inclusion in the Harlem Men's Health Cohort required residence in the 11 zip code zones that define Central Harlem, age 50–74 years, fluency in English, no reported treatment for cancer in the prior year, and no reported lifetime diagnosis or treatment for prostate carcinoma. The intent was to enroll a cohort of men presumably at risk for prostate carcinoma but without direct experience of the disease. The men ultimately will be observed in prospective studies. Aware of the high prevalence of nontelephone households in Harlem (23.9% in the recent Harlem Health Survey; D McLean, personal communication), we enrolled men into the study from clinic settings to represent men without telephones (of whom 17.5% reported no telephone). A second sample of community-based men was identified using a random-digit dial (RDD) telephone methods. Men were enrolled into the study from February 1995 to June 1996. The study was reviewed and approved by the New York State Department of Health and Harlem Hospital Center institutional review boards.
The clinic-based sample of 404 men was recruited from general medical and urologic clinics in Central Harlem, covering all providers in this geographic area. These included medical and medical subspecialty clinics at Harlem Hospital Center, North General Hospital, Robeson Clinic, and the Renaissance Health Care Network, which consists of five community-based ambulatory care centers. Recruitment from urology clinics contributed less than 5% of the clinic sample. Clinic patients were recruited by trained project interviewers based at each site who approached every adult male present at selected clinic sessions. Clinic recruitment sessions were chosen to capture a representative volume of service use. If subjects met study criteria, informed consent was obtained, and interviews took place in a private, adjoining room. For the 10% of patients who preferred another time or location for the interview (usually because patients had other appointments scheduled), contact information was obtained and interviews were conducted within 1–2 days, usually at patient homes.
The 319 men in the community sample were enrolled using a RDD telephone approach conducted by a private research firm. The firm called numbers within all telephone exchanges in the catchment area to identify potential respondents and then forwarded names to the research team, which followed up by telephone to assess study eligibility. If respondents met eligibility criteria and consented to participate, an in-person interview was scheduled and most often was completed in the man's home.
Project interviewers were all African-American research assistants who received extensive training. Each was observed conducting interviews and monitored by a field coordinator. In addition, a 10% random sample of respondents was contacted again by the project coordinator and interviewed on a select set of measures to ensure reliability of data.
The in-person interview involved a structured 60–90-minute assessment of respondent demographics (e.g., age, education, living conditions, income), medical access (e.g., insurance coverage, types of care providers, number of physician visits during the prior year), and physical and mental health, in addition to an extensive set of measures assessing prostate screening attitudes, knowledge of prostate carcinoma risk, and screening behaviors. Self-report health measures included the Medical Outcomes Study (MOS) short form (SF)-36,18, 19 and the American Urological Association (AUA) urinary symptom scale for benign prostatic hypertrophy.20–22 Although we are unaware of studies that have explicitly validated the measures in African-American populations, these are commonly used measures that have shown value in a variety of population studies.
Attitudes toward prostate carcinoma screening and knowledge of prostate carcinoma risk were assessed using 12 core questions developed in a joint effort involving Harlem Hospital, the New York Department of Health, and Louisiana State University as part of a prostate carcinoma demonstration project funded by the Centers for Disease Control and Prevention. These questions were developed primarily for public health surveillance and were based on the health belief model.23 In accord with the model, “perceived susceptibility” to prostate carcinoma was assessed in terms of knowledge about the risk of the disease; “perceived personal benefit” from screening was assessed in terms of favorable attitudes toward the efficacy of screening to detect prostate carcinoma and confidence that early treatment is warranted. Questions used to assess attitudes toward prostate carcinoma screening and knowledge of prostate carcinoma risk were extensively pretested in a series of focus groups.
Composite measures were constructed for items assessing attitudes toward prostate carcinoma screening and knowledge of prostate carcinoma risk by using separate factor analyses (varimax rotation, selecting first principal component) for a priori defined sets of variables. Eight items formed a reliable prostate carcinoma screening attitude scale (internal consistency Cronbach alpha, 0.77). The scale included items with both positive and negative valence; items were recoded accordingly to form a scale indicating more favorable attitude toward cancer screening. Subject responses ranged from strongly agree (1) to strongly disagree (5); these were summed and then divided by the number of items to give a scale with a range of 1–5. Items assessed topics such as whether men would want to know if they have prostate carcinoma and whether they would avoid screening for fear a physician may find prostate carcinoma (see Table 2 below).
Because the attitude measure was newly designed for this research and because no guidelines were available for interpreting this composite, screening attitude scores were divided into tertiles based on the distribution of responses to scale items.
We also conducted exploratory factor analyses of the items (principal components extraction, with varimax rotation). Two factors emerged, which together accounted for greater than 50% of the variance in item responses. The first factor identified attitudes associated with the perceived benefit of screening: “Screening gives peace of mind,” “People important to me would be reassured if I were screened,” “I want tests early to find out about prostate cancer,” “Finding prostate cancer is worth the cost of screening,” and “I would want to know if I had prostate cancer as early as possible.” The second factor identified attitudes associated with fear of the disease, denial of risk, and threat related to a positive test: “Leave well enough alone: better not to know if you have prostate cancer”, “I'm healthy: do not feel a need to be screened,” and “I don't want tests: I'm afraid MD might find prostate cancer”. This classification is in accord with the health belief model's broad distinction between the perceived benefit and threat of a health behavior.23
Knowledge of prostate carcinoma risk was measured in terms of the number of correct responses to 4 items: that men older than age 50 years are more likely to develop prostate carcinoma than younger men, that 1 in 10 men are likely to develop prostate carcinoma, that a man is more likely to develop prostate carcinoma if his father had it, and that African-American men are at higher risk than whites. A simple sum of correct answers was calculated, such that knowledge scores ranged from 0 to 4. For analytic purposes, knowledge scale scores were divided into three categories: none correct, one correct, and more than one correct.
From prior clinical experience, we were aware that elicitation of PSA screening status is complex. The simple, direct question included in our survey—“Have you ever had a PSA blood test for prostate cancer?”—elicited a high number of “unsure” responses (n = 117, or 16.2% of the total sample). To explore men's responses to this question in more detail, we relied on cognitive survey methodology,24 using the consistency of men's responses to four related items to check, and, if warranted, reassign men's responses on this outcome measure. The four items included the following: 1) “Have you ever heard of prostate cancer?” 2) “Have you ever heard of any tests for prostate cancer?” 3) “Have you ever been tested for prostate cancer?” 4) “Have you ever had a PSA blood test for prostate cancer?” Use of the four items is superior to the single item because it allowed us to identify men who inconsistently answered items, for example, those who asserted that they had been tested for prostate carcinoma while earlier reporting that they were unaware of such tests, or who stated they had never heard of prostate carcinoma but then said that they had been screened.
While recognizing the limitations of this approach (see “Discussion,”), we used this procedure to adjust self-reported prevalence of PSA screening in two ways. First, it allowed us to identify inconsistent responders who likely reported PSA screening out of social desirability in the interview context or who simply were unlikely to be reliable in their self-reports. Second, the procedure allowed us to examine more carefully men who answered the four questions consistently but were uncertain if they had been screened.
For assessing correlates of PSA screening, we introduced additional measures that might affect screening behaviors. One set of measures involved genitourinary function, which we included because little information on genitourinary function is available for older, African-American men and also because we wished to examine whether genitourinary symptoms might be associated with prostate carcinoma screening. We used the AUA urinary symptoms scale.22 Scores on the AUA scale were categorized according to published guidelines: scores of 0–7 indicate no or mild symptoms, 8–19 moderate symptoms, and 20–35 severe symptoms. A second set of measures assessed overall functioning and well-being. We used the MOS SF-36 and computed summary scales according to suggested guidelines; these range from 0 to 100, with 100 considered optimal health in each domain.19 The SF-36 is useful for identifying the effects of medical conditions that might affect cancer screening behavior (i.e., severe physical debility). These measures, along with sociodemographic indicators, were included in multiple logistic models as covariates that might be related to PSA screening.
As mentioned earlier, the Harlem Men's Health Cohort includes two different sample populations, a clinic sample and RDD community sample. The two samples were ascertained in settings likely to be correlated with the PSA screening outcome (i.e., clinic men reported less income and education, as well as worse health and less access to regular medical providers). Hence, all analyses were stratified by ascertainment site (clinic or community). Prevalence estimates were calculated separately for clinic and community men, and significance tests between the two samples are not presented. Multiple logistic regression models were constructed to examine correlates of the dichotomous PSA screening outcome separately in the clinic and community samples. All potential correlates were entered, and those not meeting entry criteria (P > 0.05) were deleted from models. Interaction terms of specific interest, such as the interaction between screening attitudes and knowledge regarding prostate carcinoma risk, were introduced into logistic models. Odds ratios and 95% confidence intervals were calculated.
To estimate the prevalence of PSA screening in Central Harlem, we adjusted self-reported prevalence in the RDD survey group according to the prevalence of nontelephone residences (as reported in the Harlem Household Survey (D. McLean, personal communication) in two ways: using the difference in the prevalence of reported PSA screening between clinic patients with and without telephones and using the ratio between these two prevalence estimates.
The response rate for the clinic sample was quite high; less than 5% of eligible men declined to participate. In the community sample, participation was lower. Interviews were completed by approximately 65% of the men referred to the research team by the private survey firm. The primary reason for nonresponse in this group was inability to locate respondents who had responded to a first call from the survey firm commissioned to identify respondents (20%), and further discovery that referred men were not, in fact, residents at these addresses and were accordingly ineligible (10%). An additional 5% refused when contacted by the research team.
The sociodemographic characteristics of participants in the Harlem Men's Health Cohort are shown in Table 1. The clinic and community samples show important differences. The community sample was more highly educated (27.8% with postsecondary education, compared with 14.9%), less likely to rely on Medicaid for medical insurance (31.2% vs. 61.2%), more likely to be employed (22.9% vs. 13.3%), and less impoverished (26.5% with incomes greater than $20,000, compared with 9.6%).
|Characteristic||Clinic (n = 404)||Community (n = 319)|
|Age (yrs [standard deviation])||60.3 (±7.21)||61.4 (±7.21)|
|Low (0–8 yrs)||27.5||22.5|
|Mid (9–12 yrs)||57.6||49.7|
|High (13+ yrs)||14.9||27.8|
|Medical insurance coverage (%)a|
|Current marital status (%)|
Health Status of the Cohort
The self-reported health of cohort respondents showed important differences between the clinic and community samples. In the community sample, 80.6% of the men had no or mild urinary symptoms compared with only 62.7% in the clinic sample; 1.9% of community men had severe symptoms compared with 5.5% among clinic men. The mean MOS physical function score for community men was 80.0% compared with 61.1% among clinic men. 62.0% of the clinic sample rated their health as fair or poor as compared with 43.4% of the community sample.
Indicators of Screening Behaviors, Prostate Carcinoma Risk, and Attitudes toward Prostate Carcinoma Screening
Respondents in the clinic and community samples answered questions regarding screening behaviors (and awareness of screening), knowledge of prostate carcinoma risk, and attitudes toward screening similarly (Table 2). In both samples, approximately half of the men reported they were aware that there were tests to screen for prostate carcinoma. A high proportion of men in both samples were favorable toward screening, but knowledge about prostate carcinoma risk was low.
|Awareness of prostate carcinoma and screening (%)|
|Ever heard of prostate carcinoma||82||91|
|Ever heard of tests to screen for prostate carcinoma||47||59|
|Ever tested for prostate carcinoma||42||54|
|Ever had PSA testa||8||17|
|Knowledge items (% agree)b|
|Man is more likely to develop PCa if father had it||34||29|
|Men older than age 50 yrs are more likely to develop PCa||53||52|
|One in every 10 men will develop PCa in lifetime||38||33|
|African-American men are at higher risk than whites||29||30|
|Attitude items (% agree)b|
|Leave well enough alone: better not to know if you have PCa||16||7|
|Would want tests early to find out about PCa||94||96|
|Doesn't want tests: afraid MD might find PCa||15||11|
|Finding PCa is worth cost of screening||90||93|
|People important to me would be reassured if I were screened for PCa||92||92|
|I'm healthy: do not feel need to be screened||13||10|
|Screening gives peace of mind||89||91|
|Would want to know if I had PCa as soon as possible||96||97|
Inconsistent or Uncertain Responses to Prostate Screening Questions
The breakdown of the sample on the four hierarchic measures of screening used in the survey is shown in Figure 1. One hundred two men reported that they had never heard of prostate carcinoma, even though 15 of these men reported that they had or might have had a PSA test. We considered these men not to have been screened with PSA. Similarly, 255 men had heard of prostate carcinoma but had not heard of tests to screen for it; yet, 56 of these men reported that they had been screened, including several with PSA. We again did not consider these valid reports of PSA. Conversely, of 257 men who reported that they had heard of prostate carcinoma, that they were aware of tests for it, and that they had been screened for it, 49 were uncertain if they had a PSA test but were assumed to have had PSA screening.
In 6.7% of the total sample, a valid response to 1 or more of the 4 questions was missing. Accordingly, these men could not be assigned a PSA screening status, leaving a sample of 675 for analysis (n = 377, clinic; n = 298, community). The proportion of men who could not be assigned PSA screening status was similar between clinic and community samples (6.7% in the clinic sample, 6.6% in the community sample).
With the above procedure, approximately one third of the clinic men and approximately half of the community men who said they were “unsure” of their PSA screening status were reclassified as PSA screened, and the remainder was assigned to nonscreened status. In the remainder of the sample, 18% of clinic men and 14% of community men were reclassified. Although we could not validate this reclassification against actual screening status (established, for example, through medical chart review), we were able to assess the validity of the reclassification by examining expected correlates of screening. For example, education was more strongly related to reclassified than initially reported PSA status. Self-reported PSA screening in the community sample was higher in those with 0–8 years of school (17.5%) than in those with 9–12 years (14.5%), contrary to evidence showing a strong relation between education and preventive health behaviors. In the reclassification,18.0% of the least educated and 22.1% of the more educated group were considered to have been screened.
Prevalence of Self-Reported PSA Screening in Central Harlem
The prevalence of PSA screening, after reclassification derived from the cognitive survey methodology, was 25.5% in the community sample and 11.1% in the clinic sample. We estimated the PSA prevalence for community men without telephones by reducing the RDD sample PSA estimate by the difference in the PSA screening prevalence estimates for clinic patients with telephones (self-reported PSA prevalence 12.4%) and without telephones (7.9%): 25.5% − (12.4% − 7.9%) = 21.0%. We then derived an estimate (24.7%) of the overall community PSA prevalence as a weighted average of the RDD prevalence (25.5%) and prevalence for men without telephones (21.0%), weighting by the proportions of households with and without telephones in the Harlem Health Survey. To see how sensitive the result was to this specific technique, we also tried an adjustment based on the ratio of prevalences (12.4%/7.9%), yielding estimates of 25.5% / (12.4%/7.9%) = 16.2% for community men without telephones and 23.6% for the community as a whole.
Of those screened with PSA, 79.1% of clinic men and 86.8% of community men reported having a digital rectal exam (DRE). Digital rectal exam screening without PSA was also common in both samples: 25.1% of clinic men and 35.5% of the community men reported a DRE in the absence of a PSA test.
Multiple Logistic Regression Models of PSA Screening in Central Harlem
Correlates of PSA screening were examined separately for clinic and community samples (Tables 3 and 4). Potential correlates meeting model selection criteria (P < 0.05) are shown, along with sociodemographic and health access variables selected on a priori grounds. We assessed all variables for interaction effects; however, no interaction term was significant. Substituting other indicators of sociodemographic status (e.g., income, any insurance coverage) did not alter results. For the clinic sample, 326 of 377 respondents with assigned PSA status (86%) had valid responses to all correlate variables and were used in the final logistic model. For the community sample, 270 of 298 respondents with assigned PSA status (91%) were available for this analysis.
|Predictor (n)||Odds ratio||95% CI|
|High (13+) (51)||Ref|
|Mid (9–12) (186)||0.5||0.2–1.1|
|Low (0–8) (89)||0.2a||0.1–0.8|
|Regular medical provider|
|Attitudes toward PCa screening|
|High (most favorable) (89)||Ref|
|Knowledge of PSA|
|Know someone with PCa|
|Predictor (n)||Odds Ratio||95% CI|
|High (13+) (99)||Ref|
|Mid (9–12) (135)||0.6||0.3–1.2|
|Low (0–8) (56)||0.4a||0.1–1.0|
|Regular medical provider|
|Attitudes toward PCa screening|
|High (most favorable) (42)||Ref|
|Knowledge of PCa|
|Know someone with PCa|
In the clinic sample, age was not significantly associated with increased likelihood of screening; however, low-educated (≤ 8 years of school) men were significantly less likely to report PSA screening than men with more than 12 years of school (low education, odds ratio [OR], 0.2; 95% confidence interval [CI], 0.1–0.8). In the clinic sample, less favorable attitudes toward screening also were associated with a reduced likelihood of screening. Respondents scoring in the middle tertile of the attitude composite were significantly less likely to report screening than respondents with the most favorable attitudes (OR, 0.3; 95% CI, 0.1–0.8). Respondents with the least favorable attitude were also less likely to report PSA screening (OR, 0.5; 95% CI, 0.2–1.3), but this relation did not achieve statistical significance. Finally, respondents knowing someone with prostate carcinoma were significantly more likely to report PSA screening than respondents who said they did not know anyone with prostate carcinoma (OR, 3.0; 95% CI, 1.4–6.4).
In the community sample, higher education and more favorable attitudes toward screening were also significant and independent correlates of PSA screening, whereas knowledge of prostate carcinoma risk and Medicaid status were not independent correlates. In the community sample, however, age and having a regular source of medical care were significant correlates. Respondents in the oldest age group (age 70–74 years) were more likely to report screening than younger men (OR, 2.8; 95% CI, 1.2–6.3, when compared with men aged 50–59 years). Respondents reporting a single, regular source of medical care were also more likely to report PSA screening than those without a regular source of care (OR, 1.9; 95% CI, 1.0–3.6). Among men in the community sample, knowing someone with prostate carcinoma was not a significant, independent correlate of PSA screening.
We also examined the sets of items indicating the perceived “benefit” and “threat” of screening. Using this approach in logistic regression models, in which the two scales were entered as continuous variables, we found that perceived threat was the more important correlate of self-reported screening. In both clinic and community samples, greater perceived threat was significantly associated with increased likelihood of self-reported screening (P < 0.05). Perceived benefit was not a significant predictor.
This report presents new information about the knowledge, attitudes, and screening behavior of innercity, African-American men regarding prostate carcinoma. Accuracy in knowledge regarding prostate carcinoma risk was low among Harlem men and was not a significant correlate of reported PSA screening in multiple logistic models. For three of the four knowledge items, only one third of the respondents gave correct answers. The finding of low levels of knowledge regarding prostate carcinoma risk has been reported for other innercity African-American samples25 and should be considered when interpreting the favorable attitudes toward screening elicited in surveys. Also, it is important to note that 42% in the clinic sample and 59% of the community sample were aware of tests to screen for prostate carcinoma, and many were unsure if they had ever been screened for the disease.
Overall, Harlem men, as has been reported for other African-American men,4, 5, 26 held favorable attitudes toward prostate carcinoma screening. In multiple logistic models, which adjust for sociodemographic differences and variation in access to medical care, favorable attitudes toward prostate carcinoma screening remained a significant correlate of reported screening. However, in the clinic sample, the observed association was not monotonic. In both samples, perceived threat was a stronger correlate of self-reported screening than perceived benefit.
We estimate the prevalence of self-reported PSA screening among men residing in Central Harlem in 1995–96 as 23.6–24.7%. This is higher than an age-adjusted estimate of 18% in 1991 by Demers et al.27 and Potosky et al.,28 based on data from four SEER areas linked to Medicare claims files. However, the Potosky estimate was developed for men aged 65 and older, whereas most of the men of our study population were younger than 65 years, and our maximum age was 74 years. Using administrative data, Williams et al. report a prevalence of 25% for PSA screening in primary care practices,29 a prevalence more than double that of the Harlem clinic sample.
Within this low-income community, we found that prostate carcinoma screening was associated with indices of sociodemographic status. In the community sample, older and more educated men were more likely to report screening in models that controlled for source of medical insurance, having a regular medical provider, knowledge of prostate carcinoma risk, and whether they knew someone with prostate carcinoma. Of note, men aged 70–74 years were more likely to be screened, an age group less likely to benefit from early detection and aggressive management, given competing comorbidities.
Strengths of the research include its comprehensive self-reported measures, the face-to-face interview format using interviewers from the community, and sampling from both community and clinic men. Limitations involve sampling issues for the community component and reliance on self-report for the screening outcome. Random-digit dialing obviously misses people who do not have a telephone and undersamples people whose telephone access is intermittent or who are rarely at home during normal calling hours. Also, we do not know how many working telephone numbers with an eligible subject existed, so that an overall response rate cannot be estimated. We adjusted the overall prevalence estimate in an attempt to take into account a likely lower prevalence of PSA screening among men without access to a telephone, but in the absence of more data on the relation of telephone access to PSA prevalence our adjustment is necessarily imprecise.
The validity of self-reports cannot be assumed. Although validation of the self-reports is not possible given our study design, which involved personal interviews only, we did include several questions in the survey that at least allowed us to identify inconsistent responders (whose reports of PSA screening are less likely to be reliable) and to reclassify men who reported some testing but who were unsure if it included a PSA test. This cognitive survey methodology should not be seen as a substitute for external validation of self-reported screening. However, in the absence of such a criterion, it is likely to offer greater accuracy than reliance on self-reports to a single item on screening. The last consideration is also important in light of missing data in the sample. As described earlier, 16.2% of the sample were unsure if they had been screened with a PSA test and would otherwise be lost to analysis without some kind of imputation.
Information from respondents about their physicians also would have been helpful, because research has shown that the physician is the single most influential factor in whether patients make use of other screening modalities, such as mammography.30 In this survey, we were unable to assess the role of primary care physicians (for example, academic-based practitioners vs. community physicians) as an additional determinant of the likelihood of prostate carcinoma screening. We currently are investigating how physicians make information on PSA screening available and how patients decide to follow through on recommendations for screening. At this point, it is still unclear what physicians say about the pros and cons of testing and how patients interpret this information.
Prostate carcinoma screening remains one of the most controversial subjects in medical practice today. Scientific and medical organizations differ or disagree regarding recommendations for screening and early detection. Given that high proportions of African-American men in Central Harlem have inaccurate impressions about prostate carcinoma risk, yet have highly favorable attitudes toward screening, efforts should be made to increase awareness in the community of the true risks of the disease. It would be valuable to increase African Americans' knowledge about the potential benefits, serious risks, and uncertainties of existing clinical interventions for prostate carcinoma. Also, differences between clinic and community samples suggest that health promotion approaches will need to be sensitively targeted even within a single community.
The American Cancer Society recently revised its guidelines regarding prostate carcinoma screening to recommend that information should be provided to patients regarding potential risks and benefits of intervention. Our study examined awareness of PSA tests but not awareness of the risks of screening—additional research worthy of further investigation. African-American men are at the highest risk for prostate carcinoma and death from prostate carcinoma. Medical providers and public health officials need to be prepared to offer objective information to African-American men about prostate carcinoma and prostate carcinoma screening.
The authors gratefully acknowledge David Momrow, M.P.H., Susan True, M.Ed., and Joyce Vanna, Ph.D., at New York State Department of Health for their technical support and Gerald Thomson, M.D., Diane McLean, Ph.D., and Diana Gurley, Ph.D., of the Harlem Disease Prevention Center for their involvement in early stages of the study.
- 2American Cancer Society. Cancer facts and figures, 1998. Atlanta, GA: American Cancer Society, 1998.
- 1SEER. Cancer statistics review, 1973–1994. SEER Available from URL: http://www-seer.ims.nci.nih.gov [accessed June 1, 2000].
- 4RiesLAG, MillerBA, HankeyBF, KosaryCL, HarrasA, EdwardsBK, eds. SEER cancer statistics review, 1973–1991: table and graphs, 1994. Bethesda, MD: National Cancer Institute, 1994.
- 12U.S. Congress, Office of Technology Assessment. Costs and effectiveness of prostate cancer screening in elderly men, 1995. Washington, DC: U.S. Government Printing Office, 1995.
- 18Measuring functioning and well-being: the Medical Outcomes Study, 1992. Durham, NC: Duke University Press, 1992., .
- 19Health Outcomes Institute. Health status questionnaire 2.0: user guide. Boston, MA: New England Medical Center, 1993.
- 23The health belief model. In: GlanzK, LewisFM, RomerBK, editors. Health behavior and health education, 2nd ed. San Francisco: Jossey-Bass, 1997: 41–59., .
- 24JabineTB, StrafML, TanurJM, TourangeauR, editors. Cognitive aspects of survey methodology: bulding a bridge between disciplines. Washington, DC: National Academy Press, 1984.