Age-Specific Patterns of Prostate-Specific Antigen Testing Among Primary Care Physician Visits

Although direct evidence on the effectiveness of prostate cancer screening is not available, a decision model to estimate its benefits and risks provides indirect evidence. That model uses favorable but unproven assumptions to demonstrate that if prostate cancer screening proves effective, men aged 50 to 69 years will benefit most while those aged 70 years and older will benefit least. A decision model by Krahn and colleagues found that prostate cancer screening may result in worse health outcomes, particularly for men aged older than 70 years. To better characterize physician practices that provide the context for the ongoing prostate cancer screening debate, we focused on a large database of outpatient visits to US primary care physicians to examine national age-specific patterns of prostate-specific antigen (PSA) testing.

Methods

Data for this study were obtained from the National Ambulatory Medical Care Surveys (NAMCS) for 1995 and 1996. Those surveys, conducted annually by the National Center for Health Statistics, provide an ongoing assessment of the practices of US office-based physicians. The sampling process made use of the master lists of all US practicing physicians from the American Medical Association and the American Osteopathic Association. A subset of physicians from these lists was randomly selected and stratified by geographic area and specialty. For each participating physician in each year, patient visits during a random week were sampled systematically. Recorded information included patient demographics, reasons for the visit, diagnoses, and tests ordered. Among eligible physicians, annual participation rates averaged 72%. The unit of analysis was the physician-patient visit.

We examined the NAMCS surveys for 1995 and 1996. We focused on primary care physician visits by adult men aged 18 years and older. Primary care physicians included physicians in the specialties of family practice, internal medicine, general practice, and geriatrics. Prostate cancer was identified by an International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic code 185.00. We excluded men with prostate cancer, to limit PSA testing to that done for screening or diagnosis. In 1995 and 1996, 708 primary care physicians provided information on 6397 ambulatory visits by adult men without prostate cancer. Estimates of national practices were calculated by weighting data from the sampled visits. Our outcome measure was the probability of a primary care physician ordering a PSA test at any given visit. We also analyzed visits that might be associated with an increased frequency of PSA testing, such as those by men with lower urinary tract symptoms, those who had been diagnosed with benign prostatic hypertrophy (BPH), or those who were seeking a general medical examination. We employed National Center for Health Statistics relative standard error figures to calculate 95% confidence intervals (CIs) for national estimates. Comparisons were tested using continuity adjusted chi-square tests as implemented with Statistical Analysis System (SAS) software. We modified survey weights using proportional scaling with the method of Pothoff and colleagues to determine effective sample sizes.

Results

Extrapolating nationally, there were 7.7 million (95% CI, 6.2 million to 9.2 million) primary care physician visits per year by adult men in the United States during which a screening or diagnostic PSA test was ordered. PSA testing occurred during 3.9% of primary care physician visits by adult men. Of all PSA tests ordered during primary care physician visits, 17% were for men aged younger than 50 years, 50% for men aged 50 to 69 years, and 33% for men aged 70 years and older.

The probability of PSA testing was highest among men aged 60 to 64 years (7.1%) and remained relatively constant with increasing age, until age 80 years when the probability declined to 3.1% Table 1. Among those aged younger than 50 years, black men were 3.4 times more likely than white men to have a PSA test, but the probability did not differ later in life Table 2. For visits by men aged younger than 50 years, however, 63% of the PSA tests were for white men.

We hypothesized that visits by men with lower urinary tract symptoms, a diagnosis of BPH, and for a general medical examination would be associated with more PSA testing. We found higher probabilities of testing in men with lower urinary tract symptoms compared with those without (11% vs 3.8%, P<.01), much higher probabilities in men with a diagnosis of BPH than those without (38% vs 3.0%, P<.001), and higher probabilities during a general medical examination than routine visits (13% vs 3.0%, P<.001). The observed PSA testing patterns by age and race persisted when we stratified visits by these characteristics. In particular, rates of PSA testing remained relatively constant until a patient was older than 80 years. Also, the high frequency of PSA testing in older men was not simply a matter of increased incidence of BPH or lower urinary tract symptoms Table 1.

Discussion

Although the 1997 American College of Physicians (ACP) clinical guideline on prostate cancer screening stated that men aged 50 to 69 years will benefit most if prostate cancer screening ultimately proves effective and the official clinical guideline of the American Academy of Family Physicians (AAFP) recommended counseling men aged 50 to 65 years about the known risks and uncertain benefits of screening for prostate cancer, we found that half of all PSA testing by primary care physicians in 1995 and 1996 was for patients whose ages made them less likely to benefit. The relatively high rates of PSA testing in men aged 70 years and older (even after excluding men with BPH or lower urinary tract symptoms) implies, in part, that primary care physicians may not have been sensitive to patient age when making their decisions about ordering a PSA test. The rate of PSA testing did not decline until patients aged 80 years, yet at 3.1% of visits it was still high considering the greater risk of competing causes of mortality in these men, as well as higher rates of treatment complications and a shorter time span for any potential benefit from prostate cancer screening.

Although the American Cancer Society and the American Urological Association have recommended that clinicians begin screening high-risk men (black men and men with a family history of prostate cancer) at age 40 years, the more recently released ACP clinical guideline did not recommend screening high-risk men any differently, because of the lack of direct and indirect evidence quantifying the value of earlier or more aggressive screening. We found that among men aged younger than 50 years, black men were more than 3 times more likely than white men to have PSA testing. This observation may suggest that during our 1995 and 1996 study period primary care physicians were following the recommendations of the American Cancer Society or those of the American Urological Association. Although early PSA testing in white men may have been because of positive family history, that particular risk factor could not be determined using NAMCS data.

There have been recent efforts to show that the practice patterns of PSA testing may not be optimal. Our results from a nationally representative sample of primary care physicians from varying disciplines add to the data from previous studies on the use of PSA testing that have shown that a substantial proportion occurs among men whose ages make them less likely to benefit from screening.

Limitations

Our findings must be interpreted with caution because the cross-sectional nature of the NAMCS data does not allow us to follow the care provided to individuals over time. Nevertheless, these data provide the opportunity to analyze the office practices of physicians across the nation during a typical workweek. Also, the national figures that we report are extrapolated from a sample of visits to study physicians. That extrapolation, however, is in accordance with the weighting and statistical aggregation processes provided by the National Center for Health Statistics. Additionally, the episodic aspect of PSA screening means that our rates in older men may be underestimated. Because older men visit their physicians more frequently, they have more opportunities to be screened and may be less likely to be screened at any given visit. Finally, our focus on the PSA screening controversy in older men does not imply a lack of controversy in screening younger men.

Conclusions

Our findings suggest that current prostate cancer screening decisions by primary care physicians are not sensitive to age. Some investigators argue that age alone should not guide decision making for screening because of the importance of health status. However, the overall age-specific mortality rate increases with age, decreasing the overall impact of a reduction in any disease-specific mortality rate.

Prostate cancer screening remains widespread, despite recommendations against routine screening by the United States Preventive Services Task Force and the ACP, and recommendations by the AAFP for counseling about the known risks and uncertain benefits of screening for prostate cancer. Also, the American Cancer Society and the American Urological Association have softened their previously strong recommendations for routine annual screening; they now suggest screening be offered annually. Whether these recommendations will have an impact on future primary care physician practice patterns for PSA testing is yet to be shown.

Acknowledgments

This work was supported in part by grants HS 08397 and HS 09538 from the Agency for Health Care Policy and Research.

Although direct evidence on the effectiveness of prostate cancer screening is not available, a decision model to estimate its benefits and risks provides indirect evidence. That model uses favorable but unproven assumptions to demonstrate that if prostate cancer screening proves effective, men aged 50 to 69 years will benefit most while those aged 70 years and older will benefit least. A decision model by Krahn and colleagues found that prostate cancer screening may result in worse health outcomes, particularly for men aged older than 70 years. To better characterize physician practices that provide the context for the ongoing prostate cancer screening debate, we focused on a large database of outpatient visits to US primary care physicians to examine national age-specific patterns of prostate-specific antigen (PSA) testing.

Methods

Data for this study were obtained from the National Ambulatory Medical Care Surveys (NAMCS) for 1995 and 1996. Those surveys, conducted annually by the National Center for Health Statistics, provide an ongoing assessment of the practices of US office-based physicians. The sampling process made use of the master lists of all US practicing physicians from the American Medical Association and the American Osteopathic Association. A subset of physicians from these lists was randomly selected and stratified by geographic area and specialty. For each participating physician in each year, patient visits during a random week were sampled systematically. Recorded information included patient demographics, reasons for the visit, diagnoses, and tests ordered. Among eligible physicians, annual participation rates averaged 72%. The unit of analysis was the physician-patient visit.

We examined the NAMCS surveys for 1995 and 1996. We focused on primary care physician visits by adult men aged 18 years and older. Primary care physicians included physicians in the specialties of family practice, internal medicine, general practice, and geriatrics. Prostate cancer was identified by an International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic code 185.00. We excluded men with prostate cancer, to limit PSA testing to that done for screening or diagnosis. In 1995 and 1996, 708 primary care physicians provided information on 6397 ambulatory visits by adult men without prostate cancer. Estimates of national practices were calculated by weighting data from the sampled visits. Our outcome measure was the probability of a primary care physician ordering a PSA test at any given visit. We also analyzed visits that might be associated with an increased frequency of PSA testing, such as those by men with lower urinary tract symptoms, those who had been diagnosed with benign prostatic hypertrophy (BPH), or those who were seeking a general medical examination. We employed National Center for Health Statistics relative standard error figures to calculate 95% confidence intervals (CIs) for national estimates. Comparisons were tested using continuity adjusted chi-square tests as implemented with Statistical Analysis System (SAS) software. We modified survey weights using proportional scaling with the method of Pothoff and colleagues to determine effective sample sizes.

Results

Extrapolating nationally, there were 7.7 million (95% CI, 6.2 million to 9.2 million) primary care physician visits per year by adult men in the United States during which a screening or diagnostic PSA test was ordered. PSA testing occurred during 3.9% of primary care physician visits by adult men. Of all PSA tests ordered during primary care physician visits, 17% were for men aged younger than 50 years, 50% for men aged 50 to 69 years, and 33% for men aged 70 years and older.

The probability of PSA testing was highest among men aged 60 to 64 years (7.1%) and remained relatively constant with increasing age, until age 80 years when the probability declined to 3.1% Table 1. Among those aged younger than 50 years, black men were 3.4 times more likely than white men to have a PSA test, but the probability did not differ later in life Table 2. For visits by men aged younger than 50 years, however, 63% of the PSA tests were for white men.

We hypothesized that visits by men with lower urinary tract symptoms, a diagnosis of BPH, and for a general medical examination would be associated with more PSA testing. We found higher probabilities of testing in men with lower urinary tract symptoms compared with those without (11% vs 3.8%, P<.01), much higher probabilities in men with a diagnosis of BPH than those without (38% vs 3.0%, P<.001), and higher probabilities during a general medical examination than routine visits (13% vs 3.0%, P<.001). The observed PSA testing patterns by age and race persisted when we stratified visits by these characteristics. In particular, rates of PSA testing remained relatively constant until a patient was older than 80 years. Also, the high frequency of PSA testing in older men was not simply a matter of increased incidence of BPH or lower urinary tract symptoms Table 1.

Discussion

Although the 1997 American College of Physicians (ACP) clinical guideline on prostate cancer screening stated that men aged 50 to 69 years will benefit most if prostate cancer screening ultimately proves effective and the official clinical guideline of the American Academy of Family Physicians (AAFP) recommended counseling men aged 50 to 65 years about the known risks and uncertain benefits of screening for prostate cancer, we found that half of all PSA testing by primary care physicians in 1995 and 1996 was for patients whose ages made them less likely to benefit. The relatively high rates of PSA testing in men aged 70 years and older (even after excluding men with BPH or lower urinary tract symptoms) implies, in part, that primary care physicians may not have been sensitive to patient age when making their decisions about ordering a PSA test. The rate of PSA testing did not decline until patients aged 80 years, yet at 3.1% of visits it was still high considering the greater risk of competing causes of mortality in these men, as well as higher rates of treatment complications and a shorter time span for any potential benefit from prostate cancer screening.

Although the American Cancer Society and the American Urological Association have recommended that clinicians begin screening high-risk men (black men and men with a family history of prostate cancer) at age 40 years, the more recently released ACP clinical guideline did not recommend screening high-risk men any differently, because of the lack of direct and indirect evidence quantifying the value of earlier or more aggressive screening. We found that among men aged younger than 50 years, black men were more than 3 times more likely than white men to have PSA testing. This observation may suggest that during our 1995 and 1996 study period primary care physicians were following the recommendations of the American Cancer Society or those of the American Urological Association. Although early PSA testing in white men may have been because of positive family history, that particular risk factor could not be determined using NAMCS data.

There have been recent efforts to show that the practice patterns of PSA testing may not be optimal. Our results from a nationally representative sample of primary care physicians from varying disciplines add to the data from previous studies on the use of PSA testing that have shown that a substantial proportion occurs among men whose ages make them less likely to benefit from screening.

Limitations

Our findings must be interpreted with caution because the cross-sectional nature of the NAMCS data does not allow us to follow the care provided to individuals over time. Nevertheless, these data provide the opportunity to analyze the office practices of physicians across the nation during a typical workweek. Also, the national figures that we report are extrapolated from a sample of visits to study physicians. That extrapolation, however, is in accordance with the weighting and statistical aggregation processes provided by the National Center for Health Statistics. Additionally, the episodic aspect of PSA screening means that our rates in older men may be underestimated. Because older men visit their physicians more frequently, they have more opportunities to be screened and may be less likely to be screened at any given visit. Finally, our focus on the PSA screening controversy in older men does not imply a lack of controversy in screening younger men.

Conclusions

Our findings suggest that current prostate cancer screening decisions by primary care physicians are not sensitive to age. Some investigators argue that age alone should not guide decision making for screening because of the importance of health status. However, the overall age-specific mortality rate increases with age, decreasing the overall impact of a reduction in any disease-specific mortality rate.

Prostate cancer screening remains widespread, despite recommendations against routine screening by the United States Preventive Services Task Force and the ACP, and recommendations by the AAFP for counseling about the known risks and uncertain benefits of screening for prostate cancer. Also, the American Cancer Society and the American Urological Association have softened their previously strong recommendations for routine annual screening; they now suggest screening be offered annually. Whether these recommendations will have an impact on future primary care physician practice patterns for PSA testing is yet to be shown.

Acknowledgments

This work was supported in part by grants HS 08397 and HS 09538 from the Agency for Health Care Policy and Research.

Although direct evidence on the effectiveness of prostate cancer screening is not available, a decision model to estimate its benefits and risks provides indirect evidence. That model uses favorable but unproven assumptions to demonstrate that if prostate cancer screening proves effective, men aged 50 to 69 years will benefit most while those aged 70 years and older will benefit least. A decision model by Krahn and colleagues found that prostate cancer screening may result in worse health outcomes, particularly for men aged older than 70 years. To better characterize physician practices that provide the context for the ongoing prostate cancer screening debate, we focused on a large database of outpatient visits to US primary care physicians to examine national age-specific patterns of prostate-specific antigen (PSA) testing.

Methods

Data for this study were obtained from the National Ambulatory Medical Care Surveys (NAMCS) for 1995 and 1996. Those surveys, conducted annually by the National Center for Health Statistics, provide an ongoing assessment of the practices of US office-based physicians. The sampling process made use of the master lists of all US practicing physicians from the American Medical Association and the American Osteopathic Association. A subset of physicians from these lists was randomly selected and stratified by geographic area and specialty. For each participating physician in each year, patient visits during a random week were sampled systematically. Recorded information included patient demographics, reasons for the visit, diagnoses, and tests ordered. Among eligible physicians, annual participation rates averaged 72%. The unit of analysis was the physician-patient visit.

We examined the NAMCS surveys for 1995 and 1996. We focused on primary care physician visits by adult men aged 18 years and older. Primary care physicians included physicians in the specialties of family practice, internal medicine, general practice, and geriatrics. Prostate cancer was identified by an International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic code 185.00. We excluded men with prostate cancer, to limit PSA testing to that done for screening or diagnosis. In 1995 and 1996, 708 primary care physicians provided information on 6397 ambulatory visits by adult men without prostate cancer. Estimates of national practices were calculated by weighting data from the sampled visits. Our outcome measure was the probability of a primary care physician ordering a PSA test at any given visit. We also analyzed visits that might be associated with an increased frequency of PSA testing, such as those by men with lower urinary tract symptoms, those who had been diagnosed with benign prostatic hypertrophy (BPH), or those who were seeking a general medical examination. We employed National Center for Health Statistics relative standard error figures to calculate 95% confidence intervals (CIs) for national estimates. Comparisons were tested using continuity adjusted chi-square tests as implemented with Statistical Analysis System (SAS) software. We modified survey weights using proportional scaling with the method of Pothoff and colleagues to determine effective sample sizes.

Results

Extrapolating nationally, there were 7.7 million (95% CI, 6.2 million to 9.2 million) primary care physician visits per year by adult men in the United States during which a screening or diagnostic PSA test was ordered. PSA testing occurred during 3.9% of primary care physician visits by adult men. Of all PSA tests ordered during primary care physician visits, 17% were for men aged younger than 50 years, 50% for men aged 50 to 69 years, and 33% for men aged 70 years and older.

The probability of PSA testing was highest among men aged 60 to 64 years (7.1%) and remained relatively constant with increasing age, until age 80 years when the probability declined to 3.1% Table 1. Among those aged younger than 50 years, black men were 3.4 times more likely than white men to have a PSA test, but the probability did not differ later in life Table 2. For visits by men aged younger than 50 years, however, 63% of the PSA tests were for white men.

We hypothesized that visits by men with lower urinary tract symptoms, a diagnosis of BPH, and for a general medical examination would be associated with more PSA testing. We found higher probabilities of testing in men with lower urinary tract symptoms compared with those without (11% vs 3.8%, P<.01), much higher probabilities in men with a diagnosis of BPH than those without (38% vs 3.0%, P<.001), and higher probabilities during a general medical examination than routine visits (13% vs 3.0%, P<.001). The observed PSA testing patterns by age and race persisted when we stratified visits by these characteristics. In particular, rates of PSA testing remained relatively constant until a patient was older than 80 years. Also, the high frequency of PSA testing in older men was not simply a matter of increased incidence of BPH or lower urinary tract symptoms Table 1.

Discussion

Although the 1997 American College of Physicians (ACP) clinical guideline on prostate cancer screening stated that men aged 50 to 69 years will benefit most if prostate cancer screening ultimately proves effective and the official clinical guideline of the American Academy of Family Physicians (AAFP) recommended counseling men aged 50 to 65 years about the known risks and uncertain benefits of screening for prostate cancer, we found that half of all PSA testing by primary care physicians in 1995 and 1996 was for patients whose ages made them less likely to benefit. The relatively high rates of PSA testing in men aged 70 years and older (even after excluding men with BPH or lower urinary tract symptoms) implies, in part, that primary care physicians may not have been sensitive to patient age when making their decisions about ordering a PSA test. The rate of PSA testing did not decline until patients aged 80 years, yet at 3.1% of visits it was still high considering the greater risk of competing causes of mortality in these men, as well as higher rates of treatment complications and a shorter time span for any potential benefit from prostate cancer screening.

Although the American Cancer Society and the American Urological Association have recommended that clinicians begin screening high-risk men (black men and men with a family history of prostate cancer) at age 40 years, the more recently released ACP clinical guideline did not recommend screening high-risk men any differently, because of the lack of direct and indirect evidence quantifying the value of earlier or more aggressive screening. We found that among men aged younger than 50 years, black men were more than 3 times more likely than white men to have PSA testing. This observation may suggest that during our 1995 and 1996 study period primary care physicians were following the recommendations of the American Cancer Society or those of the American Urological Association. Although early PSA testing in white men may have been because of positive family history, that particular risk factor could not be determined using NAMCS data.

There have been recent efforts to show that the practice patterns of PSA testing may not be optimal. Our results from a nationally representative sample of primary care physicians from varying disciplines add to the data from previous studies on the use of PSA testing that have shown that a substantial proportion occurs among men whose ages make them less likely to benefit from screening.

Limitations

Our findings must be interpreted with caution because the cross-sectional nature of the NAMCS data does not allow us to follow the care provided to individuals over time. Nevertheless, these data provide the opportunity to analyze the office practices of physicians across the nation during a typical workweek. Also, the national figures that we report are extrapolated from a sample of visits to study physicians. That extrapolation, however, is in accordance with the weighting and statistical aggregation processes provided by the National Center for Health Statistics. Additionally, the episodic aspect of PSA screening means that our rates in older men may be underestimated. Because older men visit their physicians more frequently, they have more opportunities to be screened and may be less likely to be screened at any given visit. Finally, our focus on the PSA screening controversy in older men does not imply a lack of controversy in screening younger men.

Conclusions

Our findings suggest that current prostate cancer screening decisions by primary care physicians are not sensitive to age. Some investigators argue that age alone should not guide decision making for screening because of the importance of health status. However, the overall age-specific mortality rate increases with age, decreasing the overall impact of a reduction in any disease-specific mortality rate.

Prostate cancer screening remains widespread, despite recommendations against routine screening by the United States Preventive Services Task Force and the ACP, and recommendations by the AAFP for counseling about the known risks and uncertain benefits of screening for prostate cancer. Also, the American Cancer Society and the American Urological Association have softened their previously strong recommendations for routine annual screening; they now suggest screening be offered annually. Whether these recommendations will have an impact on future primary care physician practice patterns for PSA testing is yet to be shown.

Acknowledgments

This work was supported in part by grants HS 08397 and HS 09538 from the Agency for Health Care Policy and Research.