Given the greater intensity of screening early in the PSA era, and the resultant higher cumulative incidence of a prostate cancer diagnosis and attempted curative treatment particularly with radical prostatectomy among cohort members in the Seattle–Puget Sound area than in Connecticut over the entire 15 years of follow-up, a lower prostate cancer mortality in the Seattle region might have been expected over time if these interventions were effective. However, no significant difference in prostate cancer mortality was found over 15 years, and the 95% confidence interval around the rate ratio is inconsistent with more than a 4% reduction in prostate cancer mortality in the Seattle area compared to Connecticut. Interestingly, the overall drop in prostate cancer mortality has been similar in the two regions (Fig. ). These results suggest that the higher intensity of screening and treatment with radiotherapy or surgery in the Seattle area compared to Connecticut early in the PSA era did not translate into any further reduction in prostate cancer mortality for men over age 65.
The subgroups in our cohorts age 65–69 at baseline are of particular interest, as these men had life expectancies above ten years during the early PSA era, and might be hypothesized to have been most likely to have benefited from more aggressive screening and treatment. During the exposure window of 1988–1990, men in this age group in the Seattle area cohort were about 5.6 times as likely to have a PSA test and 2.2 times more likely to undergo a prostate biopsy than men in the Connecticut cohort;9
while over the entire follow-up period, they were 4.4 times as likely to have a radical prostatectomy and equally likely to have undergone radiation therapy. Nevertheless, they did not have a significantly lower prostate cancer mortality risk over the next 15 years.
How much of an impact on population-based prostate cancer mortality might have been expected from the difference in the use of radical prostatectomy or external beam radiotherapy among these younger men? Based on the observed treatment patterns in the two study cohorts, for the youngest men age 65–69 in 1987, about 961 fewer radical prostatectomies or courses of radiotherapy were observed in the Connecticut cohort. If just 10% of these 961 men were destined to die of prostate cancer without treatment and instead died of other causes, then the population-based prostate cancer mortality rate in Connecticut would have increased from about 125 to 141/100,000 and the difference would have reached statistical significance for this subgroup.
The subgroups of men age 75–79 at baseline are also important to consider, as these men had less than a ten-year life expectancy at baseline, and as a result, are not being included in current screening trials. During the exposure window of 1988–1990, men this age in the Seattle area cohort were about 5.2 times as likely to have a PSA test and 1.7 times more likely to undergo a prostate biopsy than men in the Connecticut cohort;9
while over the entire follow-up period, they were 15 times as likely to have a radical prostatectomy and 1.8 times as likely to have undergone radiation therapy. Nevertheless, they did not have a significantly lower risk of prostate cancer mortality over the next 15 years.
Our findings are consistent with a case-control study carried out at ten Veterans Affairs medical centers. In that study, PSA screening was not significantly associated with lower all-cause or cause-specific mortality.16
Our results are also in accord with an ecological analysis that did not show a statistically significant relationship between the prevalence of PSA testing and percentage declines in prostate cancer mortality across all SEER areas.5
Our findings conflict with a report from Tyrol, Austria, where widespread provision of free PSA screening was reported to have resulted in lower prostate cancer mortality in Tyrol versus the rest of Austria.17
It is difficult to compare the two studies, as the frequency of PSA testing, biopsy, and attempted curative treatment have not yet been reported from Tyrol. Our findings also conflict with a recent study that described lower overall survival among older men with low-risk and intermediate-risk prostate cancer managed with observation compared to active treatment.18
However, most of the differences in overall survival occurred in the first five years of follow-up, and only a small percentage of deaths were attributable to prostate cancer, which strongly suggests residual confounding by comorbidity.
Because our cohorts only included men aged 65 and over, the results may not be applicable to younger men. However, our study findings are highly relevant to the screening debate as the majority of men are diagnosed with prostate cancer after age 65. Another limitation of our study is the lack of data on men enrolled in Health Maintenance Organizations (HMOs). During 1987–1990, about 20% of the Seattle cohort and 10% of the CT cohort were HMO members. Based on SEER data, we had previously confirmed that older men in the Seattle area had substantially higher prostate cancer incidence and radical prostatectomy rates than men in Connecticut.19
In a sensitivity analysis excluding HMO enrollees from both cohorts, no significant difference in prostate cancer mortality emerged.
A study by Newschaffer and colleagues suggested that clinicians may be more likely to code deaths due to prostate cancer among men not receiving attempted curative therapy.20
If this coding bias exists, then prostate cancer mortality in the Seattle area cohort, where men were more likely to have attempted curative procedures, should have appeared lower.
The most striking difference between our two cohorts was that the rate of radical prostatectomy was substantially higher in the Seattle area. Currently only radical prostatectomy has been shown to be an effective treatment for localized prostate cancer in a randomized trial; however, radical prostatectomy had little effect for men aged 65 or older in that study, and few of the men in that trial were detected by PSA screening.21
Ultimately, several large randomized trials of PSA screening now underway will determine whether such screening reduces overall or disease-specific mortality. Population-based studies such as ours (as well as rigorous cohort or case-control studies) that are focused on the effectiveness of interventions in the community may provide insights that will be complementary to the results of these clinical trials, particularly for men age 75 and older who have been excluded from the trials. Our results suggest that it is inappropriate to assume that more intensive efforts at PSA screening and attempted curative treatment such as practiced in the Seattle area compared to the state of Connecticut early in the PSA era further reduced prostate cancer mortality for men age 65 and older. Until results of randomized trials become available, shared decision-making regarding PSA screening should continue between clinicians and patients. It remains unclear whether PSA screening is primarily responsible for the decline in prostate cancer-specific mortality that has been observed among older men in the United States.