In a study of electronic medical records in men under the age of 70 from a large healthcare system, we found that following the 2012 USPSTF guideline change, relative screening rates for prostate cancer declined 23.4%, prostate biopsy rates declined 64.3%, incident prostate cancer detection rates declined 53.5%, and metastatic prostate cancer rates increased 36.9%; all changes over time were statistically significant. We feel that the rates in our report are accurate and can be useful to providers and patients during the shared decision-making process surrounding prostate cancer screening.
Comparison with Previous Studies
Studies have consistently found that PSA screening has declined since the 2012 USPSTF statement [7,8,9,10,11]. Consistent with these reports, our study reported a decline in screening rates from 448.6 per 1000 person-years in 2010 to 334.0 per 1000 person-years in 2015. Drazer et al., using data from the National Health Interview Survey (NHIS), compared 2010 survey respondents with 2013 respondents and found a significant decline in PSA testing in men ages 50 to 59 from 33.2 to 24.8% and in men ages 60–74 from 51.2 to 43.6% [10]. These self-reported data may be subject to recall bias. Jemal et al. also used the NHIS data and compared screening rates in men 50–74 years of age between 2010 and 2013 and found that screening declined significantly from 36.8 to 29.9% [7]. The screening rate ratio of 0.81 is comparable with the present study. In contrast to the above studies, Kim et al. studied over 5 million men using administrative claims data and concluded that PSA screening rates remained stable from the years 2008 through 2013 in men under the age of 75 [11].
Prostate biopsy rates declined 64.3% in our study which is larger than previous reports. Review of self-reported case logs of urologists undergoing re-certification by the American Board of Urology showed a 28.7% relative decline in prostate biopsy when comparing years 2009–2012 with 2013–2016 [16]. In a study using administrative claims data of over five million men spanning the years 2005 to 2014, relative biopsy rates declined 33% [12]. A Medicare data set comparing 2012 with 2014 showed a 12.4% relative decline in biopsy rates [26]. As mentioned previously, the accuracy of rate calculations in these studies is questionable because of the challenge in identifying an accurate denominator.
Few studies have examined differences in rates of incident prostate cancer since the 2012 USPSTF Statement. Jemal et al. compared the rates of all incident cancers between 2011 and 2012 in men 50–74 years of age and found the rate ratio was 0.84 [7]. The decline was driven by local/regional disease for which the rate ratio was 0.83 while metastatic rates remained stable (rate ratio = 0.99). The incidence rate of localized and regional disease further declined from 2012 to 2013 (356.5 to 335.4 per 100,000, IRR = 0.94) yet incident rates of metastatic disease were unchanged [17]. We also observed relatively stable rates of localized and metastatic prostate cancer through 2013. However, using data that extended through 2015, we observed a decline in incident cancer detection yet an increase in metastatic disease. Negoita et al. noted a rise of distant disease rates with an inflection point around the years 2010–2011 [20]. However, the age group included men up to age 75 and it is possible this was driven by higher metastatic rates in men ages 70–75, who are at greater risk. Kelly et al. using SEER data, reported an increase in metastatic rates in men aged 45–54 and men aged 55–69 in the years 2012 to 2014 but interestingly also noted an increased rate in the years 2008 and 2011 [21]. These authors had no data on PSA screening rates in this study. In contrast, we saw relatively stable metastatic rates in the years 2010–2012. To our knowledge, we are the first study to report an increase in metastatic disease in men under the age of 70, the population most relevant for counseling in shared decision-making and to policy makers.
Strengths and Limitations
We could define an “eligible population” which served as the denominator in rate calculations, thus accurate rates could be determined. Our sample size was quite large and diverse, and PSA screening and biopsy rates were determined by the PSA test and biopsy result recorded in electronic medical records, not by patient recall or administrative claims data or order history. We must emphasize that the changes in our screening rates are associated with the timing of the 2012 USPSTF Statement but we cannot prove causality in this observational study.
To be a confounder, a variable must be associated with both the exposure and the outcome. Our exposure in this study was the time period; thus a confounder would need to be associated with time as well as with the outcome. We looked at several possible confounders in our analysis including age, race, comorbidities and care utilization; Table 1 shows minimal changes in these variables over the study time period. A widely used criterion to determine confounding is that a confounder must change the effect estimate by at least 10% [27]. When we adjusted our cancer detection rates and metastatic rates for age and race, we found negligible changes (0 to 3%) in estimated incidence rates, far below the 10% threshold. It is possible that other unmeasured confounders exist, although they would need sufficiently strong changes over the study time period to meet the criteria for confounding. Screening and biopsy rates were presented unadjusted, consistent with previous studies [7,8,9,10,11,12, 16, 25].
Another limitation of our study was that we did not use family history of prostate cancer in defining our eligible population as this data element was not uniformly collected in our health record. However, we have no reason to suspect that the distribution of family history would have changed during the study period.
It is important to recognize that some of the declines in screening, biopsy and incident cancer detection may be beneficial. False-positive PSA elevations result in unnecessary biopsies that are associated with inherent risks of the procedure along with patient anxiety and costs. Additionally, the detection of indolent cancers likely decreased although we were not able to quantify this outcome. Such an analysis was beyond the scope of this project as it would require detailed pathologic review including the number of biopsy cores positive, the extent of individual core length, and prostate size. The balance between finding fewer indolent cancers yet avoiding more metastatic disease warrants further study.
We do not have any comparative mortality data at this time but we hypothesize this stage migration may result in an increase in mortality yet additional time is needed to study this endpoint. In a recent large clinical trial, the median overall survival for men with metastatic prostate cancer was approximately six years (57.6 months), and longer in men with low-volume metastatic disease [28]. Also of interest will be to monitor screening, biopsy and detection rates following the 2018 USPSTF. We hypothesize that our screening rates may increase once again which will likely result in a higher incident cancer detection rate yet a lower metastatic rate.
Our study shows that in the years following the guideline change, major reductions in PSA screening have coincided with lower overall incident cancer detection rates yet higher rates of metastatic disease in men under the age of 70. The ideal screening strategy would identify only those cancers that require treatment. At present, this goal remains elusive.