SEARCH

SEARCH BY CITATION

Keywords:

  • European Randomized Study of Screening for Prostate Cancer (ERSPC);
  • prostate-specific antigen (PSA);
  • follow-up;
  • guidelines;
  • general practice;
  • prostate cancer

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  10. Supporting Information

Objective

  • To determine the impact of the European Randomized Study of Screening for Prostate Cancer (ERSPC) publication in 2009 on prostate-specific antigen (PSA) level testing by Dutch general practitioners (GPs) in men aged ≥40 years.

Materials and Methods

  • Retrospective study with a Dutch insurance company database (containing PSA test claims) and a large district hospital-laboratory database (containing PSA-test results).
  • The difference in primary PSA-testing rate as well as follow-up testing before and after the ERSPC was tested using the chi-square test with statistical significance at P < 0.05.

Results

  • Decline in PSA tests 4 months after ERSPC publication, especially for men aged ≥60 years.
  • Primary testing as well as follow-up testing decreased, both for PSA levels of <4 ng/mL as well as for PSA levels of 4–10 ng/mL.
  • Follow-up testing after a PSA level result of >10 ng/mL moderately increased (P = 0.171).
  • Referral to a urologist after a PSA level result of >4 ng/mL decreased slightly after the ERSPC publication (P = 0.044).

Conclusions

  • After the ERSPC publication primary PSA testing as well as follow-up testing decreased.
  • Follow-up testing seemed not to be adequate after an abnormal PSA result. The reasons for this remain unclear.

Abbreviations
ERSPC

European Randomised Study of Screening for Prostate Cancer

NHG

Nederlands Huisartsen Genootschap, Dutch College of General Practitioners

PLCO

Prostate, Lung, Colorectal and Ovarian (Cancer Screening Trial)

USPSTF

USA Preventive Services Task Force

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  10. Supporting Information

Prostate cancer screening with PSA level testing is widespread. Despite the USA Preventive Services Task Force (USPSTF) recommendation against prostate cancer screening in men aged ≥75 years in 2008, PSA screening rates did not change in the USA [1]. However, Zeliadt et al. [2] showed that PSA testing in the USA decreased slightly after publication of the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial and the European Randomized Study of Screening for Prostate Cancer (ERSPC) results in March 2009. But screening remained substantial; 34% in men aged 40–54 years and 47% in men aged 55–74 years.

In the Netherlands, most patients with questions about cancer screening will visit their GP. GPs can use the Dutch GP guideline on Lower Urinary Tract Symptoms in older men, published in 2004 to advise these patients [3]. In this guideline, GPs are encouraged to discuss the pros and cons of PSA screening and to be reluctant to screen, as the impact of screening was unclear [3]. Much attention was given to the publication of the ERSPC in March 2009 [4]. All GPs received a Dutch version of this key article, written by the ERSPC study group. In a comment by the Dutch College of General Practitioners (Nederlands Huisartsen Genootschap, NHG) sent with this article, the 20% decrease of prostate cancer mortality was weighted against the high number needed to screen and treat [5]. This resulted in strong advice not to screen [5]. We investigated the effect of the ERSPC results on PSA testing rate by Dutch GPs in men aged ≥40 years.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  10. Supporting Information

We performed a retrospective study, using two separate databases. First, we used the claims database of a Dutch insurance company (Achmea), containing all PSA tests claimed for ≈3 million insured patients from 2007 onwards. This database has not been used before for scientific research. The insurance company has strict internal checks and we think that the database is sound. If any irregularity was present in the database, this would be non-specific and not related to the study outcome.

Furthermore, we analysed a database of all PSA tests from the only laboratory in the area of a large (900 beds) district general teaching hospital from 2004 until present. This laboratory has been recognised and granted accreditation by the Coordination Committee for the improvement of Quality control for Laboratory tests in health care (CCKL, http://www.cckl.nl). They have a quality system that is regularly validated and independent inspections are carried out. We therefore think that the hospital database is valid and reliable as well.

From both databases, we collected all PSA tests requested by GPs for men aged ≥40 years. PSA-test-requesting rates were estimated by dividing the number of primary tests by the number of men in the target population (derived from the Central Bureau of Statistics of The Netherlands [6]) based on the postal codes from the hospital database. For the claims database, we used the number of men in each age group registered at the insurance company.

The recruitment for the ERSPC in Rotterdam started in November 1993. The claims database included participants from the ERSPC study region (Rotterdam). That is, these participants may have had their insurance with Achmea. The ERSPC randomised 42 376 inhabitants (response rate 40%) of Rotterdam and 12 neighboring municipalities aged 55–75 years between June 1994 and 2000 into a screening and control arm [7]. For the early years of the ERSPC, Otto et al. [7] showed that the opportunistic screening rate in the control arm was considerably higher than for non-participants. Our claims database did not contain PSA tests performed at a 4-year interval for patients in the screening arm of the ERSPC. However, it did contain PSA tests for men in the control arm as well as for non-participating inhabitants in this region, as these tests were requested by their GPs. To study the impact of the aforementioned higher opportunistic screening rate in the control arm of the ERSPC on our claims database, we performed subgroup analyses. We selected all primary PSA tests for men living in the ERSPC-Rotterdam region, based on postal codes (information on specific region provided by Monique Roobol; personal communication). PSA-test-requesting rates for the ERSPC-Rotterdam region were compared with the rates in men living outside this region. Notably, we were not able to select ERSPC participants from the claims database, as the insurance company did not register participation status.

To assess the influence of the ERSPC publication, we defined the 12 months before publication as ‘before ERSPC’ and the 12 months after as ‘after ERSPC’. PSA tests performed in March 2009 (month of publication) were excluded from the analyses.

From both databases, primary PSA tests requested by a GP were selected. We considered a PSA test to be a primary test when no earlier test was available in the databases.

For men with an increased PSA level (>4 ng/mL) without a repeat test in the hospital database, we checked the medical files to see if there had been a referral to a urologist.

The proportion of men with a primary PSA test for different age groups per 1000 men per month from January 2008 to December 2010 are presented, as derived from the claims database. We compared the proportion of primary PSA tests (per 1000 men) before and after publication of the ERSPC results for different age groups and regions (ERSPC-Rotterdam vs other regions), using the chi-square test.

From the hospital database, we compared PSA test results (categories) between the two periods, as well as the number of men with repeat testing after initial PSA tests.

For the repeat tests we used a maximum follow-up time of 1 year after the initial test, as for the second period (after publication of the ERSPC study) less follow-up time was available. Moreover, especially for patients with abnormal test results, a longer period between primary test and repeat testing was considered inadequate. Repeat tests were dichotomised, as none or ≤1 year after the initial test. Furthermore, the percentages of repeat testing for different age groups and PSA level categories before and after ERSPC publication were compared, using the chi-square test. Additionally, we performed multivariable logistic regression analyses with repeat test as the dependent variable and study period, age groups, and PSA level categories as the covariates (odds ratios are presented). Nagelkerke R-square was estimated to value the percentage of variance explained by the model. A P < 0.05 was considered to indicate statistical significance.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  10. Supporting Information

The claims database included ≈715 000 insured men aged ≥40 years, of which 11% were inhabitants of the ERSPC-Rotterdam region. This database included 123 996 PSA test claims, of which 66 848 were considered primary tests requested by a GP from 2008 onwards. Of these, 9691 tests (14.5%) were performed in the ERSPC-Rotterdam region.

A decline in the incidence of PSA tests is shown ≈4 months after the ERSPC publication, especially for men aged ≥60 years (Fig. 1). The age-weighted average proportion of tested men per year was 33.09/1000 men before and 27.52/1000 men after ERSPC publication, for the total study sample (Mantel–Haenszel chi-square, P < 0.001).

figure

Figure 1. Proportion of men with a primary PSA test according to age (claims data).

Download figure to PowerPoint

In both periods the average proportion was significantly lower in younger men (age groups 40–50 and 50–60 years), compared with older men, but did not differ for older age groups (chi-square, all P > 0.05, except for age groups 60–69 vs 70–79 years, P = 0.049, Table 1). The PSA testing rate in the ERSPC-Rotterdam region was considerably higher than in the other regions (Table 1). The relative risk for PSA testing varied from 1.23 in men aged 60–70 years to 1.70 in men aged 40–50 years before ERSPC publication. After the publication, these figures were 1.13 and 1.54, respectively.

Table 1. Proportion of primary PSA tests (per 1000 men) 1 year before and after publication of the ERSPC according to age and region, derived from the claims database
 Before the ERSPCAfter the ERSPCP (chi-square)
  1. a

    The number of men before and after ERSPC publication differs due to an increase in the number of insured people.

Age group (number in group)a, years(675671)(766626) 
40–4912.1711.790.221
50–5928.8125.96<0.001
60–6951.6340.97<0.001
70–7957.3544.80<0.001
≥8052.0841.32<0.001
Within ERSPC-Rotterdam region:(78268)(82365) 
40–4914.4212.950.156
50–5940.1533.95<0.001
60–6960.8043.90<0.001
70–7970.3051.02<0.001
≥8072.1151.56<0.001
Outside ERSPC-Rotterdam region:(597404)(684261) 
40–498.508.450.736
50–5926.2824.26<0.001
60–6949.8039.20<0.001
70–7954.5141.70<0.001
≥8048.0937.75<0.001

The hospital database contained 30 558 PSA test results requested by GPs, of which 9766 were considered primary tests from 2006 until 2011. In this database, PSA categories did not differ between periods (Table 2).

Table 2. PSA test results (hospital data) before and after publication of the ERSPC study
 Before the ERSPCAfter the ERSPCP (chi-square)
Number of PSA tests109810000.539
PSA level (ng/mL) category, %   
<481.782.4 
4–1011.011.5 
>107.36.1 

Before ERSPC publication, 38.5% of all tests were followed by a repeat test within 1 year. After ERSPC publication, this dropped to 26.5% (chi-square, P < 0.001, Table 3).

Table 3. Percentage of repeat PSA tests ≤1 year after initial test, according to initial PSA level and age groups derived from the hospital database
 

Before the ERSPC

Number (%)

After the ERSPC

Number (%)

P (chi-square)
Total study population1098 (38.5)1000 (26.5)<0.001
PSA levels, ng/mL:   
<4897 (31.8)824 (18.8)<0.001
4–10121 (65.3)115 (52.2)0.028
>10107 (73.8)83 (82.0)0.171
Age groups, years:   
40–50125 (23.2)134 (11.9)0.017
50–60332 (39.5)279 (21.1)<0.001
60–70348 (46.0)328 (34.8)0.002
70–80209 (35.9)166 (33.7)0.373
≥8084 (33.3)93 (21.5)0.055

For all age groups, the chance of follow-up after a normal test result (PSA level of <4 ng/mL) dropped from 31.8%before to 18.8% after ERSPC publication (chi-square, P < 0.001, Table 3). For men with moderately increased PSA levels (4–10 ng/mL) this chance also decreased from 65.3% to 52.2% (chi-square, P = 0.028, Table 3). However, the chance of follow-up increased for men with PSA levels >10 ng/mL from 73.8% to 82.0% (chi-square, P < 0.171, Table 3).

The medical files of the 190 men with PSA level of >4 ng/mL without a repeat test were searched. Of these men, 28.3% before and 23.5% after ERSPC publication did receive follow-up by a urologist, but the urologist decided not to repeat the PSA test (Mantel–Haenszel chi-square P = 0.044). We did not look for the reason for referral, nor for outcome of the analyses performed by the urologist as this was beyond the scope of this study.

Repeat testing decreased for all age groups after the ERSPC publication (Table 3), with the largest difference for men aged 40–50 and 50–60 years (relative risk reduction of 48.7% and 46.6%, respectively) and smallest difference for men aged 70–80 years (6.1% reduction). Multivariable logistic regression analyses showed that age, initial PSA level and study period were all associated with the chance of repeat testing (Table 4). Nagelkerke R-square for this model was 0.20.

Table 4. Multivariable logistic regression analyses on repeat testing according to age, PSA level group and study period
 OR (95% CI)P
  1. OR, odds ratio.

Study period: <0.001
Before ERSPC (reference) 
After ERSPC0.55 (0.45–0.67) 
PSA level (ng/mL) categories:  
<4 (reference)<0.001
4–104.79 (3.54–6.49) 
>1013.63 (8.73–21.27) 
Age groups (years): <0.001
40–50 (reference) 
50–601.95 (1.34–2.84) 
60–702.66 (1.84–3.86) 
70–801.37 (0.90–2.07) 
≥800.72 (0.42–1.22) 

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  10. Supporting Information

The present study showed a considerable decline in primary PSA testing, as well as in follow-up testing, after the ERSPC publication. Although the present study was not designed to establish a causal relationship, we think that this decrease in PSA testing can largely be explained by the impact of the ERSPC publication on GPs' testing rates, as no other factors could be identified.

Insurance companies in the Netherlands completely reimburse PSA testing costs. Therefore, this could not account for the reduction in PSA testing. Furthermore, GPs in the Netherlands are advised to follow the NHG-guideline, which has not been changed during the study period. International guidelines are not commonly well-known by Dutch GPs. We assume that any influence of other guidelines outside the NHG is unlikely.

It might be possible that our databases included some PSA tests requested for men with prostate cancer (stable PSA under active surveillance or after radical prostatectomy). The number of such cases will be small, as in the Netherlands, such patients are mainly treated and followed by urologists. Moreover, these numbers would not differ between the studied periods and would have no relation to the ERSPC publications. Therefore, we think that the impact of such PSA tests on our conclusions is negligible.

The area of the insurance company (claims database) included the ERSPC-Rotterdam region. We showed that the PSA testing rate in this region was considerably higher. This is consistent with the figures reported by Otto et al. [7], who described the opportunistic screening rates for ERSPC participants in the early years of that study. About 40% of all men in this region were randomised. We think that the higher primary testing rate in this region cannot be the result of study participation only. It may reflect a higher awareness of non-participants as well. Moreover, based on the experience with more prostate cancer cases, GPs in this region may be more acquainted with PSA tests and might be less reluctant. Because no information on the background of these tests was available from this database, these can only be theoretical explanations.

As 11% of the men included in the claims database lived in the ERSPC-region, in our database 2.2% of the men participated in the screening arm, and 2.2% participated in the control arm. We assume that these small percentages did not influence the main results of the present study. We therefore think that the screening rates shown and the alterations after the ERSPC publication represent the screening patterns of all Dutch GPs.

We confirmed the age-specific effect of the ERSPC (and PLCO) publication on PSA testing described by Zeliadt et al. [2], but we found a considerably larger decrease in testing rates. Zeliadt et al. excluded the first 4 months after publication of the ERSPC and PLCO, while we only excluded the month of publication. Excluding the first 4 months, would make the decrease in PSA testing in the present study even more apparent. The smaller impact in the USA as shown by Zeliadt et al. may reflect differences in prostate cancer screening as a whole; the prevalence of PSA testing is much higher in the USA than in the Netherlands, as screening was never advocated in the Netherlands.

Furthermore, Zeliadt et al. [2] also evaluated the effect of an update of the USPSTF (suggesting an upper age limit of 75 years for PSA screening) published in August 2008. After this publication, a decrease in PSA testing in men aged ≥75 years was shown, as well as an increase in men aged 40–54 and 55–74 years. This might be explained by the publicity about PSA testing at the time of the USPSTF publication. After publication of the ERSPC and the PLCO results, they described that the subsequent decrease in PSA testing was statistically significant in all age groups compared with the period after the USPSTF publication. But when comparing the study period after the ERSPC and PLCO with all the previous study periods (before and after USPSTF) the decrease was only statistically significant in younger (40–54 years) and older men (≥75 years) [2]. This may in part be explained by the earlier increase in PSA testing after the USPSTF publication and by the varying PSA testing practice found among their studied practice groups [2].

Although the ERSPC analysed a core age group, it is well known that PSA tests are ordered for both younger and (much) older men. Although this may be incorrect, in daily practice, physicians may regard the ERSPC results as ‘evidence for the effect of PSA screening in general’. This is comparable to the implementation of, for example, drug treatments, which are in general tested in defined groups, but used in non-defined groups later on. Therefore, we were interested in all age groups and we have used age group comparisons.

After the ERSPC publication, primary testing as well as follow-up testing decreased, both for normal PSA results (<4 ng/mL) as well as for moderately elevated PSA results (4–10 ng/mL). GPs might regard the latter as less relevant since the ERSPC publication. Follow-up testing after an elevated PSA result of >10 ng/mL increased moderately after the ERSPC, but referral rates for this group decreased slightly. This suggests, that follow-up testing was not adequate after an abnormal PSA result (>4 ng/mL) and even less adequate after ERSPC publication. We presume that patients wanted to be informed on their likelihood of developing prostate cancer and as a consequence were counselled before testing. As abnormal PSA levels coincide with an increased chance of prostate cancer, it was surprising to see that GPs refrained from repeating such a test. The reasons for this remain unclear, but the NHG advice on restrained PSA use could be an influence.

The validated prostate cancer risk calculator (http://www.Prostatecancer-riskcalculator.com) of the Prostate Cancer Research Foundation (SWOP) could help GPs in determining their policy for follow-up testing, especially for men with elevated PSA levels [8].

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  10. Supporting Information

We would like to thank Paul Wagenaar from Achmea for providing us with the claims database and Anton Huisman from the Isala Clinics for his help with the hospital database. Furthermore, we would like to thank Monique Roobol from the ERSPC Rotterdam region for providing us with the exact region from which participants were invited.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  10. Supporting Information
  • 1
    Prasad SM, Drazer MW, Hou D, Hu JC, Eggener SE. 2008 US Preventive Services Task Force recommendations and prostate cancer screening rates. JAMA 2012; 307: 16921694
  • 2
    Zeliadt SB, Hoffman RM, Etzioni R et al. Influence of publications of US and European prostate cancer screening trials on PSA testing practices. J Natl Cancer Inst 2011; 103: 520523
  • 3
    Wolters RJ, Spigt MG, Van Reedt Dorland PF, Gercema AJ, Klomp ML. NHG-Standard ‘Difficult voiding in older men’ (second revision). Huisarts Wet 2004; 47: 571586
  • 4
    Schroder FH, Hugosson J, Roobol MJ et al. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med 2009; 360: 13201328
  • 5
    Van Weert H, Wiersma TJ. Prostate cancer screening? Not yet! Huisarts Wet 2009; 7: 350351
  • 6
    Centraal Bureau voor de Statistiek. Available at: http://www.cbs.nl. Accessed January 2013
  • 7
    Otto SJ, van der Cruijsen IW, Liem MK et al. Effective PSA contamination in the Rotterdam section of the European Randomized Study of Screening for Prostate Cancer. Int J Cancer 2003; 105: 394399
  • 8
    Roobol MJ, van Vugt HA, Loeb S et al. Prediction of prostate cancer risk: the role of prostate volume and digital rectal examination in the ERSPC risk calculators. Eur Urol 2012; 61: 577583

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
bju12029-sup-0001-si.mov53887K

Video.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.