1. ¹Finnish Cancer Registry, Liisankatu 21 B, FIN-00170 Helsinki, Finland
2. ²Tampere School of Public Health, FIN-33014 University of Tampere, Tampere, Finland
3. ³Department of Urology, Tampere University Hospital and University of Tampere, Box 2000, FIN-33521 Tampere, Finland
4. ⁴Department of Urology, Helsinki University Hospital, Box 580 FIN-00029, Helsinki, Finland
5. ⁵Department of Surgery, Jorvi Hospital, Turuntie 150, FIN-02740 Espoo, Finland
6. ⁶Department of Pathology, Tampere University Hospital, University of Tampere, Box 2000, FIN-33521 Tampere, Finland
7. ⁷Department of Clinical Chemistry, Helsinki University Hospital, Box 700, FIN-00029 Helsinki, Finland
8. ⁸Finnish Cancer Institute, Liisankatu 21 B, FIN-00170 Helsinki, Finland

Correspondence: Professor A Auvinen, Finnish Cancer Institute, Liisankatu 21 B, FIN-00170, Helsinki, Finland. E-mail: anssi.auvinen@uta.fi

Received 8 September 2006; Revised 10 November 2006; Accepted 13 November 2006.

Abstract

Specificity constitutes a component of validity for a screening test. The number of false-positive (FP) results has been regarded as one of major shortcomings in prostate cancer screening. We estimated the specificity of serum prostate-specific antigen (PSA) determination in prostate cancer screening using data from a randomised, controlled screening trial conducted in Finland with 32 000 men in the screening arm. We calculated the specificity as the proportion of men with negative findings (screen negatives, SN) relative to those with negative and FP results (SN/(SN+FP)). A SN finding was defined as either PSA4 ng ml^-1 or PSA 3.0–3.9 ng ml^-1 combined with a negative ancillary test (digital rectal examination, DRE or free/total, F/T PSA ratio). False positives were those with positive screening test followed by a negative diagnostic examination. Of the 30 194 eligible men, 20 794 (69%) attended the first screening round and 1968 (9.5%) had a screen-positive finding. A total of 508 prostate cancers were detected at screening (2.4%). Hence, the number of SN findings was 18 825 and the number of FP results 1358. Specificity was estimated as 0.933 (18 825 out of 20 183) with 95% confidence interval (CI) 0.929–0.936. Specificity decreased with age. Digital rectal examination as ancillary examination had similar or higher specificity than F/T PSA. In the second screening round, specificity was slightly lower (0.912, 95% CI 0.908–0.916). The specificity of PSA screening in the Finnish screening trial is acceptable. Further improvement in specificity could, however, improve acceptability of screening and decrease screening costs.