PSA screening prevents one prostate cancer death for every 456 men invited

After 23 years of follow-up, the largest randomised trial of PSA screening found one prostate cancer death prevented for every 456 men invited to be screened, and a 13% lower rate of prostate cancer death overall. The same trial found 30% more prostate cancers diagnosed in the screened group. Those two numbers are why guidelines disagree, and why the disagreement is not a sign that one side has misread the evidence.

The trial the argument rests on

The European Randomized Study of Screening for Prostate Cancer began in 1993 across eight European countries. Its predefined core age group is 162,236 men who were 55 to 69 at randomisation, split between a screening group offered repeated PSA testing and a control group not invited [s1]. The primary outcome was prostate cancer mortality [s1].

At a median 23 years of follow-up, prostate cancer mortality was 13% lower in the screening group (rate ratio 0.87, 95% CI 0.80 to 0.95), an absolute risk reduction of 0.22% (95% CI 0.10 to 0.34) [s1]. Cumulative prostate cancer incidence was higher in the screening group (rate ratio 1.30, 95% CI 1.26 to 1.33) [s1].

The trial reports both the number needed to invite and the number needed to diagnose, and both improved with time. At 23 years, one prostate cancer death was prevented for every 456 men invited (95% CI 306 to 943) and for every 12 men diagnosed (95% CI 8 to 26) [s1]. At 16 years, the equivalent figures were 628 men invited (95% CI 419 to 1,481) and 18 diagnosed (95% CI 12 to 45) [s1]. The trial's conclusion is that longer follow-up confirms a sustained mortality reduction alongside an improved harm-benefit ratio, and that future strategies should be risk-based to reduce overdiagnosis [s1].

What a systematic review makes of the whole field

A Cochrane review published in May 2026 updated the 2013 version, which had concluded that screening does not significantly reduce disease-specific or overall mortality [s2]. The update includes six randomised trials totalling 789,086 participants, men aged 45 to 80, conducted in Europe and North America [s2]. Three trials — CAP in the UK, PLCO in the US, and ERSPC — account for 85% of participants [s2]. Follow-up ranged from 3.2 to 23 years [s2].

The review's headline conclusions are graded rather than absolute. Screening likely reduces prostate cancer-specific mortality, at moderate-certainty evidence, derived mainly from a sensitivity analysis restricted by risk of bias to ERSPC (rate ratio 0.87, 95% CI 0.80 to 0.95) [s2]. Against an assumed baseline of 16 prostate cancer deaths per 1,000, that is 2 fewer per 1,000 (95% CI 3 fewer to 1 fewer) [s2].

On overall mortality — the outcome that captures whether screening extends life rather than reallocating the cause of death — screening may reduce it, but the confidence interval includes the possibility of little to no effect (rate ratio 0.99, 95% CI 0.97 to 1.00, four studies, 675,121 participants, low-certainty evidence) [s2].

The diagnosis figures show the trade. Screening likely increases prostate cancer diagnoses (rate ratio 1.30, 95% CI 1.27 to 1.34) and localised stage I and II disease (RR 1.53, 95% CI 1.48 to 1.59), both moderate certainty [s2]. It likely has little to no effect on advanced stage III and IV diagnosis (RR 0.90, 95% CI 0.85 to 0.95) and may reduce metastatic diagnosis (RR 0.65, 95% CI 0.59 to 0.71) [s2].

The most consequential sentence in the review is about what is missing: the authors report insufficient evidence on the potential harms of screening, such as biopsy- and treatment-related complications [s2]. On quality of life, one study of 969 participants found essentially no difference (0.777 screened against 0.779 control, P = 0.88, low certainty) [s2]. And the review adds an unusual caveat — interpretation of its findings is highly sensitive to the choice of minimal clinically important difference, the threshold the reviewers set for what counts as meaningful [s2].

Why two guidelines read the same trials differently

The US Preventive Services Task Force's 2018 recommendation gives men aged 55 to 69 a C grade: the decision to undergo periodic PSA screening should be an individual one, made after a discussion of benefits and harms, and clinicians should not screen men who do not express a preference for it [s3]. For men 70 and older it recommends against PSA screening, a D grade [s3]. Its stated harms are false positives requiring biopsy, overdiagnosis and overtreatment, and treatment complications including incontinence and erectile dysfunction [s3]. That recommendation is currently listed on the Task Force's own site as being updated [s3].

The American Urological Association's guideline, amended in 2026, is more directive. It says clinicians should offer regular prostate cancer screening every two to four years to people aged 50 to 69, a strong recommendation at evidence grade A [s4]. It says clinicians may begin screening and offer a baseline PSA between ages 45 and 50, and should offer screening from age 40 to 45 for people at increased risk based on Black race, germline mutations or a strong family history [s4]. It also states that clinicians should engage in shared decision-making and proceed based on a person's values and preferences [s4].

Both documents read the same trials. The difference is what they do with an intervention whose mortality benefit is small, real and slow, and whose principal harm — diagnosing and treating cancers that would never have caused symptoms — falls on a different set of men than the benefit does. A body that weights population harm reaches a C; a body that weights an averted death reaches an A. Neither is misreading ERSPC.

What has changed since the trials were designed

Both the trial authors and the 2026 guideline point in the same direction: away from PSA alone. ERSPC's own conclusion calls for risk-based approaches to minimise overdiagnosis while maintaining benefit [s1]. The AUA guideline states that clinicians may use MRI before an initial biopsy to increase detection of the more aggressive Grade Group 2 or higher cancers, and that when the risk of clinically significant cancer is sufficiently low on available clinical, laboratory and imaging data, clinicians and patients may forgo near-term biopsy [s4].

The Cochrane review included one trial testing that newer approach — PSA combined with a kallikrein panel and MRI — but its results are preliminary, and prostate cancer-specific mortality, overall mortality, adverse events and quality of life were all unreported [s2]. Whether risk-based screening delivers the same mortality benefit with fewer diagnoses is, at present, an untested proposition.

This article describes what the trials and guidelines report. It is not medical advice, and it does not recommend for or against screening for any reader.

Sources

• European Study of Prostate Cancer Screening — 23-Year Follow-up — New England Journal of Medicine, 2025-10-29
• Prostate-specific antigen (PSA) test for prostate cancer screening — Cochrane Database of Systematic Reviews, 2026-05-15
• Prostate Cancer: Screening — Final Recommendation Statement — US Preventive Services Task Force, 2018-05-08
• Early Detection of Prostate Cancer: AUA/SUO Guideline (2026) — American Urological Association, 2026