高齢者医療における臨床医の行動変容：前立腺特異抗原スクリーニングのための適時アラート

Prostate-specific antigen (PSA) screening in older men may lead to overdiagnosis and overtreatment. This study quantified treatment effect heterogeneity of a behavioral science-informed clinical decision support (CDS) to reduce clinician-level PSA overtesting and explored clinician-level trends in alert triggering.

A secondary analysis of a pragmatic cluster-randomized trial was performed including 338 clinicians in 59 primary care practices. Clinics received point-of-care CDS plus brief education (intervention) or education alone (control). The primary outcome was clinician-level annual PSA overtesting rate-the proportion of male patients >75 years without prostate cancer receiving PSA testing in the prior 12 months, evaluated overall and across clinician subgroups. Monthly trends in alerts were also examined. Data were collected in 2019-2022 and analyzed in 2024-2025.

Over 18 months, intervention clinicians reduced PSA overtesting by 12.1 per 100 patients versus controls (95% CI=7.5, 16.7). Reductions were greater among female (15.6, 95% CI=9.4, 21.9) than male clinicians (6.9, 95% CI=2.2, 11.6; p-interaction=0.01) and for graduates after 1999 (17.4; 95% CI=9.5, 25.3) versus earlier graduates (7.8; 95% CI=3.4, 12.3; p-interaction=0.03). The proportion of intervention clinicians triggering alerts declined by -0.6% per month (95% CI=0.3%, 0.8%). In encounters with triggered alerts, PSA ordering increased slightly (0.56 per 100 patients per month, 95% CI=0.04, 1.1).

A behavioral science-informed CDS lowered clinician-level PSA overtesting, with greater effectiveness among female and more recently trained clinicians. Falling alert rates suggested clinician learning, whereas increasing order rates among persistent alerts called for additional strategies to engage less responsive clinicians.