Men on finasteride 1mg for hair loss were prescribed PDE5 inhibitors at 13.6% vs 10.5% — and 238 days sooner
Journal: World Journal of Urology | Published: 2026-06-08 | Type: Retrospective matched-cohort database analysis | PMID: 42258011 Authors: Xu Perry, Patel Amir, Krambeck Amy (Northwestern University Feinberg School of Medicine, Chicago); Roane Ariel, Gibbs Cora, Mechlovich Danit, Bucko Remi, Stock Laura (Boston Scientific, Minnetonka, MN) Funding/COI: Not listed — but five of eight authors are employed by Boston Scientific, a manufacturer of penile prostheses and urology devices. This is not disclosed as a conflict in the abstract.
Using the Truveta EHR database, researchers matched 3,470 men on long-term finasteride 1mg for hair loss against 3,470 men with androgenic alopecia who never took the drug. Men on finasteride were more likely to be prescribed PDE5 inhibitors (sildenafil, tadalafil, etc.) and received those prescriptions nearly eight months sooner — yet their formal ED diagnosis rates were statistically identical. Finasteride users also showed lower rates of elevated PSA and prostate cancer diagnoses, consistent with the drug's known PSA-suppressing effect.
This is a retrospective matched-cohort study using a large commercial EHR database (Truveta), which aggregates real-world clinical data from multiple health systems. The matched design — pairing on age, race, ethnicity, and marital status — controls for several obvious confounders, and the 6,940-patient total is a meaningful sample for a question that has essentially no prior literature at the 1mg dose. Exclusions of patients with prior ED, BPH, antidepressant use, or prostate cancer diagnoses are appropriate and help isolate finasteride's effect.
There is a structural problem worth flagging: median follow-up was 2,554.7 days for the finasteride group and 3,129.5 days for controls — a gap of nearly 575 days (roughly 19 months). The finasteride group had substantially less observation time, yet still accumulated more PDE5i prescriptions. That makes the PDE5i finding more robust, not less. But the same asymmetry runs in reverse for prostate cancer: the control group's higher PC diagnosis rate may partly reflect ~19 more months of follow-up during which cancers could accumulate. The authors used restricted mean survival time analysis to partially address this, but the follow-up imbalance is not foregrounded in the discussion and deserves more scrutiny than it receives.
The central paradox — more PDE5i prescriptions but identical ED diagnosis rates — is the most methodologically interesting result. It suggests EHR ICD coding systematically undercaptures ED as a formal diagnosis (patients and physicians may avoid documenting it) while pharmacy records catch downstream treatment. This is a real signal and a genuine contribution to how database researchers should think about measuring sexual dysfunction.
This paper is asking a real question — what happens urologically to men who take finasteride 1mg for years — and it finds something genuinely worth knowing: they end up on ED medications significantly more often and significantly sooner than matched men who don't take it. That's a clinically relevant signal even if the mechanism is unclear and even if formal ED diagnosis rates don't track it. The methodology is reasonable for a database study. But the Boston Scientific authorship — five of eight co-authors — combined with zero COI disclosure is a serious credibility problem that the journal should have caught. A device company with a commercial stake in ED treatment progression publishing a finding that finasteride users take ED drugs eight months earlier than their peers should have disclosed that relationship on the title page. Read the finding; question whose interests funded the framing.