Suicidal risk associated with finasteride versus dutasteride among men treated for benign prostatic hyperplasia: nationwide cohort study
Moussa Laanani, Alain Weill, Fabrice Jollant, Mahmoud Zureik, Rosemary Dray‐Spira
Scientific Reports2023
In plain terms
Researchers used France's national health insurance records to follow nearly 290,000 men aged 50 or over who started finasteride, or a similar drug called dutasteride, for an enlarged prostate. While the men were taking the drug (usually for only a few months), suicide deaths and hospital stays for self-harm were rare with both: about 1 in 1,300 men on finasteride and 1 in 1,600 on dutasteride. After allowing for differences between the groups, finasteride did not show a clearly higher risk overall. In men with a past mood disorder such as depression, finasteride was linked with more suicide deaths and serious self-harm, but this rests on small numbers (8 suicide deaths among men on finasteride and 10 in the larger group on dutasteride).
What it doesn’t show: It can't show whether either drug raises risk compared with taking neither, because it only compared the two drugs, and it stopped following men once they stopped or switched, so it says nothing about effects after stopping or about younger men taking the lower hair-loss dose.
Summary (paraphrased)
Using the French national health data system, investigators compared suicidal outcomes between 69,786 new finasteride users and 217,577 new dutasteride users aged 50+ treated for prostate enlargement (2012–2016), with inverse-probability weighting for psychiatric and medical confounders. Overall, finasteride was not associated with a statistically significant excess of suicide death or self-harm hospitalization versus dutasteride, although self-harm with intensive care admission was more frequent (13 vs 16 events; HR 2.60, 95% CI 1.25–5.38). In men with a history of mood disorders, finasteride was associated with more of the composite outcome (HR 1.64, 1.00–2.68), suicide death (HR 2.71, 1.07–6.91), violent self-harm (HR 3.11, 1.01–9.61) and self-harm with intensive care admission (HR 3.97, 1.26–12.5), each based on few events (25, 8, 6 and 7 on finasteride). Follow-up ended when men stopped or switched treatment (median treatment 66 vs 87 days), so the study did not address suicidal risk after discontinuation.
Evidence
Extracted from the full text; page numbers refer to the paper. The library’s own notes are labelled as such.
- Design
- Cohort studyNationwide retrospective cohort of new users in linked claims, hospital and death-registry data, comparing finasteride with dutasteride (active comparator); Cox models with inverse probability of treatment weighting (IPTW)
- Setting
- French National Health Data System (SNDS), linking outpatient claims, hospital discharges and the causes-of-death registry; restricted to the general health insurance scheme (76% of residents). Treatment starts 1 January 2012 to 30 June 2016, follow-up to 31 December 2016.
- Population
- Men aged 50 or older newly starting finasteride 5 mg or dutasteride 0.5 mg (alone or with an alpha-blocker), with no dispensing of either drug in 2011. Men aged 49 or younger were excluded to limit hair-loss use of the reimbursed 5 mg tablet. Median age 72.0 (finasteride) vs 71.1 (dutasteride) in the abstract and on p. 4; Table 1 gives 71.7 for dutasteride. Prior psychiatric history 28.4% vs 25.6%; prior mood disorder 12.4% (8,638) vs 10.8% (23,503).
- Size
- 287,363 · 69,786 finasteride and 217,577 dutasteride new users (31,344.9 and 110,329.4 person-years), from 279,332 and 554,773 men dispensed the drugs
- Exposure
- Finasteride 5 mg (ATC G04CB01); exposure days from tablets dispensed plus a 15-day grace period
- Compared with
- Dutasteride 0.5 mg new users (ATC G04CB02, G04CA52); no unexposed group
- Outcome
- Primary, the first of suicide death (ICD-10 X60–X84 as underlying cause on the death certificate) or hospitalisation for self-harm (same codes). Secondary, each separately, and severe self-harm (violent means, X66–X83, or intensive care admission). Weighted for age, start year, prostatic and psychiatric history and treatments, and Charlson comorbidities; analyses repeated by psychiatric history and with follow-up cut at 90 days.
- Follow-up
- On treatment only, from first dispensing to event, stopping or switching, death or 31 December 2016. Median treatment duration 66 days (IQR 29–182) for finasteride and 87 days (31–220) for dutasteride.
Key results
- Whole cohort, suicide death or self-harm hospitalisation: 52 events (1.66 per 1,000 person-years) on finasteride vs 133 (1.21) on dutasteride; HR 1.21 (95% CI 0.87–1.67). Rates recompute from the counts and person-years · pp. 1, 4–5; Table 2, p. 7
- Separately: suicide death 18 (0.57) vs 47 (0.43), HR 1.25 (0.72–2.16); self-harm hospitalisation 34 (1.08) vs 87 (0.79), HR 1.17 (0.79–1.75) · Table 2, p. 7
- Self-harm with intensive care admission, whole cohort: 13 (0.41) vs 16 (0.15), HR 2.60 (1.25–5.38). Self-harm by violent means: 11 vs 21, HR 1.75 (0.84–3.64) · p. 5; Table 3, p. 8
- Men with a history of mood disorders: composite 25 vs 46 events, HR 1.64 (1.00–2.68; p = 0.049); suicide death 8 vs 10, HR 2.71 (1.07–6.91); violent self-harm 6 vs 6, HR 3.11 (1.01–9.61); self-harm with intensive care 7 vs 5, HR 3.97 (1.26–12.5) · pp. 1, 5; Tables 2–3, pp. 7–8
- Men with no psychiatric history or prior self-harm: composite 12 vs 49 events, HR 0.78 (0.41–1.47); suicide death 5 vs 21, HR 0.77 (0.29–2.05) · Table 2, p. 7
- First 90 days only: composite 32 vs 64 events, HR 1.46 (0.95–2.25). In men with prior mood disorders, suicide death 3 vs 5, HR 4.66 (1.10–19.7), and self-harm with intensive care 4 vs 1, HR 11.4 (1.29–100.1) · Tables 2–3, pp. 7–8
Limitations the authors note
- Few events in the psychiatric subgroups; those results need caution and confirmation
- Psychiatric disorders are hard to identify in SNDS, so residual confounding by their presence and severity cannot be excluded
- Possible residual confounding from missing sociodemographic data
- Self-harm codes cannot separate suicide attempts from non-suicidal self-injury; only hospitalised acts are captured
- Hair loss may be more common in the finasteride group and could not be adjusted for; this would overestimate the finasteride risk
- Cannot distinguish a raised risk with finasteride from a protective effect of dutasteride
- Restricted to the general insurance scheme (excluding, e.g., farmers), reducing power
- Results cannot be directly generalised to 1 mg finasteride for hair loss
Also worth weighing (library’s note)
- Active-comparator design, so it shows only whether finasteride differs from dutasteride; an effect shared by both drugs would not appear
- Follow-up stopped at discontinuation or switch and treatment episodes were short (mean on-treatment follow-up, from person-years / n, about 0.45 vs 0.51 years), so it addresses risk during treatment, not after stopping
- Indication is not recorded in the database; prostate enlargement is inferred from age 50 or over and the 5 mg / 0.5 mg doses
- Tables 2–3 report 45 subgroup-by-outcome rows (44 estimable) without adjustment for multiple comparisons; 10 have p < 0.05, and the significant subgroup estimates rest on 3 to 25 events in the finasteride arm
- Outcomes cover only deaths and hospital-treated self-harm, not suicidal thoughts, depression or self-harm without admission
- Exposure is inferred from pharmacy dispensing, not confirmed intake
Funding: None ("This research received no funding"); the authors are from Epiphare (French medicines agency ANSM and national health insurance CNAM) and academic centres Interests: None declared
What it can support (library’s note): That, during treatment for an enlarged prostate in men aged 50 or over, finasteride 5 mg was not associated with more suicide deaths or self-harm admissions overall than dutasteride, with a signal in men with prior mood disorders that rests on few events. It cannot address risk compared with no treatment, risk after stopping, or the 1 mg hair-loss dose in younger men.
Why it’s in the corpus
The largest comparative safety study of finasteride versus dutasteride on suicidal outcomes — critical quantitative context for the corpus's neuropsychiatric thread and for the 2026 meta-analysis (LIT-A-018).
Related records
- Peer-reviewed paper
Post-finasteride syndrome: a surmountable challenge for clinicians
Traish, Abdulmaged M. · Fertility and Sterility · 113(1):21-50
Systematic evaluation of 250+ articles (1989–2019) on finasteride/dutasteride concludes PFS is a real constellation of persistent sexual, neurological, physical and mental effects in a susceptible subset, proposing endocrine disruption plus epigenetic…
PFS-0012020PFSCore corpus - Community accountPSSD Discord pickSelf-report · N=1
80% PFS recovery (PropeciaHelp member story)
PropeciaHelp forum member (Russia/Israel, 45 y.o.) · forum.propeciahelp.com
Anecdotal member story: after ~6 months of finasteride (1 mg/day) the poster developed complete loss of libido, severe insomnia, restless leg syndrome, and depression that persisted about 10 years; a 13-day Ayurveda/Panchakarma fast with castor oil gave…
DISC-017date not captured from thread pagePFSPSSD Discord picks - Peer-reviewed paper
Persistent erectile dysfunction in men exposed to the 5α-reductase inhibitors, finasteride, or dutasteride
Tina Kiguradze, William H. Temps, Paul R. Yarnold, John Cashy, Robert E. Brannigan, Beatrice Nardone, Giuseppe Micali, Dennis Paul West… · PeerJ
Using electronic medical records from Northwestern Medicine, the authors tested whether longer 5-alpha-reductase inhibitor exposure raises the risk of persistent erectile dysfunction (PED, lasting at least 90 days after stopping the drug). Among 11,909…
LIT-0022017PFSLiterature sweep · Oct 2026