Finasteride Real-World Evidence: What Registries and RWE Studies Actually Show

What counts as real-world evidence here

Real-world evidence is data generated outside controlled trials: national prescription registers, health-system medical records, insurance claims, and reports submitted to FDA's adverse event reporting system [8]. This page covers two register cohorts on psychiatric outcomes, one electronic-record analysis of persistent erectile dysfunction, and postmarketing reports described in the finasteride labeling [6][1][2][4].

The two register cohorts: depression and suicide

The Ontario cohort followed 93,197 men aged 66 or older who started a 5-alpha reductase inhibitor (finasteride or dutasteride) between 2003 and 2013, mostly for BPH, matched to an equal number of unexposed men. Suicide hazard ratio was 0.88 (95% CI 0.53 to 1.45). Self-harm was higher only in the first 18 months (HR 1.88, 1.34 to 2.64). New depression was diagnosed at HR 1.94 (1.73 to 2.16) in the first 18 months and 1.22 (1.08 to 1.37) afterward, an absolute excess of 17 self-harm and 237 depression cases per 100,000 patient-years [1].

The Swedish register covered 2,236,876 men aged 50 to 90 between 2005 and 2018. Of them, 70,645 started finasteride, at a median age of 73. Depression HR was 1.61 (1.48 to 1.75) and stayed constant over time. Suicide showed no significant association (finasteride HR 1.22, 0.99 to 1.49; dutasteride HR 0.98, 0.62 to 1.54). The dementia association faded over follow-up, which the authors said may partly reflect more dementia detection among men with BPH [2].

Both cohorts primarily involved older men treated for BPH, so they provide limited evidence about younger men taking finasteride for hair loss [1][2].

The Propecia label carries no boxed warning. Depression and suicidal ideation sit in the postmarketing section, which states that the reports are voluntary, come from a population of uncertain size, and cannot reliably establish cause [6][7]. The wording followed a May 31, 2022 FDA Safety Labeling Change order citing suicidal ideation and behavior, additional sexual dysfunction and psychoneurocognitive events; supplement S-030, approved 08/01/2022, made the change [7].

Persistent erectile dysfunction in a health-system record

One analysis searched a single academic health system's electronic records for men exposed to finasteride or dutasteride who developed erectile dysfunction beginning after exposure, lasting at least 90 days after stopping, and treated with a PDE5 inhibitor. Of 11,909 exposed men, 167 (1.4%) met that definition, with median persistence of 1,348 days (interquartile range 631.5 to 2,320.5). Of 530 men with new erectile dysfunction, 167 (31.5%) met the persistent definition [4]. The design had no untreated comparison group, and the outcome required both a recorded diagnosis and a prescription.

When the trial safety data were audited

Thirty-four published trial reports of finasteride for androgenetic alopecia were reviewed for safety reporting quality. None had adequate safety reporting: 19 were partially adequate, 12 inadequate, and 3 reported no adverse events. 76% evaluated safety for one year or less. Funnel plots were asymmetric, skewed toward lower odds ratios for sexual adverse effects, which suggested systematic under-detection. In a separate cohort of 5,704 men treated with finasteride 1.25 mg per day or less, only 31% would have qualified for the pivotal trials [5].

Sexual side-effect rates: label, meta-analyses, expectation

In three 12-month trials, 945 men on finasteride 1 mg were compared with 934 on placebo [6]:

• Decreased libido 1.8% vs 1.3% [6]
• Erectile dysfunction 1.3% vs 0.7% [6]
• Ejaculation disorder 1.2% vs 0.7% [6]
• One or more of these: 3.8% vs 2.1% (p=0.04) [6]
• Discontinuation for a drug-related sexual event: 1.2% vs 0.9% [6]

Each individual rate fell to 0.3% or lower by year five. A self-administered sexual function questionnaire at month 12 favored placebo in three of four domains: sexual interest, erections and perception of sexual problems [6].

A meta-analysis of 15 placebo-controlled trials in 4,495 men with androgenetic alopecia put the relative risk of sexual dysfunction at 1.57 (95% CI 1.19 to 2.08) for 5-alpha reductase inhibitors, 1.66 (1.20 to 2.30) for finasteride alone, and 1.37 (0.81 to 2.32, not significant) for dutasteride [9]. A second meta-analysis of 17 trials and 17,494 patients reported pooled relative risks of 2.56 (1.48 to 4.42) in BPH and 1.21 (0.85 to 1.72) in androgenetic alopecia, the latter not statistically significant [10].

Expectation can change reporting. In one trial, 120 men with BPH were randomized to finasteride 5 mg with or without counseling about sexual side effects; 107 completed. One or more sexual side effects were reported by 43.6% of counseled men versus 15.3% of uncounseled men (P=0.03) [11].

Hair-loss effectiveness: what the trials measured

Two 1-year trials enrolled 1,553 men aged 18 to 41, and 1,215 continued into blinded second-year extensions. Hair count in a 1-inch (5.1 cm2) circle of vertex scalp, from a baseline of 876 hairs, was 107 hairs higher than placebo at 1 year and 138 higher at 2 years (P<0.001). Men on placebo kept losing hair [3].

The Propecia label specifies one 1 mg tablet daily, with or without food, and states that daily use for three months or more is usually needed before benefit is seen, that continued use sustains the benefit, and that withdrawal leads to reversal of effect within 12 months. Efficacy in bitemporal recession has not been established, and the drug is not indicated for women [6]. HealthRX.com offers finasteride 1 mg by prescription.

What these datasets can and cannot tell you

The Ontario study matched men using measured health conditions, medication use, and health-care utilization, but both observational cohorts can still be affected by differences between users and nonusers [1][2]. They cannot rule out differences in how often men on finasteride see a doctor; the Swedish authors said more dementia detection among men with BPH may partly explain their dementia finding [2]. Voluntary reports to FDA carry no denominator and cannot establish causality, a limitation the label states in its own words [6]. None of these datasets predicts how much hair any individual man will keep.