Honest Criticisms and Limitations of the Olsen Finasteride 5-year Trial

What the five-year study actually was

The 5-year report appeared in 2002 in the European Journal of Dermatology, credited to the Finasteride Male Pattern Hair Loss Study Group [1]. It followed two 1-year Phase III trials that randomized 1,553 men with male pattern hair loss to finasteride 1 mg a day or placebo; 1,215 of those men then continued into up to four 1-year, placebo-controlled extensions [1].

The 1 mg label describes the same program: three double-blind, placebo-controlled 12-month studies in 1,879 men. Two of them enrolled 1,553 men with mainly mild to moderate vertex hair loss, and 1,215 of these men elected to continue into double-blind, placebo-controlled 12-month extensions [2]. E.A. Olsen is an author of the group's 1998 report of the 1- and 2-year results in the Journal of the American Academy of Dermatology, which is where the commonly quoted hair counts come from [3].

Limitations that hold up

The extension years are a different tier of evidence

After the first year, 1,215 of the 1,553 randomized men entered the extension studies, and the label says these men elected to continue [1][2]. The later years therefore reflect the men who stayed in the program rather than everyone originally randomized.

The population was narrow

Enrolled men were 18 to 41 with mild to moderate, but not complete, hair loss, and 88% were Caucasian [2]. When researchers applied the pivotal-trial entry criteria to 5,704 men actually prescribed finasteride in practice, only 31% would have qualified [4]. Older men and men with advanced loss were not enrolled [2].

The endpoint is mostly vertex hair count

In the two trials reported in 1998, hair was counted in a 1-inch diameter circle (5.1 cm2) of balding vertex scalp, with a baseline of 876 hairs; finasteride was 107 hairs above placebo at 1 year and 138 above at 2 years (P<0.001) [3]. The 5-year report also used patient self-assessment, investigator assessment and expert panel review of photographs [1]. A separate study described in the label enrolled men with mild to moderate anterior mid-scalp loss [2]. The label states that efficacy in bitemporal recession has not been established [2].

Adverse-event reporting in this literature is weak

A review of 34 finasteride trials for androgenetic alopecia found none with adequate safety reporting, 19 partly adequate and 12 inadequate, with three reporting no adverse events at all; funnel plots pointed to under-detection of sexual adverse effects [4]. The same review found that 76% evaluated safety for a year or less, and 18 of 34 (53%) disclosed conflicts of interest [4].

Sexual side effects: small in-year rates, uncertain long term

In the 12-month controlled trials, drug-related decreased libido was 1.8% versus 1.3% on placebo, erectile dysfunction 1.3% versus 0.7%, and ejaculation disorder 1.2% versus 0.7%, in 945 and 934 men respectively; 3.8% versus 2.1% reported one or more (p=0.04), and each rate fell to 0.3% or lower by year 5 [2]. A 2019 meta-analysis of 15 placebo-controlled trials (4,495 men) put the relative risk of sexual dysfunction with finasteride at 1.66 (95% CI 1.20 to 2.30) [5]. The Propecia label's postmarketing section separately lists sexual dysfunction that continued after stopping, as well as depression and suicidal ideation. The label states that these voluntary reports come from a population of uncertain size and that a causal relationship cannot reliably be established [2].

Data on suicide and depression come from other studies. Among 93,197 Ontario men aged 66 or older starting a 5-alpha reductase inhibitor, mostly for prostate enlargement, the suicide hazard ratio was 0.88 (95% CI 0.53 to 1.45). The depression hazard ratio was 1.94 (95% CI 1.73 to 2.16) in the first 18 months [7]. These were older men, not young men treated for hair loss.

Points that are easy to misread

• Blinding of the extensions. The label describes the first 12-month extension as double-blind and placebo-controlled, and the 5-year report describes up to four placebo-controlled extensions, so the extensions were not open-label [2][1].
• Patient-reported outcomes. Patient self-assessment was one of the reported endpoints [1].
• Year-by-year numbers. The 5-year report summary gives the 1,553 and 1,215 figures but no count for each later year, so later-year retention cannot be stated here [1].
• Where the hair counts come from. The quoted hair counts of 107 and 138 hairs above placebo are 1- and 2-year figures from the 1998 report [3]. The 5-year report summary describes durable improvement across all endpoints [1].

What the five-year data still support

At five years, finasteride produced durable improvements in scalp hair across all endpoints versus placebo (P<0.001), while placebo-treated men kept losing hair [1]. The labeled schedule is one 1 mg tablet daily, with three months or more usually needed before benefit is seen. The label says continued use is needed to sustain benefit and that withdrawal leads to reversal of effect within 12 months [2].

Propecia has no boxed warning; depression, suicidal ideation and persistent sexual dysfunction appear in the postmarketing section with the causality caveat attached [2].

Bottom line

The 5-year report comes from two 1-year placebo-controlled trials in young men, followed by placebo-controlled extensions. It supports a durable hair-count benefit and reports no new safety concerns with long-term use [1]. It does not establish efficacy for bitemporal recession [2] or provide evidence for men older than 41. Persistent sexual dysfunction is addressed in the label's postmarketing reports, while the label also reports declining trial rates through year 5 [2].