Finasteride: the fourth item on this site to carry 'strong evidence' — and one of its three badges is a harm

Two trials in J Am Acad Dermatol, 1998, and a Japanese trial in Eur J Dermatol, 2004.

First, 1998.

In two 1-year trials, 1,553 men aged 18 to 41 with male pattern hair loss received oral finasteride 1 mg/day or placebo. And 1,215 men continued in blinded extension studies for a second year.

Efficacy was evaluated four ways: scalp hair counts, patient assessment, investigator assessment, and review of photographs by an expert panel.

The results.

• finasteride improved scalp hair by all evaluation techniques at 1 and 2 years (P < .001 vs placebo, all comparisons)
• hair count — measured in a 1-inch diameter circular area (5.1 cm²) of balding vertex scalp, baseline 876 hairs. Increases of 107 hairs at 1 year and 138 at 2 years versus placebo (P < .001)
• treatment with placebo resulted in progressive hair loss
• patients' self-assessment showed slowed loss, increased growth and improved appearance, and these were corroborated by investigator and photographic assessments
• adverse effects were reported as minimal

Next, the 2004 Japanese trial.

414 Japanese men with male pattern hair loss were randomised, double-blind, into three arms — finasteride 1 mg (n = 139), finasteride 0.2 mg (n = 137), placebo (n = 38) — for 48 weeks.

• all efficacy endpoints showed significant improvement by 12 weeks (p < 0.05 versus placebo)
• at 48 weeks, improved on global photographic assessment — 58%, 54% and 6% in the 1 mg, 0.2 mg and placebo groups
• at 48 weeks all endpoints were numerically superior for 1 mg over 0.2 mg

Now how to read it.

First, look at what the placebo groups did. In 1998, those on placebo went on losing hair. In 2004, 6% of the placebo group improved on photographic assessment.

That is what separates this entry from many others on this site. The problem this site keeps writing about — 'the placebo group improved too' — is absent here. Compare the pycnogenol page, where the placebo arm fell 34%. Hair loss gets worse if left alone, so here the placebo arm is closer to the problem of holding a baseline.

Second, the endpoints are good. Hairs were counted inside a defined area. And an expert panel looked at photographs. This is not a trial with only self-assessment. This site does not grant strong evidence on self-assessment alone.

Third, the numbers and the duration are among the largest on this site. 1,553 plus 414 people, with 1,215 continuing into a second year. Not a scale comparable to the cosmetic-trial standard of 20 to 40 people over 8 to 12 weeks.

Fourth, the funding. The first author of the 1998 paper is at Merck Research Laboratories. This site marks that as industry funding. The 2004 Japanese trial lists Tokyo Women's Medical University, so we left it 'unclear'. Had both been industry-funded, this badge would have stopped in the middle. That is what this site's condition says.

Fifth, read the number as written. At two years it is 138 hairs versus placebo, inside a circle whose baseline was 876. It does not mean hair returns to how it was. What the trial showed is that progression slowed and some came back.

Sixth, two years. Material beyond that is not in this entry. And this drug reverses when stopped. That is under cautions below.

Seventh, age. The 1998 trials were in men aged 18 to 41. Results for older men cannot be taken from them.

So it sits at 'strong evidence'. The fourth on this site.

A trial in J Am Acad Dermatol, 2000. Same drug, same 1 mg, same year as the entry above, and the result disappeared.

137 postmenopausal women aged 41 to 60 with androgenetic alopecia received finasteride 1 mg/day or placebo. One year, double-blind, placebo-controlled, randomised, multicentre.

Efficacy was evaluated five ways: scalp hair counts, patient assessment, investigator assessment, assessment of global photographs by a blinded expert panel, and histologic analysis of scalp biopsy specimens.

The results.

• after one year of therapy there was no significant difference in the change in hair count between the finasteride and placebo groups
• both treatment groups had significant decreases in hair count in the frontal/parietal (anterior/mid) scalp during the year
• patient, investigator and photographic assessments, as well as scalp biopsy analysis, did not demonstrate any improvement in slowing thinning, increasing growth or improving appearance in the finasteride group
• finasteride was generally well tolerated

Conclusion: in postmenopausal women with androgenetic alopecia, finasteride 1 mg/day for 12 months did not increase hair growth or slow the progression of thinning.

Now how to read it. This is why the entry belongs on this page.

First, the same drug came apart by who took it. The entry above and this one use the same drug, the same dose, the same duration. The only thing that differed is who took it. And the results diverged completely.

This is what this site keeps writing on ingredient and medicine pages: 'this ingredient works' is usually an incomplete sentence. For what, for whom, for how long — it is not a sentence until those are attached.

Second, this is not 'not measured yet'. It was measured properly. 137 people, one year, double-blind, placebo-controlled, multicentre, and with biopsies. And no difference came out. It is the same kind of badge as the endermologie page — measured, and not there.

Third, both groups lost hair. In the frontal and mid scalp both decreased significantly over the year. The same direction as the placebo arm in the male trials above. Hair loss progresses if left alone. And in this trial the treated arm did not stop it either.

Fourth, here is what this trial did not answer. It looked at postmenopausal women. These are not results for premenopausal women, nor for women with raised androgens. Material on those is not on this page.

Fifth, 137 people. Far smaller than the male trials above (1,553). Whether finding no difference means there is none, or that 137 could not catch it, cannot be settled completely by this one trial. But all five assessments pointed the same way, and one of them was a biopsy.

Sixth, if you are looking at hair loss in women, the evidence is elsewhere. The female pattern entry on minoxidil carries 'strong evidence' on this site (381 people over 48 weeks, 404 over 24). Start from the hair loss page and the cause.

So it sits at 'not established'.

A systematic review and meta-analysis in Acta Derm Venereol, 2019, from the Institute of Hair and Cosmetic Medicine, Wonju College of Medicine, Yonsei University, Korea. Unlike the two trials above, this is not from the manufacturer's side.

First, why this entry is up here as a badge. This site usually badges efficacy only. But if a drug carries 'strong evidence' for what it does, the harms should be shown at the same size. Efficacy as a badge and harm as a line under cautions makes a screen that leans.

What was pooled.

Fifteen randomised double-blinded placebo-controlled trials, 4,495 subjects, of finasteride 1 mg/day or dutasteride 0.5 mg/day for male androgenetic alopecia.

The results.

• 5-alpha-reductase inhibitors overall — a 1.57-fold risk of sexual dysfunction (95% CI 1.19–2.08)
• finasteride — 1.66 (95% CI 1.20–2.30)
• dutasteride — 1.37 (95% CI 0.81–2.32)
• both drugs were associated with an increased risk, although the increase was not statistically significant for dutasteride. The authors wrote that studies into dutasteride were limited and further trials are required
• conclusion: it is important that physicians are aware of, and assess, the possibility of sexual dysfunction in patients treated with 5-alpha-reductase inhibitors

Now how to read it.

First, look at the confidence intervals. For finasteride it is 1.20–2.30. The lower end is above 1. That is, 'no increase in risk' is not inside that interval. For dutasteride it is 0.81–2.32, which contains 1 and is therefore not significant. This site writes those two cases as different sentences.

Second, what the 1.66 is 1.66 of. It is a relative risk. How many more people actually experience it depends on the underlying rate. The abstract does not give an absolute rate. So this site does not write 'so many in so many'. We do not invent it.

Third, the endpoint is what people said. Sexual dysfunction is reported by the person. This site marks that as 'self-reported', and an endpoint like that on its own does not raise a grade. So despite 15 trials and 4,495 people, this entry stops at 'moderate'.

One thing to add about this endpoint, though. For sexual dysfunction there is no method other than what the person says. When this site writes that self-assessment is a weak endpoint, it does not mean the figure matters less; it means expectation is hard to separate out. And the 15 trials pooled here were all placebo-controlled and double-blinded. The 1.66 came out under that condition.

Fourth, the funding. This meta-analysis came from a university institute, a different side from the efficacy trials above. This site records that: for this drug, the efficacy material comes from the manufacturer's side and the harm material from an independent team.

Fifth, what this meta-analysis does not cover. Symptoms reported as persisting after stopping the drug, and data on mood, are not in it. We could not carry that material on this page. It is under 'what is not established' below.

Sixth, what to do with this number is for you and your clinician. This site does not tell you to take it or not to. It writes how far the efficacy has been established and how far the harms have been, at the same size.

So it sits at 'moderate'.