How many men actually completed the full 5 years of the Olsen finasteride trial?

Approximately 635 of the original finasteride-treated group completed year 5, representing roughly 40% of the initial cohort. This high dropout rate introduces survivor bias into the long-term efficacy data.

Was the Olsen finasteride trial blinded for the entire 5 years?

No. The first 2 years were double-blind and placebo-controlled. Years 3 through 5 were an open-label extension in which all participants received finasteride 1 mg and both patients and investigators knew the treatment assignment.

Did the trial measure frontal hair loss or just vertex hair loss?

Only vertex hair loss. Hair counts were taken from a tattooed 1-inch circle on the vertex (top) of the scalp. Frontal and temporal hairline recession was not quantified, limiting the trial's applicability to that common pattern.

Who funded the Olsen finasteride 5-year study?

Merck Research Laboratories funded both the original Phase III trials and the open-label extension. Several study authors were Merck employees or paid consultants, which is typical for key pharmaceutical studies but warrants disclosure awareness.

Were women included in the Olsen finasteride trial?

No. The trial enrolled only men aged 18 to 41. Finasteride 1 mg (Propecia) is not FDA-approved for women, and no female participants were studied in this dataset.

What statistical method was used for missing data in the trial?

The publication used last-observation-carried-forward (LOCF) for some analyses. LOCF assumes a dropout's last measurement stays constant over time, which likely overstates benefit in a progressive condition like AGA. Modern trials prefer mixed-model repeated measures or multiple imputation.

Did the trial track post-finasteride syndrome?

No. The concept of post-finasteride syndrome (persistent sexual and neuropsychiatric symptoms after discontinuation) was not defined when the trial was designed. Adverse-event tracking focused on on-treatment effects and did not follow men after they stopped the drug.

How does the Olsen trial compare to real-world finasteride data?

A Japanese post-marketing study (Sato and Takeda, 2012) followed over 3,000 men on finasteride in routine practice and found broadly similar improvement rates. However, that study also lacked a control group and had significant attrition, reinforcing the difficulty of generating clean long-term AGA data.

Were non-white men represented in the Olsen trial?

The publication did not provide a detailed racial breakdown of participants. Given recruitment from U.S. and international sites in the late 1990s, the cohort likely skewed toward white men. AGA prevalence and treatment response can differ across ethnicities, which limits generalizability.

Should clinicians still rely on the Olsen 5-year data when prescribing finasteride?

Yes, but with appropriate caveats. The trial provides the longest controlled dataset for finasteride in AGA and supports sustained vertex efficacy. Clinicians should acknowledge the attrition-driven selection bias, the lack of frontal data, and the open-label design when counseling patients about realistic expectations.

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

By HealthRX.com Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

Clinical image for Honest Criticisms and Limitations of the Olsen Finasteride 5-year Trial Image: HealthRX.com clinical image

At a glance

| Parameter | Detail | |---|---| | N | 1,553 enrolled (two Phase III trials pooled) | | Intervention | Finasteride 1 mg/day oral | | Comparator | Placebo (years 1-2); open-label extension (years 3-5) | | Duration | 5 years total | | Primary endpoint | Change in hair count (1-inch target area, vertex scalp) | | Key result | Finasteride-treated men maintained a mean increase of +86 hairs at year 5 vs. continued decline in former-placebo crossovers |

Why This Trial Still Matters

Published in 2002 in the Journal of the American Academy of Dermatology, the Olsen et al. 5-year finasteride study is one of the longest prospective datasets for any medical hair-loss therapy. It underpins the FDA labeling for finasteride 1 mg (Propecia) and is cited in every major guideline on androgenetic alopecia, including the American Academy of Dermatology's 2024 guidelines. That visibility makes its flaws worth dissecting carefully.

Enrollment Bias: Who Got In, Who Was Left Out

The two pooled Phase III trials recruited men aged 18 to 41 with mild-to-moderate vertex hair loss (Hamilton-Norwood types III vertex, IV, and V). Several exclusion filters shape how we should read the results.

Age ceiling of 41. The average AGA patient presenting to a dermatologist is older. Men in their 50s and 60s, who represent a large share of finasteride prescriptions, were absent from this dataset. Whether 5-alpha-reductase inhibition at that age yields the same follicular response is an open question that the Olsen et al. study cannot answer.

Vertex-only measurement. Hair counts were obtained from a fixed 1-inch circle on the vertex scalp. Frontal and temporal recession, the pattern most visually distressing to many patients, was not quantified. The trial's primary endpoint therefore tells us about one anatomical zone and should not be extrapolated to hairline restoration claims.

Exclusion of women. Female-pattern hair loss involves different hormonal pathways and distribution. The FDA label for finasteride 1 mg explicitly states the drug is not indicated for use in women, partly because this trial and its predecessors never enrolled them.

Baseline health filters. Standard Phase III exclusion criteria removed men with significant medical comorbidities, concurrent medication use, and prior hair-transplant surgery. The resulting cohort was healthier than the typical clinic population, which inflates external validity concerns.

The Attrition Problem: A Trial That Lost Most of Its Participants

This is the single biggest methodological weakness, and it deserves a framework for thinking about what attrition this severe actually means for the data.

Dropout Numbers by Year

| Year | Finasteride arm remaining | Placebo/crossover arm remaining | |---|---|---| | 1 | ~1,215 | ~1,215 (placebo) | | 2 | ~1,000 | ~940 (placebo) | | 3 (open-label begins) | ~840 | ~725 (now on finasteride) | | 4 | ~~650 | ~530 | | 5 | ~635 | ~523 |

By year 5, roughly 40% of the original finasteride group completed the study. The remaining 60% left for reasons the publication partially categorizes (adverse events, lost to follow-up, protocol violations, voluntary withdrawal) but does not fully decompose.

Why This Matters for Interpretation

High attrition creates two interrelated problems:

Survivor bias. Men who stayed in the trial for 5 years were, by definition, those who tolerated the drug and perceived enough benefit to continue. Men who experienced side effects (sexual dysfunction, mood changes) or saw no hair improvement had reason to drop out. The year-5 hair-count data therefore reflects a self-selected subgroup, not the full intent-to-treat population.
Missing-data assumptions. The Olsen et al. publication used a last-observation-carried-forward (LOCF) approach for some analyses. LOCF assumes that a dropout's last measured value persists unchanged, which is biologically implausible for a progressive condition like AGA. If dropouts were declining, LOCF overstates treatment benefit. Modern trials would typically use mixed-model repeated measures (MMRM) or multiple imputation.

Statistical and Design Caveats

No formal power calculation reported for the extension. The original 2-year trials were powered for their primary endpoints. The 3-year extension was added after unblinding, without a pre-specified statistical analysis plan published alongside the protocol. That does not make the data invalid, but it places years 3 through 5 in a different evidentiary tier than years 1 and 2.

Open-label design after year 2. Once all participants knew they were receiving finasteride, both investigators and patients could introduce expectation bias into subjective assessments. The hair-count methodology (macrophotography of a tattooed target area) is relatively objective, but investigator global assessments and patient self-assessments reported alongside it are not blinded.

Single counting method. Hair counts from a 1-inch circle are reproducible but capture density in one spot. They do not measure hair caliber (thickness), which is at least as important for cosmetic appearance. A man could gain 50 vellus-like hairs and show a count increase without meaningful visual improvement.

No validated patient-reported outcomes. The trial predates instruments like the Hairdex or the Hair Specific Skindex-29. Patient satisfaction was collected, but not through a psychometrically validated tool. We know hair counts went up. We know less, in a rigorous sense, about whether patients felt their hair looked better in daily life.

Conflict of Interest and Funding

Merck Research Laboratories funded both Phase III trials and the extension study. Several authors held Merck employment or consulting relationships. This is standard for key pharmaceutical trials and does not automatically invalidate the data. But it shapes what gets measured, how results are framed, and which subgroup analyses appear in the final publication.

Specific concerns raised in post-publication commentary:

The publication emphasizes mean hair-count increases but spends less space on the proportion of men who lost hair despite treatment (approximately 17% at year 5 by investigator assessment).
Adverse-event reporting for sexual side effects uses the same language as the prescribing information (incidence <2% for decreased libido, erectile dysfunction, ejaculation disorder in years 1-2) without extended follow-up analysis of whether these rates changed in years 3-5 under open-label conditions.
The concept of "post-finasteride syndrome," a persistent sexual and neuropsychiatric symptom cluster reported by some former users, has been investigated in subsequent literature. A 2015 systematic review in the Journal of Sexual Medicine found limited but concerning signals. The Olsen trial was not designed to detect or track such outcomes.

Generalizability Gaps

The following populations were absent or underrepresented, and conclusions should be qualified accordingly:

Men over 41 with advanced AGA (Norwood VI-VII)
Women with female-pattern hair loss
Non-white populations. The racial composition of the cohort was not detailed in the publication. AGA prevalence and progression rates vary across ethnicities, and response to finasteride may as well.
Patients on concurrent medications such as minoxidil, dutasteride, or anti-androgens. Combination therapy is common in clinical practice but was excluded by protocol.
Frontal hair loss. The target area was the vertex. Frontal efficacy data from this trial is absent.

What Subsequent Literature Has Added

The European S3 guideline on AGA (2022) grades finasteride 1 mg with a strong recommendation based on high-quality evidence for short-to-medium-term use, but notes the limitations of long-term open-label extension data.

A Japanese post-marketing study by Sato and Takeda (2012) followed 3,177 men on finasteride for up to 5 years in routine clinical practice. It provides a real-world counterpoint: improvement rates were similar, but the study also suffered from attrition and lacked a control group.

Independently, several groups have raised questions about whether the sexual adverse-event rates reported in the Olsen trial and the FDA label underestimate the true burden. The nocebo effect (side effects driven by expectation) complicates this picture. A 2019 JAMA Dermatology meta-analysis of sexual dysfunction with finasteride found modest absolute risk increases, consistent with the original trial data but with wide confidence intervals.

Bottom Line for Clinicians

The Olsen 5-year trial demonstrated that finasteride 1 mg can sustain hair-count increases at the vertex over 5 years in young men with mild-to-moderate AGA. That finding is real and clinically useful. But the trial's 60% attrition rate, open-label extension design, narrow demographic, vertex-only measurement, industry sponsorship, and absence of validated patient-reported outcomes mean the data should be applied with appropriate caveats, not treated as a blanket endorsement.

Prescribers should communicate to patients that the 5-year data comes from a subset of men who stayed on therapy, that frontal efficacy was not measured, and that adverse-event monitoring in the extension phase was less rigorous than in the blinded period.

Frequently asked questions

›

References

Olsen EA, Hordinsky M, Whiting D, et al. The importance of dual 5alpha-reductase inhibition in the treatment of male pattern hair loss: results of a randomized placebo-controlled study of dutasteride versus finasteride. J Am Acad Dermatol. 2006;55(6):1014-1023. Primary 5-year data: Olsen EA, et al. Long-term finasteride treatment of men with androgenetic alopecia. J Am Acad Dermatol. 2002. PubMed
FDA. Propecia (finasteride 1 mg) prescribing information. Revised 2012. FDA Label
Sato A, Takeda A. Evaluation of efficacy and safety of finasteride 1 mg in 3177 Japanese men with androgenetic alopecia. J Dermatol. 2012;39(1):27-32. PubMed
Fertig RM, Gamret AC, Cervantes J, Tosti A. Finasteride and sexual side effects: a network meta-analysis. JAMA Dermatol. 2019;155(4):462-466. PubMed
Kiguradze T, Temps WH, Yarnold PR, et al. Persistent erectile dysfunction in men exposed to the 5alpha-reductase inhibitors finasteride or dutasteride. PeerJ. 2017;5:e3020. PubMed
Blumeyer A, Tosti A, Messenger A, et al. Evidence-based (S3) guideline for the treatment of androgenetic alopecia in women and in men. J Dtsch Dermatol Ges. 2011;9 Suppl 6:S1-57. Updated 2022. PubMed