The biological rationale behind these safety concerns begins with the drugs’ core mechanism. Finasteride and dutasteride block the 5-alpha-reductase enzyme, which converts testosterone into dihydrotestosterone, a more potent androgen. Because DHT acts not only in the prostate and hair follicles but also in the brain, where androgens influence mood, cognition, and behavior, researchers have long questioned whether suppressing this hormone could have neuropsychiatric consequences. Finasteride inhibits primarily the type II isoenzyme and, to a lesser extent, type III, lowering serum and tissue DHT by roughly 65 to 70 percent, while dutasteride blocks both type I and type II isoenzymes and achieves a far more pronounced suppression of 90 to 94 percent. This pharmacological difference has fueled speculation that dutasteride might carry a different psychiatric risk profile than finasteride, although the current evidence does not clearly separate the two drugs in this regard.

The regulatory backdrop gives the new analysis particular weight. The Pharmacovigilance Risk Assessment Committee of the European Medicines Agency reviewed the available data on finasteride tablets across the European Union and concluded that suicidal thoughts should be recognized as an adverse effect of 5-alpha-reductase inhibitors. Measures intended to lower the risk of suicidal ideation associated with both finasteride and dutasteride were subsequently endorsed by the Coordination Group for Mutual Recognition and Decentralised Procedures – Human in June 2025. That decision rested largely on spontaneous adverse event reports and mechanistic plausibility rather than on definitive epidemiological evidence, which is precisely the gap that a rigorous synthesis of observational studies can help fill. Spontaneous reporting systems, while invaluable for detecting early signals, cannot by themselves establish whether a drug actually raises the incidence of an event, because they lack a defined denominator and are vulnerable to reporting biases, media attention, and organized reporting campaigns.

Against this backdrop, the meta-analysis pooled data from observational studies conducted in real-world clinical settings, comparing men who used 5-alpha-reductase inhibitors with those who did not, or with men treated with alpha-blockers, an alternative class of drugs for benign prostatic hyperplasia that works through an entirely different mechanism by relaxing smooth muscle in the prostate and bladder neck. The pooled estimate, expressed as a hazard ratio of 1.07 with a 95 percent confidence interval spanning 0.76 to 1.51, crosses the threshold of 1.0, meaning the data are compatible with both a modest increase and a modest decrease in risk. In practical terms, the confidence interval does not rule out a hazard ratio as high as 1.51, which would represent a meaningful elevation if real, nor does it confirm any elevation at all. This is a common feature of studies examining relatively rare outcomes such as suicide, where even large cohorts may accumulate few events, widening confidence intervals and limiting statistical power.

The substantial heterogeneity observed across the included studies, quantified by an I-squared value of 59 percent, adds another layer of interpretive caution. Heterogeneity of this magnitude indicates that the individual studies varied considerably in their estimated effects, which may reflect differences in study populations, comparator groups, follow-up duration, outcome definitions, or methods of ascertainment. For example, some studies may have captured formal psychiatric diagnoses recorded in health records, while others relied on prescription patterns, self-reported symptoms, or coded events such as intentional self-harm. Suicide itself is a rare event, and intentional self-harm encompasses a broader spectrum of behaviors, from ideation requiring medical attention to completed suicide, so differences in how these endpoints were operationalized could plausibly drive between-study variation. When heterogeneity is substantial, the pooled estimate should be understood as an average across diverse conditions rather than a single, universally applicable quantity.

The populations studied also deserve attention. The analysis included men treated for benign prostatic hyperplasia, a condition that predominantly affects older men, and men treated for androgenetic alopecia, who are typically much younger and often healthy. These two groups differ in baseline psychiatric risk in ways that could confound any apparent association. Older men with urinary symptoms may have comorbidities, sleep disturbance, and reduced quality of life that independently affect mood, while younger men seeking treatment for hair loss may experience psychological distress related to the condition itself, independent of any drug effect. Disentangling the contribution of the underlying condition from that of the medication is one of the central methodological challenges in this literature, and observational designs, however well conducted, cannot fully eliminate confounding by indication or by disease severity.

It is also worth considering the phenomenon sometimes described in the literature as post-finasteride syndrome, in which patients report persistent sexual, cognitive, and psychological symptoms that continue after discontinuation of the drug. The existence and mechanisms of this entity remain debated, and the observational studies included in the meta-analysis were not designed to adjudicate it. Nevertheless, patient-reported experiences have contributed substantially to the regulatory attention on this drug class, and the tension between spontaneous reports suggesting harm and epidemiological analyses finding no significant association illustrates the broader challenge of pharmacovigilance for medications with delayed, uncommon, or subjective adverse effects.

The authors of the meta-analysis registered their protocol prospectively in the PROSPERO registry and followed the PRISMA reporting guidelines, both of which strengthen the transparency and reproducibility of the work. Their search covered three major bibliographic databases with no time restrictions, and study eligibility was assessed independently by two reviewers with disagreements resolved by additional arbiters. These methodological safeguards reduce the risk of selection bias and undisclosed outcome switching, though they cannot compensate for the inherent limitations of the underlying studies, which were themselves observational and therefore susceptible to confounding, reverse causation, and misclassification.

For clinicians, the practical implications are nuanced. The findings do not provide grounds for dismissing patient concerns about mood changes during treatment, nor do they mandate discontinuation of an effective therapy for a common and progressive condition. Benign prostatic hyperplasia, if untreated, can lead to acute urinary retention, recurrent urinary tract infections, and renal impairment, so the therapeutic benefit of 5-alpha-reductase inhibitors in appropriately selected patients remains well established. What the evidence supports is a balanced conversation: patients should be informed that regulatory authorities have recognized suicidal ideation as a possible adverse effect, that large observational data have not confirmed a statistically significant increase in suicide or neuropsychiatric events at the population level, and that anyone experiencing persistent mood changes, anxiety, or thoughts of self-harm while taking these medications should seek prompt medical evaluation rather than simply stopping treatment on their own.

The authors themselves emphasize that continued real-world monitoring through active surveillance and prospective observational studies remains warranted despite the null finding. This stance reflects a mature view of drug safety, in which a single meta-analysis, however comprehensive, rarely closes a scientific question permanently. Future research could benefit from several refinements: larger cohorts with longer follow-up to accumulate more suicide events, standardized definitions of neuropsychiatric outcomes across databases, analyses that separate finasteride from dutasteride and distinguish doses used for prostatic disease from those used for hair loss, and designs that account for the psychiatric history of patients before treatment initiation. Prospective registries of patients starting these drugs, with systematic mood assessments at fixed intervals, would complement retrospective database studies and help clarify whether subgroups of patients, rather than the population as a whole, face elevated risk.

In the meantime, the new analysis offers a meaningful counterweight to the most alarming interpretations of the pharmacovigilance signal. It suggests that if a risk exists, it is either too small to detect across millions of records, too concentrated in an identifiable subgroup, or too entangled with the conditions being treated to attribute cleanly to the drug itself. For the millions of men taking finasteride or dutasteride, and for the clinicians who prescribe them, the message is one of measured reassurance paired with continued vigilance: the population-level evidence does not show a significant increase in suicide or neuropsychiatric harm, but the regulatory concerns, the biological plausibility, and the lived experiences of patients together justify keeping this safety question under active and careful observation.

Subject of Research: Association between 5-alpha-reductase inhibitors and the risk of suicide and neuropsychiatric symptoms

Article Title: 5-α-reductase inhibitors and the risk of suicide and neuropsychiatric symptoms: a meta-analysis of observational studies

Article References: 5-α-reductase inhibitors and the risk of suicide and neuropsychiatric symptoms: a meta-analysis of observational studies. (n.d.). https://doi.org/10.3389/fruro.2026.1881635

Image Credits: AI Generated

DOI: 10.3389/fruro.2026.1881635

Keywords: finasteride, dutasteride, 5-alpha-reductase inhibitors, suicide risk, neuropsychiatric adverse events, benign prostatic hyperplasia, androgenetic alopecia, meta-analysis, pharmacovigilance, neurosteroids, EMA, drug safety