St John's Wort

Ninety-six volunteers across eleven groups is about nine per arm, so a null in any one arm means no effect was detected at that size, not that none exists. Digoxin is handled almost entirely by the drug pump rather than by the liver enzyme, so these results speak to one half of the mechanism.

Forty-two men across six arms leaves about seven per arm, so each arm's estimate is imprecise. The preparations were picked to span the hyperforin range rather than to represent what sits on a shelf. Every arm showed some induction, including the near-hyperforin-free one, so this is a slope rather than a safe threshold.

The trials disagree with each other far more than a pooled number suggests, with heterogeneity running to 89 percent, and the reviewers downgraded every headline estimate because of it. Smaller trials reported roughly double the benefit of larger ones. Trials from German-speaking countries, where the extracts are manufactured, reported results more favourable than trials from elsewhere.

The large US trial most often cited as proof that it fails in severe depression enrolled people its own authors described as moderately severe, so it did not test the question either. No trial has been designed with severity as the thing being varied.

Trial size stands in for several things at once, including how carefully a trial was run and who paid for it, so this does not isolate a single cause. The 2026 estimate rests on three trials in 207 people, all of one proprietary extract.

The 35,562-patient figure comes from post-marketing surveillance never designed to detect interactions, which duly reported none. The review's searches stopped in 2003. Its conclusion is explicitly conditional on a doctor knowing the patient is taking it, which is the condition most over-the-counter use fails.

Every trial ran between 4 and 12 weeks, so nothing here describes a year on either treatment. These are comparisons against antidepressants rather than against placebo, so the equivalence inherits whatever each trial's drug arm was worth. None of the trials was built to catch rare side effects.

Eight healthy volunteers rather than people with HIV, and no clinical outcome was measured: resistance and treatment failure are inferences from the size of the drop, not events anyone observed here. A correction to the original report was published a year later and has never been reconciled with its figures.

The published abstract reports the pattern without the numbers, so this establishes that the variation is large without saying how large. The products were bought on the German market in 2001, and nothing here describes what is manufactured today.

The imatinib study used twelve healthy volunteers given single doses, not patients whose disease was being controlled. The irinotecan study had five patients and its 42 percent estimate runs anywhere from 14 to 70 percent. Neither measured whether any cancer did worse.

A guideline is an expert reading of the evidence with consensus in the loop, not a fresh analysis, and this one closes its evidence base in 2016. Its own authors name thin drug-interaction data as a caveat across the whole complementary-medicine section.

The label states the risk without giving it a size, and puts St John's Wort in the same list as rifampin, a far more powerful enzyme inducer, without ranking them. It covers one product; wording across other combined pills is similar but not identical, and progestogen-only methods are not covered at all.

Eight men, no placebo arm, each compared against himself before and after. The same experiment in rats produced increases of two to nearly four-fold, so the animal figures are not the human ones and must never stand in for them. Gut expression came from biopsy; the liver claim rests on a breath test.

The two-week figure comes from a single follow-up measurement in ten patients, so it is a lower bound on how long the effect lasts rather than a measured rate of decay. The half-life figure was produced and published by an extract manufacturer, and its abstract never states how many volunteers it came from.

This comes from a reference monograph summarising other people's work rather than from primary research, with references last updated in 2020. Which constituent produces the antidepressant effect has never been established, and that is a far less settled question than which one drives the drug interactions.

The monograph states this qualitatively and names no country on either side, so it establishes that prescription status varies internationally without documenting where. Its references were last updated in 2020.

Sensitivity was measured on small patches of back skin under standardised lamps, which is a laboratory endpoint rather than sunburn in ordinary life. The two-week arm had no placebo group and compared each person against himself. Blood levels of hypericin, the constituent products are standardised on, did not track photosensitivity at all.

Everyone randomised had already responded during a six-week run-in, so the result describes people it was working for rather than anyone starting out. The significance test was one-sided. Two of the five authors are connected to the company that makes the extract.

Twenty-five healthy volunteers split across two arms, single-blind, and none of them was taking digoxin for a heart condition where a quarter would come off a therapeutic target. Two measured timepoints establish the direction and rough timescale of onset rather than a curve.

Nobody rejected in this study, because the doses were actively raised as the levels fell; it shows the size of the correction needed, not what follows when no one makes it. Ten patients, open-label, no control group, with each person compared against himself.

Five of the eight authors work for the company that makes the extract and two more for the contract research organisation that ran the study. The design was open-label, non-randomised and single-sequence, in 20 people. The result belongs to this one proprietary extract, not to St John's Wort products in general, most of which carry far more hyperforin.

The six preparations were selected to span the hyperforin range, so the gap between them is wider than a random pair off a shelf would be. About seven men per arm. How far real products vary has been shown to be large but has never been given a number.

Both trials tested a single extract at a single dose, so neither says anything about preparations that differ in strength. The 340-person trial reports no trend toward benefit whatsoever, which an insensitive trial would not be guaranteed to produce. The smaller trial enrolled about a quarter fewer people than it planned to.

Two large US trials found no benefit where pooled trials find one, but in both the prescription antidepressant also failed to beat placebo. The open question is whether those trials could detect any effect at all.

Seventeen women, and although none ovulated, a study this small could not have detected a modest rise in ovulation if there were one. Bleeding was recorded by diary. The hyperforin content of the preparation was never reported, and that is the variable later work shows decides the size of the effect.

Whether St John's Wort lets ovulation break through a combined pill is unsettled. In both trials the pill's hormone levels fell and bleeding rose, but they split on ovulation, and none measured pregnancy.

Spontaneous reporting can establish that an interaction happens in practice but never how often, and it skews toward the severe and the newsworthy. The regulator's review closed in 2002, before the work showing that hyperforin content decides how strongly any given preparation induces anything.

Whether St John's Wort taken with serotonin-acting antidepressants or migraine triptans risks serotonin syndrome rests on case reports, which show it can happen but not how often. No controlled study has tested either pairing.

The review naming it does not state the size or even the direction of the effect for St John's Wort specifically, and its classification of 78 items leans heavily on case reports of varying quality. Warfarin is monitored by routine blood test, so the practical risk is a result drifting between checks rather than a failure nobody sees.