Medications

Caveat on this rating: The declarations of interest read literally: "KH has previously acted as a temporary consultant for Pfizer (manufacturers of sildenafil). MT has been paid to lecture and received travel expenses from Bristol-Myers Squibb (manufacturers of buspirone) and Otsuka; his spouse is an employee of GlaxoSmithKline (manufacturers of bupropion)." Two of the six authors therefore have financial ties to the makers of the two drugs the review endorses, which is why independence is scored false despite the funding being public. The reviewers also warn that partial reporting of subscale results in the source trials could bias the effect estimates upward.

Caveat on this rating: A regulatory label change is a precautionary act on a safety signal, not proof of incidence, causation, or permanence. The evidence base the PRAC weighed was EudraVigilance reports, literature, social media, and manufacturer reviews. Clomipramine and vortioxetine were part of the same signal assessment but were explicitly not covered by the labelling recommendation. US FDA labelling does not carry equivalent wording.

Caveat on this rating: The authors state that including open-label studies and pooling across different sexual-function scales "could reduce the significance of our findings," so the 25.8-80.3% band is wide partly because the source studies were not uniform. No funding statement or conflict-of-interest disclosure was accessible for this paper.

Caveat on this rating: The high trust score reflects the review's method, funding, and journal, not the strength of the underlying trials: 398 men across 11 studies, several from the 1970s and 1980s, with waitlist controls and confidence intervals touching 1.0. The single trial claiming group therapy outperformed sildenafil alone (WMD -12.40 on the IIEF) had 20 participants and was run by one of the review's own authors, which is a direct conflict on the review's most eye-catching result.

Caveat on this rating: The high trust score reflects the quality of this meta-analysis as a piece of evidence synthesis, not the strength of the effect it found. The pooled benefit rests on a subjective, dichotomised outcome in six trials with demonstrated publication bias, and the authors' own conclusion is the hedged "valerian might improve sleep quality," followed by a call for better studies. Twenty years on, those studies have largely not materialised.

Caveat on this rating: This review reaches the opposite conclusion to Bent 2006 from overlapping trials, and the difference is instructive: Taibi weighted trial quality and recency, and found that the better and newer the study, the smaller the effect. That pattern is the signature of a treatment whose apparent benefit comes from weak methods.

Caveat on this rating: This is a rigorous review of an essentially empty evidence base, and it is now twenty years old. Absence of evidence is not evidence of absence — but people buying valerian for anxiety should know that a full Cochrane search found one 36-person pilot study, and it was negative.

Caveat on this rating: The abstract's conclusion — that "valerian could be a safe and effective herb to promote sleep" — is not supported by the review's own primary pooled result, whose confidence interval crosses zero. The whole-root subgroup finding that the authors lean on rests on four trials and a confidence interval that nearly touches zero, and it is a post-hoc subgroup analysis, the weakest kind of evidence. One co-author is affiliated with Heartwood Education, a UK herbal-medicine education organisation, so the review is not independent of the herbal-medicine field it evaluates.

Caveat on this rating: A WEAK GRADE recommendation against a treatment is not a finding that it is dangerous; it means the trials are too few, too small, and too inconsistent to justify recommending it. The guideline also notes that it downgraded evidence quality across the board because most insomnia drug trials are industry-funded and carry publication-bias risk. It is now nine years old.

Caveat on this rating: The funding statement reads: "This study was supported by an independent investigator-initiated grant from Pfizer Inc, New York, New York. Pfizer Inc provided sildenafil and matching placebo," with the role-of-sponsor statement adding that "Pfizer had no other role in the design and conduct of the study." Pfizer makes sildenafil, and several authors disclose financial relationships with it. The Cochrane review of this literature concluded that for women it remains uncertain whether sildenafil is more effective than placebo, and flagged that unpublished data could change the picture — a direct disagreement a reader should know about.

Caveat on this rating: The funding statement reads: "This study was supported by an independent investigator-initiated grant from Pfizer Inc, New York, NY." Pfizer makes sildenafil. Several authors also disclose research support, consulting fees, or speakers-bureau payments from Pfizer. The trial predates routine registration, so there is no pre-registered protocol to check the reported outcomes against.

Caveat on this rating: The review's own evidence tables record that a substantial share of the included trials were industry-funded, had a manufacturer author, or were supplied with the product by the manufacturer, and that funding was simply unreported for several others. The authors also note that adverse events were poorly reported in the source trials and that none of the studies was designed to detect rare harms — including the drug interactions that are St. John's wort's main danger.

Caveat on this rating: The paper carries a published erratum (Transl Psychiatry 2021;11(1):465) and a published critique (Transl Psychiatry 2022;12(1):298); neither is a retraction, but both belong alongside it. Dose and formulation subgroups drawn out of a meta-analysis of heterogeneous small trials are exploratory — the Cochrane review reached a similar overall effect size and judged it very low certainty and likely biased upward.

Caveat on this rating: Nine small trials totalling 317 people, published in 2010, and the benefit is a shift of roughly 40 minutes to an hour in when sleep starts — real, but modest, and it delivers no extra total sleep. Effect depends heavily on taking the dose at the right time relative to one's own melatonin onset; taken at the wrong hour it can push the clock the wrong way.

Caveat on this rating: The higher depression rate was a borderline result on one of two coprimary outcomes while the other (mood scores) showed nothing at all; the authors did not claim omega-3 causes depression, and this was a prevention trial in people without clinically relevant depressive symptoms at baseline, so it does not test omega-3 as a treatment for existing depression. Study capsules and matching placebo were donated by Pronova BioPharma, an industry in-kind contribution, though the trial itself was NIH-funded, the funders had no role, and the result was unfavourable to the product.

Caveat on this rating: A 39-person trial has limited power, and the authors discuss whether the frequent study visits and psychoeducation inflated the placebo response — so this is a failed replication rather than proof that NAC does nothing in children. It is included precisely because the positive adult trial is usually cited without it.

Caveat on this rating: All 14 recommendations in this guideline are WEAK, including the ones in favour of prescription hypnotics, and the task force explicitly notes that evidence quality is routinely downgraded because most hypnotic trials are industry-funded. So the guidance against melatonin sits in a field of uniformly thin evidence rather than reflecting a uniquely poor showing. Several task-force members disclosed industry consulting relationships and recused themselves from the corresponding drug recommendations.

Caveat on this rating: Entirely observational. Mostly cross-sectional studies, which cannot establish that low magnesium came before the depression rather than after it, and dietary magnesium intake is a proxy for overall diet quality, socioeconomic position, and physical activity. No supplement was tested in any of the pooled studies.

Caveat on this rating: Blinding was almost certainly compromised: 95.6% of methylene blue participants reported blue-green urine discoloration versus 21.1% on placebo. The subgroup finding that drives the headline result was predicted in advance but comes from splitting 42 people by post-training fear level, so it needs replication before it means much.

Caveat on this rating: A null result across a wide credible interval means "not proven to work", not "proven not to work" — the L-theanine arm of this network rests on few small trials, so a modest real effect cannot be ruled out. The analysis came from a traditional Chinese medicine research institute, which if anything would bias it toward finding herbal benefits, making the null finding for L-theanine harder to dismiss.

Caveat on this rating: With 34 participants the study is underpowered to detect small true effects, and the authors' own Bayesian analysis in the successfully blinded subgroup was inconclusive rather than clearly negative.

Caveat on this rating: All three review authors disclosed financial relationships with Schwabe, a hypericum manufacturer: one had received a research grant and conference speaking fees, and all three had accepted travel reimbursement. The review itself flags that country of origin and trial size both predicted the result, which the authors say "complicates the interpretation" — plain language for probable bias in the underlying literature. The review is from 2008 and has not been updated.

Caveat on this rating: The competing-interests statement records that the work was "partly funded by APIRE/Eli Lilly Psychiatric Research Fellowship" and "the AACAP/Eli Lilly Pilot Research Award" alongside NIH grants, so the funding is mixed rather than fully independent — though Eli Lilly markets no melatonin product, so there is no proponent conflict on this specific question. The pooled 7-minute and 8-minute effects are statistically significant and clinically trivial; readers who see melatonin described as "effective" are usually seeing this paper's conclusion divorced from its numbers.

Caveat on this rating: The authors' own stated limitation is the small sample: 37 people cannot rule out a modest real benefit, so this is an absence of evidence rather than firm evidence of absence. The 120 mg/day dose is also low relative to the 248 mg used in the Tarleton trial, which the authors themselves flag as a possible explanation.

Caveat on this rating: Open-label with a no-treatment control rather than a placebo. That single design choice makes the headline 6-point PHQ-9 improvement uninterpretable as a drug effect: placebo responses of that magnitude are routine in depression trials, and here nothing separated the magnesium effect from the effect of knowingly starting a treatment. The authors' conclusion that "magnesium is effective for mild-to-moderate depression" outruns what an unblinded design can show.

Caveat on this rating: Only four eligible studies existed, none of them powered to measure pregnancy rates, so the evidence establishes a mechanism and a warning signal rather than a quantified failure rate. Different St. John's wort products produced different degrees of interaction, which means a negative result for one brand does not clear the others.

Caveat on this rating: The abstract reports p<0.001 alongside pooled risk ratios whose 95% confidence intervals comfortably include 1.0, which is internally inconsistent; read the confidence intervals, which show equivalence rather than superiority. This meta-analysis also inherits the industry funding and country-of-origin biases present in the underlying trial literature, and it did not assess drug interactions at all.

Caveat on this rating: This is a failed trial, not simply a negative one: escitalopram, a drug with established efficacy, also failed to beat placebo, which means the study lacked assay sensitivity and cannot cleanly rule SAMe out. Two later papers re-analysed subsets of the same dataset and reported positive findings — a single-site subsample (Sarris 2014, J Affect Disord, PMID 24856557, n=144, SAMe superior to placebo) and a male-only subgroup (Sarris 2015, Pharmacopsychiatry, PMID 26011569, significant in 51 men, not in 62 women). Post-hoc subgroups from a failed trial are hypotheses, not results.

Caveat on this rating: The meta-analytic estimates rest on very little randomised data — two parallel RCTs (three comparisons, n=117) for efficacy and two RCTs (n=109) for adverse events — so the confidence intervals are wide and the review is better read as demonstrating absence of evidence than proving absence of effect. Several authors are affiliated with psychedelic research centres and one is chief medical officer of a psychedelics company (disclosed), so the null result is not coming from sceptics.

Caveat on this rating: Participants sourced and weighed their own material, so actual doses were unverified, and the sample was self-selected enthusiasts rather than people with a diagnosed condition. The authors are based at Imperial College's Centre for Psychedelic Research with a co-author from the Beckley Foundation, both proponents of psychedelic research — which cuts against, not toward, a bias explanation for this null result.

Caveat on this rating: The mineral analysis pooled iron, zinc, and magnesium together rather than testing magnesium alone, and the population was perinatal, so this is a null result in a specific setting rather than a definitive negative for magnesium in depression generally. The wide confidence interval reflects few trials.

Caveat on this rating: One trial, 50 people, one site, and a very large effect — the combination that most often fails to replicate, and in this case did fail to replicate in children four years later. No funding or disclosure statement could be read (the journal page returned HTTP 403 on 2026-08-18), so funding is UNKNOWN, conflicts false, and independence left null.

Caveat on this rating: A clean null on the primary outcome, with post-discontinuation secondary signals that the authors themselves flag as needing explanation. Study medications and placebos were supplied by commercial nutraceutical companies (Nutrition Care, BioCeuticals and Catalent, Australia) alongside public NHMRC and CRC funding, hence MIXED; the senior author's disclosed consultancy to Bioadvantex, a NAC commercialiser, is why independence is marked false. A published erratum (BMC Med 2019;17(1):35) applies.

Caveat on this rating: Uncontrolled and unblinded: participants knew they were dosing, sourced their own substances, and were self-selected enthusiasts, so nothing here can separate drug effect from expectation. The authors say as much and explicitly call for dose- controlled research.

Caveat on this rating: Two of the authors have direct commercial ties to SAMe: the disclosure statement records that Dr Bottiglieri chaired the advisory board of and holds stock options in Methylation Sciences Inc. and is a scientific advisor to Gnosis S.p.A., and Dr Gerbarg receives royalties from books covering SAMe. The review received no funding, and it reports the mania risk plainly, but it is not written by parties independent of the product. It also pools trials of widely varying quality, dose, and route without a meta-analysis, so "promising but limited" is a judgement rather than an effect size.

Caveat on this rating: The comparator was not inert: the "placebo" arm was 15 mg/day of methylene blue, chosen to preserve blinding because the drug discolours urine. Since a 1987 trial by Naylor reported that 15 mg/day itself acted as an antidepressant, the true drug-versus- nothing effect could be either larger or smaller than the reported difference.

Caveat on this rating: The review did not pool the trials or grade certainty of evidence, so "may assist" is a narrative judgement rather than a measured effect; none of the nine trials studied people with a diagnosed anxiety disorder, and several were manufacturer-funded.

Caveat on this rating: 73 participants across two arms is small enough that a handful of responders moves the result, and the trial has not been independently replicated in the sixteen years since — the Cochrane review, published six years later, still cites it as a single study. No funding statement or disclosure statement was read, so both funding and conflicts are recorded conservatively.

Caveat on this rating: This is a systematic review of case reports, not of trials. Case reports capture only the people who reached a hospital and were written up, so they show what phenibut can do but cannot tell you how often it does it.

Caveat on this rating: ISNPR is a society organised around nutritional psychiatry, so it is the proponent of the intervention it is writing guidelines for, and the panel includes investigators whose own omega-3 trials form part of the evidence base being summarised. Two letters challenging the guideline appeared in the same journal (Psychother Psychosom 2020;89(1):48, PMID 31655818, and the authors' reply at 89(1):49, PMID 31743919). No funding statement was located, so funding is recorded UNKNOWN.

Caveat on this rating: The positive signals appeared at week 16, after treatment had been discontinued at week 12, and in secondary outcomes and a severity subgroup — the classic shape of findings that do not replicate. The trial is from the research group that has published most of the positive NAC literature, and its senior author's disclosures elsewhere record consultancy to Bioadvantex, a NAC commercialiser, hence independent false. No funding or disclosure statement was readable for this paper.

Caveat on this rating: Two of the five authors — Berk and Dean — are investigators on trials included in the pooled estimate, so this is not an independent synthesis of other people's work, and Berk's disclosures elsewhere in this literature record consultancy to Bioadvantex, a company commercialising NAC. The search ended in November 2014 and so predates the larger negative bipolar trial listed below. No funding statement was readable, so funding is UNKNOWN.

Caveat on this rating: Thirty-one samples of Canadian products bought in 2016; it does not measure how common the problem is across the whole market, and the later US gummy analysis (Cohen 2023, below) did not detect serotonin in any product it tested, so the serotonin contamination finding has not been replicated in the US supply.

Caveat on this rating: The published abstract gives no sample size, no effect size, and no confidence intervals, so the strength of the result cannot be assessed from the record; sample size is recorded as unknown rather than guessed. The dose it found effective, 15 mg/day, is the same dose the same group's 1986 trial and Alda's 2017 trial both used as the comparator — a direct internal contradiction in this small literature that has never been resolved.

Caveat on this rating: Nearly half the enrolled participants did not complete, the comparator was a low dose of the same drug rather than an inert placebo, and methylene blue's blue-green urine makes blinding difficult at any dose — all conceded in the paper. Everyone was also on lithium, so this tests methylene blue only as an add-on.

Caveat on this rating: Poison-centre data are passive surveillance: exposures are self-reported by callers, are not laboratory-confirmed (phenibut is not detected on routine urine drug screens), and most people who use phenibut never call, so these numbers are a floor, not an incidence rate. Some of the rise after 2015 reflects "phenibut" being added as a searchable term that year.

Caveat on this rating: The authors state plainly that the whole data set is case reports and is therefore subject to publication bias, that reported doses may not reflect what unregulated products actually contained, and that people describing single-substance use may have been using other drugs too.

Caveat on this rating: The authors state their own limits: only 25 products, one sample per brand, gummies only, and no test of whether tablets and capsules behave the same or whether a given brand varies batch to batch. It measures label accuracy, not clinical harm.

Caveat on this rating: The acknowledgments state that two of the three authors "received funding from Sanofi to conduct the initial systematic publication database search" and that "Sanofi also provided access to data from 3 unpublished Mg intervention studies"; the senior author separately discloses membership of the Sanofi Consumer Healthcare Advisory Board. Sanofi markets magnesium supplements, so the sponsor had a direct commercial interest in the review's conclusion, and unpublished sponsor-held data entered the evidence base without independent scrutiny. Many included studies tested magnesium combined with up to five other ingredients.

Caveat on this rating: "This study and the journal's Rapid Service Fee was funded by Ethical Naturals Inc., California, USA", the company that sells the AlphaWave L-theanine tested; the trial was run at a contract research organisation. With 16 participants and one participant lost after period 1, the trial is far too small to support the conclusion in its own discussion that the product "should be considered as a nutraceutical supplement to manage acute stress."

Caveat on this rating: "The study was funded by a grant from Neurobrands CA, USA", a beverage company with a commercial interest in the L-theanine drink tested; the authors state the funder had no input into design, conduct, or interpretation. The tested product was a multi-ingredient drink, not isolated L-theanine, so the effects cannot be attributed to L-theanine alone.

Caveat on this rating: Funded by Taiyo Kagaku Co., Ltd., which manufactures the Suntheanine-brand L-theanine; two of the seven authors are employees of that company and it supplied both the active and placebo tablets. Several outcomes were measured within-person before-and-after rather than against placebo, which inflates apparent effects.

Caveat on this rating: Only four brands met the inclusion criteria, so this is a small snapshot rather than a market-wide survey; it demonstrates that labelled dose cannot be trusted, not the exact proportion of products that are mislabelled.

Caveat on this rating: A single-state case series of 56 calls, with coingestants documented in 36% of patients, cannot separate phenibut's effects from other drugs and cannot estimate how common these outcomes are among people who use phenibut. It scores low on the evidence rubric by design — it is a description of hospital cases, not a study of whether phenibut helps or harms.

Caveat on this rating: Three of the four authors are employees of Max Zeller Söhne AG, a manufacturer of a St. John's wort herbal medicinal product, and this is a narrative review rather than a systematic one — so it sits at the bottom of the evidence rubric and its framing should be read with the authors' commercial interest in mind. Note the direction of that interest: authors paid by an SJW manufacturer nonetheless document the interactions, which makes the interaction findings harder, not softer.

Caveat on this rating: Open-label by necessity (neither ketamine nor ECT can be masked), so subjective outcomes may favor the treatment patients preferred; population excluded psychotic depression, where ECT performs best.

Caveat on this rating: One co-author (Ware) became an employee of Canopy Growth Corporation, a licensed cannabis producer, after the work was completed, and the senior author discloses an unrelated cannabidiol grant involving Aurora Cannabis; both were disclosed. The pooled estimates come from observational studies, so residual confounding by shared risk factors for cannabis use and depression cannot be excluded.

Caveat on this rating: A 5-point PHQ-9 difference is large for a supplement, which is itself a reason for caution in a 100-person pilot that was explicitly powered for feasibility rather than efficacy and was conducted at a single site in a setting where access to other care is limited. No trial registration appears in the PubMed record, so the primary outcome cannot be checked against a pre-specified protocol. It has not been replicated.

Caveat on this rating: The credible interval runs from -9.70 to -0.17, meaning the true effect could be near zero; the ashwagandha estimate rests on a handful of small trials, most of which were supplied or funded by extract manufacturers. The analysis was conducted at the Beijing Research Institute of Chinese Medicine, an institution with an institutional interest in herbal medicine generally, though its funding came from the university and Chinese national science programmes rather than from any supplement company.

Caveat on this rating: The rubric score overstates this trial badly. It was single-blinded with a no-supplement control rather than a placebo capsule, so expectancy is uncontrolled; it enrolled 30 people; and the mood change it reports is half a point on a scale where the group started at 1.2 out of 15, which is not clinically interpretable as treating depression. It is included as the most recent randomised 5-HTP evidence that exists, which is itself the point: this is what "current evidence" looks like for 5-HTP.

Caveat on this rating: The racemic-vs-esketamine difference rests on indirect comparison across trials with different designs and populations; one coauthor (Zarate/NIMH) is an inventor on ketamine-related patents. An erratum was issued in 2021 (checked: not a retraction).

Caveat on this rating: This umbrella review's STRONG tier reflects its independent, systematic method — its actual conclusion is that the underlying MDMA evidence is weak. Read the tier as confidence in the review, not in MDMA.

Caveat on this rating: The disclosure statement records that the senior author, Dr Renshaw, "is also an inventor on related patent applications assigned to McLean Hospital and the University of Utah" — an intellectual-property interest in the intervention being tested, which is why independence is marked false here even though the funding is public. The trial was women-only, single-country, and n=52; a later trial in female adolescents found no difference from placebo.

Caveat on this rating: Diphenhydramine is a weak psychoactive control; the paper's own blinding assessment showed most participants correctly guessed their assignment, so expectancy effects cannot be excluded.

Caveat on this rating: Investigators are prominent psychedelic-research advocates; blinding was imperfect in both arms. The null primary endpoint is often glossed over in media coverage.

Caveat on this rating: The declaration of conflicting interests records that co-author Dr Candow "serves on the Scientific Advisory Board for Alzchem and Create (companies that manufacture creatine)" and has received industry creatine research funding and product donations, and that Dr Ostojic "is the co-founder of KRE-ALL, a company which develops food products enriched with creatine" — commercial ties to the intervention being reviewed, which is why independence is marked false despite the review itself being unfunded and its conclusions being cautious. No PROSPERO registration was found on the published page.

Caveat on this rating: Waitlist-controlled (no placebo), very small N, single site with strong psychedelic-research identity; effect sizes this large rarely survive larger placebo-controlled replication.

Caveat on this rating: The remission odds ratio of 3.6 looks dramatic but rests on three trials and is contradicted by the response outcome from two trials pointing the other way — a pattern that usually means too few events, not a real dissociation. The reviewers' own bias analyses point the same direction. No funding or competing-interests statement could be read (the Cambridge Core page returned only the abstract), so both are recorded conservatively.

Caveat on this rating: Single site, N=29, 7-day primary horizon, and ayahuasca's intense somatic effects (nausea, vomiting) make blinding doubtful despite the inert placebo.

Caveat on this rating: The pooled odds ratio looks impressive but rests on 64 patients with a confidence interval running from 1.28 to 13.15 — statistically positive and practically uninformative. The review is from 2002 and predates almost nothing, because almost nothing usable has been published since.

Caveat on this rating: Crossover design with a discriminable active dose means most participants likely knew their condition; Heffter-funded investigators are proponents of the therapy.

Caveat on this rating: This is a case-control design, so it establishes association and a dose-response pattern rather than proving causation; reverse causation (early psychotic symptoms driving heavier use) and residual confounding cannot be fully excluded, which is why the authors flag the population-attributable fractions as conditional on causality.

Caveat on this rating: An SMD of -1.55 is roughly three times the effect size seen for SSRIs in anxiety trials, which is implausible on its face; combined with I-squared of 93.8% and the authors' own "low certainty" GRADE rating, the pooled number should be read as a summary of a biased literature, not as an estimate of benefit. The great majority of the pooled trials were run on manufacturer-supplied branded extracts.

Caveat on this rating: A pooled remission rate from mostly uncontrolled trials measures how many people improved, not how many improved because of 5-HTP; depression remits substantially on placebo and with time. Classified here as a systematic review of largely non-randomised evidence rather than a meta-analysis of RCTs for exactly that reason.

Caveat on this rating: Sponsor-employee coauthors and shareholders; N=34; the 3-month durability claim comes from an uncontrolled open-label phase.

Caveat on this rating: Small N, discriminable drug (unblinding likely), academic two-center trial; independent of MindMed/Definium, which is a strength.

Caveat on this rating: This was the single positive short-term pivotal trial; two sibling trials (TRANSFORM-1, TRANSFORM-3) missed their primary endpoints, which the approval package acknowledged. Sponsor employees authored the paper.

Caveat on this rating: Built entirely from observational studies with heterogeneous definitions of "heavy" use; the pooled odds ratio was generated by simulating individual data points from summary statistics, a method that depends on assumptions about the underlying distributions.

Caveat on this rating: Sponsor-funded with sponsor employees as authors; central raters were blinded, but participants were functionally unblinded by design of the drug itself. Phase 3 toplines (Voyage, Emerge, 2026) are positive but press-release-stage, not yet peer-reviewed.

Caveat on this rating: Randomized-withdrawal designs can overstate maintenance benefit (abrupt discontinuation effects count as "relapse"); sponsor-funded and sponsor-authored.

Caveat on this rating: Usona is a 501(c)(3), not a commercial company, but it is the sponsor-developer seeking FDA approval of this exact product — scored as industry funding for that reason. Niacin does not mimic psychedelic effects, so unblinding was likely.

Caveat on this rating: Cross-sectional design: this cannot establish that creatine intake reduces depression risk, and reverse causation is at least as plausible as the causal reading. The paper states the authors have no conflict of interest, but co-author Dr Renshaw is disclosed elsewhere in this literature (in the 2012 AJP creatine trial above) as an inventor on related patent applications, which is why independence is marked false here.

Caveat on this rating: The disclosure states that STI-Pharm supplied the CBD at no cost and that four authors are co-inventors of a fluorinated-CBD patent; the University of São Paulo has licensed that patent and holds an agreement with Prati-Donaduzzi to develop a synthetic CBD product for indications including anxiety disorders. Public grant money plus donated study drug and patent interests is why funding is recorded MIXED.

Caveat on this rating: No placebo arm and no power calculation for equivalence: a trial in which both arms improve and neither wins cannot distinguish "both work" from "neither works and people recovered anyway", and 60 completers is far too few to establish non-inferiority. A single-centre trial that has not been replicated.

Caveat on this rating: N=7; included for historical significance only.

Caveat on this rating: Same sponsor-developer structure and unblinding problem as MAPP1; FDA's CRL asked for a further trial, which the Aug 2026 resubmission does not include (it relies on an audit, a phase 1 cardiac study and VA follow-up data instead).

Caveat on this rating: Sponsor-developer-run trial. FDA's 2024 advisory committee voted 9-2 that effectiveness was not established, citing near-total functional unblinding and allegations of trial-conduct problems in the wider program; FDA issued a CRL in Aug 2024. The paper itself is not retracted, but three related phase 2 papers are.

Caveat on this rating: Sponsor-run trial by the company seeking FDA approval; benefit at week 3 faded for many by week 12 (sustained response was a minority outcome in the full paper).

Caveat on this rating: Conducted by the University of São Paulo CBD group (Zuardi, Crippa, and colleagues). The 2011 record carries no conflict-of-interest statement, but the same group's later publications disclose that several of these authors are co-inventors on a CBD patent that the university has licensed to pharmaceutical companies, so independence from a commercial interest in CBD cannot be assumed.

Caveat on this rating: Funded by Arjuna Natural Extracts Ltd, the manufacturer of the Shoden extract tested. The authors state the trial was independently managed by the site investigators and the lead author discloses prior funding from the same company for unrelated work, but the sponsor's commercial interest in a positive result is direct, and the study's own secondary psychological outcome (DASS-21) did not reach statistical significance.

Caveat on this rating: Treatment was delivered by Ambio Life Sciences, whose shareholders are coauthors and patent applicants; Stanford analysis was philanthropically funded and independent of the clinic. Uncontrolled design caps what this can show regardless.

Caveat on this rating: N=12 with 8 vs 4 allocation; MAPS-affiliated authorship; included for historical significance.

Caveat on this rating: The paper's formal declaration reads "Source of Support: Nil" and "Conflict of Interest: None", but its own text states that the high-concentration full-spectrum extract tested "is the KSM-66 Ashwagandha extract, provided by Ixoreal Biomed, Hyderabad, India" — in-kind supply of the investigational product by its manufacturer, which is industry support whatever the declaration says. Recorded here as INDUSTRY on that basis. The trial was also not preregistered and the reported p-values are implausibly uniform across every scale.

Caveat on this rating: Funded by the Institute for Research on Cannabinoids, a nonprofit on whose board three of the authors sit (unpaid), and two authors disclose personal fees from numerous cannabis and cannabinoid companies. The result nonetheless cuts against those commercial interests, which weakens the usual direction-of-bias concern.

Caveat on this rating: The CONTESTED tier here reflects rock-bottom design score (uncontrolled, unregistered, self-report), not an active dispute; there is no controlled US evidence for ibogaine in opioid withdrawal at all.

Caveat on this rating: Funded by the American Kratom Association, a pro-kratom advocacy organization, and respondents were recruited through its channels — a design that samples satisfied current users and is structurally biased toward favorable results.

Caveat on this rating: The low score reflects study design, not disputed findings: this is a narrative review with no systematic search, no pooled estimate, and no readable funding statement, so it carries little independent evidentiary weight even though its cautious conclusion has held up.

Caveat on this rating: Many of the included trials were sponsored by sertraline's manufacturer; the review authors ran sensitivity analyses excluding manufacturer-funded studies.

Caveat on this rating: Publicly funded (NIHR and Japan Society for the Promotion of Science); some co-authors disclosed lecture or consultancy fees from drug manufacturers.

Caveat on this rating: This analysis covers trazodone at antidepressant doses for depression; the much more common low-dose use for insomnia is off-label and was not what was measured here. Several authors disclosed pharmaceutical lecture/consulting fees, though the study itself was publicly funded (NIHR and JSPS).

Caveat on this rating: Trazodone is the most commonly prescribed antidepressant for insomnia in the USA, but this use is off-label; the review found no evidence on long-term use and rated adverse-event evidence low certainty.

Caveat on this rating: Publicly funded (NIHR/JSPS), but several coauthors disclosed lecture or consultancy fees from drug manufacturers, including Pfizer.

Caveat on this rating: The trazodone evidence in this Cochrane review comes from one trial of 30 people over two weeks; effects on night-time awakenings and daytime sleep appeared negligible, and longer-term safety in dementia is untested.

Caveat on this rating: Funders were independent, but several co-authors disclose consulting or speaker fees from ADHD drug manufacturers; head-to-head evidence at 12 weeks says little about long-term outcomes.

Caveat on this rating: Publicly funded (NIHR and Japan Society for the Promotion of Science); some co-authors disclosed lecture or consultancy fees from drug manufacturers.

Caveat on this rating: Publicly funded (NIHR, JSPS) and PROSPERO-registered, but most of the underlying trials were industry-funded and several authors disclosed pharmaceutical-company fees.

Caveat on this rating: Later independent reanalyses of the broader STAR*D program have argued its published cumulative remission rates were inflated by protocol deviations; the level-2 switch comparison itself found no significant between-drug differences.

Caveat on this rating: One review author disclosed speaking honoraria from several drug manufacturers; the review itself had no external funding.

Caveat on this rating: No external funding, but several authors (Moncrieff, Horowitz, Kirsch, Hengartner) are prominent critics of antidepressants and disclosed royalties/consulting tied to antidepressant-deprescribing work — a perspective readers should weigh.

Caveat on this rating: Sleep-related adverse effects were about four times more frequent than with placebo (RR 4.31); funding and conflict-of-interest statements were behind the journal paywall and could not be read.

Caveat on this rating: Funding statement could not be retrieved (paywalled full text), so funding is recorded as unknown; superseded in scope by the same group's 2018 analysis.

Caveat on this rating: Includes unpublished regulatory data, so its estimates are less flattering than published-only analyses; one author disclosed industry speaker fees.

Caveat on this rating: Two reanalysis authors have served as paid expert witnesses against GSK in paroxetine litigation (disclosed); the original GSK-funded 2001 report claimed efficacy and remains unretracted but disputed.

Caveat on this rating: 41% of included trials were industry-funded, and this review series' very-low-certainty GRADE ratings are disputed by many ADHD researchers who consider them overly conservative relative to clinical consensus.

Caveat on this rating: Observational claims data; some co-authors disclose consulting fees or equity in health-analytics and pharmaceutical-adjacent companies, though the study itself was NIMH-funded.

Caveat on this rating: Small evidence base (7 trials) limited to short-term use; using trazodone for insomnia is off-label.

Caveat on this rating: The review notes 6 of 20 included trials had pharmaceutical-industry funding and most were underpowered with inadequate randomization reporting.

Caveat on this rating: Association, not proven causation; one senior author reports grants and personal fees from Takeda (an ADHD drug maker) outside the work. Contrasts with Habel 2011, which found no short-term rise in serious cardiovascular events.

Caveat on this rating: Co-funded in part by Angelini Pharma alongside public funders (NIHR Oxford Health BRC, MRC, Swiss NSF); registration status could not be read from the paywalled full text.

Caveat on this rating: Medication use after the 14-month randomized phase was self-selected, so height and symptom comparisons are observational; several senior authors disclose past consulting for stimulant manufacturers.

Caveat on this rating: All five underlying trials were conducted and funded by the manufacturer (GlaxoSmithKline); the statistical analysis was independent, and one co-author sat on the manufacturer's advisory board.

Caveat on this rating: Limited to published trials, so publication bias cannot be excluded; funding source not verifiable from fetched records.

Caveat on this rating: Funding statement could not be read (paywalled); this BOLDER I registration trial supported the manufacturer's bipolar-depression indication, so sponsor involvement is plausible but unverified.

Caveat on this rating: Funding statement is behind a paywall and could not be read; the two component maintenance trials are widely reported as manufacturer-sponsored, so weigh sponsorship accordingly.

Caveat on this rating: Government-funded, but several authors disclosed consulting or research ties to pharmaceutical companies; observational within-individual design.

Caveat on this rating: Observational studies only — people prescribed hypnotics differ from those who aren't, so confounding by insomnia itself and by frailty cannot be excluded; fall results were heterogeneous and not statistically significant.

Caveat on this rating: Observational study; despite propensity-score adjustment, people prescribed lithium may differ from those prescribed the comparison drugs.

Caveat on this rating: Open-label design and part-funded by Sanofi-Aventis (a valproate manufacturer), though the results favored lithium over valproate.

Caveat on this rating: Included evidence is mixed (RCT discontinuation phases, open trials, and case reports), so withdrawal incidence for venlafaxine cannot be precisely quantified; design scored conservatively as a review including observational data.

Caveat on this rating: Funded by Janssen-Cilag, which markets competing antipsychotics (paliperidone, risperidone), though the funder reportedly had no role in design or analysis; observational design cannot fully exclude confounding by indication.

Caveat on this rating: Included studies were small and observational; the same group's companion meta-analysis (PMID 15033227) found impairment improved but did not fully resolve after withdrawal.

Caveat on this rating: Included non-placebo-controlled studies (compared against a "virtual placebo" sample), total participant count and funding not verifiable from accessible text — scored conservatively; direction of the mirtazapine weight finding is consistent with the placebo-controlled Kamp 2025 meta-analysis.

Caveat on this rating: Publicly funded (NIHR and Japan Society for the Promotion of Science); some co-authors disclosed lecture or consultancy fees from drug manufacturers.

Caveat on this rating: Publicly funded (NIHR and Japan Society for the Promotion of Science); some co-authors disclosed lecture or consultancy fees from drug manufacturers.

Caveat on this rating: One review author disclosed speaking honoraria from several drug manufacturers; the review itself had no external funding.

Caveat on this rating: Most included trials were manufacturer-sponsored comparisons of escitalopram against rival drugs, a known source of sponsorship bias the review discusses.

Caveat on this rating: Review itself was publicly funded, but several authors declared lecture or consultancy fees from pharmaceutical companies, and most included trials were industry sponsored.

Caveat on this rating: Funders were independent, but several co-authors disclose consulting or speaker fees from ADHD drug manufacturers; head-to-head evidence at 12 weeks says little about long-term outcomes.

Caveat on this rating: Authors are FDA scientists analysing the agency's own trial database; no external funding was received.

Caveat on this rating: The reviewers note that 8 of the 11 included trials were funded by the same manufacturer, so industry sponsorship bias in the underlying trials cannot be ruled out.

Caveat on this rating: Analysis of EMA clinical study reports by Nordic Cochrane Centre authors who are prominent critics of industry trial reporting; stress urinary incontinence is not an FDA-approved duloxetine indication, and the authors concluded harms outweighed benefits for that use.

Caveat on this rating: Of 146 included studies reporting funding, 72 were funded by pharmaceutical companies; safety evidence was rated very low certainty.

Caveat on this rating: Government-funded (VA Cooperative Studies Program), but Bristol-Myers Squibb supplied the aripiprazole used in the comparison arm.

Caveat on this rating: Several authors reported grants from or ties to pharmaceutical companies; confidence intervals could not rule out a modest risk increase, healthy-user bias may flatter the results, and the later Zhang 2024 study found rising cardiovascular risk with more than 3 years of cumulative use.

Caveat on this rating: Observational despite target-trial-emulation design; one senior author reports grants and personal fees from Shire/Takeda outside the work; results cover ADHD medications as a class, mostly stimulants.

Caveat on this rating: Observational claims data; some co-authors disclose consulting fees or equity in health-analytics and pharmaceutical-adjacent companies, though the study itself was NIMH-funded.

Caveat on this rating: 16 of the 19 included trials were industry-funded and the review graded the evidence low to very low quality, so effect sizes may be optimistic.

Caveat on this rating: This is an EU regulatory action; the US FDA label does not impose the same 100 mg daily cap.

Caveat on this rating: NIH-funded, but Forest Laboratories (escitalopram's patent holder) supplied the study drug and placebo, and several authors disclosed industry relationships.

Caveat on this rating: All included studies were industry funded, and only four were not sponsored by duloxetine's manufacturer; the review authors cautioned that sponsorship bias may overestimate treatment effects.

Caveat on this rating: Eli Lilly, duloxetine's manufacturer, sponsored all but one of the included trials.

Caveat on this rating: The journal funding footnote sits behind JAMA's paywall, so funding is recorded as unknown; ClinicalTrials.gov (NCT00006286) lists the National Institute of Mental Health as the study sponsor.

Caveat on this rating: Funding statement could not be retrieved (paywalled full text), so funding is recorded as unknown; superseded in scope by the same group's 2018 analysis.

Caveat on this rating: Directly contradicts the dose-response signal in Billioti de Gage 2014; the dementia question is unsettled, and dependence/fall risks in older adults are established regardless.

Caveat on this rating: This trial tested buspirone as an add-on for depression, not for anxiety — an off-label use of the drug.

Caveat on this rating: Slightly elevated risk appeared at low but not high exposure - a pattern inconsistent with causation - yet the finding conflicts with the Billioti de Gage 2014 case-control study; the question remains unsettled.

Caveat on this rating: Contested: the prospective Gray 2016 BMJ cohort found no dose-response link between benzodiazepine use and dementia, so reverse causation (early dementia symptoms leading to prescriptions) remains a live alternative explanation.

Caveat on this rating: Full funding statement is paywalled and could not be read; the pooled trials were of the manufacturer's drug, so sponsorship could not be ruled out.

Caveat on this rating: Association, not proven causation - anxiety and insomnia can be early dementia symptoms that trigger benzodiazepine prescriptions years before diagnosis, and the prospective Gray 2016 cohort (BMJ) found no dose-response supporting a causal link.

Caveat on this rating: Funded by Pfizer and GlaxoSmithKline, the manufacturers of the drugs studied; several authors were Pfizer or GSK employees.

Caveat on this rating: Contemporary critiques in the same journal (e.g. PMID 2660772, 1989) disputed how durable the advantage over placebo was beyond the first weeks of this manufacturer-era trial and emphasized rebound and dependence on discontinuation.

Caveat on this rating: Funding and author industry relationships for this meta-analysis could not be read from open sources; several included trials were manufacturer-run.

Caveat on this rating: Pre-registration-era trial; the funding source could not be read from available records.

Caveat on this rating: Pre-registration-era trial; the funding source could not be read from available records.

Caveat on this rating: Small single-site pilot RCT from 1993; effect size is large but the study predates modern trial registration and reporting standards.

Caveat on this rating: Open-label design — patients and clinicians knew which drug was taken, which can inflate between-drug differences.

Caveat on this rating: Several authors disclosed speaker fees and grants from pharmaceutical companies (Janssen, Lundbeck, Otsuka, Angelini, and others).

Caveat on this rating: One author holds a position at Lund University funded by AB Svenska Spel, the state-owned Swedish gambling operator, as part of its responsible-gambling policy, and reports research funding from the same source alongside public research councils. The conflict is with the outcome being studied rather than with the drug, and it is disclosed in full in the paper. It is repeated here because a reader deciding how much weight to give a gambling finding should know who funds the chair that produced it.

Caveat on this rating: Study funding is public, but the author group carries extensive pharmaceutical relationships, all declared as unrelated to this work: honoraria from Lundbeck and Otsuka, consulting fees from Bristol-Myers Squibb, Novartis, Roche, Merck and Pfizer, stock options in several device and pharmaceutical companies, and participation in regulator-mandated phase IV studies funded by Alcon, Almirall, Astellas, AstraZeneca, Boehringer-Ingelheim, Novo Nordisk, Servier and LEO Pharma with funds paid to the institution. None of those are the makers of quetiapine, and the finding runs against the drug rather than for it, which is the direction sponsorship bias does not usually run. Recorded here so a reader can weigh it rather than discover it.