The Evidence-Based Truth About Mental Health: Understanding Anxiety, Depression, Insomnia, and Emerging Therapies

The Evidence-Based Truth About Mental Health: From Treatment Evidence to Marketing Claims

The conclusion in one sentence

Mental health care should be grounded in diagnosis, controlled clinical evidence, and individualized risk assessment. A signal of benefit does not mean a treatment works for everyone; association does not establish causation; and supplements or alternative therapies without reliable controls cannot replace established care. PMID: 34817851

Separate “associated with” from “able to treat”

Depression may occur alongside markers of infection, nutritional status, loneliness, or living conditions. Observing a relationship between two factors, however, does not show that either causes the other, much less that a particular supplement or intervention can treat the illness. Research on herpesviruses and depression has a long history, and later data have continued to show a statistical association between seropositivity and depression. These findings still do not establish a causal chain that can be translated into treatment. PMID: 40633509

Higher SITH-1 antibody levels among people with depression are an association or biomarker signal worthy of further study, not proof that a single definitive cause of depression has been found. Professor Lin’s objection at the time to media reports presenting a research signal as a “confirmed cause” was consistent with the cautious language of the original study. PMID: 32534440

Meta-analyses support an association between loneliness and a higher risk of death. Yet loneliness, social isolation, and the broader quality of social relationships are not identical exposures. The evidence does not show that loneliness must outweigh any particular conventional risk factor, nor does it justify assigning fixed health weights arbitrarily to diet, exercise, and social activity. PMID: 25910392

Evidence-based treatments still have limits of appropriate use

Buspirone has evidence of efficacy over placebo for generalized anxiety disorder, but its clinical effects do not appear immediately. It should therefore not be understood as a fast-acting medication that can relieve acute anxiety on demand. The central points in Professor Lin’s original explanation of its primary use and onset of action are supported by controlled trials. PMID: 10485635

For irritability or aggressive behavior in people with dementia, buspirone has shown retrospective signals of improvement, but prospective, double-blind studies are still needed. Rare case reports have also described worsening psychosis. Such reports can prompt clinical monitoring, but they cannot estimate the incidence in the general patient population. PMID: 28124634

Psychological and behavioral therapies can improve insomnia and related psychological symptoms. They are not equivalent to poorly defined commercial therapies that claim to “balance mind-body energy.” Without verifiable clinical research, the latter can only be described as having unproven efficacy; they cannot borrow the evidence for psychotherapy merely because their names sound similar. PMID: 41934053

Many meta-analyses of TMS for major depressive disorder report signals of efficacy and safety, but effect estimates are affected by heterogeneity, small-study effects, and limitations in review quality. TMS may be an option following professional assessment, but it is not a package that guarantees results. Professor Lin’s placement of TMS in the clinical context of considering it after common treatments have failed is broadly reasonable. Current data, however, are insufficient to classify every earlier use as inappropriate. PMID: 36587461

“Natural” does not automatically make a supplement a treatment

For fish oil in adults with major depressive disorder, the overall evidence shows, at most, a small to moderate statistical signal with low certainty, and the improvement may not reach clinical importance. Fish oil therefore should not be marketed as a reliably proven antidepressant treatment. PMID: 34817851

Larger prevention trials likewise do not support using marine omega-3 supplements to prevent depression or improve mood in older adults without depression. Formulation differences also matter: existing trials more strongly suggest that any potential benefit is driven by EPA. Changes observed in human studies without placebo controls cannot establish definite efficacy for a DHA formulation. PMID: 34932079

A large, long-term randomized trial of vitamin D3 found no evidence that supplementation prevents depression in adults or improves mood scores. A controlled trial in people with early psychosis likewise found no improvement in core psychiatric symptoms, functioning, depression, or metabolic outcomes. These results weigh against specific supplementation strategies and uses; they should not be extrapolated to mean that vitamin D has no role in every physiological or skeletal context. PMID: 32749491

Low ferritin, iron deficiency, and depression also cannot simply be joined by an equals sign of causation. Current evidence does not consistently show that iron supplementation treats depression, and no accepted standard supports defining everyone below a single high ferritin threshold as iron deficient. Suspected iron deficiency should be assessed in light of the person’s symptoms, laboratory results, and clinical context—not self-treated with iron as though it were an antidepressant prescription. PMID: 40945632

L-tyrosine likewise lacks reliable clinical evidence as an effective treatment for ADHD. Early small trials did not establish sustained or significant benefit, and biochemical observational studies in children do not support reducing the problem to a deficiency that can be corrected with aromatic amino acid supplementation. PMID: 26938936

Psychedelic research: a signal is not permission for unsupervised use

LSD-assisted therapy has produced efficacy signals in some psychiatric studies, with short-term findings for alcohol use disorder receiving particular attention. Study quality, overlapping samples, long-term benefits, and safety data remain limited, however, so the evidence currently does not support routine use outside strict supervision. PMID: 41517502

Popular claims about psychedelic microdosing are even more dependent on placebo-controlled evidence. In a self-blinding study, psychological outcomes improved both among participants taking psychedelic microdoses and among those taking placebo, with no significant difference between groups. This suggests that expectancy effects can explain many of the reported benefits. PMID: 33648632

At the same time, it would be scientifically inappropriate to declare every possible effect a placebo response. Controlled studies have observed mixed cognitive changes, and another study found an effect on time perception. What has not been demonstrated is a robust improvement in attention or work performance that can be separated from expectancy effects. That is precisely the difference between “a measurable change” and “therapeutic value.” PMID: 30478716

New risk signals must be updated as the literature evolves

Professor Lin’s original article accurately recorded that regulators had initiated a review of suicide and self-harm risks associated with semaglutide／GLP-1 receptor agonists after receiving safety reports. The literature has since been updated, and subsequent studies have not confirmed an increased risk. The more appropriate statement today is that monitoring should continue and assessment should reflect each person’s medical history—not that the drugs have an established harmful effect. PMID: 40105856

This update does not rewrite Professor Lin’s original article or his historical position. It distinguishes “a safety signal emerged at the time” from “whether that signal was later confirmed.” Evidence-based knowledge evolves with new research. Respect for the original text also requires us to state clearly how far the current evidence can take us. PMID: 41302345

A practical way to evaluate health claims

First ask whether the study actually investigated the product, population, and outcome being claimed. Then check for random assignment, placebo controls, blinding, and adequate follow-up. Cell experiments can explore mechanisms, single-arm studies can generate hypotheses, and case reports can raise safety alerts. None of them alone can directly establish therapeutic benefit for the general population or quantify the magnitude of risk. PMID: 34131267

Finally, return the choice to the individual context. Symptom severity, comorbidities, current medications, affordability, and personal preferences all affect a decision. If symptoms persist, impair functioning, or include thoughts of self-harm, the priority is not to find another supplement promoted by influencers. It is to seek qualified medical and mental health care promptly.