Chinese Medicine, Folk Remedies, and Health Claims: Where In Vitro Research Ends and Clinical Evidence Begins

An Evidence-Based Reading of Chinese Medicine, Folk Remedies, and Health Claims

The conclusion in one sentence

Chinese medicines and other natural therapies must each be judged by human evidence, the design of the control comparison, clinical outcomes, and adverse effects. In vitro activity, animal findings, and a small number of cases cannot directly establish that a treatment cures disease, prevents infection, or strengthens immunity as a whole. PMID: 33249281

First distinguish “studied” from “proven effective”

Professor Ching-Shun Lin’s original articles consistently ask not whether a therapy has a traditional narrative or a plausible mechanism, but whether it has actually been compared in patients in a way that can adequately rule out the natural course of illness, concomitant treatment, and selection bias. The case of Qingguan Yihao makes this boundary especially clear. Early evidence included in vitro pharmacology and uncontrolled observations in a small number of hospitalized patients. Even if a patient’s viral test became negative, those data alone cannot attribute the change to NRICM101, much less recast a laboratory test result as symptom relief or clinical recovery. PMID: 33249281

By the same reasoning, the direct studies catalogued by ECU contain no controlled prevention data showing that uninfected people can take Qingguan Yihao to prevent COVID-19 or “increase resistance.” The evidence must therefore be classified as insufficient and cannot support a health-maintenance recommendation. PMID: 33249281

The professor contrasted oral antiviral drugs tested in randomized trials with therapies supported only by observational data in order to show that the levels of evidence are different. That methodological comparison remains valid. The literature has, however, been updated: later meta-analytic research indicates that the benefits of Paxlovid and molnupiravir are constrained by the prevailing variant and the population’s pre-existing immune context. Results from early trials cannot be extrapolated indefinitely without qualification. PMID: 39286035

Cell, animal, and ingredient studies cannot be extrapolated across evidentiary levels

Finding that a substance inhibits cancer cells, a virus, or inflammatory signaling in cultured cells means that further investigation may be worthwhile. It does not demonstrate less pain, better quality of life, longer survival, or tumor shrinkage in patients. The functional-genomics paper used to promote Compound Kushen employed methods and outcomes confined to cells in vitro; it therefore does not constitute clinical evidence of an anticancer effect. PMID: 27602759

Updates to the literature should also be stated accurately. Randomized trials in patients were subsequently included in a systematic review of Compound Kushen, so it would no longer be appropriate to say that “there are no clinical studies at all.” Yet the review still calls for larger, higher-quality trials with long-term clinical outcomes. It does not support promoting Compound Kushen as a miraculous cancer cure. PMID: 40248100

The same logic applies to gua sha. A mouse response to an intradermal antigen cannot be extrapolated into claims that the practice broadly strengthens human “immunity,” detoxifies the body, or promotes self-healing. For musculoskeletal pain, existing systematic reviews still find the controlled clinical evidence insufficient. PMID: 27672506；PMID: 20205902

Risks must be assessed for Chinese medicines and natural products too

“Natural” does not mean nontoxic. Black nightshade contains glycoalkaloids, and the literature supports a risk of toxicity in both humans and animals. “Nontoxic and safe to eat in large quantities” is therefore not a valid safety conclusion. As for treating cancer, hypertension, or diabetes, the available evidence remains predominantly preclinical and is not sufficient to support clinical use. PMID: 41819152；PMID: 41743169

Although a meta-analysis of Bofutsushosan observed a small reduction in BMI among people with obesity, it found no significant improvement in waist circumference, glucose or lipid measures, or blood pressure. A limited change cannot be packaged as a comprehensive weight-loss benefit. PMID: 35417488

For safety, cases of pneumonia and liver injury have been reported in association with Bofutsushosan, alongside signals from spontaneous adverse-event reports. These data cannot directly determine the incidence among all users, but they are sufficient reason to avoid excessive use and to seek medical care if symptoms occur. PMID: 12692947；PMID: 18679001；PMID: 35370270

Limited signals for supplements are not the miraculous effects promised in advertising

The pooled clinical evidence concerning ejiao does not consist of trials testing ejiao alone as an anticancer treatment, and its results are highly heterogeneous. It cannot establish broad claims such as fighting cancer, enriching the blood, or improving appearance. PMID: 41015792

Scientific progress has produced a more nuanced answer for oral collagen. Later reviews integrating randomized trials report that oral peptides may yield limited findings for skin hydration, radiance, and wrinkles, so “it has absolutely no effect” is no longer an appropriate blanket statement. But the effects are inconsistent and limited in magnitude; they cannot in turn validate sweeping claims that ejiao or collagen treats a multitude of diseases. PMID: 41924746

The presence of amino acids or trace elements in pearl powder does not directly prove that taking it orally whitens or refines the skin. Existing human trials measured antioxidant markers in blood, not skin tone or texture, so their findings cannot be extrapolated to the advertised cosmetic effects. PMID: 29389568

Edible bird’s nest and sialic acid must likewise be evaluated separately. The available evidence consists mainly of proposed applications and preclinical studies; it has not yet provided randomized trials confirming broad health benefits in healthy adults. PMID: 34025406；PMID: 37764670

Based on the double-blind Phase II trial available at the time, the professor noted a signal for upper-limb strength with extended-release sialic acid in GNE myopathy. The literature has since been updated: the later and larger Phase III trial did not confirm efficacy. The early signal can therefore no longer be described as an established benefit. PMID: 27854209；PMID: 31036580

Do not replace an available standard treatment with a folk remedy

Psychological support can be valuable in care, but current evidence does not show that it generally and reliably extends survival in cancer, much less that psychological balance accounts for half of a person’s lifespan. PMID: 33359756

When an actionable standard cancer treatment is available, rejecting it and substituting a natural or alternative therapy is associated with worse outcomes. This does not mean that all supportive care is the same; it means that an unproven therapy must not delay an intervention supported by clinical evidence. PMID: 42329504；PMID: 40627185

Understand association, causation, and the evolution of science

The sweeping statement “milk causes cancer” ignores differences among cancer types. Modern evidence syntheses find that dairy consumption is either unrelated to most cancers or associated with lower risk. They do not support directly classifying milk consumption as carcinogenic. PMID: 40715472

When cancer types are examined separately, milk or dairy intake is inversely associated with colorectal cancer risk, and Mendelian-randomization research supports a protective direction. On the other hand, there is a small positive association with prostate cancer, although biases such as screening must still be considered. An association cannot be simplified into the claim that everyone who drinks milk will develop cancer. PMID: 42116288；PMID: 25527754

The professor’s emphasis that an observational association is not the same as causation remains important. What the updated literature adds is that causal-inference tools can now provide part of the answer when long-term carcinogenic feeding trials cannot be conducted. Absolute statements that the question can “never be answered” should therefore be avoided. PMID: 42116288

Overall, evidence-based appraisal does not decide what is true by first applying a label such as “Chinese medicine” or “Western medicine.” It examines the study population, comparator, outcomes, bias, and harms for each claim. When a claim lacks human evidence, the most honest conclusion is “not yet established.” When new data later emerge, we should say that the literature has been updated and science continues to evolve, while preserving the historical context in which Professor Lin wrote the original article.