Authors: Changzhen Gong1, Arthur Yin Fan2
1. Changzhen Gong, PhD, Editor-in-Chief, International Journal of Clinical Acupuncture;
American TCM Association (ATCMA)
2. Arthur Yin Fan, MD, PhD, Associate Editor, Journal of Integrative Medicine;
Licensed Acupuncturist, American TCM Association (ATCMA)
Abstract
This article provides a critical appraisal of a randomized, double-blind, placebo-controlled pilot trial of Peaceful Palace Bovine Bezoar Pill (Angong Niuhuang Wan) led by Peking Union Medical College Hospital, with a focus on study design, interpretation of results, and distortions in public dissemination. The trial showed that among 120 patients with moderate to severe acute ischemic stroke, adjunctive use of Peaceful Palace Bovine Bezoar Pill did not achieve statistical significance in primary endpoints such as infarct volume and cerebral edema at 14 days; however, favorable trends were observed in certain subgroups and in 90-day functional outcomes. The authors argue that these findings should not be simplistically interpreted as “equivalent to placebo.”
The paper further examines key methodological issues, including misalignment of indications, marginal effects and the ceiling of therapeutic benefit, mismatch between evaluation metrics and the drug’s traditional role, insufficient sample size and statistical power, and suboptimal dosage and treatment duration. It emphasizes that “lack of statistical significance” does not equate to “proof of ineffectiveness.”
Keywords: Peaceful Palace Bovine Bezoar Pill; Acute Ischemic Stroke; Randomized Controlled Trial; Traditional Chinese Medicine; Statistical Significance; Clinical Trial Methodology
I. The Peking Union Medical College Hospital–Led Peaceful Palace Bovine Bezoar Pill Trial: Design, Findings, and Social Impact
Recently, a clinical trial of Peaceful Palace Bovine Bezoar Pill (ān gōng niú huáng wán) led by Peking Union Medical College Hospital was published in the English edition of the Chinese Medical Journal¹. The study was designed as a multicenter, randomized, double-blind, placebo-controlled pilot trial. A total of 120 patients with moderate to severe acute ischemic stroke (NIHSS score 10–20) were enrolled across 17 hospitals. All patients received guideline-recommended standard modern stroke care, alongside which Peaceful Palace Bovine Bezoar Pill (treatment group) or a visually identical placebo (control group) was administered at a dose of 3 g (1 pill) daily for 5 consecutive days. The primary endpoint was not a clinical functional outcome, but rather changes in infarct volume and cerebral edema volume at 14 days. Secondary endpoints included 90-day imaging outcomes, changes in NIHSS scores, and the proportion of patients achieving a modified Rankin Scale (mRS) score of 0–2 at 90 days.
According to the original report¹, the median age of participants was 66.0 years, and the median baseline NIHSS score was 12.0. In the modified intention-to-treat (mITT) population, 57 patients were assigned to the treatment group and 60 to the placebo group. At day 14, changes in infarct volume were 0.3 mL in the treatment group and 0.4 mL in the placebo group (median difference: –7.1 mL; interquartile range [IQR]: –18.3 to 2.3 mL; P = 0.30). Changes in cerebral edema volume at day 14 were 11.4 mL and 4.0 mL, respectively (median difference: 3.0 mL; IQR: –1.3 to 9.9 mL; P = 0.15), neither reaching statistical significance. Within 90 days, the incidence of serious adverse events was comparable between the treatment group (3/57, 5%) and the placebo group (7/60, 12%) (P = 0.36). Changes in serum mercury and arsenic levels were also similar between groups. Notably, in patients with large-artery atherosclerosis (LAA), the treatment group showed a reduction in infarct volume at day 14 (median difference: –12.3 mL; IQR: –27.7 to –0.3 mL; P = 0.03).
Regarding 90-day functional outcomes, the proportion of patients achieving mRS 0–2 was 38% in the treatment group versus 24% in the placebo group, corresponding to an odds ratio of approximately 1.9; however, this difference did not reach statistical significance due to limited sample size. The investigators concluded that “in patients with moderate to severe stroke, Peaceful Palace Bovine Bezoar Pill demonstrated a safety profile comparable to placebo and reduced infarct volume, although the statistical results did not reach significance”. They further noted that in the LAA subgroup, the reduction in infarct volume at 14 days achieved statistical significance (MD –12.3 mL, P = 0.03).
Importantly, the discussion section explicitly acknowledged that this study was an exploratory pilot trial. The sample size of 120 was predetermined without formal power calculation or Type I error estimation prior to trial initiation, and the study had relatively high rates of loss to follow-up and dropout. Therefore, the findings require cautious interpretation. However, in dissemination on certain mass media² and social media platforms³, this “pilot study with non-significant differences” was simplified into extreme narratives such as “Peaceful Palace Bovine Bezoar Pill ≈ placebo”, “a fallen miracle drug”, or “the collapse of a legend”. These interpretations spread rapidly across social media platforms, transforming what should have been a rational discussion of a clinical trial into an emotionally charged public judgment of a traditional formula.
We observe a dual nature in the study results. On the one hand, most primary and secondary endpoints did not achieve statistical significance. On the other hand, both at 14 days and 90 days, indicators such as infarct volume and the proportion of mRS 0–2 consistently trended in favor of the treatment group. Particularly in the LAA subgroup, a statistically significant imaging benefit was observed. These findings are difficult to reconcile with a simplistic conclusion of complete ineffectiveness.
II. Indication Misalignment: From a Targeted Remedy for “Heat-Block with Clouded Spirit” to a “Universal Stroke Intervention”
From the perspective of classical Chinese medicine and real-world clinical practice, Peaceful Palace Bovine Bezoar Pill is traditionally a renowned formula for clearing heat, resolving toxin, opening the orifices, and arousing the spirit. Its core indication lies in conditions characterized by heat entering the pericardium and heat-block with clouded spirit”, commonly seen in severe warm-heat disease or critical illnesses. These presentations include high fever, delirious speech, stupor, vexation, and agitation, falling under the category of “heat-block” disturbance of the spirit-mind. Its therapeutic value lies in opening the orifices, promoting arousal, and addressing acute emergencies, rather than serving as a comprehensive treatment for the entire course of ischemic stroke.
In the inclusion and exclusion criteria of the clinical trial, the investigators did note that patients deemed unsuitable for Peaceful Palace Bovine Bezoar Pill based on traditional Chinese medicine (TCM) pattern identification would be excluded. However, the study did not explicitly define what constituted an appropriate indication, nor did it provide any systematic data on TCM pattern distribution. Crucially, heat-block with clouded spirit was not established as a required inclusion criterion. This implies that, in practice, a substantial number of patients with moderate to severe ischemic stroke who did not exhibit heat-block patterns could still be enrolled and randomized to receive either Peaceful Palace Bovine Bezoar Pill or placebo, as long as no clear contraindications were present.
Such an approach may appear methodologically sound within the framework of modern randomized controlled trials (RCT), in which a single disease entity is treated with a uniform intervention. Yet from the standpoint of TCM’s principle of pattern identification as the basis for determining treatment, this represents a form of “Westernized use of Chinese medicine”. A classical formula, originally dependent on precise pattern identification and temporal dynamics of disease progression, is instead applied indiscriminately to a heterogeneous population of acute ischemic stroke patients. The resulting conclusion of no significant overall effect is then derived from this mismatch.
One might ask: if a study on the indications of ACE inhibitors were arbitrarily expanded to include all individuals with slightly elevated or even borderline-low blood pressure, combined with diverse cardiac and renal conditions, and the drug were administered uniformly without stratification, would Western medical practitioners accept a conclusion that dismisses the efficacy of ACE inhibitors based on an overall non-significant result?
In the case of Peaceful Palace Bovine Bezoar Pill, a more appropriate interpretation of the current trial findings would be as follows: among patients with moderate to severe ischemic stroke defined by NIHSS scores and imaging criteria, but not strictly limited to the classical pattern of heat-block with clouded spirit, short-term adjunctive use of Peaceful Palace Bovine Bezoar Pill did not demonstrate a statistically significant group difference in the primary imaging endpoint at 14 days. This conclusion remains fundamentally distinct from claiming ineffectiveness when used according to its classical indications and in an appropriate clinical context.
III. Marginal Effects and the “Ceiling” of Efficacy: A Single Pill on Top of Comprehensive Modern Therapy
Another defining feature of the Peaceful Palace Bovine Bezoar Pill trial is that all participants had already received standard modern stroke care¹. Every enrolled patient was treated according to the guidelines of the Chinese Society of Neurology, including intravenous thrombolysis, endovascular intervention, antiplatelet therapy, anticoagulation, neuroprotective strategies, lipid-lowering therapy, as well as blood pressure and glucose management, and structured rehabilitation. The investigators noted that these key treatments were evenly distributed between the 2 groups, and explicitly stated that concomitant therapies were “comparable.”
In other words, Peaceful Palace Bovine Bezoar Pill was not evaluated in a setting of “no treatment” or “TCM-only treatment”, but rather as the N-th intervention added on top of a highly saturated combination of evidence-based modern therapies.
Within such a context, at least 3 issues merit careful consideration.
First, one must consider the decreasing marginal effects⁴. When a patient has already received more than a dozen effective medical interventions, many of which independently influence infarct volume, cerebral edema, and functional outcomes, the incremental contribution of adding another medicinal is inevitably “diluted”. Its effect is easily submerged within statistical noise. Under these conditions, attempting to detect differences on the order of a few milliliters in infarct volume with a sample size of only 120 patients, as in this trial, makes the emergence of a measurable effect intrinsically unlikely.
Second, there is the ceiling effect of the rapeutic benefit4. Any treatment for acute stroke is subject to a physiological and biological “upper limit”. When multiple interventions such as thrombolysis, endovascular therapy, blood pressure control, lipid lowering, and rehabilitation have already brought some patients close to that limit, adding yet another treatment will not necessarily produce a linear increase in benefit. Instead, it may simply add cost, monitoring burden, and the risk of drug interactions. For a medicinal traditionally used to restore consciousness and clear heat, placing it at the very top of a “maximal modern treatment regimen” and expecting it to produce a statistically significant additional reduction in infarct volume within 14 days is, in itself, an exacting demand.
Third, there is also a mismatch between the evaluation system and the therapeutic positioning of the medicinal. For Peaceful Palace Bovine Bezoar Pill, outcomes such as “rousing the spirit, improving mental status, shortening the duration of coma or delirium”, and “relieving high fever and associated disturbances of spirit-mind” are much closer to its traditional role than expecting it to bear the primary burden of reducing infarct volume on its own. Yet in this trial, the main focus was fixed on imaging indicators, while measures of consciousness such as the Glasgow Coma Scale (GCS) were treated only as secondary analyses. Even though the changes in GCS at 7 and 14 days showed near-significant differences, with P values both around 0.05, this signal was not meaningfully explored. Instead, it was largely submerged beneath the broader statement that the primary endpoint was negative.
This raises a rather thought-provoking question: what if the trial were designed in reverse? That is, both groups would be required to take Peaceful Palace Bovine Bezoar Pill, assuming pattern correspondence, and on that basis be randomized to receive or not receive a particular Western medical intervention, after which outcomes would be compared. Would the result then also be summarized as “that Western drug is equivalent to placebo”? This hypothetical study is, of course, not realistically implementable, but it reveals the one-way filter embedded in the current trial design: Western medical treatment is always assumed to be the baseline, while Chinese medicinals are treated as a passive add-on, an optional accessory whose presence or absence is presumed to be of secondary importance.
IV. “No Statistical Significance” Is Not “Falsification” : Triple Shifts in Sample Size, Endpoints, and Interpretation
From a rigorous methodological standpoint, several design choices in the Peaceful Palace Bovine Bezoar Pill trial mean that it can at most support a conclusion of “failure to demonstrate efficacy”, but cannot justify the stronger claim of “proven ineffectiveness”.
The first issue concerns sample size and statistical power. The investigators explicitly acknowledged that the total sample size of 120 patients was an “exploratory choice”, without formal pretrial calculation of statistical power or alpha level, and without a clearly defined assumption regarding expected effect size. By contrast, major stroke trials published in leading international journals⁴, ⁵ typically enroll between 1,648 and 2,839 patients. In this context, a study with only 120 participants is inherently prone to falling into the statistical trap of “no significant difference”, regardless of whether a true effect exists.
Moreover, in the modified intention-to-treat (mITT) population, the number of patients actually included in the primary endpoint analysis was further reduced due to missing imaging data. In subgroups such as large-artery atherosclerosis (LAA), the effective sample size became even smaller. Under such conditions, the study is structurally underpowered to detect even moderate effect sizes, let alone the subtle incremental benefits of Peaceful Palace Bovine Bezoar Pill within the context of multiple concurrent modern therapies.
The second issue lies in the limitations of endpoint selection. While choosing changes in infarct volume and cerebral edema volume at 14 days as dual primary endpoints may have scientific rationale, for clinicians and patients, the truly meaningful outcome is functional recovery at 90 days, typically measured by the modified Rankin Scale (mRS 0–2). On this clinically relevant endpoint, the treatment group achieved 38% compared to 24% in the control group, with an odds ratio approaching 2.0. The lack of statistical significance here is far more plausibly attributable to insufficient sample size than to absence of effect.
To reduce such findings to the simple claim that Peaceful Palace Bovine Bezoar Pill is no different from a placebo is, from both clinical intuition and statistical reasoning, at the very least an oversimplification.
A further issue lies in the conceptual conflation between “non-significance” and “no difference”. In public discourse, a result of “P > 0.05” is quickly translated into “no difference from placebo”, and then amplified into claims of proven ineffectiveness or the “collapse of a miracle drug.” Yet in statistical terms, “failure to detect a significant difference” and “evidence of no difference” are fundamentally distinct. The former indicates that, under the current sample size, study design, and level of noise, the possibility of a true effect cannot be excluded. The latter, however, requires adequate statistical power, a well-defined non-inferiority margin, or even an equivalence trial design before such a conclusion can be cautiously drawn. Notably, the trial itself maintained a measured tone in its discussion, using terms and phrases such as “trend”, “potential effect”, and “further research is needed”¹. By contrast, certain media narratives and anti-TCM rhetoric bypassed this intermediate step altogether, leaping directly from a negative trial to the sweeping conclusion that the intervention had been definitively disproven.
V. Suboptimal Dosage and Insufficient Treatment Duration
In this trial, Peaceful Palace Bovine Bezoar Pill was administered at a dosage of 1 pill (3 g) per day for 5 consecutive days. For patients with moderate to severe ischemic stroke, such a regimen may be considered both under-dosed and too short in duration.
The renowned TCM physician Zhou Zhongying has reported using Peaceful Palace Bovine Bezoar Pill in severe pneumonia cases with high fever and coma at a dosage of 6–9 pills per day for up to 2 weeks, an approach personally observed by Dr. Arthur Yin Fan. Similarly, Dr. Fan has treated trauma-induced coma with a dosage of 6 pills daily for more than 1 week. Modern derivatives inspired by Peaceful Palace Bovine Bezoar Pill, such as Qingkailing Injection, do not rigidly follow a “1 pill per day” paradigm, but instead recalibrate dosing based on clinical realities, adjusting dosage according to different diseases and individual patient conditions⁶.
To apply Peaceful Palace Bovine Bezoar Pill in a “Westernized” framework, shifting from syndrome-based (pattern identification) use to disease-based (biomedical diagnosis) application for moderate to severe acute ischemic stroke, it is essential to first conduct a series of preliminary studies, or what would correspond to Phase I trials. These should aim to determine the optimal daily dosage and treatment duration, as well as to identify the minimum clinically important difference (MCID) for evaluating efficacy between treatment and control groups⁷ ⁸. Only after these foundational parameters have been established can subsequent Phase II and III clinical trials be meaningfully designed and interpreted.
VI. How Should We Properly Interpret the Clinical Trial and Peaceful Palace Bovine Bezoar Pill Itself?
We do not deny that this trial carries a constructive value in demystifying Peaceful Palace Bovine Bezoar Pill. It serves as a reminder to both clinicians and the public that:
• Peaceful Palace Bovine Bezoar Pill is not a health supplement, nor a universal remedy to be taken routinely for all conditions.
• In the context of acute ischemic stroke, it cannot replace standard treatments such as thrombolysis, thrombectomy, antiplatelet therapy, lipid-lowering interventions, or structured rehabilitation.
• Moreover, given that it contains mineral components such as mercury and arsenic, its safety must be ensured through appropriate monitoring and rational duration of use, rather than arbitrary dose escalation or prolonged administration.
At the same time, it is equally important to emphasize that the current trial¹ neither demonstrates nor is capable of demonstrating the ineffectiveness of Peaceful Palace Bovine Bezoar Pill. What the study shows is more limited and specific: under conditions of a relatively small sample size, broad inclusion criteria, comparatively low dosage and short treatment duration, and the given statistical framework, Peaceful Palace Bovine Bezoar Pill did not exhibit a statistically significant advantage in the 14-day imaging-based primary endpoint. However, in certain subgroups such as large-artery atherosclerosis (LAA), and in 90-day functional outcomes, signals suggestive of potential benefit did emerge and warrant further investigation. This remains fundamentally distinct from concluding equivalence to placebo or complete lack of efficacy.
What truly calls for reflection is the way research on TCM is often shaped by what might be termed an “imperialist lens”. When Chinese medicinals and acupuncture are routinely evaluated within frameworks defined entirely by Western biomedicine—where disease labels replace pattern identification, single imaging endpoints substitute for holistic clinical outcomes, and fixed short-term dosing replaces dynamic, course-based adjustment—what emerges is not the full reality of TCM, but rather a flattened version compelled to mimic the logic of pharmaceuticals. In this sense, the real value of the Peaceful Palace Bovine Bezoar Pill trial lies precisely in exposing this structural tension. What has been tested is not the entirety of Peaceful Palace Bovine Bezoar Pill as understood within its own medical system, but a version flattened into the shadow of a Western drug paradigm.
VII. What Should Come Next?
If the genuine goal is to evaluate the value of Peaceful Palace Bovine Bezoar Pill in encephalopathic conditions through evidence-based methods, then at least the following measures are needed.
• The study design should strictly limit enrollment to core TCM patterns such as “heat-block with clouded spirit” to avoid diluting indications.
• Outcomes more closely aligned with the medicinal’s traditional function, such as restoring consciousness and shortening the duration of coma or delirium, should be included among the primary endpoints, rather than relying mainly on the hope of an additional reduction in infarct volume at 14 days.
• Within the context of standard modern treatment, investigators should realistically assess marginal effects and clinically acceptable effect sizes, using proper sample-size calculations and clinically meaningful MCIDs⁷, ⁸, instead of expecting a 120-patient study to bear the double burden of both “disproving” and “endorsing” the medicinal.
In sum, the Peaceful Palace Bovine Bezoar Pill trial may, in a certain sense, be regarded as a worthwhile pilot study. It provides relatively rigorous safety data for Peaceful Palace Bovine Bezoar Pill and leaves clues in certain subgroups and longer-term outcomes that deserve further follow-up. However, it is by no means a final verdict on the efficacy or nature of the medicinal, nor can it support the simplistic slogan that its effect is equivalent to placebo. Rather than using this study to hang up a public denunciation poster against a so-called TCM “miracle drug”, it would be better to regard it as a necessary yet imperfect starting point: a mirror reflecting the biases, inertia, and filters we bring to the ways we study, understand, and evaluate TCM.
What truly needs to collapse is not Peaceful Palace Bovine Bezoar Pill itself, but the arrogance of using a single standard to cut down a complex medical system.
References
1. Li S, Wang A, Shi L, Liu Q, Guo X, Liu K, et al. Safety and efficacy of Peaceful Palace Bovine Bezoar Pills in patients with moderate-to-severe acute ischemic stroke (ANGONG TRIAL): A randomized double-blind placebo-controlled pilot clinical trial. Chin Med J (Engl). 2025 Mar 5;138(5):579-588. doi: 10.1097/CM9.0000000000003133. Epub 2024 Nov 6.
PMID: 39501831
2. 陳輝,林清清,朱嘉樂,張華. 安宮牛黃丸療效遇科研“挑戰”-北京協和醫院研究稱其在減少腦梗死體積和腦水腫方面效果與安慰劑組無異，業界發出不同聲音.羊城晚報.
2025-09-10.https://news.ycwb.com/ikinvkbtjk/content_53660025.htm
3. 老爸講科學.安宮牛黃丸—無效且危險的安慰劑.知乎.
2025-09-22.https://zhuanlan.zhihu.com/p/1953403504132612538
4. Fan AY, Gu S, Gong C, Xiao S, Tan Y, Tian H, et al.Traditional Chinese Medicine FYTF-919 (Zhongfeng Xingnao Oral Prescription) for the Treatment of Acute Intracerebral Hemorrhage: A Methodological Appraisal of the CHAIN Trial. Altern Ther Health Med. 2025 Oct 8:AT11728. Online ahead of print. PMID: 41056394
5. Anderson CS, Heeley E, Huang Y, Wang J, Stapf C, Delcourt C, et al; INTERACT2 Investigators. Rapid blood-pressure lowering in patients with acute intracerebral hemorrhage. N Engl J Med. 2013;368:2355–2365. doi: 10.1056/NEJMoa1214609
6. Ma C, Wang X, Xu T, Zhang S, Liu S, Zhai C, et al. An Integrative Pharmacology-Based Analysis of Refined Qingkailing Injection Against Cerebral Ischemic Stroke: A Novel Combination of Baicalin, Geniposide, Cholic Acid, and Hyodeoxycholic Acid.Front Pharmacol. 2020 May 8;11:519. doi: 10.3389/fphar.2020.00519. eCollection 2020.PMID: 32457601
7. Fan AY. Comment letter on Tongxinluo and Functional Outcomes Among Patients With Acute Ischemic Stroke published by Dong Y, Jiang K, Li Z, et al. DOI: 10.13140/RG.2.2.26209.98406. Minimal clinical Important Difference (MCID) should be included in this Tongxinluo RCT. Comment on
https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2824104, JAMA Netw Open. 2023; September 28.
8. Gong C. Beyond p-values: Assessing clinical significance in acupuncture research. Adv Chin Med. First published: July 8, 2025. https://doi.org/10.1002/acm4.34
安宮牛黃丸等同安慰劑嗎？——對安宮牛黃丸臨床試驗的評議
作者：鞏昌鎮1 ，樊鎣2（
1.鞏昌鎮( 博士)，《國際臨床針灸雜誌》
主編；全美中醫藥學會(ATCMA)
2.樊鎣(博士)，《結合醫學雜誌》副主編；
美國中醫針灸師，全美中醫藥學會(ATCMA)
摘要
本文圍繞北京協和醫院牽頭的安宮牛黃丸隨機、雙盲、安慰劑對照預試驗展開評議，系統分析其研究設計、結果解讀及社會傳播中的偏差。研究顯示，在120例中重度急性缺血性腦卒中患者中，加用安宮牛黃丸在14天腦梗死體積和腦水腫等主要終點上未達統計學顯著，但在部分亞組及90天功能結局中呈現有利趨勢。作者指出，該結果不能簡單等同於「療效等於安慰劑」。
文章進一步從適應證錯位、邊際效應與療效天花板、評價指標與藥物定位不匹配、樣本量與統計效能不足、劑量與療程不合理等多個方面進行方法學反思，強調「未達顯著」並不等於「證實無效」。同時指出，當前研究將中醫方劑置於西醫範式之下進行單一指標評價，可能導致結論失真。
作者認為，該試驗作為探索性研究具有一定價值，但不足以對安宮牛黃丸療效作出終局判斷。未來研究應嚴格限定中醫證候、優化終點設置、合理設計樣本量與劑量方案，從而更客觀評價其在腦病治療中的真實作用。
關鍵詞：安宮牛黃丸；急性缺血性腦卒中；隨機對照試驗；中醫藥；統計學顯著性；臨床試驗方法學
一、北京協和醫院牽頭的安宮牛黃丸試驗：內容、結論及其社會影響
近期，北京協和醫院牽頭的安宮牛黃丸臨床試驗發表於《中國醫學雜誌》英文版1。該研究被設計為一項多中心、隨機、雙盲、安慰劑對照的預試驗，在17家醫院納入 120 例中度到重度急性缺血性腦卒中（NIHSS 10–20 分）患者，所有患者均接受指南推薦的現代標準腦卒中治療，在此基礎上再加用安宮牛黃丸（安宮組）或外觀相同的安慰劑（安慰組），每日 3 g（1粒），連續 5 天。主要終點並非臨床功能結局，而是14天腦梗死體積與腦水腫體積的變化；次要終點包括 90 天影像學指標、NIHSS 改變和90天mRS 0–2級百分比等。
根據原文描述1，中位年齡為 66.0 歲，基線時 NIHSS 評分中位數為 12.0。納入改良 ITT 分析的樣本數、安宮組和安慰組分別是 57 與 60 例。第 14 天時，安宮組和安慰組的腦梗死體積變化分別為 0.3 mL 和 0.4 mL（中位數差異：-7.1 mL；四分位距 [IQR]：-18.3 至 2.3 mL，P = 0.30）。第 14 天時，安宮組和安慰組的腦水腫體積變化分別為 11.4 mL 和 4.0 mL（中位數差異：3.0 mL，IQR：-1.3 至 9.9 mL，P = 0.15），未達統計學顯著。90 天內嚴重不良事件的發生率在安宮組（3/57，5%）和安慰組（7/60，12%）中相似（P = 0.36）。血清汞和砷濃度的變化也相似。在患有大動脈粥樣硬化的患者中，安宮組降低了第 14 天的腦梗死體積（中位數差異：-12.3 mL；IQR：-27.7 至 -0.3 mL，P = 0.03）。
90 天功能結局方面，mRS 0–2 比例為安宮組38%、安慰組24%，比值比約 1.9，但同樣因樣本量有限而未達統計學顯著。研究者在結論中將整體結果定位為“在患有中度至重度中風的患者中，安宮牛黃丸組的安全性與安慰劑相似，它降低了腦梗死體積、但統計數據未達顯著”、並且指出在大動脈粥樣硬化（LAA）亞組中，14 天梗死體積減少達到統計學顯著性（MD –12.3 mL，P=0.03）。
值得注意的是，文章討論部分明確承認：本研究為“探索性預試驗”，樣本量 120 例是事先人為設定，試驗啟動前並未進行形式上的效能與 I 型錯誤率計算，且失訪與脫落率偏高，結果需要謹慎解讀。然而，在某些大眾媒體2與自媒體等3的傳播中，這一“預試驗、無顯著差異”的結果被概括為“安宮牛黃丸療效≒安慰劑”、“神藥塌房”、“跌落神壇”等極端話語，迅速在社交平臺上擴散，將一個本應理性討論的臨床試驗，變成圍繞傳統名方的情緒化公審。
我們觀察到這一研究結果的兩面性：一方面，多數主要和次要終點在統計學意義上“未達標”；另一方面，無論是 14 天還是 90 天，腦梗死體積、mRS 0–2級的比例等指標在數值上都傾向於安宮組更好，尤其是在 LAA 亞組中出現了統計學顯著的影像學獲益，這些都很難被簡單歸結為完全無效。
二、適應證錯位：從「熱閉神昏」專藥到「中風全程藥」
從中醫經典與方劑學角度和中醫臨床實際看，安宮牛黃丸原本是一味清熱解毒、開竅醒神的名方，其核心適應證是溫熱病或多種急重病中的熱入心包、熱閉神昏、包括高熱、譫語、昏迷、煩躁等證候，屬“熱閉”一類神志障礙。它的價值在於開關、促醒、救急，本不是從頭到尾包攬所有缺血性腦卒中的治療。
這種做法在現代 RCT 語境中看似合理——“同一種病，統一用方”，但從中醫辨證論治的角度看，卻是在“中藥西用”：把一個原本高度依賴證候選擇和病程節律的經典方劑，硬性塞進急性缺血性卒中一籃子患者的框架里，然後得出整體無顯著療效的結論。
在安宮牛黃丸臨床試驗的入排標準中，研究者確實提到，若經中醫師辨證後認為不適宜服用安宮牛黃丸者予以排除，但文中既未正面交代何為適宜證，也未提供任何系統性的中醫證候分佈數據，更沒有將熱閉神昏設為入組必要條件。這就意味着：只要沒有明顯禁忌，大量並非熱閉的中重度缺血性卒中患者，都可以被納入試驗並隨機服用安宮牛黃丸或安慰劑。
試問，如果將針對 ACE 抑制劑適應證的研究，任意擴展到所有血壓值略高或略低、合併各種不同心腎狀態的人群，而不加區分、統一投藥，再用總體無顯著差異去否定 ACE 抑制劑本身的價值，西醫同行會接受這樣的設計和結論嗎？
對安宮牛黃丸而言，安宮牛黃丸臨床試驗當前的結果，更合理的表述是：在未嚴格限定為熱閉神昏證候、而僅以 NIHSS 評分與影像學標準確定的中重度缺血性卒中人群中，短程加用安宮牛黃丸，在 14 天影像學主要終點上未顯示出顯著的群體差異。這與在其經典適應證和合理使用情境下無效之間，仍存在本質差別。
三、邊際效應與療效“天花板”：在標準現代治療之上的一粒藥丸
換言之，安宮牛黃丸不是在“無治療”和“僅中醫治療”的背景下被檢驗，而是在多項循證有效現代療法疊加之後，作為第 N 個干預，被添加到一個已經非常擁擠的治療組合中。
安宮牛黃丸臨床試驗的另一特點，是“全員已接受現代標準卒中治療”1。所有入組患者均按中國神經內科學會指南接受靜脈溶栓治療、血管內介入治療、抗血小板治療、抗凝治療、神經保護治療、降脂、血壓與血糖管理以及規范康復。研究者指出，關鍵治療在兩組間均衡分佈；研究者也明確指出，兩組的伴隨治療是“可比”的。
在這種情境下，至少有三個問題值得警惕：
邊際效應銳減4：如果一個患者已經接受了十幾項現代醫學的有效干預，其中多項本身對梗死體積、腦水腫和功能結局有影響，那麼再額外增加一味藥的邊際貢獻，必然被“平均攤薄”，很容易掉到統計噪音之中。安宮牛黃丸臨床試驗選擇以 120 例的樣本量，在高度複雜的聯合治療背景下觀察數 mL 級別的梗死體積差異，本身就注定了效應極難顯現。
療效天花板效應4：任何急性腦卒中治療都存在生理和生物學的“上限”。當溶栓、血管內治療、降壓、降脂、康復等多項干預已經把部分患者推到接近該上限時，再加入更多治療並不會線性增加好處，反而可能只增加成本、監測負擔與藥物相互作用。對於一味本就定位為促醒、清熱的藥物，把它塞進“現代綜合治療大全”的最高層，要求它在 14 天讓梗死體積出現顯著額外縮小，本身就是苛刻的要求。
評價體系與藥物定位的錯位：對安宮牛黃丸而言，“醒神、改善意識狀態、縮短昏迷或譫妄時間”“緩解高熱及相關神志症狀”更接近它的傳統定位，而不是“單獨負擔梗死體積縮小”。然而，安宮牛黃丸臨床試驗將主要關注點鎖定在影像學指標上，而對 GCS 等“清醒度”指標僅作次要分析。即便在 GCS 7 天和 14 天的變化中已出現接近顯著的差異（P 值均為 0.05 左右），這一信號也沒有被認真展開，而是被淹沒在“主要終點陰性”的表述之中。
我們提出這樣一個頗具啟發性的問題：如果我們反過來設計一個試驗——兩組患者都必須服用安宮牛黃丸（前提是證候相符），在此基礎上再隨機加用或不加用某一西藥干預，再比較結局——那麼結果是否還會是“（某一）西藥等同安慰劑”？這個假設當然不可能真正實施，但它恰恰揭示了當前試驗設計中的單向濾鏡：永遠假定西醫治療是基線，中藥只是被動疊加的可有可無的附屬品。
四、統計學上“無顯著差異”並不等於“證偽”：樣本量、終點與解讀的三重偏移
從嚴格的臨床試驗方法學角度看，安宮牛黃丸臨床試驗的若干設計選擇，決定了它最多只能支持“未能證明有效”，而遠不能推出“被證實無效”。
樣本量與效能問題：研究團隊在文中坦承，本試驗樣本量120例是“探索性設定”，事前沒有進行規範的效能 (power) 和 α 水準計算，也未針對預期效應大小做出清晰假設。值得注意的是，在國際重要醫學雜誌上發表的兩組比較的中風臨床研究4,5，其樣本量達到1,648-2,839，而本研究僅納入120例、使得研究的結論極大概率會落入統計學比較沒有差異的陷阱。
在改良 ITT 集中，真正納入主要終點分析的病例更少（影像學數據缺失）；在 LAA 等亞組中的有效樣本數則進一步降低。這樣的設計，幾乎注定試驗很難對“中等效應量”以上的差異做出可靠判斷，更不用說在多種現代治療疊加背景下去捕捉安宮牛黃丸的細微增益。
終點設定的局限：以 14 天梗死體積和腦水腫體積變化作為雙重主要終點，固然有其科學考慮，但對臨床醫生和患者而言，90 天功能結局（mRS 0–2）才是真正關心的硬終點。在這一更具有臨床意義的指標上，安宮組 38% 對比對照組 24%，比值比接近 2.0，只是因樣本量不足未達統計學顯著。
把這樣的結果簡單歸入與安慰劑無異，無論從臨床直覺還是統計學常識來看，都至少是過度簡化。
非顯著與無差異的概念偷換：在輿論場中，“P>0.05”被快速翻譯為與安慰劑無異，再進一步被演繹為證明無效、神藥塌房。但統計學上，“未發現顯著差異”與“已經證明沒有差異”是兩個概念。前者意味着：在當前樣本量、當前設計與噪音水平下，我們沒能排除真效應存在的可能；後者則需要有足夠的效能、合理的非劣界值甚至等效設計，才能謹慎下結論。安宮牛黃丸臨床試驗本身在討論中尚且保持了“趨勢”、“潛在效應”、“尚需進一步研究的表述”1，而部分媒體與中醫黑話語則直接跨越了這一步，完成了從陰性試驗到證偽中藥的話語飛躍。
五、安宮牛黃丸的使用劑量過小、療程過短
在這項安宮牛黃丸試驗中，安宮牛黃丸劑量是每日1丸（3g）、連用5天。這個劑量對於治療中重度缺血性中風可能是劑量過小、療程過短。
著名中醫周仲瑛在使用安宮牛黃丸治療重症肺炎、引起的高熱、昏迷病人，每日使用6-9粒，連續使用兩週（樊鎣醫生見證）。作者之一的樊鎣醫生治療車禍導致的昏迷，也是每日使用6粒，連用一週以上。由安宮牛黃丸改良而來的現代制劑清開靈注射液，也不是按照安宮牛黃丸每日1粒這個劑量推算而來，而是根據臨床實際重新摸索劑量，治療不同疾病、不同病人，其使用劑量也由不同。6
將安宮牛黃丸中藥西用、由辨證使用改為辨病使用，用以治療中度到重度急性缺血性腦卒中（腦梗塞），先做一系列的預實驗，或者稱為I期臨床試驗，摸索最適合的每日劑量及其療程，摸索與對照組之間判斷療效的最小臨床顯著差距（MCID）7,8，然後才能正式進行臨床試驗（II、III期）。
六、應當如何正確地看待安宮牛黃丸臨床試驗與安宮牛黃丸？
我們並不否認，這項試驗對安宮牛黃丸“去神化”具有積極意義。它提醒臨床醫生和公眾：
• 安宮牛黃丸不是保健品，更不是定期服一丸、包治百病的養生神藥；
• 在急性缺血性卒中中，它不能替代溶栓、取栓、抗血小板、降脂和規范康復等標準治療；
• 含汞、砷等礦物藥的安全性需要在規范監測和合理療程內使用，不能隨意加大劑量或長期服用。
與此同時，我們也必須明確：當前的安宮牛黃丸臨床試驗1並沒有、也無力“推翻”安宮牛黃丸地療效。這項實驗只是表明：在當前較小樣本量、當前入組標準、當前較小用藥劑量及用藥方式、以及當前統計設定之下，安宮牛黃丸在 14 天影像學主要終點上未顯示顯著優勢；在特定亞組（如 LAA）和 90 天功能結局上，則呈現出值得進一步驗證的積極信號。這與療效等同安慰劑、徹底無效之間，相距甚遠。
真正需要反思的，是中醫藥研究被“帝國濾鏡”塑形的方式。當我們習慣於讓中藥和針灸在西醫制訂的話語規則下接受“審判”—以單一疾病標簽取代辨證分型，以單一影像學指標取代整體療效體驗，以短程固定劑量取代病程動態調整——那麼得到的往往只是被迫扮演西藥的失敗者的中醫。安宮牛黃丸臨床試驗的價值，恰恰在於暴露了這一結構性問題：不是安宮牛黃丸的全部，而是被壓扁成西藥影子之後的安宮牛黃丸。
七、下一步應當是什麼？
如果真正希望通過循證方法評價安宮牛黃丸在腦病中的價值，至少需要：
• 在設計中嚴格限定“熱閉神昏”等核心中醫證候，避免適應證被稀釋；
• 將醒神、縮短昏迷或譫妄時間等更貼近藥物本位功用的指標納入主要終點，而不是僅僅寄希望於 14 天梗死體積的附加縮小；
• 在標準治療背景下，現實地評估邊際效應和可接受的效應量，採用合理的樣本量估算與臨床上有意義的 MCID 7,8，而不是用 120 例地樣本去承擔“證偽”與“背書”的雙重任務。
總之，在某種意義上，安宮牛黃丸臨床試驗是一項值得肯定的預試驗，它對安宮牛黃丸的安全性提供了較為嚴謹的數據，也在特定亞組與遠期結局上留下了可供追蹤的線索；但它絕不是給安宮牛黃丸療效和定性的終審判決，更不足以支撐療效等同安慰劑的簡單口號。與其借此掛起中醫“神藥”的大字報，不如把它看作一次必要而不完美的起點：一面鏡子，照出的是我們在如何研究、如何理解、如何評價中醫藥時所帶着的偏見、惰性和濾鏡。
真正需要塌的，不是安宮牛黃丸這味藥，而是那種用單一標準裁剪複雜醫學體系的傲慢。
參考文獻
1. Li S, Wang A, Shi L, Liu Q, Guo X, Liu K, et al. Safety and efficacy of Peaceful Palace Bovine Bezoar Pills in patients with moderate-to-severe acute ischemic stroke (ANGONG TRIAL): A randomized double-blind placebo-controlled pilot clinical trial. Chin Med J. 2025;138(5):579-588.
2. 陳輝, 林清清, 朱嘉樂, 張華. 安宮牛黃丸療效遇科研「挑戰」. 羊城晚報. 2025-09-10.
3. 老爸講科學. 安宮牛黃丸——無效且危險的安慰劑. 知乎. 2025-09-22.
4. Fan AY, Gu S, Gong C, et al. TCM FYTF-919 for Acute Intracerebral Hemorrhage: A Methodological Appraisal of the CHAIN Trial. Altern Ther Health Med. 2025.
5. Anderson CS, Heeley E, Huang Y, et al. Rapid blood-pressure lowering in acute intracerebral hemorrhage. N Engl J Med. 2013;368:2355-2365.
6. Ma C, Wang X, Xu T, et al. An Integrative Pharmacology-Based Analysis of Refined Qingkailing Injection Against Cerebral Ischemic Stroke. Front Pharmacol. 2020;11:519.
7. Fan AY. Comment letter on Tongxinluo and Functional Outcomes. Comment on JAMA Netw Open. 2023.
8. Gong C. Beyond p-values: Assessing clinical significance in acupuncture research. Adv Chin Med. 2025.
