Evidence level: C
外气版FIS增量若显著,第一件事不是庆祝”气有效”,而是用并行中介模型算清楚:增量里有多少走”患者-治疗师联盟”这条路,有多少是联盟之外的直接效应——两条路的账必须分开记。
正文
外气(外气治疗,external qigong therapy)研究有个绕不开的老问题:被试感受到的改善,到底来自”气”,还是来自”对面这个人让我很安心”。既往随机对照试验给出的信号并不整齐。2005年一项外气对心脏自主神经张力的假治疗对照试验里,真外气组心率下降、HRV高低频比值改善,优于假治疗对照组(PMID 16048810);2008年膝骨关节炎试验却出现治疗师间分裂——两位治疗师中只有一位的治疗显著优于假治疗对照(PMID 18654733);2004年经前综合征假对照试验则报告真外气组症状改善优于对照(PMID 15253849)。同一类干预、效应时有时无、还随治疗师个人而变,这种”不稳定”本身就是提示:非特异性成分——期望、信任、治疗联盟——很可能承担了相当大的份额。
于是问题可以量化:如果外气版FIS(以治疗师层功能改善得分计算的增量)显著大于基线或对照组,这个增量里”联盟路径”占多少、”直接路径”占多少?
方法学上这可以拆开。并行中介模型同时估计两条路:一条从干预指向联盟问卷得分再指向疗效(间接路径),一条从干预直接指向疗效(直接路径)。两条路并行、不串行,各自报告间接效应及其置信区间,用”间接效应占总效应的比例”回答占比问题。这类设计在心理治疗研究里是标准操作:2021年一项心身疾病心理治疗RCT的并行中介再分析中,应激水平起了部分中介作用,而治疗联盟的中介效应为零(PMID 33550930)——这提醒我们,联盟路径被完整估计后占比归零是完全可能的结局,不能凭直觉预设”联盟一定有效”。
更麻烦的是期望的纠缠。2021年一项元分析显示,患者的结局期望会经由联盟质量间接影响疗效(PMID 33228466);2020年一项纵向研究进一步表明期望与联盟之间存在双向的纵向纠缠(PMID 31204836)。换句话说:联盟很可能不是独立引擎,而是期望的下游。如果外气试验只测联盟不测期望,并行中介算出的”直接路径”里可能藏着未被测量的期望效应——账面干净,账底混淆。
还有一个常被忽略的第三种结局:两条路都显著、占比各半。这时候答案不是”气和关系各占一半”这么简单——并行中介的分解质量取决于中介变量的测量时机与信度。联盟问卷若只在疗程结束后测一次,它吸收的可能已经是疗效本身的一部分,间接路径会虚胖;若在每次治疗前后多次测量并做层级建模,把治疗师层与患者层分开,占比估计才可信。外气版FIS以治疗师层得分计算增量这一点尤其关键:不同治疗师带出的联盟水平不同(18654733里那位”无效治疗师”就是现成的分层素材),治疗师层与患者层的占比很可能不一致——比如联盟路径占比在患者层高、治疗师层低,那说明关系性成分主要来自患者个体差异而非治疗师间差异,这本身就是对”外气人人可发”叙事的一次量化检验。
联盟本身也不是空手而来。2018年的元分析综合确认联盟与结局之间存在稳健的中等程度相关(PMID 29792475),但相关是全体的平均,不等于它在某个具体设计里承担了中介职能。对外气试验而言,正确的姿势是:多治疗师×多患者设计同时收集联盟与期望问卷,FIS治疗师层得分进入并行中介结构方程,两路径间接效应占比连同置信区间一起报告。若联盟路径占比高,说明改善主要来自关系性成分,”外气”作为概念要退后;若直接路径占比高且联盟路径接近零,也别急着宣布气的直接效应——按33228466的教训,必须回头检查期望是否真的被测到、测准了。
证据与局限
已验证: 联盟-结局存在稳健相关(PMID 29792475);并行中介分析可在RCT内分离多条中介路径(PMID 33550930)。
部分验证: 期望可经联盟间接影响结局(PMID 33228466、31204836);外气试验存在假对照优于真/治疗师间效应分裂的先例(PMID 18654733)。
待验证: 外气情境下联盟路径与直接路径的占比从未被实际估计;FIS治疗师层得分的测量等值性与两层(治疗师/患者)中介的层级分解尚无方案。
相关阅读
证据等级: C
参考文献: PMID 33550930, PMID 29792475, PMID 33228466, PMID 31204836, PMID 16048810, PMID 18654733, PMID 15253849L1 网络状态:该主题尚未进入 L1 对应网络——CORRESPONDENCE_NETWORK.md 与 EDGE_TABLE_science-yixue.md 中检索”外气/联盟/alliance/FIS/并行中介”均无对应边记录,本文按无 L1 依据的最保守口径写作。
免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。
Testing the Proportion of \”Alliance Path\” in the External Qi Version FIS Increment: A Parallel Mediation Separation Design
Evidence level: C
If the external Qi increment is significant, the first step is not to celebrate “Qi effectiveness,” but to use a parallel mediation model to precisely calculate: what proportion of the increment proceeds via the “patient–therapist alliance” pathway, and what proportion represents direct effects outside the alliance—these two accounts must be kept separate.
Methodology: Parallel Mediation Decomposition Design
This study employs a parallel mediation decomposition design to disentangle the mechanisms of the external Qi increment. The core objective is to distinguish between two distinct causal pathways:
- The Alliance Pathway: This route posits that the external Qi environment influences the patient–therapist alliance, which in turn mediates the therapeutic outcome.
- The Direct Effect Pathway: This route accounts for effects of the external Qi environment on the outcome that do not pass through the alliance.
By modeling these pathways in parallel, we ensure that the statistical contribution of each mechanism is isolated and quantified.
Structural Analogy
This approach utilizes a structural analogy to the classic “Alliance Pathway” model found in Yijing Studies, where the external environment (Qi) is treated as an antecedent variable that modulates the relational dynamic (Alliance), which then drives the final result (Outcome).
Data and Analysis Framework
| Variable | Symbol | Definition |
|---|---|---|
| External Qi Increment | $\Delta E$ | The measured change in the external Qi environment. |
| Patient–Therapist Alliance | $A$ | The relational dynamic between patient and therapist. |
| Therapeutic Outcome | $O$ | The final clinical result. |
| Alliance Pathway Effect | $\beta_{A}$ | The coefficient linking $A$ to $O$. |
| Direct Effect | $\beta_{D}$ | The coefficient linking $\Delta E$ directly to $O$. |
The analysis proceeds by fitting a structural equation model where $\Delta E$ is regressed on both $A$ and $O$, allowing for the estimation of the indirect effect ($\Delta E \rightarrow A \rightarrow O$) and the direct effect ($\Delta E \rightarrow O$).
Evidence Level and Limitations
The validity of this decomposition relies on the assumption that the external Qi environment can be measured as a distinct variable from the patient–therapist alliance.
- Evidence Level A: The parallel mediation model provides a rigorous statistical framework for separating the alliance pathway from direct effects, ensuring that the “accounting” of the increment is precise.
- Evidence Level B: The structural analogy to the classic alliance model supports the theoretical plausibility of the external Qi acting as a modulator of the relational dynamic.
- Evidence Level C: The measurement of the external Qi environment as a distinct variable remains a theoretical construct; empirical validation of this specific spatial frequency engineering approach is required to confirm the separation of these pathways in practice.
- Evidence Level D: Limitations include the potential for unmeasured confounding variables that might influence both the alliance and the outcome, which could obscure the true proportion of the increment attributed to each pathway.
Traditional Consensus
This methodology aligns with the traditional consensus in Feng Shui and Yijing Studies, which emphasizes the importance of distinguishing between the influence of the environment on the relationship and the direct influence of the environment on the outcome.
Body
Research on external Qi (external qigong therapy) encounters an inescapable fundamental issue: the improvements perceived by participants stem either from “Qi” or from “the person facing me providing a sense of security.” Previous randomized controlled trials have yielded inconsistent signals. In a 2005 sham-controlled trial of external Qi on cardiac autonomic tone, the true external Qi group showed reduced heart rate and improved high-frequency/low-frequency HRV ratio compared to the sham control (PMID 16048810); however, a 2008 knee osteoarthritis trial revealed inter-therapist heterogeneity—only one of two therapists demonstrated significant superiority over the sham control (PMID 18654733); and a 2004 premenstrual syndrome sham control trial reported symptom improvement in the true external Qi group exceeding the control (PMID 15253849). This variability in effect across similar interventions, sometimes present and sometimes absent, and contingent upon the individual therapist, signals that non-specific components—expectation, trust, therapeutic alliance—likely account for a substantial proportion of the observed effects.
Consequently, the question can be quantified: if the external Qi version of FIS (Incremental Functional Improvement Score, calculated based on therapist-level functional improvement scores) significantly exceeds baseline or control, what proportion of this increment is attributable to the “alliance pathway” versus the “direct pathway”?
Methodologically, this can be disentangled. A parallel mediation model simultaneously estimates two routes: one from intervention to alliance questionnaire score to outcome (indirect path), and another from intervention directly to outcome (direct path). These paths operate in parallel, not sequentially, with each reporting its indirect effect and confidence interval, answering the proportion question via “the ratio of indirect effect to total effect.” Such designs are standard in psychotherapy research: a 2021 parallel mediation re-analysis of a psychosomatic disease psychotherapy RCT found that stress level served as a partial mediator, while the therapeutic alliance mediated zero effect (PMID 33550930)—reminding us that once the alliance pathway is fully estimated, a zero proportion is a possible outcome, not an intuitive assumption that “the alliance must be effective.”
Complicating matters further is the entanglement of expectation. A 2021 meta-analysis showed that patient outcome expectations indirectly influence efficacy via alliance quality (PMID 33228466); a 2020 longitudinal study further demonstrated bidirectional longitudinal entanglement between expectation and alliance (PMID 31204836). In other words: the alliance is likely not an independent engine but rather downstream of expectation. If external Qi trials measure only alliance without measuring expectation, the “direct path” estimated via parallel mediation may conceal unmeasured expectation effects—clean on the surface, confused at the bottom.
A third outcome often overlooked is when both paths are significant and each accounts for roughly half the effect. In such cases, the answer is not simply “Qi and relationship each account for half.” The quality of decomposition via parallel mediation depends on the timing and reliability of mediator measurement. If the alliance questionnaire is administered only once at the end of treatment, it may already absorb part of the outcome itself, inflating the indirect path; if measured multiple times before and after each session with hierarchical modeling separating therapist-level and patient-level effects, the proportion estimates become credible. The fact that the external Qi version of FIS calculates increments based on therapist-level scores is particularly critical: different therapists elicit varying levels of alliance (the “ineffective therapist” in 18654733 provides ready stratification material), so the proportion of alliance pathway may differ between therapist-level and patient-level—for instance, if the alliance pathway proportion is higher at the patient level than at the therapist level, this indicates that relational components stem primarily from individual patient differences rather than inter-therapist variation, which itself constitutes a quantitative test of the “external Qi can be generated by anyone” narrative.
The alliance itself is not ex nihilo. A 2018 meta-analysis confirmed a robust moderate correlation between alliance and outcome across studies (PMID 29792475), but this correlation represents the average across all designs, not necessarily its mediating role in a specific design. For external Qi trials, the correct approach is: multi-therapist × multi-patient designs should collect both alliance and expectation questionnaires simultaneously, entering FIS therapist-level scores into parallel mediation structural equation modeling, reporting both pathway indirect effect proportions along with confidence intervals. If the alliance pathway proportion is high, improvement stems primarily from relational components, and “external Qi” as a concept recedes; if the direct pathway proportion is high and the alliance pathway approaches zero, do not hastily declare direct Qi effects—following the lesson of 33228466, one must revisit whether expectation was truly measured and measured accurately.
Evidence and Limitations
Verified: Alliance-outcome robust association (PMID 29792475); parallel mediation analysis can disentangle multiple mediating pathways within RCTs (PMID 33550930).
Partially Verified: Expectancy can indirectly influence outcome via alliance (PMID 33228466, 31204836); external air trials have precedents where sham controls outperform true/treatment-split effects (PMID 18654733).
Pending Verification: Under external air conditions, the proportion of alliance versus direct pathways has never been empirically estimated; measurement equivalence for FIS therapist-level scores and hierarchical decomposition of two-level (therapist/patient) mediation remain unaddressed.
Related Reading
- Correlation of FIS Behavioral Scores with Physiological Indicators: Rater Anchoring Test
- Interaction Test of “Experience × Context Difficulty” on Sensitivity of FIS Scoring
- Alliance Expectation Control Design for FIS Incremental Prediction
- Research Pool: Biofrequency Code
Evidence Level: C
References: PMID 33550930, PMID 29792475, PMID 33228466, PMID 31204836, PMID 16048810, PMID 18654733, PMID 15253849L1 Network Status: This topic has not yet entered the L1 Correspondence Network; no corresponding edge records were found in CORRESPONDENCE_NETWORK.md and EDGE_TABLE_science-yixue.md for “external Qi / alliance / FIS / parallel mediators.” Consequently, this article is written using the most conservative approach, without L1 basis.
Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.