Reply to Zaman et al.
Reply to Zaman et al.
复制标题
回复扎曼等人。
DOI：
10.1097/j.pain.0000000000001572
复制
发表时间：
2019
期刊：
影响因子：
7.4
通讯作者：
Atlas,LaurenY
中科院分区：
文献类型：
--
作者：
Mischkowski,Dominik;Atlas,LaurenY
我们感谢Zaman et al. 16对我们最近文章的回应。我们这篇文章的主要目的是评估诸如评价之类的心理过程是否介导温度对自主神经反应的影响，或者是否如IASP定义所提出的那样，对有害热量的自主神经唤起纯粹是反射性的（这是我们多次强调的目标）。我们认为这个问题比Zaman等人关注的疼痛和自主神经反应之间的方向性问题更基本，但次要。这封信还强调了我们在讨论11中明确讨论过的两个要点：(1)对相关调解的警告；(2)心理因素对疼痛和疼痛诱发的自主反应的影响。因此，尽管有许多观点是一致的，但不幸的是，Zaman等人错误地描述了我们工作的中心目标。我们同意中介分析不足以建立因果关系，这在统计学、实验心理学和相关领域已经被广泛讨论过。4,8,9,13 - 15在主流的中介方法中，研究人员检验被操纵或测量的因果变量与假设的效应变量之间的关联，并检验测量的中介变量是否在统计上解释了这种关联。理想情况下，可以通过实验操纵中介来建立因果机制。在我们的讨论中，我们建议未来的研究使用药理学干预来调节疼痛，并测试强度对自主神经反应的影响是否被消除。这将补充和扩展我们的统计中介，并表明因果关系。类似的方法（即，疼痛与交感神经系统活动的药理学调节）可以正式测试Zaman等人提出的躯体标记假设5和疼痛与自主神经反应之间的方向性问题16。当介质被测量而不是被操纵时，这种方法是有限的，并且可能倾向于替代解释。正如Zaman等人所讨论的那样：(1)中介与中介模型中其他变量之间的反向或互惠关系8,9,14和(2)混淆可能解释中介模型中变量之间相关性的第三个变量。4,6,10我们同意Zaman等人的观点，即我们对疼痛在热效应中对自主神经反应的作用的研究在明确揭示疼痛作为中介的因果结论方面是有限的。这就是为什么我们在讨论中广泛讨论了局限性并提供了实验性解决方案。然而，我们不同意Zaman等人的建议，即我们误用了“调解”一词。Zaman等人假设时间优先，即在被提议的因变量之前测量被提议的中介是将变量标记为中介的必要条件。因为我们必须在自主神经反应之后测量疼痛，我们应该违反了这个标签标准。然而，即使在测量所提出的原因(X)时，时间优先于所提出的结果(Y)，也不能排除
We appreciate the letter by Zaman et al. 16 in response to our recent article. 11 The principal objective of our article11 was to evaluate whether psychological processes such as appraisal mediate the effects of temperature on autonomic responses, or whether autonomic arousal in response to noxious heat is purely reflexive, as proposed by the IASP definition7—a goal which we assert multiple times (eg, Introduction11). We feel that this question is more fundamental than the related, but secondary, question of directionality between pain and autonomic responses, on which Zaman et al. 16 focus. The letter also highlights 2 important points that were discussed explicitly in our Discussion11:(1) caveats to correlational mediation and (2) the influence of psychological factors on pain and pain-evoked autonomic responses. Thus, although there are many points of agreement, Zaman et al. have unfortunately mischaracterized the central goal of our work. We agree that mediation analysis is not sufficient to establish causality, as has been discussed extensively in statistics, experimental psychology, and related fields. 4, 8, 9, 13–15 In the dominant approach to mediation, researchers test the association of a manipulated or measured proposed causal variable with a putative effect variable and test whether a measured mediator variable statistically accounts for this association. Ideally, the mediator would then be experimentally manipulated to establish a causal mechanism. 13 In our Discussion, 11 we propose that future investigations use pharmacological interventions to modulate pain and test whether intensity effects on autonomic responses are abolished. This would supplement and extend our statistical mediation and indicate causality. A similar approach (ie, pharmacological modulation of pain vs sympathetic nervous system activity) could formally test the somatic marker hypothesis5 and questions of directionality between pain and autonomic responses raised by Zaman et al. 16 When a mediator is measured and not manipulated, this approach is limited and can be prone to alternative explanations, as discussed by Zaman et al. 16:(1) reversed or reciprocal relationships between the mediator and other variables in the mediation model8, 9, 14 and (2) confounding third variables that may account for correlations between variables in the mediation model. 4, 6, 10 We agree with Zaman et al. that our research on the role of pain in heat effects on autonomic responses is limited in its ability to definitely reveal causal conclusions about pain as a mediator. This is why we extensively discussed limitations and offered experimental solutions in our Discussion. 11 However, we disagree with Zaman et al.’s suggestion that we misuse the term “mediation.” Zaman et al. postulate that time precedence, ie, measuring the proposed mediator before the proposed dependent variable is a sine qua non condition for labeling a variable as mediator. 8 Because we had to measure pain after the autonomic response, we supposedly violated this labeling standard. However, even time precedence in measuring the proposed cause (X) before the proposed effect (Y) does not rule out the possibility of
影响因子：
2.4
作者：
Thoemmes, Felix
通讯作者：
Thoemmes, Felix
影响因子：
3.5
作者：
Lemmer, Gunnar;Gollwitzer, Mario
通讯作者：
Gollwitzer, Mario
影响因子：
3.5
作者：
Fiedler, Klaus;Schott, Malte;Meiser, Thorsten
通讯作者：
Meiser, Thorsten