Background: In neck pain treatment, many therapies are focused on etiology, while it is well-known that placebo analgesia is also present in these therapies. The specific efficacy for etiology may be underestimated by ignoring their actual placebo effect. In this study, a logistic regression analysis is used to explore the risk factors causing different placebo responses in patients with neck pain among two RCTs. The probability of the placebo effect is predicted based on these risk factors.

Methods: Trial A and Trial B were similarly designed, randomized, double-/single-blind, placebo-controlled trials in patients treating neck pain with Qishe pill or Shi-style manipulation. Both studies set a placebo pill twice a day or traction for every other day as control. For further analyses on the placebo effect in neck pain management, logistic regression was used to assess subgroup-placebo interactions. The odds ratio assessed a significant influence on the placebo effect.

Results: In this pooled analysis, the total number of patients recruited for these two studies was 284, of which 162 patients received placebo treatment (placebo drug or traction for every other day). No statistically significant differences are found at baseline between the participants with placebo effect and non-placebo effect in the gender, age, and disease duration except in VAS and NDI at the initial time. There are numerically more patients with placebo effect in the shorter disease duration subgroup (< 4 months [76%]), higher initial VAS subgroup (>60 mm [90%]), and worse initial NDI subgroup (>24 [72%]) compared with the gender and age subgroup. An ROC curve is established to assess the model-data fit, which shows an area under the curve of 0.755 and a 95% confidence interval of 0.677–0.830. Participants who show placebo effect after 2 weeks have significantly lower VAS scores after 4 weeks, while there is no significant difference in NDI improvement between the two groups after 4 weeks.

Conclusion: Neck pain patients with shorter disease duration are more likely to overscore their pain severity, because of their less experience in pain perception, tolerance, and analgesia expectation.

背景：在颈部疼痛的治疗中，许多疗法都集中在病因上，而众所周知，这些疗法中也存在安慰剂镇痛剂。忽略其实际的安慰剂效应，可能会低估病因的具体功效。在这项研究中，使用逻辑回归分析来探索在两项 RCT 中导致颈痛患者出现不同安慰剂反应的危险因素。根据这些风险因素预测安慰剂效应的概率。

方法：试验 A 和试验 B 是设计相似的、随机、双/单盲、安慰剂对照试验，患者使用七蛇丸或施式手法治疗颈部疼痛。这两项研究都设定了每天两次的安慰剂药丸或每隔一天服用一次作为对照。为了进一步分析颈部疼痛管理中的安慰剂效应，使用逻辑回归评估亚组-安慰剂相互作用。优势比评估了对安慰剂效应的显着影响。

结果：在这项汇总分析中，这两项研究招募的患者总数为 284 名，其中 162 名患者接受了安慰剂治疗（每隔一天服用安慰剂或牵引）。除了初始时间的 VAS 和 NDI 外，安慰剂效应和非安慰剂效应的参与者在基线时在性别、年龄和疾病持续时间方面没有发现统计学上的显着差异。在较短病程亚组（<4 个月 [76%]）、较高初始 VAS 亚组（>60 mm [90%]）和较差初始 NDI 亚组（>24 [72%]）中，安慰剂效应患者数量较多) 与性别和年龄亚组相比。建立 ROC 曲线来评估模型数据拟合，曲线下面积为 0.755，95% 置信区间为 0.677-0.830。

结论：病程较短的颈痛患者更容易夸大其疼痛严重程度，因为他们在疼痛感知、耐受性和镇痛期望方面的经验较少。