Department of Neurosurgery, Leiden University Medical Centre (LUMC), PO Box 9600, 2300 RC Leiden, The Netherlands.
J Clin Epidemiol. 2013 Dec;66(12):1347-55.e3. doi: 10.1016/j.jclinepi.2013.06.007. Epub 2013 Sep 7.
The goal of this systematic review was to evaluate if the influence of methodological features on treatment effect differs between types of intervention.
MEDLINE, Embase, Web of Science, Cochrane methodology register, and reference lists were searched for meta-epidemiologic studies on the influence of methodological features on treatment effect. Studies analyzing influence of methodological features related to internal validity were included. We made a distinction among surgical, pharmaceutical, and therapeutical as separate types of intervention. Heterogeneity was calculated to identify differences among these types.
Fourteen meta-epidemiologic studies were found with 51 estimates of influence of methodological features on treatment effect. Heterogeneity was observed among the intervention types for randomization. Surgical intervention studies showed a larger treatment effect when randomized; this was in contrast to pharmaceutical studies that found the opposite. For allocation concealment and double blinding, the influence of methodological features on the treatment effect was comparable across different types of intervention. For the remaining methodological features, there were insufficient observations.
The influence of allocation concealment and double blinding on the treatment effect is consistent across studies of different interventional types. The influence of randomization although, may be different between surgical and nonsurgical studies.
本系统评价旨在评估方法学特征对治疗效果的影响是否因干预类型而异。
检索了 MEDLINE、Embase、Web of Science、Cochrane 方法学登记册和参考文献列表，以寻找关于方法学特征对治疗效果影响的元流行病学研究。纳入了分析与内部有效性相关的方法学特征影响的研究。我们将手术、药物和治疗三种干预类型分别区分开来。计算异质性以确定这些类型之间的差异。
共发现 14 项元流行病学研究，其中有 51 项关于方法学特征对治疗效果影响的估计。在随机分组方面，干预类型之间存在异质性。随机分组的手术干预研究显示出更大的治疗效果；这与药物研究的结果相反，药物研究发现了相反的结果。对于分配隐藏和双盲，不同干预类型的方法学特征对治疗效果的影响是可比的。对于其余的方法学特征，观察到的结果不足。
在不同干预类型的研究中，分配隐藏和双盲对治疗效果的影响是一致的。尽管随机分组的影响可能在手术和非手术研究之间有所不同。
