Rahlfs Volker W, Zimmermann Helmuth
idv-Data Analysis and Study Planning, Gauting, Germany.
Front Neurol. 2021 May 5;12:601749. doi: 10.3389/fneur.2021.601749. eCollection 2021.
The Mean Vertigo Score (MVS) is a composite score for defining the burden of disease of patients suffering from vestibular disorders. It has been used in clinical research for about 30 years. This study investigates discriminant validity of the MVS and describes structural relationships of the 12 single criteria used for construction of the MVS. The statistical analyses are based on the raw data of an earlier conducted randomized, doubleblind, placebo-controlled clinical trial, which compared the following four randomized treatment groups: a fixed combination of cinnarizine and dimenhydrinate (Arlevert), two groups with only one of the two study drugs, and a group with placebo. The method used for the statistical calculations is the Wei-Lachin procedure, a multivariate generalization of the Mann-Whitney test, which takes into account correlations among the 12 single symptoms of the composite score. All 12 single symptoms of the composite endpoint proved to be useful for detecting differences (Mann-Whitney effect size measures: 0.58-0.73) and thus for discriminating between treatment groups. Their Pearson product-moment correlations are all positive (range 0.07-0.71) and point to the same direction, which indicates one-dimensionality and good internal consistency of the composite index MVS. Furthermore, our statistical calculations revealed that successively increasing the number of single items of the MVS to up to twelve enhances its reliability ( = 0.923), which leads to a substantially higher test power and reduction of the number of patients needed (sample size) in a clinical trial. The use of the multivariate Wei-Lachin procedure provides further evidence of the validity of the 12-item composite score MVS, based on the efficacy data of its 12 single vertigo symptoms. The present findings demonstrate that the MVS is a powerful tool, which can be used to adequately describe the patients' self-perceived vertigo complaints, both qualitatively and quantitatively. It may therefore be regarded as a clinically meaningful alternative to other questionnaires that are presently used in vestibular research.
平均眩晕评分（MVS）是用于定义前庭疾病患者疾病负担的综合评分。它已在临床研究中使用了约30年。本研究调查了MVS的判别效度，并描述了用于构建MVS的12个单一标准的结构关系。统计分析基于先前进行的一项随机、双盲、安慰剂对照临床试验的原始数据，该试验比较了以下四个随机治疗组：桂利嗪和茶苯海明的固定组合（阿乐韦特）、两组仅使用两种研究药物中的一种，以及一组使用安慰剂。用于统计计算的方法是魏-拉钦程序，这是曼-惠特尼检验的多变量推广，它考虑了综合评分的12个单一症状之间的相关性。复合终点的所有12个单一症状都被证明可用于检测差异（曼-惠特尼效应量测量：0.58 - 0.73），从而用于区分治疗组。它们的皮尔逊积矩相关性均为正（范围0.07 - 0.71）且指向同一方向，这表明复合指数MVS具有一维性和良好的内部一致性。此外，我们的统计计算表明，将MVS的单一项目数量依次增加至多达12个可提高其可靠性（ = 0.923），这会显著提高检验效能并减少临床试验所需的患者数量（样本量）。基于其12个单一眩晕症状的疗效数据，多变量魏-拉钦程序的使用为12项复合评分MVS的效度提供了进一步证据。目前的研究结果表明，MVS是一个强大的工具，可用于定性和定量地充分描述患者自我感知的眩晕症状。因此，它可被视为目前前庭研究中使用的其他问卷在临床上有意义的替代方法。
