Gates Kelly E, Mefferd Antje S, Stipancic Kaila L
Department of Communicative Disorders and Sciences, University at Buffalo, NY.
Department of Hearing and Speech Sciences, Vanderbilt University Medical Center, Nashville, TN.
J Speech Lang Hear Res. 2025 Aug 12;68(8):3771-3788. doi: 10.1044/2025_JSLHR-24-00899. Epub 2025 Jul 21.
The minimally detectable change (MDC), widely used in rehabilitation sciences to interpret changes in outcome measures, is calculated using a reliability method, reliability statistic, and standard error of measurement (). This study examined how different methodological choices affect MDC thresholds of speech intelligibility in speakers with dysarthria. The goals of this study were to compare MDCs calculated using (a) three different reliability methods, (b) two different reliability statistics, and (c) three different calculations.
Recordings of the Speech Intelligibility Test from 200 speakers including speakers with amyotrophic lateral sclerosis ( = 16), Huntington's disease ( = 44), multiple sclerosis ( = 60), and Parkinson's disease ( = 40), along with healthy controls ( = 40), were drawn from two databases. Thirty inexperienced listeners completed two sessions, providing orthographic transcriptions of 20 speakers. MDCs of intelligibility were calculated using (a) three reliability methods (i.e., test-retest, split-half, and intrarater), (b) two reliability statistics (i.e., Pearson and intraclass correlation coefficients [ICCs]), and (c) three different formulas for calculating the . Kruskal-Wallis tests were used to assess the effects of reliability methods, statistics, and calculations.
Significant differences were found between the MDCs when using split-half and test-retest reliability, when using Pearson and ICC, and when using two of the three calculations.
Results demonstrate that methodological decisions can impact MDCs of speech intelligibility in speakers with dysarthria, highlighting the need for specific, detailed reporting of methodology used to calculate MDCs in future work. Findings can provide methodological guidance for future studies and contextualize existing research on intelligibility changes.
最小可检测变化(MDC)在康复科学中被广泛用于解释结果测量的变化,它是使用一种可靠性方法、可靠性统计量和测量标准误差来计算的。本研究考察了不同的方法选择如何影响构音障碍患者言语清晰度的MDC阈值。本研究的目的是比较使用以下方法计算的MDC:(a)三种不同的可靠性方法;(b)两种不同的可靠性统计量;(c)三种不同的计算方法。
从两个数据库中选取了200名受试者的言语清晰度测试录音,包括肌萎缩侧索硬化症患者(n = 16)、亨廷顿舞蹈症患者(n = 44)、多发性硬化症患者(n = 60)、帕金森病患者(n = 40)以及健康对照者(n = 40)。30名没有经验的听众完成了两个阶段的测试,对20名受试者的言语进行正字法转录。使用以下方法计算言语清晰度的MDC:(a)三种可靠性方法(即重测法、分半法和评分者内法);(b)两种可靠性统计量(即皮尔逊相关系数和组内相关系数[ICC]);(c)三种不同的测量标准误差计算公式。使用Kruskal-Wallis检验来评估可靠性方法、统计量和测量标准误差计算方法的影响。
在使用分半法和重测法的可靠性时、使用皮尔逊相关系数和ICC时以及使用三种测量标准误差计算方法中的两种时,MDC之间存在显著差异。
结果表明,方法学决策会影响构音障碍患者言语清晰度的MDC,这突出了在未来工作中需要具体、详细地报告用于计算MDC的方法。研究结果可为未来研究提供方法学指导,并将现有的关于清晰度变化的研究置于具体情境中。