Intra- and Interjudge Reliability of the Apraxia of Speech Rating Scale in Early Stroke Patients.

Purpose: There is a recognized need for a reliable assessment instrument for apraxia of speech (AOS) diagnosis for post stroke patients. In 2014, Strand and colleagues reported high to excellent intra- and interjudge reliability of the Apraxia of Speech Rating Scale (ASRS) in neurodegenerative speech and language disorders. Excellent interjudge reliability of the ASRS total score has also recently been reported in another study of individuals with chronic AOS after stroke, where the ratings were carried out by two experienced researchers not involved in the development of the instrument. However, it is still not fully determined whether the ASRS is a reliable instrument in assessment of patients in an early phase after stroke, where severe AOS is not uncommon. It is also not determined whether ASRS ratings can be performed reliably by practicing speech-language pathologists (SLPs) without long common experience and joint training. This study therefore addresses these questions.

Method: The ASRS was administered to thirty-six individuals in the first six months after stroke. The assessment procedures were video recorded. Ten of the recordings were selected for the reliability study, representative of patients typically seen at the actual inpatient ward with varying degrees of AOS severity. Five SLPs from different hospital departments participated as raters. The raters viewed each video and independently rated the presence and severity of AOS using the ASRS. To study intrajudge reliability, a rescoring was performed after a minimum of three weeks. Reliability was calculated using the intraclass correlation coefficient (ICC).

Results: Intrajudge agreement for the ASRS total score varied from moderate to excellent (mean ICC = 0.69, 95 % CI [0.60, 0.77]) with most of the mean item level agreements within the categories 'moderate' or 'good. Interjudge reliability was poor for the ASRS total score (ICC = 0.42, 95 % CI [0.35, 0.50]). The item level results varied between moderate and poor, with lack of agreement on several items.

Conclusions: For clinicians without expert knowledge of AOS and limited training using the ASRS, intra- and interjudge reliability of the ASRS is not satisfactory. Also, since some items in the protocol require a certain level of speech production to target the diagnostic marker, findings indicate that the ASRS in its present design has limitations in assessment of severe AOS. As suggested by Strand and colleagues, video examples that illustrate the ASRS characteristics could be one helpful alternative to support clinician training. A minor revision of response definitions of the scale may improve the applicability of the ASRS in severe AOS.