Article
Verbal learning memory testing for longitudinal assessment of German brain tumour patients
Analyse und Weiterentwicklung eines verbalen Gedächtnistests für die Verlaufsbeobachtung deutschsprachiger Hirntumorpatienten
Search Medline for
Authors
Published: | June 26, 2020 |
---|
Outline
Text
Objective: Due to its relative simplicity in comparison to other verbal learning memory tests (VLMT) the revised Hopkins VLMT (HVLMT-R) is used most often for the assessment of brain tumour patients. The German version of the HVLMT-R consists of a translation of the original English HVLMT-R, using exactly the same items – without acknowledging language-specific linguistic parameters of standardization. We analysed the comparability of the parallel German word lists regarding different levels of word processing and memorization.
Methods: Ninety patients with primary glioblastoma (m = 53, age range: 24-81 years) were included. Two different parallel versions of the HVLMT-R consisting of twelve nouns grouped in three semantic categories were administered to the patients on two consecutive days. After three learning trials with immediate recall, delayed recall and word recognition (from a read-out list of 24 items including 12 distractors) were assessed. Kendall’s rank correlation coefficient b was calculated to compare the test results between parallel versions.
Results: In a descriptive linguistic analysis the German wordlists of the HVLMT did not conform to parallelized item sets concerning linguistic parameters (e.g. number of syllables, word frequency, morphological and/or phonological complexity, lexicosemantic ambiguity) e.g., the total number of syllables varies between 18 and 30 and mean word frequency ranges from 5.25 to 28.9. The retest-reliability for the overall learning trial as well as for the late recall trial showed a moderate correlation (b=0.5 – 0.6; p<0.001), whereas the retest-reliability for the percentage of retention was weak (b=0.40, p<0.001).
Conclusion: Our preliminary results support our hypothesis that the translated parallel versions of the HVLMT-R do not compare well regarding linguistic parameters, which might be one major reason for a only weak to moderate parallel-test reliability. This underlines the need for a linguistically standardized construction of native-language based word lists. Such alternative, parallelized word lists developed by our group are currently under pre-clinical evaluation.