Interchangeability of two versions of the Montreal Cognitive Assessment for the longitudinal evaluation of patients with breast cancer.

PURPOSE The cognitive performance of patients with breast cancer (BCa) may be affected by cancer and its treatments. The Montreal Cognitive Assessment (MoCA) is a widely used cognitive impairment screening tool, but practice effects must be considered for longitudinal assessments. Since learning effects could be overcome through the alternate use of two versions of the MoCA, we aimed to explore their interchangeability by comparing their overall, and domain- and task-specific, scores among patients with BCa. METHODS BCa patients from the NEON-BC cohort were evaluated with the MoCA, version 7.1, after diagnosis and after 1 year. At the 3-year follow-up (n = 422), the 7.1 and 7.3 versions were applied at the beginning and at the end (approximately 1 h later) of this evaluation, respectively. Agreements between versions, regarding total, sub-domain, and task scores, were assessed using Bland-Altman plots and intraclass correlation coefficients (ICC). RESULTS The mean total scores were not statistically different between versions and the ICC was 0.890. The Bland-Altman limits of agreement were - 3.70 to 3.88. For women with midrange scores, total scores were significantly higher in version 7.1. There were significant differences in the percentage of correct answers in 7 out of 12 tasks, being the highest for the copy of a geometric figure (more than twofold higher with version 7.3). In version 7.1, the language and memory domains presented higher scores and lower visuospatial ability. CONCLUSION Despite similar overall scores being obtained with the two versions of the MoCA, there were item-specific differences that may compromise their interchangeable use.