Comparative Efficacy of Cognitive Training for Post-Stroke Aphasia: A Systematic Review and Network Meta-Analysis.

Cognitive training was defined loosely. The 11 trials used six different add-ons aimed at different abilities (working memory, attention, general cognition, virtual reality), so grouping them together hides real differences, and the authors say generalising from the result is difficult. Some trials did not describe how they generated the random sequence or hid the allocation, and neither participants nor therapists could be blinded, so the reported effects may be larger than the true ones. The trials were small and varied in dose, duration and how often therapy was given. With so few studies the authors could not run subgroup analyses to find out whether any of that mattered. Every trial used the same language test, so nothing here speaks to everyday communication, participation or quality of life. Nine of the eleven trials recruited people whose first language was Chinese, and six were published in Chinese, so it is unclear how far the results carry to other languages and health systems. Several of the confidence intervals were wide and came close to zero. For repetition the interval ran from 0.08 to 2.93, so the true effect could be very small.

The authors declared no competing interests and no potential conflicts of interest with respect to the research, authorship or publication of the article. The work was funded by Chinese government and institutional sources: the National Natural Science Foundation of China (81973790), the National Administration of Traditional Chinese Medicine's International Cooperation Department, and a Beijing University of Chinese Medicine Dongzhimen Hospital clinical research project (DZMG-XZYY-23008).

Do not read the ranking of working memory training as proof that it is the best add-on to speech and language therapy. That ranking comes from a statistical ordering built largely on indirect comparisons between small trials, not from trials that tested the add-ons against each other, and all 11 studies were rated at some concerns for risk of bias. Likewise, an add-on failing to separate from speech and language therapy alone here does not show it does nothing; the trials were few and small.