Evaluation of the SWAN Game-Based Approach to Re-Building Numeracy Skills in Aphasia: Feasibility and Preliminary Findings.

There was no control group, no randomisation and no blinding, so nothing separates the effect of the game from natural recovery, the extra attention, or practice at the tests themselves. The group's scores were already improving between the two tests done before treatment began, and the non-numerical control task improved as well. Both point to change that is not specific to the game. Only 18 people took part and the authors' own post-hoc power calculation suggests the study was underpowered for counting (64%), so a real effect there could have been missed. The number tests were designed by the authors for this study and are not fully validated, and the control task kept improving across every time point, which weakens it as a check on specificity. Participants averaged younger than is typical for stroke, and no one had Wernicke's aphasia, so the results may not hold for older adults or people with severe comprehension problems. The dose was low, 3 hours 45 minutes requested against the 20 hours or more used in other number therapy studies, and the game forced people through levels in order, which many found too easy. Five people showed no change at all and three got measurably worse, and the authors could not tell whether the people who reported feeling more confident were the ones who improved on testing.

The work was funded by a research grant from the Nuffield Foundation. The authors declare no conflicts of interest. The game was built by the research team, and the paper does not state who owns it or whether it will be sold.

Do not read the improvements as proof that this game improves number skills in aphasia. Scores were already rising between the two tests done before anyone played, and a non-numerical puzzle task improved at the same time, so practice, attention or natural recovery could explain the change. Everyday number tasks did not improve.