The scene repeats itself in speech and language clinics on both sides of the Atlantic. A person with primary progressive aphasia arrives with a printout: a case report, a conference abstract, a video of someone with a rubber headband and two damp electrodes naming pictures faster than they could a month earlier. The question that follows is always some version of the same one. Could we buy that? It is a fair question, and until recently the honest answer was that the field had lots of single cases and not much arithmetic. Now there is arithmetic.

A systematic review and meta-analysis published in Neuroscience and Biobehavioral Reviews (Godoi and colleagues, with Cardiff University's David McGonigle as senior author, online 18 June 2025) pooled every randomised controlled trial it could find that compared transcranial direct current stimulation, tDCS, against sham stimulation in people with PPA. The searches ran to July 2024, the protocol was registered in advance on PROSPERO (CRD42024499012), and the yield was ten parallel and cross-over randomised trials, 178 patients, 218 observations. Two results cleared the bar. General naming ability improved, with a standardised mean difference of 0.37 (95 percent confidence interval 0.07 to 0.67). Spelling improved more, at 0.65 (0.10 to 1.20). A standardised mean difference is a way of expressing benefit in units of how much people's scores vary anyway; 0.37 is small, real, and a long way from transformative.

What did not move matters just as much, and this is the part that rarely survives the journey into a clinic waiting room. Pooled across the same trials, there was no significant difference for naming of trained items or untrained items, no significant difference in global language, none in working memory, none in semantic fluency. So the pooled signal sits on general naming and on spelling, and thins out everywhere else the researchers looked. It is also worth being precise about what the comparison was: active stimulation against sham stimulation, in trials where both groups were being tested and worked with. The question the evidence answers is what the current adds, not whether therapy is worth doing.

The picture is not perfectly tidy, which is itself informative. A separate systematic review and updated meta-analysis in Neurological Sciences (Alhwaishel and colleagues, 2025), drawing on ten randomised trials and 167 participants with searches to June 2024, reported a significant benefit for trained naming tasks (0.96, 95 percent CI 0.27 to 1.65) with effects persisting up to two months, alongside a borderline overall improvement immediately after stimulation, and again found nothing significant for cognitive tasks, untrained comprehension, or trained semantic tasks. Two teams, overlapping trials, different groupings of outcomes, different answers about trained words. That is what a small, heterogeneous literature looks like when you squeeze it, and it is why neither review reads as a green light.

Meanwhile the delivery model is moving faster than the effect size. A German pilot case series published in Brain Sciences on 10 July 2025 put stimulation in people's homes, self-administered with a caregiver's help alongside daily teletherapy over ten consecutive workdays: two participants, a 95 percent completion rate, well tolerated, with improvements in naming and communication that the authors describe as preliminary and needing a larger trial. At the University of Texas at Austin, with the UCSF Memory and Aging Center as a site, a randomised trial of remotely supervised tDCS plus virtual speech-language therapy in logopenic variant PPA (NCT05901233) began in July 2023 and is registered to finish in September 2026. A larger NIH-funded successor at the same lab (NCT07260253) is listed as recruiting, with primary completion estimated for 2029. Feasibility is close to settled. Efficacy is not.

The regulatory position is the last piece a family should have in hand. On 8 December 2025 the US Food and Drug Administration granted premarket approval to the Flow FL-100 tDCS device (P230024), the first home-use tDCS approval of its kind, and it is for major depressive disorder. Not for aphasia, not for dementia, not for word finding. Devices sold for wellness or sold offshore carry no such review at all, and the trial protocols that do exist screen carefully: histories of seizure, skull surgery, metal in the head, implanted stimulators and skin conditions are exclusions, which is a clue about why the supervision in these studies is called supervision.