An evidence review has placed exercise, reminiscence and several forms of cognitive training among the clearest recommendations for non-drug care in Alzheimer’s disease. It also gave a Grade A recommendation to high-frequency rTMS for people with mild-to-moderate disease, while placing adaptive digital cognitive therapy in Grade B and marking it for revision; simple acupuncture and combined scalp-abdominal acupuncture were also Grade B. The review does not provide a fixed estimate of cognitive improvement: its recommendations combine evidence quality with feasibility, resources, patient preferences and clinical context, rather than treatment effect size alone.
How the evidence was assembled
The review was an integrative scoping review and best-evidence summary. It used a registered protocol under the JBI framework and PRISMA-ScR, along with a five-level 5S model to prioritize pre-appraised evidence. Eligible sources concerned people with Alzheimer’s disease dementia, or Alzheimer’s subgroups in mixed populations, and used recognized global cognitive scales and specified diagnostic criteria.
The source base comprised 15 evidence documents: one guideline, seven expert consensus reports, six systematic reviews and one randomized controlled trial. Researchers assessed each source with tools matched to its design, checking disagreements through discussion or consultation with a third researcher. The evidence was then combined qualitatively, without pooling effect sizes; the JBI system assigned evidence levels from Level 1 to Level 5 by study design.
The review extracted 18 evidence items and assigned 10 Grade A and eight Grade B recommendations. It counted seven Level 1 items and 11 Level 5 items. Expert ratings showed basic agreement: Kendall’s concordance coefficient, a measure of how closely ratings matched, was 0.562, and the reported P value for the expert-rating concordance test was less than 0.001. Those grades should not be read as a simple effectiveness ranking, because the review also considered feasibility, resources, patient preferences and clinical context.
What received the strongest recommendations
For cognitive approaches, paper-and-card cognitive training, computer-assisted cognitive rehabilitation and cognitive stimulation were all adopted as Grade A. Adaptive digital cognitive therapy was Grade B and marked for revision. The review says digital therapeutics were supported mainly by expert consensus and require rigorous clinical verification.
For mild-to-moderate Alzheimer’s disease, high-frequency rTMS targeting the left or bilateral DLPFC was adopted as Grade A. tDCS alone was not recommended, and 40-Hz gamma-tACS was a revised Grade B option that could be used at home. Moderate-intensity aerobic exercise and strength training also received Grade A recommendations, with supervision advised for patients with balance impairment.
Other recommendations were more mixed. A multimodal lifestyle intervention was Grade B, whereas reminiscence and memory guidance with family, friends or caregivers were Grade A. Dance and horticultural interventions were stratified by disease severity. A combined intervention using music, reminiscence and neuroregulation was Grade A, while simple acupuncture and combined scalp-abdominal acupuncture were Grade B. Acupuncture evidence was described as methodologically heterogeneous and potentially confounded by medication and lack of blinding.
Why the list needs a careful reading
The review’s central caution is that its recommendations cannot be converted into a single effect size. Because intervention parameters, exercise regimens and controls were heterogeneous, it did not perform a quantitative meta-analysis or report pooled effect sizes. Nor did it provide pooled confidence intervals or standardized adverse-event rates. Adverse events were described as rare and mild, but potential problems included headaches, irritability, fatigue and falls.
That caution extends to the evidence base itself. The authors cited uneven evidence quality, intervention and outcome heterogeneity, possible missed studies, overlapping review data and inconsistent definitions of adverse events, and called for multicenter, high-standard randomized trials.
Taken together, the paper is best read as a stage- and setting-sensitive guide to implementation, not as a pooled comparison proving that one non-drug treatment is best. It points clinicians toward cognitive training, exercise, reminiscence and selected neuromodulation or sensory approaches, while leaving digital therapy, acupuncture and long-term comparative effects as areas requiring stronger verification. The study reported financial support from the China Medical Foundation’s Scientific Research Promotion Program for Clinical Medicine.
Paper data and sources
Original title: Non-pharmacological interventions for cognitive function improvement in patients with Alzheimer's disease: a best evidence summary
Authors: Wang N, Dang L, Chang W et al.
Journal/Repository: Frontiers in neurology
Status: Peer-reviewed
First online: 2026-10-10
DOI: Not available
Original paper · Full text