The effect of music therapy on cognitive functions in patients with dementia: a systematic review and meta-analysis.

The researchers tested whether music therapy (MT) improves cognitive function in people with dementia compared to standard care or other non-musical activities.

**Intervention:** Music therapy, defined as any music intervention conducted by or in consultation with a trained music therapist. This included two types:

**Active MT:** Patients actively participate by singing, playing instruments, or moving to music.

**Receptive MT:** Patients listen to pre-recorded or live music, often with guided relaxation or discussion.

**Comparators:** Standard care (usual treatment without music), or other non-musical activities (e.g., reading, conversation, resting).

**Outcomes measured:** Seven cognitive domains:

Global cognition (overall mental function)

Complex attention (ability to focus on multiple things)

Executive function (planning, decision-making, impulse control)

Learning and memory

Language

Perceptual-motor skills (coordination, spatial awareness)

Social cognition (understanding others' emotions/intentions)

**Total participants:** 330 (111 males, 33.6%)

**Sample sizes per study:** ranged from 30 to 104 participants

**Mean age range:** 78.8 to 86.3 years

**Diagnosis:** Dementia (Alzheimer's disease and other types), diagnosed by a physician using DSM or ICD criteria

**Severity:** Highly heterogeneous — mean Mini-Mental State Examination (MMSE) scores at baseline ranged from 6.4 (severe impairment) to 20.25 (moderate impairment). MMSE is a 30-point scale where lower scores = worse cognition.

**Setting:** Nursing homes, residential care facilities, and outpatient clinics across Europe (Italy, France, Finland), Asia (Taiwan), and the USA

**Excluded:** People without a formal dementia diagnosis, those receiving music interventions not led by a music therapist

The included studies used a variety of validated cognitive assessment tools. The meta-analysis grouped these into cognitive domains:

**Global cognition:** Mini-Mental State Examination (MMSE, 0–30, higher = better), Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog, 0–70, lower = better), and other dementia-specific batteries

**Complex attention:** Trail Making Test Part A (time to complete, lower = better), Digit Span Forward (number of digits recalled, higher = better)

**Executive function:** Trail Making Test Part B, Stroop Test, Verbal Fluency Tests (category and letter fluency)

**Learning and memory:** Rey Auditory Verbal Learning Test, Wechsler Memory Scale subtests, story recall tasks

**Language:** Boston Naming Test, verbal fluency tasks

**Perceptual-motor skills:** Clock Drawing Test, constructional praxis tasks

**Social cognition:** Emotion recognition tasks (used in only one study)

When a study used multiple tests for the same cognitive domain, the researchers averaged the effect sizes across those tests.

**Study design:** Systematic review and meta-analysis of randomized controlled trials (RCTs). The protocol was pre-registered on PROSPERO (CRD42016048144).

**Search strategy:** The researchers searched five databases (Medline, PsycINFO, Embase, CINAHL, RILM) from inception to September 8, 2016, using terms for music, cognition, randomization, and dementia. They also hand-searched reference lists. No language restrictions were applied.

**Inclusion criteria:**

Participants: Patients with physician-diagnosed dementia (DSM or ICD criteria)

Intervention: Music therapy conducted by or in consultation with a music therapist

Comparison: Standard care or non-musical activities

Outcomes: At least one cognitive outcome reported

Design: RCT (parallel or crossover)

**Exclusion criteria:** Non-randomized studies, studies without a music therapist, studies not reporting cognitive outcomes.

**Data extraction and quality assessment:** One reviewer extracted data; uncertainties were resolved by consultation with two additional reviewers. Risk of bias was assessed using the Cochrane Risk of Bias Tool (6 domains: selection bias, performance bias, detection bias, attrition bias, reporting bias, other bias). A study was classified as "high risk of bias" if it scored high in at least one domain.

**Statistical analysis:**

Effect sizes were calculated as standardized mean differences (SMD) using a random-effects model

SMD interpretation: 0.2 = small, 0.5 = medium, 0.8 = large effect

Heterogeneity assessed using I² statistic (I² > 50% = high heterogeneity)

Subgroup analyses planned for: MT approach (active vs. receptive), type (individual vs. group), and duration (<12 weeks vs. ≥12 weeks)

**What this design can prove:**

A meta-analysis of RCTs provides the strongest evidence for causal effects, because RCTs minimize confounding through random allocation

The systematic review approach reduces publication bias by searching multiple databases and including unpublished data

Subgroup analyses can suggest which specific types of MT might work better

**What this design cannot prove:**

Cannot determine mechanisms — why MT might or might not work

Cannot generalize beyond the specific populations studied (mostly moderate-to-severe dementia in residential care)

Small number of studies (6) limits statistical power and the ability to detect small effects

Cannot rule out that unmeasured factors (e.g., social interaction during MT, not the music itself) drove any observed effects

Cannot assess long-term effects beyond the study durations (6–16 weeks)

**Major methodological weaknesses:**

Only 6 studies met inclusion criteria, limiting generalizability

High heterogeneity in dementia types, stages, and MT protocols

Most studies had small sample sizes (30–104 participants)

Blinding was often impossible (participants and therapists knew they were receiving music therapy), introducing potential performance bias

Only one reviewer extracted data (though uncertainties were discussed with others)

Some studies had high risk of bias in at least one domain

**Primary analysis — overall effect of music therapy on all cognitive domains:**

No significant effect was found for any cognitive domain when all MT types were combined

Global cognition: SMD = 0.15, 95% CI -0.09 to 0.39, p = 0.22 (not significant)

Complex attention: SMD = 0.10, 95% CI -0.18 to 0.38, p = 0.48 (not significant)

Executive function: SMD = 0.14, 95% CI -0.16 to 0.44, p = 0.36 (not significant)

Learning and memory: SMD = 0.08, 95% CI -0.20 to 0.36, p = 0.57 (not significant)

Language: SMD = 0.13, 95% CI -0.15 to 0.41, p = 0.37 (not significant)

Perceptual-motor skills: SMD = 0.09, 95% CI -0.19 to 0.37, p = 0.53 (not significant)

Social cognition: Only one study reported this; insufficient data for meta-analysis

**Subgroup analysis — active vs. receptive music therapy:**

**Active MT showed a significant small-to-moderate effect on global cognition:**

- SMD = 0.29, 95% CI 0.02 to 0.57, p = 0.04

- This means active MT improved global cognition by about 0.29 standard deviations compared to control

Receptive MT showed no significant effect on any cognitive domain

The difference between active and receptive MT was not directly tested statistically

**Subgroup analysis — duration of treatment:**

No significant difference between interventions lasting <12 weeks vs. ≥12 weeks

Both short and longer durations showed non-significant effects overall

**Subgroup analysis — individual vs. group MT:**

No significant difference between individual and group formats

Both showed non-significant effects overall

**Heterogeneity:**

I² values ranged from 0% to 45% across analyses, indicating low-to-moderate heterogeneity

This suggests the studies were reasonably consistent in their findings

**Risk of bias:**

Most studies had unclear or high risk of bias in at least one domain

Blinding was the most common weakness (participants and therapists knew the intervention)

Some studies had incomplete outcome data or selective reporting

The only statistically significant finding was a small-to-moderate effect of active music therapy on global cognition (SMD = 0.29). To translate this into practical terms:

On the Mini-Mental State Examination (MMSE, 0–30 scale), an SMD of 0.29 corresponds to roughly a 2–3 point improvement. For context, the average annual decline in untreated Alzheimer's disease is about 2–4 points per year. So active MT might offset about one year's worth of cognitive decline.

On the Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog, 0–70 scale, lower is better), an SMD of 0.29 corresponds to roughly a 3–4 point improvement. For context, donepezil (Aricept), a common dementia medication, typically produces a 2–3 point improvement over 6 months.

However, this effect was small and barely reached statistical significance (p = 0.04, just below the conventional 0.05 threshold). The confidence interval was wide (0.02 to 0.57), meaning the true effect could be anywhere from negligible to moderate.

For all other cognitive domains and for receptive MT, the effects were essentially zero — no meaningful improvement was detected.

**What the authors acknowledge:**

Very small number of included studies (6), limiting statistical power

High heterogeneity in dementia types, stages, and MT protocols across studies

Most studies had small sample sizes

Blinding was often impossible due to the nature of music therapy

Only one reviewer performed data extraction (though uncertainties were discussed)

Some studies had high risk of bias

**What a critical reader would note:**

**Publication bias:** The authors searched for unpublished studies but did not formally test for publication bias (e.g., funnel plot) due to the small number of studies

**Control group quality:** Some studies compared MT to "standard care" which may include no structured activity. This means any benefit could be due to social interaction or attention, not music specifically

**Dose variation:** Session frequency ranged from once weekly to daily, and session duration from 30 to 60 minutes. The meta-analysis did not analyze dose-response

**Outcome measurement:** Different studies used different tests for the same cognitive domain, making comparisons difficult

**Follow-up duration:** Maximum follow-up was 16 weeks. No data on whether effects persist after MT stops

**Population limits:** Mostly moderate-to-severe dementia in residential care. Results may not apply to mild dementia or community-dwelling individuals

**Music therapist definition:** "In consultation with a music therapist" is vague — some studies may have had minimal therapist involvement

**Industry funding:** Not explicitly reported, but some studies may have received funding from music therapy organizations

For someone running their own n=1 experiment (e.g., a caregiver trying music activities with a family member with dementia):

### What to test

**Active music therapy** (not just listening): Try structured singing sessions, playing simple percussion instruments (drums, shakers, bells), or moving/dancing to music. The evidence suggests active participation is more effective than passive listening.

**Dose:** Based on the studies, aim for 30–45 minute sessions, 2–3 times per week. The median study duration was 12 weeks, but effects might appear earlier.

**Music selection:** Use familiar, preferred music from the person's youth or early adulthood. Musical memory is often preserved in dementia.

### Minimum meaningful duration

Run the experiment for at least 8–12 weeks. The studies ranged from 6–16 weeks, with the median being 12 weeks. Shorter periods may not show effects.

Assess at baseline, then every 4 weeks. If no trend appears by 12 weeks, the intervention is unlikely to work.

### What to measure

**Primary metric:** Global cognition. Use a simple, validated tool like the Mini-Mental State Examination (MMSE) if available through a healthcare provider. Alternatively, use the Montreal Cognitive Assessment (MoCA, 0–30, higher = better).

**Secondary metrics:**

- Attention: How long can the person focus on the music activity? (minutes of engagement)

- Memory: Can they recall song lyrics or melodies from previous sessions?

- Mood/behavior: Track agitation, irritability, or apathy using a simple 1–5 scale daily

- Quality of life: Use the DEMQOL (Dementia Quality of Life) or a simple 1–10 self-report scale

**Objective measure:** Video record sessions (with consent) and have a blinded rater assess engagement and cognitive responses

### Key confounds to control for

**Social interaction:** The benefit might come from one-on-one attention, not music. Control for this by having a non-music social activity (e.g., conversation, card game) on alternate days.

**Time of day:** Cognitive performance varies by time of day. Test at the same time each session.

**Medication changes:** Note any changes in dementia medications (donepezil, memantine) or psychotropic drugs during the experiment.

**Sleep quality:** Poor sleep worsens cognition. Track sleep duration and quality daily.

**Illness:** Acute infections (UTIs are common in dementia) can cause temporary cognitive decline. Exclude days when the person is ill.

**Rater bias:** If possible, have someone who doesn't know whether the person is in the "music" or "control" phase assess cognitive outcomes.

### What a positive result would look like

A 2–3 point improvement on the MMSE (0–30 scale) after 8–12 weeks of active music sessions, compared to baseline

Alternatively, a slower rate of decline than expected (e.g., stable scores over 12 weeks when the person was previously declining 1–2 points per month)

Improved attention during sessions (e.g., from 5 minutes to 15 minutes of sustained engagement)

Reduced agitation or improved mood on days with music sessions vs. days without

The person initiating singing or humming outside of sessions — a sign of carryover effects

**Bottom line:** The evidence for music therapy improving cognition in dementia is weak overall, but active participation (singing, playing instruments) shows a small, potentially meaningful effect. For an n=1 experiment, focus on active music-making, run for at least 12 weeks, and measure both cognition and quality of life. The effect is likely modest — don't expect dramatic improvements, but even slowing decline by 2–3 points on the MMSE over 3 months would be clinically meaningful.