Elmsword
← Sleep

Sleep · Research · 5 min read

Music therapy and sleep in older adults: a promising signal, very low certainty

A meta-analysis of 10 studies with 602 older adults found better pooled sleep quality with music therapy, but results varied widely and GRADE certainty was very low.

Sleep disturbances are common in older adults; the review cites figures of up to 50%. Music therapy has been proposed as a non-pharmacological option. A 2025 systematic review and meta-analysis in PLOS One pooled 10 studies with 602 participants. Across all studies, sleep quality improved more in the music groups than in the control groups. That result was significant in the randomized trials but not in the non-randomized ones. Results also varied a lot from study to study, and the authors rated the certainty of the evidence as very low. So this is a promising signal, not proof. [1]

What the researchers did

The authors followed PRISMA guidelines and searched PubMed, Embase, Web of Science, Cochrane Library, CINAHL, PsycINFO and China National Knowledge Infrastructure (CNKI) for studies of music interventions and sleep quality in older adults. The search retrieved 473 articles. After removing duplicates and screening, 18 full texts were evaluated and 10 studies, published between 2010 and 2023, were included: 6 randomized controlled trials (RCTs) and 4 non-randomized controlled trials (non-RCTs). Together they had 602 participants, as the authors counted them, 297 in intervention groups and 305 in control groups. In the paper's Table 1 and forest plot, study sizes ranged from 31 to 100. [1]

Participants were recruited from community centers, long-term care facilities, nursing homes and hospitals. Several studies focused on people with poor sleep quality; one included older patients with hematologic cancer. Passive music listening was the main approach, with some studies using active music participation. The interventions in brief: [1]

  • Music: classical (Western and Chinese), instrumental, nature sounds, religious music and culturally specific genres. Studies that reported tempo used slow music, 60–85 BPM (4 of the 10 studies in the paper's Table 1).
  • Sessions: mostly 20–60 minutes, with frequencies from daily to weekly.
  • Duration: from a single session to three months of regular application.
  • Outcome: sleep quality, mainly measured with the Pittsburgh Sleep Quality Index (PSQI), an instrument for subjective sleep quality. Some studies added polysomnography, visual analog scales or biochemical markers.

Results were pooled as standardized mean differences (SMDs) with 95% confidence intervals (CIs), with the random effects model as the primary analysis. Risk of bias was rated with RoB 2 for RCTs and ROBINS-I for non-RCTs, and the certainty of evidence with GRADE. [1]

Key findings

Pooled sleep quality, random effects model. The authors describe the negative pooled SMD as an improvement in sleep quality with music therapy. * Statistically significant. Test for subgroup differences: P = 0.28. [1]
AnalysisSMD (95% CI)Heterogeneity
All 10 studies (602 participants)−0.79 (−1.25 to −0.33)*I² = 79%
6 RCTs−0.59 (−1.11 to −0.07)*I² = 67%
4 non-RCTs−1.08 (−2.36 to 0.19)I² = 87%
Prediction interval, all studies−2.22 to 0.64–
GRADE certaintyVery low (RCTs and non-RCTs)–
Music therapy and sleep quality in older adultsOf 473 articles retrieved, 18 full texts were evaluated and 10 studies with 602 participants were pooled (6 RCTs and 4 non-RCTs). Pooled standardized mean difference for sleep quality: all studies −0.79 (95% CI −1.25 to −0.33, significant); RCTs −0.59 (−1.11 to −0.07, significant); non-RCTs −1.08 (−2.36 to 0.19, not significant). Prediction interval −2.22 to 0.64. I² = 79%; GRADE certainty very low.Music therapy and sleep quality in older adultsSystematic review and meta-analysis: 10 studies, 602 participantsA. Which studies were pooled473 articlesretrieved by the search18 full textsevaluated in detail10 studies, 602 participants6 RCTs · 4 non-RCTsTempo, where reported (4 of 10 studies): slow, 60–85 BPM; sessions mostly 20–60 minutes, daily to weeklyB. Sleep quality, standardized mean difference (SMD) with 95% CI, random effectsAll 10 studiesSMD −0.79 (−1.25 to −0.33)*6 RCTsSMD −0.59 (−1.11 to −0.07)*4 non-RCTsSMD −1.08 (−2.36 to 0.19)Prediction interval−2.22 to 0.64−2−101Negative SMD = improvement in sleep quality, as reported by the authors* Statistically significant. Test for subgroup differences: P = 0.28.Heterogeneity I² = 79%. GRADE certainty: very low. Sleep quality mostly self-reported (PSQI).Source: Li et al., PLOS One 2025, Results (Meta analysis; Subgroup analysis).
Source: Li et al., PLOS One 2025, Results (Meta analysis; Subgroup analysis). Squares show the pooled SMD, lines the 95% CI; the dashed line is the prediction interval. [1]

Overall, the music groups did better. Pooled across all 10 studies, the SMD was −0.79 (95% CI −1.25 to −0.33; P < 0.001). The authors call this a moderate to large effect size. [1]

“The random effects model (Fig 6) showed a significant improvement in sleep quality with music therapy (SMD: −0.79; 95% CI, −1.25 to −0.33; P < 0.001).”

Li et al. · Results · full text

The randomized trials showed a smaller, significant effect. In the 6 RCTs, the SMD was −0.59 (95% CI −1.11 to −0.07). In the 4 non-RCTs, it was −1.08 (95% CI −2.36 to 0.19), which was not statistically significant. The test for a difference between the two subgroups was not significant either (P = 0.28). Two of the four non-RCTs were assessed as having serious risk of bias. [1]

Where tempo was reported, the music was slow. The studies varied a lot, but they shared a few features. In the paper's Table 1, 4 of the 10 studies give a tempo (60–80 or 60–85 BPM), 3 say tempo was not specified, and 3 do not mention it. In one trial, the control group heard the same music without binaural beats. [1]

“Most interventions utilized slow tempo music (60–85 BPM) with session durations of 20–60 minutes, applied with frequencies varying from daily to weekly.”

Li et al. · Results · full text

No single study drove the result. In leave-one-out sensitivity analyses, the SMD stayed between −0.62 and −0.82, and every run remained statistically significant in the common effect model. Tests for publication bias found no significant asymmetry. [1]

Results varied widely between studies. Heterogeneity was substantial (I² = 79%). The prediction interval, which estimates the range of true effects in future studies, ran from −2.22 to 0.64, so it includes zero. Meta-regression on intervention duration, session duration, frequency and total sessions explained little of this variation (R² range: 0–2.97%). [1]

“However, substantial heterogeneity and the wide prediction interval (−2.22 to 0.64) indicate individual responses vary considerably, highlighting the need for personalized approaches.”

Li et al. · Discussion · full text

What this doesn't tell us

Under GRADE, the certainty of the evidence was very low for both the RCTs and the non-RCTs. The authors downgraded it for risk of bias and for high heterogeneity. They say future research may substantially change the picture. [1]

“However, substantial heterogeneity and very low certainty of evidence require cautious interpretation.”

Li et al. · Conclusions · full text

Sleep quality was mostly self-reported. The authors note that the PSQI is susceptible to recall bias and social desirability bias, which limits the review's ability to tell whether music produces measurable physiological changes in sleep. [1]

“A notable limitation across the included studies was the predominant reliance on subjective sleep quality measures, particularly the PSQI, rather than objective sleep assessment tools such as polysomnography or actigraphy.”

Li et al. · Discussion · full text

The evidence base is small: 10 studies, each with 31 to 100 participants (per the paper's Table 1 and forest plot). Settings ranged from community centers to hospitals, and one study enrolled older patients with hematologic cancer, so the pooled number mixes quite different groups. [1]

The paper did not analyze when the music was played, for example at bedtime. The authors list timing, along with music type and duration, as intervention parameters for future research to investigate. They also call for objective sleep measures and high-quality randomized controlled trials with standardized interventions. [1]

What to take away

Opinion, not a study result: If I were weighing this for an older family member, I would see slow tempo music as a possible option worth trying, which the authors describe as potentially safe and non-invasive. I would not call it proven, because the certainty of the evidence is very low and results varied a lot between studies. If sleep problems persist, talk to a doctor.

Reference

  1. PLOS One 2025[1] Li C, Kumar AP, Wankhar D, Kuppusamy M, Govindasamy K. Effect of music therapy on sleep quality in elderly: A systematic review and meta-analysis. PLOS One. 2025;20(11):e0334356.
    DOI: 10.1371/journal.pone.0334356 · Full text

This is a research explainer for general education, not medical advice.