Music therapy reduces stress and anxiety in critically ill patients: a systematic review of randomized clinical trials.

The researchers tested **music therapy** as a non-pharmacological intervention to reduce anxiety and stress in adult patients admitted to intensive care units (ICUs). Music therapy was defined as an interpersonal process where a therapist uses music to help patients improve or maintain health — though in practice, most studies simply played pre-recorded music through headphones.

**Intervention:** A single session of music therapy, typically lasting 30 minutes (range: 15–60 minutes). Only 2 out of 11 studies repeated the intervention more than once daily. Music was delivered via headphones or speakers at the bedside.

**Comparators:** Control groups received one of the following:

Standard care (usual ICU treatment with no music)

Relaxation instructions without music

Headphones with no music (sham control)

Noise-cancelling headphones (to block ICU noise)

**Outcomes measured:**

**Primary:** State anxiety (self-reported using validated scales)

**Secondary:** Physiological stress markers (heart rate, respiratory rate, blood pressure), sedation levels, pain scores, and amount of sedative drugs administered

**Total sample:** 959 patients across 11 studies (10 randomized controlled trials and 1 quasi-experimental study). Individual study sample sizes ranged from 17 to 373 patients (median: 60 patients).

**Population:** Critically ill adult patients admitted to medical or surgical ICUs. This included:

Mechanically ventilated patients (on breathing machines)

Non-ventilated ICU patients

Patients with varying levels of consciousness and sedation

**Exclusion criteria across studies:** Patients under 18 years old, those with known psychiatric disorders, and patients receiving interventions other than music therapy (such as nature sounds).

**Setting:** ICUs in hospitals — the specific countries and hospital types varied across studies but were not individually detailed in the review.

The review compiled data from studies using several validated instruments:

**Anxiety scales:**

**Spielberger State-Trait Anxiety Inventory (STAI) – short version:** A 6-item self-report scale (abbreviated from the full 20-item version). Scores range from 6 to 24, with higher scores indicating greater anxiety. Used in 5 studies.

**Faces Anxiety Scale:** A single-item visual scale showing faces from calm to anxious. Used in 1 study.

**100-mm Visual Analog Scale (VAS) for anxiety:** A line where patients mark their anxiety level from "no anxiety" to "worst possible anxiety." Used in 2 studies.

**Physiological measures:**

Heart rate (beats per minute)

Respiratory rate (breaths per minute)

Blood pressure (systolic and diastolic, mmHg)

These were measured before, during, and after the music intervention

**Sedation/agitation scales:**

**Richmond Agitation Sedation Scale (RASS):** A 10-point scale from -5 (unarousable) to +4 (combative), with 0 being alert and calm. Used in 1 study.

**Ramsay Sedation Score:** A 6-point scale from 1 (anxious/agitated) to 6 (unresponsive). Used in 1 study.

**Pain scales:**

**Numerical Rating Scale (NRS):** 0–10 scale where 0 = no pain and 10 = worst possible pain. Used in 2 studies.

**Drug consumption:** The amount of sedative medications (e.g., benzodiazepines, propofol) administered was recorded in some studies.

**Study design:** This is a **systematic review** — a method that searches for, evaluates, and synthesizes all available studies on a topic using a pre-specified protocol. The review was registered on PROSPERO (CRD42018100036) and followed PRISMA guidelines for reporting.

**Search strategy:** The authors searched 6 databases (PubMed, CINAHL, Cochrane Library, Scopus, Web of Science, and an Italian nursing literature database) from inception through April 30, 2018. They also manually screened reference lists of included papers. Search terms included "music therapy," "intensive care unit," "anxiety," "stress," and related keywords.

**Inclusion criteria:** Randomized controlled trials (RCTs) and quasi-experimental studies of critically ill adult ICU patients receiving music therapy, with outcomes measuring anxiety, stress, pain, or agitation.

**Exclusion criteria:** Studies of patients under 18, those with psychiatric disorders, studies using nature sounds instead of music, and studies measuring only biochemical stress markers (hormones like cortisol) without behavioral or physiological outcomes.

**Quality assessment:** The authors used the PEDro scale (Physiotherapy Evidence Database) to rate study quality. Since blinding patients and therapists to music therapy is impossible, the maximum achievable score was 8 out of 11. Quality was described as "satisfactory" overall.

**Data synthesis:** The authors did **not** perform a pooled meta-analysis (combining all data statistically) because the studies were too heterogeneous — they used different music types, durations, control conditions, outcome measures, and patient populations. Instead, they narratively summarized the findings.

**What this design can and cannot prove:**

**Can prove:** That across multiple independent studies, music therapy is consistently associated with reduced anxiety and physiological stress in ICU patients. The systematic review approach increases confidence because it aggregates evidence rather than relying on a single study.

**Cannot prove:**

**Causality definitively:** Without blinding (impossible here), placebo effects cannot be ruled out. Patients who know they're getting music may report less anxiety simply because they expect to feel better.

**Optimal dose:** Because studies used different music durations, types, and frequencies, the review cannot tell you the "best" way to deliver music therapy.

**Long-term effects:** Most studies measured outcomes immediately after a single session. The review cannot speak to whether benefits persist for hours, days, or beyond ICU discharge.

**Mechanism:** The review cannot determine whether effects are due to music itself, distraction, reduced environmental noise, or the therapeutic relationship with a music therapist.

**Major methodological weaknesses flagged by the authors:**

**Lack of blinding in all studies** (unavoidable but still a bias risk)

**Small sample sizes** in many individual studies (ranging from 17 to 60 patients in most)

**Single-session interventions** in most studies (only 2 of 11 repeated music more than once)

**Short follow-up** — outcomes measured immediately post-intervention only

**Heterogeneity** — different music types, durations, control conditions, and outcome measures made direct comparison impossible

**Potential halo effect** in crossover studies (where patients who received music first might have biased responses in the control phase)

**Low baseline anxiety** in some studies — patients who weren't very anxious to begin with had less room for improvement

**Primary outcome — Anxiety reduction:**

All 11 studies reported a statistically significant reduction in anxiety levels after music therapy compared to control conditions.

Specific effect sizes were not pooled (no meta-analysis), but individual studies reported:

- STAI scores decreasing by approximately 3–6 points on the 6–24 scale (exact numbers varied by study)

- VAS anxiety scores decreasing by 15–30 mm on the 100-mm scale

The reduction was consistent across mechanically ventilated and non-ventilated patients.

**Secondary outcomes — Physiological parameters:**

**Heart rate:** Decreased by approximately 5–15 beats per minute in music groups compared to controls (reported in multiple studies)

**Respiratory rate:** Decreased by approximately 3–8 breaths per minute

**Blood pressure:** Systolic blood pressure decreased by approximately 5–15 mmHg in music groups

These changes were statistically significant in most studies that measured them.

**Sedation and drug use:**

Music therapy led to increased sedation levels (measured by RASS or Ramsay scores) or reduced the amount of sedative drugs administered.

One study reported a reduction in sedative medication requirements, though exact percentages were not provided in the review.

**Pain:**

Lower pain scores were reported in music groups compared to controls (measured by NRS or other scales).

Specific effect sizes were not reported in the review summary.

**Adverse effects:**

No adverse effects were reported in any study.

Only 2 studies reported mortality rates, and neither found a difference between music and control groups.

**Quality of evidence:**

PEDro scores ranged from 4 to 7 out of a possible 8 (since blinding was impossible). This is considered "satisfactory" but not high-quality evidence.

Because the authors did not perform a pooled meta-analysis, exact average effect sizes across all studies are not available. However, based on the individual study results described:

**Anxiety reduction:** The decrease in anxiety was roughly equivalent to moving from "moderately anxious" to "slightly anxious" on self-report scales. For context, a 15–30 mm drop on a 100-mm VAS is similar to the anxiety reduction seen after a single dose of a mild anxiolytic medication like lorazepam (0.5–1 mg) in other clinical studies.

**Heart rate reduction:** A decrease of 5–15 beats per minute is roughly equivalent to the effect of a low dose of a beta-blocker medication (e.g., 25 mg of metoprolol) or the relaxation response from 10 minutes of deep breathing.

**Blood pressure reduction:** A 5–15 mmHg drop in systolic blood pressure is comparable to the effect of a single session of progressive muscle relaxation or about half the effect of a standard dose of an antihypertensive medication.

**Practical translation:** If you're an ICU patient, listening to 30 minutes of music through headphones might lower your heart rate by about 10 beats per minute and reduce your self-rated anxiety by about 20% — enough to feel noticeably calmer, but not enough to replace sedative medications entirely.

**What the authors acknowledge:**

**Lack of blinding:** None of the studies could blind patients or therapists to the intervention, which introduces potential placebo effects and performance bias.

**Heterogeneity:** Wide variation in music type, duration, frequency, delivery method, and control conditions made pooled analysis impossible.

**Single-session design:** Most studies tested only one music session, so the effects of repeated or ongoing music therapy are unknown.

**Small sample sizes:** Many studies had fewer than 60 patients, limiting statistical power and generalizability.

**Short follow-up:** Outcomes were measured immediately after the intervention, with no data on how long benefits lasted.

**Publication bias:** The authors did not formally test for publication bias (e.g., funnel plot analysis), so studies showing no effect may be missing from the literature.

**Language restriction:** Only English and Italian language studies were included, potentially missing relevant research in other languages.

**Additional critical limitations:**

**No standardized music protocol:** Some studies used classical music, others used "sedative" music selected by researchers, and others allowed patients to choose. This makes it impossible to recommend a specific type of music.

**Self-report bias in ICU patients:** Critically ill patients, especially those on sedatives or mechanical ventilation, may have impaired cognition and unreliable self-reporting.

**Confounding by ICU environment:** Music therapy often involves headphones, which also block out ICU noise. The benefit could be from noise reduction rather than music itself. Only one study used noise-cancelling headphones as a control to address this.

**No assessment of music therapist involvement:** The review defined music therapy as involving a therapist, but most studies simply played recorded music. The added value of a live therapist versus recorded music was not tested.

**Industry funding not reported:** The review did not disclose whether any studies had funding from music industry or device manufacturers.

**No long-term or post-discharge outcomes:** No data on whether music therapy during ICU stay affects PTSD, anxiety disorders, or quality of life after hospital discharge.

For someone running their own n=1 experiment:

**What to test:**

Listen to 30 minutes of slow-tempo instrumental music (60–80 beats per minute, no lyrics, no sudden changes in volume or tempo) through noise-cancelling headphones while resting in a quiet or semi-quiet environment.

Test this once daily for at least 7 days.

Compare against a control condition of 30 minutes of silence with noise-cancelling headphones (to isolate the effect of music from the effect of noise reduction).

**Minimum meaningful duration:**

7 days minimum to see if effects are consistent and not just novelty.

14–21 days would be better to assess whether benefits persist or diminish with repeated exposure.

Each session should be exactly 30 minutes (based on the typical intervention in these studies).

**What to measure (specific metrics):**

**Anxiety:** Use the short-form State-Trait Anxiety Inventory (STAI-6) — a 6-item questionnaire taking 2 minutes to complete. Measure immediately before and after each music session. Score range: 6–24, lower = less anxious.

**Heart rate:** Measure using a pulse oximeter or smartwatch. Record resting heart rate for 2 minutes before the session and 2 minutes after. Aim for a decrease of 5+ beats per minute.

**Blood pressure:** If you have a home blood pressure monitor, measure systolic and diastolic pressure before and after. Aim for a decrease of 5+ mmHg systolic.

**Subjective relaxation:** Rate on a 0–10 scale (0 = completely tense, 10 = completely relaxed) before and after each session.

**Sleep quality the following night:** Rate on a 0–10 scale (0 = terrible sleep, 10 = perfect sleep) each morning.

**Key confounds to control for:**

**Time of day:** Always do the music session at the same time of day (e.g., 8 PM before bed) to control for circadian variation in anxiety and heart rate.

**Pre-session state:** Record your baseline anxiety, heart rate, and any stressors that day (work stress, caffeine, alcohol, exercise) before each session.

**Caffeine and alcohol:** Avoid caffeine for 4 hours before and alcohol for 2 hours before each session.

**Medications:** If you take any anxiety or blood pressure medications, keep the dose and timing consistent throughout the experiment.

**Headphone type:** Use the same headphones every time. Noise-cancelling is preferred to control for environmental noise.

**Music selection:** Use the same playlist every session. Do not change songs or genres mid-experiment.

**Position:** Sit or lie in the same position (preferably reclining) for each session.

**Expectation bias:** If possible, have someone else randomize your sessions (music vs. silence) without telling you which is which until after you record your outcomes. This is difficult to do perfectly but reduces placebo effects.

**What a positive result would look like:**

A consistent decrease in STAI-6 scores of 3+ points after music sessions compared to silence sessions.

A consistent decrease in heart rate of 5+ beats per minute after music compared to silence.

A consistent decrease in systolic blood pressure of 5+ mmHg after music compared to silence.

Subjective relaxation ratings increase by 2+ points on the 0–10 scale after music compared to silence.

These effects should appear within the first few sessions and persist or strengthen over the 7–21 day experiment.

If you see no difference between music and silence conditions, the benefit may be from quiet rest with noise reduction rather than music specifically — which is still a useful finding for your personal practice.