Effects of Non-Pharmacological Interventions on Pain and Psychological Outcomes in Patients with Burns: A Systematic Review and Meta-Analysis

Abstract

Background

Non-pharmacological interventions are used alongside analgesia in burn care, but the evidence comes from clinically diverse interventions, populations, and settings.

Methods

We searched nine databases from inception to July 1, 2026. Controlled studies of virtual reality, music, massage, hypnosis, relaxation, distraction, or structured psychological interventions in patients with burns were eligible. Continuous outcomes measuring the same construct were synthesized as Hedges’g standardized mean differences (SMDs) using random-effects models. Multi-arm trials contributed one independent study-level effect by combining eligible intervention arms sharing a common control. Risk of bias was assessed with RoB 2 or ROBINS-I, and certainty with GRADE.

Results

Sixteen studies were included; 10 contributed to meta-analysis. Pain was lower with non-pharmacological interventions (4 studies, n=354; SMD=−0.78, 95% CI −1.27 to −0.29; I²=78.8%). A restricted pain analysis remained favorable (2 studies, n=206; SMD=−0.61, 95% CI −1.16 to −0.06). Randomized-trial analyses favored interventions for general anxiety (3 studies, n=261; SMD=−1.03, 95% CI −1.63 to −0.43) and depressive symptoms (2 studies, n=141; SMD=−1.31, 95% CI −1.91 to −0.72). Pain-related anxiety was lower in 2 studies (n=360; SMD=−0.83, 95% CI −1.28 to −0.38; I²=70.3%). Certainty was low for pain and pain-related anxiety and very low for general anxiety and depressive symptoms.

Conclusion

Selected interventions were associated with lower procedural pain, pain-related anxiety, and some short-term psychological symptoms. Substantial clinical and methodological heterogeneity limits the certainty and generalizability of the pooled estimates. These interventions should be considered adjuncts to, not replacements for, pharmacological analgesia and appropriate psychological care.

Introduction

Burn injuries cause not only tissue damage but also persistent, procedural, and breakthrough pain, with lasting consequences for functional recovery and social participation.^1,2 Debridement, dressing changes, donor-site care, scar stretching, and rehabilitation exercises can all provoke substantial pain. Guidelines for pain management in adults and children with burns emphasize timely, individualized, multimodal analgesia; non-pharmacological interventions should be used as adjuncts rather than substitutes for adequate pharmacological analgesia and appropriate wound care.^3,4 Pain, anxiety, and depression are interrelated throughout burn recovery, and memories of previous pain together with threat anticipation before subsequent procedures may further reduce treatment cooperation and adherence to rehabilitation.^5–7

Recent evidence syntheses have focused predominantly on virtual reality (VR). A large systematic review spanning multiple procedural settings suggested that VR may reduce procedure-related pain, although effects varied according to device characteristics, degree of interaction, and comparator design.⁸ Burn-specific reviews similarly indicate a potential adjunctive analgesic effect during dressing changes and physical therapy, but substantial heterogeneity remains across adults, children, and clinical procedures.^9–14 A network meta-analysis has also compared VR with music-based interventions; however, rankings based largely on indirect comparisons and small studies should not be treated as a direct guide to clinical priority.¹⁵

Evidence on music-based interventions has also expanded. Systematic reviews in adults, a meta-analysis in pediatric burn populations, and updated evidence syntheses published in 2025 suggest possible reductions in pain and anxiety; however, researcher-delivered recorded music, patient-preferred music, and therapist-led music-assisted relaxation are clinically distinct interventions.^16–20 Evidence for massage, aromatherapy, and hypnosis is more limited. One randomized study suggested that a 20-minute hand shiatsu massage may reduce pain-related anxiety in patients with burns, but the conclusion requires caution because it is based on a single study with a small sample and short-term assessment.²¹ Reviews of aromatherapy and broader non-pharmacological interventions have likewise emphasized small samples, risk of bias, and heterogeneity in intervention protocols.^22,23 Evidence on acute procedural pain in broader populations provides a rationale for further investigation of hypnosis, but burn-specific trials remain limited.²⁴

Most previous syntheses have examined a single modality, particularly virtual reality or music, or have focused on a specific procedure or outcome. As a result, they provide limited information on how the available controlled evidence relates simultaneously to pain, general anxiety, depressive symptoms, and pain-related anxiety across burn-care settings. Comparator intensity, multi-arm and crossover designs, and the distinction between procedure-specific pain-related anxiety and broader psychological symptoms have also been handled inconsistently. We therefore undertook an updated synthesis across several non-pharmacological modalities while keeping these outcome constructs separate. The review examines study design, comparator type, intervention class, burn-care context, risk of bias, and GRADE certainty, with sensitivity analyses and descriptive intervention-type stratification used to assess the stability and clinical interpretation of pooled effects.

Materials and Methods

Study Design and Reporting Guidelines

This systematic review and meta-analysis was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement²⁵ and informed by the Cochrane Handbook for Systematic Reviews of Interventions.²⁶ We also considered pain-medicine-specific guidance on systematic review protocol elements, evidence synthesis, heterogeneity, and transparent reporting.^27,28 The review question was formulated using the Population, Intervention, Comparator, Outcomes, and Study design (PICOS) framework. Pain was the primary outcome; general anxiety, depressive symptoms, pain-related anxiety, and other clinically relevant outcomes were secondary outcomes. The review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD420261463436).

Information Sources and Search Strategy

PubMed/MEDLINE, Embase, Web of Science Core Collection, the Cochrane Central Register of Controlled Trials (CENTRAL), the Cumulative Index to Nursing and Allied Health Literature (CINAHL), SinoMed, China National Knowledge Infrastructure, Wanfang Data, and VIP Database were searched from inception to July 1, 2026. Search terms covered burns, scalds, virtual reality, music, massage, aromatherapy, hypnosis, guided imagery, relaxation, distraction, psychological interventions, pain, anxiety, depression, dressing changes, debridement, and rehabilitation. Controlled vocabulary, search fields, and truncation operators were adapted for each database.

Eligibility Criteria

Studies were eligible if they: (1) enrolled patients with burn or scald injuries; (2) evaluated virtual reality, music, massage, hypnosis, relaxation training, distraction, structured psychological interventions, or an identifiable combination of these approaches; (3) included usual care, usual analgesia, an attention control, an active comparator, or a within-participant control; (4) reported pain, general anxiety, depressive symptoms, pain-related anxiety, or another prespecified outcome; and (5) used a randomized parallel-group, randomized crossover, randomized repeated-measures, quasi-experimental, or non-randomized controlled design. The intervention categories were grouped under non-pharmacological adjunctive care because they were non-drug strategies intended to modify pain perception, anticipatory threat, arousal, attention, or coping during burn care; they were not assumed to be mechanistically interchangeable. No review-level minimum or maximum threshold for total body surface area (TBSA) or burn depth was imposed. Burn depth, TBSA, anatomical site, phase of care, and ICU or procedural setting were extracted when reported and were considered potential sources of clinical heterogeneity. Reviews, guidelines, protocols, conference abstracts, case reports, animal studies, uncontrolled single-arm studies, studies in which the independent effect of the non-pharmacological component could not be separated from pharmacological co-interventions, studies without extractable key outcome data, and duplicate publications were excluded.

Study Selection and Data Extraction

Two reviewers independently screened titles and abstracts, assessed full-text eligibility, and extracted data using prespecified eligibility criteria. Disagreements were resolved through discussion, with arbitration by a third reviewer when necessary. Extracted data included study design, sample size, age, burn depth or severity, TBSA when reported, anatomical site, clinical setting, intervention components, frequency and duration, comparator conditions, analgesic co-interventions, pain and psychological outcome instruments, assessment time points or follow-up, means, standard deviations, event counts, and numbers analyzed. If multiple instruments measured the same construct, the measure most directly aligned with the prespecified construct and principal care procedure was selected to avoid double counting. Multiple reports arising from the same trial were linked according to the underlying independent participant sample and grouped into a single StudyID-defined report family to avoid duplicate weighting by publication.

Risk of Bias Assessment

Randomized parallel-group trials were assessed using the revised Cochrane risk-of-bias tool for randomized trials (RoB 2).²⁹ Appropriate RoB 2 variants were used for crossover or within-participant trials, with additional consideration of sequence, period, and carryover effects and the handling of paired data. Non-randomized intervention studies were assessed using the Risk Of Bias In Non-randomized Studies of Interventions tool (ROBINS-I).³⁰ Two reviewers independently assessed risk of bias. Disagreements were resolved by consensus, with arbitration by a third reviewer when necessary.

Certainty of Evidence

The certainty of evidence for pain, general anxiety, depressive symptoms, and pain-related anxiety was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach.³¹ Evidence was downgraded for risk of bias, inconsistency, indirectness, imprecision, and publication bias. For general anxiety and depressive symptoms, GRADE judgments were based primarily on sensitivity analyses restricted to randomized controlled trials.

Statistical Analysis

Analyses were performed using Python 3.13.5 with NumPy, SciPy, statsmodels, and Matplotlib for effect-size estimation, random-effects modeling, and figure generation. Continuous outcomes measuring the same prespecified construct on different instruments (for example, VAS and NRS pain scales) were harmonized using standardized mean differences corrected with Hedges’ g and 95% confidence intervals (CIs). Scale directions were aligned so that negative values favored the non-pharmacological intervention. Pain-related anxiety was analyzed separately from general anxiety and depressive symptoms because it represents a procedure-specific construct. Dichotomous outcomes were expressed as risk ratios (RRs) with 95% CIs. The primary synthesis used inverse-variance random-effects models, with between-study variance estimated using the DerSimonian-Laird method. For the overall class-effect question, when a multi-arm trial contained more than one eligible non-pharmacological intervention sharing the same control group, eligible intervention arms were combined so that the trial contributed one independent study-level effect estimate and the shared control was not reused as independent information. Arm-level contrasts were retained for descriptive intervention-type interpretation but were not used for formal between-subgroup tests when categories were sparse or dependent. Crossover and within-participant studies were pooled only when paired effect estimates were reported or could be estimated. Statistical heterogeneity was described using Cochran Q, I², and between-study variance (tau²). Because fewer than 10 independent studies contributed to each outcome, funnel plots and formal small-study-effect tests were not performed; publication bias therefore could not be assessed reliably.

Results

Study Selection

The searches, covering database inception through July 1, 2026, identified 1039 records. After deduplication and manual checking, 716 records underwent title and abstract screening, of which 695 were excluded. Full texts were sought for 21 reports, but 3 could not be retrieved. Eighteen reports were assessed for eligibility; 2 were excluded, leaving 16 studies for qualitative synthesis. Ten studies contributed data to at least one meta-analysis. The study selection process is shown in Figure 1.

Figure 1.

PRISMA 2020 flow diagram of study selection.

Characteristics of Included Studies

The 16 included studies were published between 2001 and 2025 and were conducted in China, Iran, the United States, Türkiye, and India. Both pediatric and adult populations were represented. Burn severity was heterogeneous: some reports enrolled deep partial-thickness or deep facial burns, some included second- and third-degree burns, and several did not provide a numeric TBSA. Where reported, TBSA ranged from small outpatient burns to substantially larger inpatient burns. Clinical settings included dressing changes, debridement, donor-site care, physiotherapy or rehabilitation, pre-discharge psychological care, and other inpatient or outpatient burn-care procedures. Interventions included virtual reality, music-based approaches, massage, hypnosis, and structured psychological interventions. Detailed burn characteristics, intervention dose, comparator, outcome instruments, and follow-up are presented in Table 1 and Table 2.

Table 1.

Characteristics of Included Studies, Including Burn Severity and Intervention Details

Study	Country/Design	Sample and Burn Severity/TBSA	Intervention	Intervention Dose/Duration	Comparator
Shu 202532	China; controlled study; allocation unclear	105; deep second-degree burns; TBSA 14.23±3.45% vs 13.87±3.24%	VR-based recovery video psychological intervention	One module daily for 7 days, beginning 1 week before discharge	Usual care
He 202133	China; prospective RCT	81 analyzed; deep facial burns; eligibility TBSA ≤20%; mean TBSA ~13±3% vs 14±4%	VR-based video psychological intervention	Once daily for 7 days before discharge	Standard video plus usual care
Zhang 202034	China; parallel-group RCT	86; deep II/III-degree burns; TBSA 7–21% vs 9–25%	VR plus music	First 3 dressing changes; initiated 5 min before dressing	Usual care
Chen 201935	China; non-randomized controlled study	112 adults; severe large-area burns excluded; numeric TBSA NR	VR distraction	Started 5 min before and continued during dressing	Music distraction
Yang Q 202336	China; parallel-group RCT	94 pediatric outpatients (6–12 y); burn/scald severity and TBSA NR	VR during dressing	Single dressing episode (principally second dressing)	Usual pain management
Fu 202537	China; parallel-group RCT	60; burn rehabilitation phase with residual wounds; numeric TBSA NR	Acupoint massage plus music	1-month treatment course	Acupoint massage
Yang 201138	China; non-randomized controlled study	46; mild-to-moderate burns; TBSA NR	Music therapy	20-30 min, twice daily; total course NR	Usual care
Najafi 201839	Iran; 2×2 factorial RCT	240 inpatients; mean TBSA 23.92±9.75% (range 14–47%)	Patient-preferred music, Swedish massage, or combined intervention	Single 20-min session before wound care	Usual care
Whitehead-Pleaux 200640	United States; randomized pilot study	14 pediatric burn patients; burn severity/TBSA NR	Live music therapy	During donor-site dressing (procedure 5–40 min)	Verbal interaction
Fratianne 200141	United States; randomized repeated-measures study	25; mean TBSA 9.96% (range 1–43%)	Music-based imagery and musical alternate engagement	Two consecutive debridement days; ~15-30 min before through 15–20 min after procedure	Within-participant no-music control
Görücü 202542	Türkiye; three-arm RCT	120; 67.5% second-degree; eligibility TBSA <20% (2nd degree) or <15% (3rd degree)	VR or music	20 min during each of 2 dressing procedures, 1 week apart	Usual care
Rohilla 201843	India; quasi-experimental crossover pilot	10 patients; 104 dressings; mean TBSA 27±19% (range 5–60%)	Patient-selected music during dressing	Alternate dressing sessions every other day over ~2 months	Standard dressing-change care
Harandi 200444	Iran; randomized clinical trial	44 women; 66% second/third degree; TBSA 27.1±9.3% vs 30.2±11.3%	Hypnotherapy	4 sessions over approximately 3 days	Usual physiotherapy
Wang HY 202345	China; parallel-group RCT	88 children (5–12 y); numeric TBSA NR	VR video	30 min, 3 times/day for 5 days, including one session during dressing	Mobile-phone video
Yin 201646	China; parallel-group RCT	40 children with head/facial burns; burn area 1–25%	Comprehensive psychological intervention	Intervention duration NR	Usual treatment and nursing care
Wang Q 202347	China; randomized crossover study	34 school-aged children; mild deep II/III-degree burns; TBSA mean 1.91% (range 1–6%; all <10%)	Adjunctive VR	First/second dressing crossover; initiated ~5 min before dressing	Standard care

Note: Values are reported as stated in the source studies; when a study did not report numeric TBSA or duration, NR is shown.

Abbreviations: NR, not reported; RCT, randomized controlled trial; TBSA, total body surface area; VR, virtual reality.

Table 2.

Clinical Setting, Outcome Instruments, Follow-Up, and Contribution to Quantitative Synthesis

Study	Clinical Setting	Pain/Psychological Outcome Instruments	Follow-Up/Assessment Period	Quantitative Contribution
Shu 202532	Pre-discharge/post-treatment psychological rehabilitation	SAS; SDS; coping/QoL; aesthetic outcome	Intervention 7 days; aesthetic outcome at 3 months	Anxiety, depression, aesthetic outcome
He 202133	Pre-discharge psychological intervention	SAS; SDS; Social Avoidance and Distress	1 month after discharge	Anxiety, depression
Zhang 202034	Wound dressing	VAS/NRS pain; heart rate; blood pressure	Across first 3 dressing changes	Pain
Chen 201935	Dressing-change procedural pain	Worst pain; pain unpleasantness; attention-to-pain measures	Immediate procedural assessment	Narrative synthesis; active comparator
Yang Q 202336	Outpatient pediatric dressing	NRS-based pain occurrence; cooperation; dressing time; satisfaction	Single dressing episode	Pain incidence, satisfaction
Fu 202537	Burn rehabilitation	VAS; SAS; SDS; quality of life	After 1-month course	Pain, anxiety, depression
Yang 201138	Inpatient burn care	SAS; SDS	End of treatment; total course NR	Anxiety, depression
Najafi 201839	Before painful burn care procedure	Persian BSPAS; anticipated anxiety; relaxation/pain measures	Immediately around single 20-min intervention	Pain-related anxiety
Whitehead-Pleaux 200640	Donor-site dressing change	Wong-Baker FACES; Fear Thermometer; Nursing Assessment of Pain Index	During/immediately after dressing	Narrative synthesis
Fratianne 200141	Burn debridement	Combined Faces/VAS self-report pain/anxiety; pulse; observed tension	Across 2 consecutive treatment days	Narrative synthesis
Görücü 202542	Burn dressing/care procedure	VAS; STAI-I/II; BSPAS	Pre/post assessments across 2 dressings 1 week apart	Pain, anxiety, pain-related anxiety
Rohilla 201843	Dressing change	NRS pain; state-trait anxiety measure; opioid use	Repeated alternating dressings over ~2 months	Opioid use; dressing episodes as unit
Harandi 200444	Physiotherapy	VAS procedural pain; state anxiety	Across 4 sessions over ~3 days	Narrative synthesis
Wang HY 202345	Dressing change	Wong-Baker FACES pain; serum cortisol	Day 1 to day 5	Pain
Yin 201646	Head/facial burn psychological care	Psychological status/post-traumatic growth measures; cooperation	Every 3 months for 1 year after discharge	Narrative synthesis
Wang Q 202347	Burn wound dressing	Wong-Baker FACES; FLACC; vital signs/adverse effects	10 min before, during, and 10 min after dressing	Narrative synthesis; paired correlation unavailable

Abbreviations: BSPAS, Burn Specific Pain Anxiety Scale; FLACC, Face, Legs, Activity, Cry, Consolability; NRS, Numeric Rating Scale; NR, not reported; SAS, Self-Rating Anxiety Scale; SDS, Self-Rating Depression Scale; STAI, State-Trait Anxiety Inventory; VAS, Visual Analog Scale.

Risk of Bias

Among the 12 randomized, crossover, or repeated-measures studies, 8 were judged to have some concerns and 4 to be at high overall risk of bias; none was rated low risk. Common concerns were incomplete reporting of randomization or allocation concealment, inability to blind participants and intervention personnel, reliance on subjective pain and psychological outcomes, and incomplete reporting of prespecified protocols or paired-data handling. The 4 non-randomized studies were judged to have serious overall risk of bias, principally because of confounding, participant selection, and subjective outcome measurement (Figures 2 and 3).

Figure 2.

Risk of bias in randomized studies assessed using RoB 2.

Figure 3.

Risk of bias in non-randomized studies assessed using ROBINS-I.

Pain

Four independent studies involving 354 participants contributed to the pain meta-analysis. After eligible intervention arms in the three-arm trial were combined to avoid repeated use of the shared control, non-pharmacological interventions were associated with lower pain scores (SMD = −0.78; 95% CI, −1.27 to −0.29; p = 0.002; I² = 78.8%; tau² = 0.194; Cochran Q = 14.16, p = 0.003). After excluding the mobile-video active-comparator study and the add-on music comparison against acupoint massage, 2 independent studies involving 206 participants remained. The effect remained favorable (SMD = −0.61; 95% CI, −1.16 to −0.06; p = 0.030; I² = 71.7%; tau² = 0.113; Q = 3.54, p = 0.060). Leave-one-out analyses yielded pooled effects ranging from −0.94 to −0.55, with no change in direction. The pooled pain effect is shown in Figure 4.

Figure 4.

Study-level random-effects meta-analysis of non-pharmacological interventions for pain.

General Anxiety and Depressive Symptoms

Five independent studies involving 412 participants contributed to the general-anxiety meta-analysis. After combining eligible intervention arms within the three-arm trial, the overall analysis favored non-pharmacological interventions (SMD = −1.52; 95% CI, −2.22 to −0.82; p < 0.001; I² = 89.3%; tau² = 0.556; Q = 37.40, p < 0.001). When restricted to 3 randomized controlled trials involving 261 participants, the pooled effect was SMD = −1.03 (95% CI, −1.63 to −0.43; p = 0.001; I² = 79.4%; tau² = 0.222; Q = 9.71, p = 0.008). The direction of effect was unchanged in leave-one-out analyses, with pooled effects ranging from −1.77 to −1.31. The pooled general-anxiety effect is shown in Figure 5.

Figure 5.

Study-level random-effects meta-analysis of non-pharmacological interventions for general anxiety.

Four independent studies involving 292 participants reported depressive symptoms. The overall analysis favored non-pharmacological interventions (SMD = −1.60; 95% CI, −2.06 to −1.13; p < 0.001; I² = 65.7%; tau² = 0.144; Q = 8.73, p = 0.033). Restricting the analysis to 2 randomized controlled trials involving 141 participants yielded an SMD of −1.31 (95% CI, −1.91 to −0.72; p < 0.001; I² = 60.8%; tau² = 0.113; Q = 2.55, p = 0.110). Leave-one-out estimates ranged from −1.77 to −1.42 and remained favorable. The pooled depressive-symptom effect is shown in Figure 6.

Figure 6.

Random-effects meta-analysis of non-pharmacological interventions for depressive symptoms.

Pain-Related Anxiety and Other Outcomes

Two independent multi-arm studies involving 360 unique participants contributed to the pain-related-anxiety meta-analysis. Eligible intervention arms sharing a common control were combined within each study before pooling, so that each trial contributed one independent effect estimate. The pooled result favored non-pharmacological interventions (SMD = −0.83; 95% CI, −1.28 to −0.38; p < 0.001; I² = 70.3%; tau² = 0.075; Q = 3.37, p = 0.066). The two study-level estimates were approximately −0.58 and −1.04, indicating a consistent direction but clinically important between-study variation. The pooled pain-related-anxiety effect is shown in Figure 7. A summary of all meta-analysis results is provided in Table 3. The distribution of intervention types and the dependency issues relevant to subgroup interpretation are summarized in Table 4.

Figure 7.

Study-level random-effects meta-analysis of non-pharmacological interventions for pain-related anxiety.

Table 3.

Summary of Meta-Analysis Results and Heterogeneity Statistics

Outcome	Independent Studies	Participants	SMD (95% CI)	p	I2 (%)	tau2	Cochran Q (p)
Pain: overall	4	354	−0.78 (−1.27 to −0.29)	0.002	78.8	0.194	14.16 (0.003)
Pain: restricted sensitivity	2	206	−0.61 (−1.16 to −0.06)	0.030	71.7	0.113	3.54 (0.060)
General anxiety: overall	5	412	−1.52 (−2.22 to −0.82)	<0.001	89.3	0.556	37.40 (<0.001)
General anxiety: RCTs only	3	261	−1.03 (−1.63 to −0.43)	0.001	79.4	0.222	9.71 (0.008)
Depressive symptoms: overall	4	292	−1.60 (−2.06 to −1.13)	<0.001	65.7	0.144	8.73 (0.033)
Depressive symptoms: RCTs only	2	141	−1.31 (−1.91 to −0.72)	<0.001	60.8	0.113	2.55 (0.110)
Pain-related anxiety	2	360	−0.83 (−1.28 to −0.38)	<0.001	70.3	0.075	3.37 (0.066)

Note: Negative SMD values favor the non-pharmacological intervention.

Abbreviations: CI, confidence interval; RCT, randomized controlled trial; SMD, standardized mean difference.

Table 4.

Descriptive Intervention-Type Heterogeneity and Limitations of Subgroup Inference

Outcome	Intervention-Type Pattern	Independent Studies	Dependency/Sparsity Issue	Interpretation
Pain	VR, music, VR+music, acupoint massage+music	4	Several modalities represented once; Görücü includes VR and music arms sharing one control	Modality may contribute to heterogeneity; no reliable formal between-subgroup test
General anxiety	Music-based, VR/video psychological, combined interventions	5	Sparse modality groups; Görücü includes VR and music arms sharing one control	Descriptive differences only; no comparative ranking
Depressive symptoms	Music-based, VR/video psychological, combined interventions	4	Most modality categories represented by one or two studies	Formal subgroup inference underpowered
Pain-related anxiety	VR, music, preferred music, Swedish massage, combined music+massage	2	Both contributing studies are multi-arm trials with shared controls	Formal intervention-type subgroup meta-analysis not statistically reliable

Notes: Intervention-type patterns are presented descriptively. Formal between-subgroup tests were not emphasized because several categories were represented by only one independent study and some arm-level comparisons shared a common control.

Other outcomes were each reported by a single study and were not pooled. Yang Q 2023 reported a lower incidence of pain and a higher overall satisfaction rate in the virtual reality group. Shu 2025 reported a higher rate of good or excellent aesthetic outcomes. Rohilla 2018 reported less opioid use during music-assisted dressing-change sessions; however, dressing-change episodes rather than independent participants were used as the unit of analysis, so this finding should be interpreted cautiously. The individual dichotomous outcomes are summarized in Table 5.

Table 5.

Dichotomous Outcomes Reported by Individual Studies

Outcome	Study	Events/Total, Intervention	Events/Total, Comparator	RR (95% CI)	Definition
Pain incidence	Yang Q 202336	2/47	18/47	0.11 (0.03 to 0.45)	Event defined as pain
Overall satisfaction	Yang Q 202336	46/47	39/47	1.18 (1.03 to 1.35)	Event defined as satisfaction
Good/excellent aesthetic outcome	Shu 202532	50/53	40/52	1.23 (1.04 to 1.44)	Good/excellent outcome
Opioid use during dressing	Rohilla 201843	2/52	10/52	0.20 (0.05 to 0.87)	Dressing episodes; observations not independent

Intervention-Type Heterogeneity

Intervention type was examined descriptively as a potential source of heterogeneity. Formal between-subgroup tests were not considered reliable because several modality categories were represented by only one independent study and the Görücü three-arm trial contributed both virtual-reality and music intervention arms against the same control group. Arm-level estimates varied across virtual reality, music, combined VR-plus-music, acupoint-massage-plus-music, massage, and structured psychological interventions, supporting the view that modality may contribute to heterogeneity but not permitting a statistically defensible ranking of intervention categories. The distribution of intervention types and the dependency issues relevant to subgroup interpretation are summarized in Table 4.

Certainty of Evidence

GRADE assessment indicated low-certainty evidence for pain because of risk of bias and inconsistency. Evidence for general anxiety and depressive symptoms from randomized controlled trials was rated very low certainty because of risk of bias, inconsistency, and imprecision. Evidence for pain-related anxiety was downgraded to low certainty because of risk of bias and inconsistency after the multi-arm studies were analyzed at the independent study level. Because fewer than 10 independent studies contributed to each outcome, publication bias could not be assessed reliably. The GRADE summary of findings is presented in Table 6.

Table 6.

GRADE Summary of Findings

Outcome	Studies/Participants	Effect Estimate (95% CI)	Reasons for Downgrading	Certainty	Interpretation
Pain	4/354	SMD −0.78 (−1.27 to −0.29)	Risk of bias; inconsistency	Low	May reduce pain
General anxiety (RCTs only)	3/261	SMD −1.03 (−1.63 to −0.43)	Risk of bias; inconsistency; imprecision	Very low	May reduce anxiety; magnitude uncertain
Depressive symptoms (RCTs only)	2/141	SMD −1.31 (−1.91 to −0.72)	Risk of bias; inconsistency; imprecision	Very low	May reduce short-term symptoms; magnitude highly uncertain
Pain-related anxiety	2/360	SMD −0.83 (−1.28 to −0.38)	Risk of bias; inconsistency	Low	May reduce pain-related anxiety; magnitude uncertain

Abbreviations: GRADE, Grading of Recommendations Assessment, Development and Evaluation; RCT, randomized controlled trial; SMD, standardized mean difference.

Discussion

Principal Findings and Clinical Interpretation

The pooled results should not be interpreted as evidence of a uniform class effect for non-pharmacological interventions. After the dependency structure of multi-arm trials was re-examined, the direction of the pooled effects remained favorable, but heterogeneity was substantial for pain, general anxiety, and pain-related anxiety. Previous burn-specific reviews and meta-analyses have also reported variation associated with device immersion, mode of delivery, comparator conditions, and clinical setting.^8–15 The present review extends those modality-specific syntheses by evaluating pain, general anxiety, depressive symptoms, and procedure-specific pain-related anxiety within the same review, using design-specific risk-of-bias tools, comparator and design sensitivity analyses, and GRADE certainty ratings. The available intervention categories were too sparse and partly dependent to support a reliable ranking of modalities.

The pooled continuous effects were expressed as SMDs because the included studies used different instruments to measure the same constructs. This permits quantitative synthesis but limits direct clinical interpretation because a pooled SMD cannot be converted into one common number of points on the VAS, NRS, SAS, SDS, or BSPAS. Pain-medicine-specific guidance recommends prespecifying clinically important thresholds for primary outcomes when appropriate;²⁷ however, the included burn studies rarely reported a consistent minimal clinically important difference or responder threshold that could be applied across procedures and instruments. Statistical significance should therefore not be treated as evidence of clinically important benefit for every patient. Effect size, certainty, heterogeneity, intervention burden, safety, and patient preference need to be considered together.

In burn care, repeated procedures may be as important as the pain score recorded during a single session. Dressing changes, debridement, and rehabilitation stretching recur throughout treatment, and anticipatory distress can affect cooperation with later procedures. Adjunctive measures such as clear procedural explanations, patient-preferred music, relaxation training, or virtual reality may be useful in selected patients after appropriate safety assessment, but they should be evaluated in relation to the specific procedure and concomitant analgesia.

Pain Findings Support Procedure-Specific Selection of Adjunctive Interventions

The high heterogeneity in the overall pain analysis, which persisted in the restricted analysis, indicates that the included studies addressed related but not identical clinical questions. Comparisons with usual care may capture effects of additional attention, novelty, or interaction as well as the intervention itself, whereas active comparators more directly estimate incremental modality-specific effects. Descriptive stratification suggested variation by intervention type, but formal subgroup-difference testing was not reliable because most categories contained only one independent study and some arm-level contrasts shared a control group. Adult and pediatric burn pain guidelines place non-pharmacological approaches within timely, individualized, multimodal analgesia.^3,4 Intervention selection should therefore depend on clinical context, feasibility, and safety rather than on subgroup point estimates alone.

For brief procedures such as dressing changes and debridement, patient-preferred music, relaxation, interactive distraction, or virtual reality may be considered as adjuncts when they are acceptable and feasible, with adequate pharmacological analgesia maintained. Evidence from rehabilitation and intensive care settings is more limited and should not be extrapolated directly from dressing-change studies. Hoffman et al reported the feasibility of immersive virtual reality during wound cleaning in children with large, severe burns.⁴⁸ In adults receiving intensive care, Ettenberger et al found that music-assisted relaxation may reduce early background pain, whereas clear between-group effects on general anxiety and depression were not observed.⁴⁹ These studies support procedure- and setting-specific interpretation rather than a single protocol for all burn pain.

General Anxiety and Depressive Symptoms Require Formal Screening and Ongoing Management

The pooled effects for general anxiety and depressive symptoms were large, but these findings are particularly susceptible to small-sample effects, lack of blinding, differential attention, and subjective outcome measurement. Pain, anxiety, and depression remain interrelated over the longer course of burn recovery, and short-term reductions in questionnaire scores do not demonstrate that persistent psychological disorders have been treated.^5–7 Some studies reassessed the Self-Rating Anxiety Scale, Self-Rating Depression Scale, or Hospital Anxiety and Depression Scale shortly after the intervention, although these instruments typically capture symptoms over a defined period. When the assessment time does not align with the recall period, lower scores may primarily reflect immediate emotional relief or a better care experience. Recent meta-analyses of music interventions in adults and children likewise reported large effects with substantial heterogeneity and emphasized the need to standardize intervention protocols, providers, and assessment timing.^16–20

Bedside non-pharmacological measures should complement, not substitute for, psychological assessment when symptoms persist. Low mood, sleep disturbance, marked social withdrawal, appearance-related distress, or avoidance of treatment may warrant formal assessment with validated instruments and referral according to local clinical pathways. Deng et al reported favorable effects of a structured dignity-based intervention on selected psychological and quality-of-life outcomes in people with burns.⁵⁰ Such interventions address different clinical goals from brief distraction during dressing changes and should be evaluated separately.

Pain-Related Anxiety May Be a Useful Indicator for Managing Repeated Procedures

Pain-related anxiety is linked to an anticipated dressing change, debridement, or stretching procedure. Anticipation may heighten vigilance and avoidance and can make repeated procedures more difficult. After the shared-control dependency was handled at the study level, heterogeneity for this outcome was substantial rather than minimal. The pooled estimate therefore suggests a favorable direction, but the magnitude is likely to vary with intervention composition, patient population, and comparator context.

Only two multi-arm studies contributed to pain-related anxiety, and both used shared comparators. The revised study-level synthesis is therefore more conservative and should not be interpreted as evidence of a consistent modality-specific effect or an established causal mechanism. Patient-preferred music, massage, and virtual reality showed favorable directions in the available trials, but intervention components and comparator structures differed.^39,42 Future trials should report pre-procedural pain-related anxiety, peak procedural pain, rescue analgesia, procedure completion, and anxiety before the next procedure, together with covariance information for multi-arm or repeated-measures designs.

Heterogeneity and Risk of Bias Support Local Evaluation Frameworks

Between-study differences involved age, burn size and depth, TBSA, phase of care, concomitant analgesia, intervention dose, provider, comparator intensity, outcome instrument, and measurement timing. These differences are plausible contributors to the moderate-to-high I² values and non-zero tau² estimates. Risk-of-bias domains also affect confidence in the pooled estimates. Incomplete reporting of randomization and allocation concealment may introduce selection bias. Blinding of participants and intervention personnel was often not possible, which may increase expectation and attention effects for subjective outcomes. Incomplete outcome data may bias estimates when attrition is related to symptom burden or intervention tolerability, and the absence of prespecified protocols increases concern about selective outcome reporting. Serious confounding and selection concerns in non-randomized studies further reduce causal confidence. These limitations informed the GRADE downgrades; the pooled estimates are therefore more useful for judging direction of effect than for defining precise treatment magnitudes.

Where burn units use these interventions, routine documentation would strengthen local evaluation. Relevant items include procedure type, timing of pharmacological analgesia, intervention type and duration, pain and pain-related anxiety before and after the procedure, rescue analgesia, procedure completion, adherence, and adverse events. Consistent recording of these variables would make it easier to identify which patients appear to benefit and would provide data for local quality-improvement work or future trials.

Practical Implications for Patients and Burn Services

Discussion with patients should remain limited to outcomes that were actually evaluated in the included studies, including procedural pain, pain-related anxiety, selected short-term psychological symptoms, cooperation, and satisfaction. When more than one acceptable option is available, patient preference can guide the choice between approaches such as music, virtual reality, or relaxation. Alternatives that do not require devices may be preferable for patients who decline virtual reality or who have dizziness, sensory hypersensitivity, or facial wounds. These measures remain adjunctive and do not replace appropriate pharmacological analgesia, wound care, or specialist psychological support.^3,4,23

Implementation should reflect the procedure, patient needs, staff skills, and local resources. Patient-preferred music and brief relaxation require little equipment, whereas virtual reality, music therapy, hypnosis, and structured psychological interventions may require devices or trained staff. Local protocols should specify equipment cleaning, monitoring for dizziness or nausea, contraindications to massage, stopping criteria, and referral thresholds. These details are important if the interventions are to be used consistently and evaluated in routine burn care.

Limitations and Directions for Future Research

The included studies were generally small and clinically heterogeneous, and blinding was difficult or impossible for several interventions. Burn depth and TBSA were incompletely reported in some studies, which limited severity-stratified interpretation. Multi-arm and crossover designs also complicated synthesis. After shared-control dependencies were handled at the independent study level, several pooled estimates changed modestly and heterogeneity for pain-related anxiety increased substantially. Most intervention categories contained only one independent study, making formal subgroup-difference tests or meta-regression underpowered and potentially misleading. Fewer than 10 independent studies contributed to each outcome, so publication bias and small-study effects could not be assessed reliably with funnel plots or formal asymmetry tests. Pain-medicine-specific guidance similarly notes the limited power of sparse meta-analyses for evaluating heterogeneity, subgroup effects, and small-study effects.²⁸

Future multicenter randomized trials should prespecify primary outcomes and assessment time points and report sequence generation, allocation concealment, concomitant analgesia, attrition, paired variances, and adverse events in sufficient detail. In addition to pain intensity, clinically useful outcomes include rescue analgesia, procedure completion, anxiety before subsequent procedures, adherence to rehabilitation, nursing workload, equipment maintenance, and cost.

Conclusion

Selected non-pharmacological interventions were associated with lower procedural pain and pain-related anxiety and with improvements in some short-term psychological symptoms in patients with burns. The certainty and generalizability of these findings are limited by heterogeneity in intervention modality, burn severity, comparator, outcome measurement, and risk of bias. The evidence does not establish uniform efficacy or superiority of any intervention category. Non-pharmacological strategies should be used as adjuncts to adequate pharmacological analgesia, wound care, rehabilitation, and psychological assessment or treatment when indicated.

Funding Statement

This study was supported by the Ningxia Natural Science Foundation (Grant No. 2026AAC031111).

Ethics Statement

Not applicable. This review used data from published studies, did not directly recruit participants, and did not access identifiable individual-level information.

Author Contributions

Miaomiao Ye and Jing Yao contributed equally to this work and share first authorship.

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Miaomiao Ye and Jing Yao contributed equally to this work and share first authorship.

Disclosure

The authors report no conflicts of interest in relation to this work.