Objective: To conduct a systematic review of the efficacy of music for medical indications in term or preterm neonates.
Methods: We searched 17 electronic databases, subject bibliographies, reference lists and trials registries. Two reviewers independently screened studies for inclusion, assessed methodological quality, and extracted data. Meta-analysis was not feasible due to heterogeneity in outcomes so a qualitative analysis is presented.
Results: Nine randomised trials were included. The methodological quality was generally poor (median Jadad score = 1). The outcomes most commonly reported were physiological measures (heart rate (HR), respiratory rate, oxygen saturation (Sao2)), behavioural state and pain. Six studies evaluated music for the painful procedures circumcision (three trials) and heel prick (three trials). For circumcisions, one high quality pilot study (n = 23) showed benefits of music for the outcomes of HR, Sao2 and pain, while two low quality studies showed no difference. For heel prick, three low quality studies provided some evidence that music may be beneficial primarily for measures of behaviour and pain. The remaining studies evaluated the use of music in preterm infants to improve physiological and behavioural parameters (n = 31; benefits observed for behavioural parameters), to reinforce non-nutritive sucking via use of a pacifier activated lullaby (n = 32; significant increase in feeding rates), and to influence physiological stability and behaviours in infants with chronic lung disease (n = 22; no significant differences for outcomes assessed).
Conclusions: The heterogeneity in study populations, interventions and outcomes precludes definitive conclusions around efficacy. There is preliminary evidence for some therapeutic benefits of music for specific indications; however, these findings need to be confirmed in methodologically rigorous trials.
Statistics from Altmetric.com
Music is increasingly being used in neonatal units to improve behavioural or physiological outcomes or to manage pain during common procedures. The benefits listed are calmer infants and parents, stable physiological parameters, higher oxygen saturation, faster weight gain and shorter hospital stay.1 2 3 4
What this study adds
This methodologically rigorous systematic review identified nine randomised controlled trials that have studied the effects of music in the neonatal population for diverse indications.
Music may have beneficial effects on physiological parameters and behavioural states, and may reduce pain and may improve oral feeding rates among premature infants.
Although provision of effective pain relief is now standard for neonates during and after a major surgical procedure, pain-reducing therapies are often underused for the numerous minor procedures such as blood sampling that are frequently needed for routine medical care and decision-making in neonatal intensive care units (NICUs).5 A variety of non-pharmacological interventions such as oral sucrose/glucose administration,6 breastfeeding,7 facilitated tucking,8 and containment and swaddling,9 have been evaluated for the treatment of acute procedural pain in neonates. These measures have been shown to be useful in preterm and term neonates in reducing pain and are generally more effective when used in combination than when used alone. The opiates are the mainstay of treatment among the pharmacological interventions. However, the administration of opiates is associated with substantial risks including decreased oxygenation and respiratory drive, and thus their use is mostly restricted to invasive procedures and postoperative pain.10
The prevention of pain is important not only because it is an ethical expectation but also because repeated painful exposures can have deleterious consequences. Pain can cause short-term effects such as excessive crying, choking, gagging and vomiting. Long-term effects of noxious stimuli include altered pain sensitivity and the possibility of permanent neuroanatomical and behavioural abnormalities.5
As music and other physiological sounds provide an auditory stimulus which may modulate pain perception, music has been evaluated as therapy for pain. Music could obviate or decrease the need for pharmacological agents, thus reducing related harm and may result in savings in drug or provider costs in the healthcare system.
A meta-analysis published in 2002 concluded that music was beneficial for a range of outcomes among premature infants in the NICU.11 The review had substantial methodological flaws based on standard criteria for assessing research syntheses,12 including unclear reporting of how bias in the selection of studies was avoided, no assessment of the validity of the included studies, and inappropriate combining of results for different outcomes. Moreover, the results were limited by the poor methodological quality of the included studies (few employed methods of randomisation), and the conclusions and recommendations were not well supported by the data.
Our objective was to review the evidence base for music as a therapeutic modality among neonates by conducting a systematic review according to defined methodological standards.13 Our intent was to synthesise the evidence from randomised controlled trials on the efficacy of music for medical indications during the neonatal period.
A research librarian developed comprehensive search strategies relevant to music for neonates. A sample search strategy is provided online in appendix A. In July 2007, we conducted systematic searches in 15 electronic databases which included MEDLINE, MEDLINE In-Process & Other Non-indexed Citations, ERIC, EBM Reviews – Cochrane Central Register of Controlled Trials, OCLC Articles First, RILM Abstracts of Music Literature, RIPM Music Periodicals, CAIRSS (Computer-Assisted Information Retrieval Service System) for Music, PsycINFO, AMED (Allied and Complementary Medicine Database), EMBASE (Excerpta Medica), LILACS, Academic Search Premier, CINAHL (Cumulative Index to Nursing and Allied Health Literature), and Web of Science. Searches for grey literature were conducted in Doctoral Dissertations in Musicology, Proquest Dissertations and Theses: Fulltext, OCLC Proceedings and OCLC Papers First. We scanned subject bibliographies including the Bibliography of Australian Music Education Research (BAMER) as well as reference lists of relevant studies. We also searched for ongoing trials in Current Controlled Trials and the National Research Register. In May 2008, an update was conducted in databases that had originally yielded a large number of results: MEDLINE, OCLC Papers First, CINAHL, Proquest Dissertations and Theses: Fulltext and EBM Reviews – Cochrane Central Register of Controlled Trials. We did not impose any language restrictions.
The results of the search strategies were screened independently by two reviewers (MS, RL). The full manuscripts of potentially relevant studies were retrieved and screened for inclusion independently by two reviewers (MS, RL). Studies were included in the review if they met the following criteria: they were randomised controlled trials, examined subjects less than 1 month of age, used music as an intervention, and measured pain or physiological or behavioural parameters. Discrepancies regarding inclusion were resolved through discussion among the review team.
Quality assessment was performed independently by two reviewers (MS, LH). Quality was assessed based on allocation concealment, blinding of outcome assessors, and the Jadad score (randomisation, double-blinding, reporting of withdrawals and dropouts).14 Discrepancies were resolved by consensus.
Data were extracted by one reviewer (MS) using a standardised form and checked for accuracy by a second reviewer (LH). The following information was extracted: characteristics of the study (eg, language of publication, country, funding), characteristics of the study population and setting, description of the intervention and comparisons (eg, type of music, method of administration), outcome measures and measurements tools, and results.
Meta-analysis was planned but was not feasible due to inconsistent reporting or marked clinical heterogeneity in study populations, interventions, and the outcomes studied. A qualitative analysis is presented.
Description of included studies
Figure 1 describes the flow of studies through the selection process. The searches identified 1111 studies of which 180 were examined in detail for inclusion. Nine studies met the inclusion criteria, three of which were cross-over trials.15 16 17 The studies were reported between 1989 and 2006 (median year 2001). Eight studies were published in peer-reviewed journals, while one study was reported as a doctoral dissertation.18 The studies were conducted in a variety of countries around the world: USA (n = 5),18 19 20 21 22 Canada (n = 1),16 China (n = 1),15 Israel (n = 1)17 and Australia (n = 1).23
The studies varied in their study populations and indications for therapy, interventions and comparisons, and timing and measurement of outcomes (tables 1 and 2). Five studies included preterm infants,16 17 18 22 23 three studies evaluated term infants,19 20 21 and one study included both term and preterm infants.15 Six studies evaluated the use of music during medical procedures, including circumcision (three trials19 20 21) and heel prick (three trials15 16 18). Three studies evaluated music for a variety of other indications: to improve physiological and behavioural parameters (one trial17), to reinforce non-nutritive sucking for better oral feeding among poor-feeding preterm neonates (one trial22), and to influence physiological stability and behaviours in infants with chronic lung disease (one trial23).
There is marked variation across studies in the type of musical selection with lullabies being the most common (six trials16 17 18 19 22 23), in the methods of administration which were more often recorded (nine trials15 16 17 18 19 20 21 22 23) than live (one trial17), in the use of music alone (four trials16 20 21 23) or combined with other sounds or modalities (eg, intrauterine sounds,15 21 heartbeat,19 pacifier,15 18 21 22) (five trials), and in the number and type of comparison group(s). Comparison group(s) in the included trials varied from routine care (eight trials15 16 17 18 19 21 22 23), pacifier (three trials15 18 21) and intrauterine sounds (two trials20 21) to combinations of these modalities (one trial21). Two trials compared different types of music: live versus recorded17 and vocal versus instrumental.16
All but one study collected information on physiological measures with respiratory rate (RR), heart rate (HR) and oxygen saturation (Sao2) being the most common. The majority of studies also collected information on pain and behavioural state; however, different methods were used to measure these outcomes. Pain was measured using the Riley Infant Pain Scale,19 Neonatal Facial Coding System,16 Neonatal Infant Pain Scale15 and Izard’s Maximally Discriminative Facial Movement Coding System.20 Behavioural state was measured based on length of cry,19 Brazelton Neonatal Behavioral Assessment Scale (alertness section),21 modified Brazelton’s state-of-arousal,16 numerical score defined by Als et al,17 18 24 and an adapted version of the Physiological and Behavioural Assessment Form.23 In addition, one study evaluated stress levels based on criteria developed by Burns et al.25
Methodological quality of included studies
The methodological quality was generally poor: no studies reported adequate concealment of allocation and the median Jadad score was 1 (interquartile range: 1–2) (table 3). The outcome assessor was blind to the intervention for at least one outcome in four studies19 20 21 23; in one study the nurses who were recording data were blind to the purpose of the study.22
Qualitative synthesis of individual studies
Music during medical procedures
The three studies that evaluated music during circumcision showed different results. Joyce et al compared music (lullabies and nursery rhymes metronomised to a real human heartbeat) to no music among 23 term infants.19 The music group demonstrated benefits in terms of significantly lower pain scores and higher oxygen saturation at the end of the procedure, with no increased heart rate as was seen in the no-music control group. No effect was seen for RR or salivary cortisol levels. This study also simultaneously evaluated the effect of EMLA (Eutectic Mixture of Local Anesthetics) on pain reduction during circumcision. The authors do not state whether their trial had a factorial design and do not present data comparing music with EMLA or music combined with EMLA. Nevertheless, this study was assessed as high quality based on the Jadad score and blinding of pain assessments. Two earlier studies showed no clinically or statistically significant effects for recorded classical music versus various comparison groups (see table 1) in terms of physiological parameters or behavioural state.20 21 However, these two studies were poorly reported which suggests potential for bias in their results.
The three studies that evaluated music during heel prick were all of low methodological quality. In a randomised, cross-over design, Bo et al compared four groups, including a group that received music combined with intrauterine maternal pulse sounds; results showed that all three interventions had a significant effect on HR, oxygen saturation and pain.15 While HR was most affected by the music/intrauterine sounds intervention, the other two variables were most influenced when music/intrauterine sounds were combined with non-nutritive sucking (NNS). Butt et al compared recorded music involving both vocal and instrumental versions of Brahms’ lullaby versus no music in a cross-over trial of 14 preterm infants.16 The authors found a differential effect by age where infants older than 31 weeks demonstrated significant benefits from music in terms of HR, behavioural state and pain, while infants younger than 31 weeks showed no significant effects. Whipple compared a pacifier activated lullaby (PAL) mechanism versus pacifier only versus standard care among 60 preterm infants.18 The author found no clinically or statistically significant differences across groups in terms of physiological measures; however, the PAL and pacifier only groups showed significant benefits over standard care for multiple comparisons related to behavioural states and stress levels. No significant differences were observed between the PAL and pacifier only groups.
Music for other indications
Calabro et al evaluated the effects of recorded instrumental lullabies in 22 preterm infants with chronic lung disease.23 They found no differences compared to a no-music control group in physiological or behavioural outcomes, and concluded that the intervention posed no risk to the infants. The study was methodologically strong based on components assessed by the Jadad scale, as well as blinding of outcome assessors. Standley et al evaluated the use of music-reinforced non-nutritive sucking via PAL among 32 poorly feeding preterm infants and found a significant increase in oral feeding rates compared to a no-contact control group.22 In a randomised, cross-over design, Arnon et al compared live music versus recorded music versus no music in stable preterm infants.17 Live music showed significant benefits in terms of heart rate and behavioural scores during the post-intervention period. No statistically significant changes were observed for the recorded music and control groups. This was the only study to measure the preferences of medical personnel and parents: both groups preferred live music and parents perceived live music to be significantly more beneficial than recorded or no music.
This review demonstrates that music is being considered as an alternative therapeutic modality for a variety of purposes in the neonatal population. The heterogeneity in study populations, interventions and indications precluded a meta-analysis and definitive conclusions around efficacy. In general, the studies demonstrated that music may be beneficial in terms of behavioural states and pain, but showed inconsistent effects overall on physiological measures. While music demonstrated large effects in terms of pain reduction for heel prick procedures, music may not be sufficient in itself to offset the pain associated with more intense medical procedures such as circumcision.21 The results were more often significant when music was compared to a control group as opposed to another active intervention.
While there is preliminary evidence for some therapeutic benefits of music for specific indications, these benefits need to be confirmed in well-designed, high quality trials. The two major methodological flaws in the included studies were unclear concealment of allocation and lack of double-blinding. As it is always possible to adequately conceal allocation, this should be routinely implemented in future trials. Blinding for this type of intervention poses more of a logistical dilemma, as it is not always possible to blind those who may influence the care the infants receive. Calabro et al described in detail the method they used to ensure blinding of the data collectors, which should serve as a minimum standard for future trials.23 In brief, the data collectors themselves listened to recorded music via headphones and a portable cassette player while collecting data so that they would not be aware whether or not the study infant was receiving music. An evaluation of whether blinding of the outcome assessors was effective would further support this mechanism (eg, whether the outcome assessors can guess the study group to which a patient belongs).26
Another limitation of the studies included in this review was the relatively small sample sizes, although several studies were described as pilot or exploratory.16 19 20 23 The mean sample size per intervention group was 18 and ranged from 11 to 31. The lack of significant findings in some instances may be due to insufficient power rather than lack of effectiveness.
The outcomes reported in the studies included in this review were limited to immediate changes and the interventions were applied over a short period of time. Future studies could evaluate the effectiveness of music over a longer duration of time and for longer-term outcomes. For instance, Arnon et al called for more research to assess the “long-term effect on the development of preterm infants”.17
It would be of particular interest as regards the neonatal population to confirm the effects seen with the use of PAL to improve oral feeding rates among the poor-feeding preterm or near term infants.22 Most very low birth weight babies require many days to weeks of gavage feeding before they are ready to take all their feeding requirements via the oral route. This often leads to delay in the time to first discharge home. While one study demonstrated an increased rate of oral feeding with the intervention, it did not report on some more meaningful clinical outcomes such as number of days to full oral feeding or corrected post-menstrual age of the infant when this milestone was achieved. Any benefits stated in future trials in terms of these outcomes would have the potential of saving finite useful healthcare resources (eg, early discharge from hospital could save healthcare resources).
This systematic review was conducted and reported according to rigorous methodological standards.13 The limitations of the review stem from the methodological weaknesses of the existing studies, as well as the heterogeneity across studies which prohibited combining data in order to generate more precise effect estimates. However, one of the contributions of this review is that it has highlighted the paucity and heterogeneity of the existing research which can serve as a guide for future investigations in this area.
In summary, there is preliminary evidence to suggest that music may have beneficial effects in terms of physiological parameters, behavioural states and pain reduction during painful medical procedures. Music may also improve oral feeding among premature infants when combined with other non-invasive modalities, such as non-nutritive sucking. However, most trials conducted to date are of poor methodological quality. Additional methodologically rigorous, randomised controlled trials are warranted to confirm and to further elucidate the benefits of music for neonates before any specific recommendation for the use of music can be made in the neonatal population.
We thank Jeffrey Klassen for assistance with screening of studies for inclusion and Karalee Ratzlaff for assisting with article retrieval.
Appendix A Search strategy for MEDLINE Version: OvidSP_UI01.01.00
This search is comprised of three sections. Section one, lines 1–22, includes subject headings and terms for music. Section two, lines 23–37, includes terms for neonates. Section three, lines 38–76, contains terms for the old and new clinical trials filters endorsed by the Cochrane Collaboration. (At the time of the initial search, the new filter had not yet been incorporated into our regular searching practice, thus we used both to ensure we did not miss any studies). Line 77 is the combination of all three sections and ensures that at least one eligible term from each section is included in the results.
1. exp Music Therapy/
2. exp Music/
3. exp Audioanalgesia/
4. exp Dance Therapy/
5. exp Dancing/
7. (dance$ or dancing).mp.
12. Audio analgesi$.mp.
13. (melodic$ or melody or melodies).mp.
14. (lyric or lyrics or lyrical).mp.
16. Medical resonance therapy music$.mp.
17. “Guided Imagery and Music”.mp.
20. Music Based Dynamic Therapy.mp.
23. exp Infant/
24. exp Nurseries/
29. (neonat$ or perinat$).mp.
32. (preemie$ or VLBW or LBW).mp.
33. (baby or babies).mp.
34. (nursery or nurseries).mp.
36. (neonat$ or perinat$).jw.
38. RANDOMIZED CONTROLLED TRIAL.pt.
39. CONTROLLED CLINICAL TRIAL.pt.
40. RANDOMIZED CONTROLLED TRIALS/
41. RANDOM ALLOCATION/
42. DOUBLE BLIND METHOD/
43. SINGLE-BLIND METHOD/
45. ANIMAL/not HUMAN/
46. 44 not 45
47. CLINICAL TRIAL.pt.
48. exp CLINICAL TRIALS/
49. (clin$ adj25 trial$).ti,ab.
50. ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).ti,ab.
54. RESEARCH DESIGN/
56. 55 not 45
57. 56 not 46
58. COMPARATIVE STUDY/
59. exp EVALUATION STUDIES/
60. FOLLOW UP STUDIES/
61. PROSPECTIVE STUDIES/
62. (control$ or prospectiv$ or volunteer$).ti,ab.
64. 63 not 45
65. 64 not (46 or 57)
66. 46 or 57 or 65
67. clinical trial.pt.
75. 74 not 45
76. 66 or 75
77. 22 and 37 and 76
Competing interests None.
Author contributions: LH provided methodological advice, assessed study quality, extracted and analysed data, and drafted the manuscript. MS reviewed studies for inclusion, extracted data and assessed study quality. LT conducted the literature searches and assisted with manuscript preparation. RL reviewed studies for inclusion, and assisted with data extraction and manuscript preparation. YL provided statistical advice. MK provided technical advice at all stages of the review, assisted with assessment of study quality and data extraction, and helped with drafting of the manuscript. All authors read and approved the final manuscript.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.