Music Therapy for Premature Babies: What Singing in the Neonatal Unit Has and Has Not Been Shown to Do

Music therapy in a neonatal unit usually means a parent singing quietly to their own premature baby, with a music therapist helping them read the baby’s responses; small trials suggest brief calming effects on heart rate and breathing, but the largest trial found no lasting effect on bonding or on development at two years. That is a narrower result than the warm articles about it suggest, and a more useful one, because it tells a parent what the singing is for.

I came to this one sideways. A woman who sings in one of my care home groups told me, between hymns, that her granddaughter had been born at 29 weeks and that a music therapist on the neonatal unit had “taught her to sing to him through the porthole”. She asked me whether it would make him cleverer. I did not know, and nothing on this site could tell her. Where a claim about effect appears below it carries a tier (Tier 1 supported, Tier 2 promising but limited, Tier 3 practice and anecdote), and reading arts in health research explains how those are assigned.

What music therapy on a neonatal unit actually is

It is mostly a parent’s voice, shaped by a therapist, rather than music played to a baby.

In the version tested by the largest trial in the field, the intervention was parent led, infant directed singing: a mother or father singing softly to their own baby, adjusted moment to moment to how the baby was responding, with a music therapist supporting them. It ran three times a week while the baby was in hospital, or as seven sessions across six months after discharge¹. The therapist’s job in that model is less about performing and more about helping a frightened parent notice when the baby is settling, when a sound is too much, and when to stop.

That is a long way from the image of a speaker in an incubator, and the difference matters for the same reason it matters everywhere on this site. A registered therapist working with a family to an agreed purpose is doing clinical work. A playlist left running is not, however soothing it looks, and a result about one does not transfer to the other. Music therapy versus community music sets out that line in full.

The short term effects: what the small trials found

Pooled small trials show modest changes in heart rate, breathing rate, feeding and maternal anxiety while the music is happening, all at low to very low certainty, which is Tier 2.

A 2021 overview in Acta Paediatrica identified 12 systematic reviews on the question, whose methodological quality ranged from moderate to critically low, and then pooled the 14 randomised trials in which the intervention actually fitted a definition of music therapy. It found a mean difference in heart rate of -3.21 (95% CI -5.22 to -1.19), in respiratory rate of -2.93 (95% CI -5.65 to -0.22), a larger feeding volume (MD 29.59, 95% CI 12.79 to 46.38), and lower maternal anxiety (MD -17.50, 95% CI -20.10 to -14.90)².

The same overview rated every one of those outcomes at low to very low certainty, downgraded for high risk of bias, inconsistency and imprecision, and noted that no long term developmental or safety outcomes were reported in the trials it pooled². So these are Tier 2 findings: real signals, measured over minutes and days, in studies that were too small or too flawed to bear much weight. A slightly slower heart rate during a song is a plausible sign of a calmer baby. It is not, on its own, evidence of anything that lasts.

The big trial: what LongSTEP was built to test

LongSTEP randomised 206 preterm babies and their parents in five countries to find out whether singing changed bonding and later development, and on its main outcomes the answer was no.

The Longitudinal Study of Music Therapy’s Effectiveness for Premature Infants and their Caregivers ran from 2018 to 2022 in level III and IV neonatal units in Argentina, Colombia, Israel, Norway and Poland. Babies born before 35 weeks were randomised twice, once at enrolment and again at discharge, so the trial could compare singing in hospital, singing after going home, both, or neither¹. It is the kind of design this field rarely gets: multicentre, pragmatic and powered for a named primary outcome.

The primary outcome was mother and baby bonding at 6 months’ corrected age, measured with the Postpartum Bonding Questionnaire. The differences were 0.55 (95% CI -2.20 to 3.30) for music therapy in hospital and 1.02 (95% CI -1.72 to 3.76) for music therapy after discharge, and the authors concluded that parent led singing did not have clinically important effects on bonding. They also reported that it was safe and well accepted by families¹.

On this site that bonding result is Tier 3 in the stronger of its two senses. Tier 3 usually means nobody has run an adequate trial. Here somebody has, and it came out null. Those are different statements, and the second is the more informative one.

Development at two years and weight gain

Follow up at 24 months found no effect on language, cognitive or motor development, and a separate analysis found no effect on weight gain in hospital.

A predefined secondary analysis followed the LongSTEP children to 24 months’ corrected age. Language, the primary outcome of that analysis, showed mean differences of -2.36 (95% CI -12.60 to 7.88) for singing in hospital, 2.65 (95% CI -7.94 to 13.23) for singing after discharge and -3.77 (95% CI -13.97 to 6.43) for both, with no significant effects on cognitive or motor development either. Only 112 children (54%) were retained to that visit, and most scored in the normal range³. The authors were careful to add that the wide intervals mean potential effects cannot be excluded, which is the honest reading of a null result from a trial that lost nearly half its sample by the end.

A second analysis of 201 babies in the trial looked at weight. It found no significant differences in weight, weight percentiles or weight gain between the music therapy and standard care groups at birth, enrolment or discharge; the degree of prematurity remained the main factor linked to poor growth⁴. That sits awkwardly against the larger feeding volumes in the pooled small trials, and the trial with the stronger design is the one to weigh more heavily.

What a parent might reasonably take from this

Sing if it helps you and the nurses agree, and do not expect it to change the long run.

The temptation with a null trial is to turn it into “it does not work”. That is not what LongSTEP found. It found that singing, delivered this way, did not move bonding or development measured this way, while being safe and welcomed by the parents who did it. The short term calm in the small trials is Tier 2 and may well be real. A parent who sings to a baby in an incubator because it is the one thing they can do with their hands and voice is not doing something the evidence argues against.

What the evidence argues against is the promise. Any service, product or post that tells a parent that music will make a premature baby develop faster, feed better or bond more securely is going further than the best trial supports. As with every result on this site, a group average is not a prediction for one baby either way.

The other caution is the unit itself. Neonatal staff manage light and sound carefully, and very small or unwell babies can be overstimulated by things that seem gentle. Ask the nurse caring for your baby before adding any sound, including your own voice, and follow their lead on timing.

Finding a neonatal music therapist

Ask the unit, check the register, and expect provision to be patchy.

Many neonatal units do not have a music therapist, and where one exists the post may depend on local or time limited funding, so ask rather than assume. If one is offered, the checks are the same as anywhere else. In the UK, music therapist is one of four titles protected in law, and the HCPC register covered 6,103 arts therapists in July 2026⁵; search it yourself rather than relying on a job title. In the United States the credential is MT-BC, held by 10,699 people as at 30 June 2026⁶. The British Association for Music Therapy is the professional body in the UK and a useful starting point for questions about the work⁷. How to find a music therapist covers the routes in more detail, and what happens in a music therapy session describes how the work is structured with older children and adults.

The grandmother from my group, for what it is worth, did not need any of the numbers above. When I told her roughly what the trial found, she said her daughter had never thought the singing would make him cleverer; it had made the fortnight in the unit bearable, and it had given her something to do through the porthole besides watch the monitor. None of that is on a bonding questionnaire. None of it is a reason to stop or skip any care the neonatal team recommends either, and whether a particular baby is ready for any of it is their call.