Skip to content
MyFreud

Sleep Training: Does It Harm Attachment?

Two randomised trials measured the thing parents worry about, using cortisol and the strange situation. What they found, and where the evidence actually stops.

6 min read

Pop-art illustration of a baby in a sleepsuit resting across an adult lap, eyes closed.

Key takeaways

  • The strongest objection to sleep training is that it damages the bond between parent and infant. One randomised trial measured that directly, using infant saliva cortisol and the strange situation procedure, and found no adverse stress response and no effect on attachment.
  • That trial randomised 43 infants. It is the study doing most of the work in this debate and it is small, which means it can rule out a large effect and cannot rule out a modest one.
  • A separate cluster-randomised trial followed 326 children to age six and found no differences between intervention and control families on emotional and behavioural outcomes, sleep, psychosocial functioning or chronic stress.
  • The benefit that gets least attention is to mothers. The trial programme these studies come from was designed around maternal mental health, and improving infant sleep improved it.
  • None of this says you should sleep train. It says the specific claim that it harms the child is not supported by the trials that looked for harm.

The trials that set out to measure whether sleep training harms the parent-infant bond did not find harm. They measured it properly, using infant stress hormones and an observational assessment of attachment rather than a questionnaire, and they followed children for years afterwards.

They are also fewer and smaller than the confidence of the public argument suggests, and the honest version of this answer includes that.

What sleep training actually refers to

It is an umbrella term covering several different things, which is part of why the argument about it is so unproductive.

MethodWhat it involvesWhat the trials tested
Graduated extinctionReturning to settle at gradually increasing intervalsYes, directly
Bedtime fadingMoving bedtime later to match natural sleep onset, then advancing itYes, directly
Camping outStaying in the room, withdrawing presence in stagesStudied elsewhere, not in these trials
Unmodified extinctionNot returning at all until morningRarely tested; not what these trials studied

Most public argument is conducted about the last row and most research has been done on the first two. Someone defending sleep training and someone attacking it are frequently describing different practices.

The trial that measured the actual worry

The objection that matters is not that a baby cries. It is that repeated unanswered distress raises stress hormones and damages the attachment relationship. That is a specific, testable claim, and Gradisar and colleagues tested it.

Forty-three infants aged 6 to 16 months were randomised to graduated extinction, bedtime fading, or a sleep education control. Sleep was measured by parent diaries and by actigraphy, so not only by parental report. Infant stress was measured through saliva cortisol sampled in the morning and afternoon. Maternal stress was self-reported. Twelve months later, attachment was assessed using the strange situation procedure, alongside child emotional and behavioural problems.

Both interventions improved infant sleep relative to control. There were no adverse cortisol responses, and no effect on attachment or on the children’s emotions and behaviour at follow-up. [gradisar-2016-rct]

That is the right set of measurements. It is also 43 infants, which is the single most important thing to hold on to about this literature.

The shape of what the sleep measures showed Illustrative
0 25 50 75 100 Night wakings per week Baseline Day 3 Week 1 Week 2 Month 3 Intervention Control

A schematic of the pattern reported in Gradisar et al. (2016), drawn to show direction and timing rather than measured values.

What happened five years later

The other useful study is longer and larger. Price and colleagues followed up a cluster-randomised trial in which 326 children whose parents reported a sleep problem at seven months received behavioural techniques over one to three consultations at eight to ten months, against usual care.

At age six, 225 families took part, around 69 percent of the original sample. There was no evidence of differences between intervention and control on children’s emotional and conduct behaviour, sleep problems, sleep habits, parent-reported or child-reported psychosocial functioning, or chronic stress. [price-2012-five-year]

Two things about that. Sixty-nine percent follow-up over five years is decent and still leaves a third unaccounted for. And no evidence of a difference is not the same statement as evidence of no difference, though with 225 families it is reasonably informative.

The half that usually goes missing

These trials were not designed to settle an internet argument. The programme they belong to was built around maternal mental health, and Hiscock and colleagues reported that improving infant sleep improved maternal depression symptoms. [hiscock-2007-maternal]

That matters for how the decision is framed. The common framing sets the child’s interests against the parent’s convenience, which assumes the two are separable. A parent who has not slept properly for eight months is not a comfort issue running alongside the child’s wellbeing; it is part of it.

What this evidence does not establish

It does not say you should. It addresses one claim, that these methods harm the child, and finds it unsupported. Deciding what to do with that is a different question and the research does not answer it.

The sample doing the heavy lifting is small. Forty-three infants can rule out a large effect on attachment and cannot rule out a modest one. Anybody citing this as settled, in either direction, is overstating it.

The age range is narrow. Six to sixteen months in one trial, eight to ten months at intervention in the other. None of it transfers to newborns, and the reasoning for a three-month-old is not the same reasoning.

The families had asked for help. Both trials recruited parents who reported a sleep problem and wanted it addressed. That is a different situation from applying a method to a baby whose parents are reluctant, and the trials say nothing about the second.

Both were Australian samples. Infant sleep practices vary enormously between cultures, and so do expectations about what a normal night looks like.

What helps

Rule out the physical causes first. Reflux, allergy, ear infection, sleep apnoea and iron deficiency all produce night waking, and no behavioural method will fix any of them.

Pick the method you can actually run for a fortnight. Consistency does more than the choice between techniques, and an approach abandoned on night three because it felt unbearable has taught the child that persistence works.

Decide in advance what you will do, while you are not exhausted. Three in the morning is the worst possible moment to make a judgement call about your own parenting.

Count your own sleep as data. If you are heading into depression, that is a reason to act rather than a reason to feel guilty about acting. Our guide to depression covers what the early signs look like.

Do not sleep train if you do not want to. Plenty of children sleep through eventually without any of this. The evidence says the methods are not harmful; it does not say the alternative is.

For what attachment actually is, and how little it depends on any single practice, our guide to attachment theory covers the research the word comes from. And our overview of parenting and mental health covers the wider question of which parenting decisions the evidence says actually move outcomes.

When to seek help

Speak to a doctor if your child’s night waking comes with snoring, pauses in breathing, pain, or poor growth, or if it starts suddenly after a period of settled sleep. Speak to a doctor about yourself if low mood, anxiety or hopelessness has lasted more than two weeks, which is common after months of broken sleep and is treatable.

If you are having thoughts of harming yourself or your baby, treat that as urgent and contact your local emergency services or a crisis helpline. Thoughts like that are more common than people are told and they are a reason to get help quickly rather than a reason to be ashamed.

How MyFreud can help

MyFreud tracks your own mood against what your nights actually looked like, which is the measurement most missing from this decision. Parents are asked to weigh their child’s wellbeing against their own and given no way to see what the second is doing. A log makes it visible.

Download MyFreud and start today: App Store or Google Play.

Frequently asked questions

Does sleep training damage attachment?

The one randomised trial that measured attachment directly found no effect. Infants were randomised to graduated extinction, bedtime fading or a sleep education control, and attachment was assessed twelve months later using the strange situation procedure, which is the standard observational method rather than a questionnaire. Infant stress was measured through saliva cortisol. Neither showed an adverse effect. The important limit is that the trial included 43 infants, so it is better evidence against a large effect than against a small one.

Is controlled crying harmful long term?

A five-year follow-up of a much larger trial found no evidence of harm. Researchers followed 326 children whose parents had reported a sleep problem at seven months, with the intervention delivered at eight to ten months, and assessed them at age six on emotional and conduct behaviour, sleep, psychosocial functioning and chronic stress. There were no differences between intervention and control. Around a third of families were lost to follow-up, which is a real limitation and a fairly ordinary one for a study of that length.

What age is sleep training appropriate?

The trials that found no harm studied infants from roughly six months, and one intervened at eight to ten months. That is the range the evidence covers and it should not be stretched. Nothing here supports applying these methods to a newborn, whose night waking is a feeding requirement rather than a habit, and where the entire framing of the question is different. If your baby is under about six months, this literature is not about you.

Which sleep training method is best?

The trial that compared two of them found both worked. Graduated extinction, where you return at increasing intervals, and bedtime fading, where you move bedtime later to match when the child is actually sleepy and then bring it forward, both improved sleep relative to a control group. Neither beat the other clearly. The practical implication is to choose the one you can carry out consistently, because inconsistency is what turns a short unpleasant process into a long one.

Is it wrong to sleep train because I am exhausted?

Parental exhaustion is a legitimate reason and treating it as selfish misreads what these trials were built to study. The research programme behind this evidence was designed around maternal mental health, and it found that improving infant sleep improved maternal depression symptoms. A parent who is not sleeping is not a separate issue from the child's wellbeing, and the framing that sets one against the other is not supported by the data.

References

  1. 1.Gradisar M, Jackson K, Spurrier NJ, Gibson J, Whitham J, Sved Williams A, Dolby R, Kennaway DJ ( 2016). Behavioral interventions for infant sleep problems: a randomized controlled trial. Pediatrics. doi:10.1542/peds.2015-1486
  2. 2.Price AMH, Wake M, Ukoumunne OC, Hiscock H ( 2012). Five-year follow-up of harms and benefits of behavioral infant sleep intervention: randomized trial. Pediatrics. doi:10.1542/peds.2011-3467
  3. 3.Hiscock H, Bayer J, Gold L, Hampton A, Ukoumunne OC, Wake M ( 2007). Improving infant sleep and maternal mental health: a cluster randomised trial. Archives of Disease in Childhood. doi:10.1136/adc.2006.099812