Pregnancy and child

Leaving babies to cry 'will improve their sleep', study says

"Babies do sleep better if you leave them to cry," the Daily Mail reports.

A small study suggests that "graduated extinction" – better known as controlled crying in this country – increased sleep length and reduced the number of times babies woke up during the night.

Controlled crying involves waiting a set number of minutes while your baby is crying, without picking them up, to see if they drop off again.

The study compared this approach with a standard sleep education approach based on the principle of setting a standard bedtime routine, as well as a different approach known as bedtime fading.

This involves pushing back your baby's bedtime by 30 minutes if they took a while to settle the previous night.

The results suggest these two approaches work better than a sleep education only control group approach.

This didn't lead to any increases in stress to the infant or affect parental-child bonds a year later.

A problem with the study is its size – there were only 14 to 15 infants in each of the three test conditions at the start of the study.

There were even less after three months, when most of the results were analysed – only seven in each group. This isn't enough to make reliable statements about which sleep method works best.

There may not be a one-size-fits-all "trick" to getting your baby to sleep. Some babies may respond to controlled crying, others may prefer bedtime fading or a set bedtime routine.

Where did the story come from?

The study was led by researchers from Flinders University, Australia, and was funded by the Australian Rotary Health Fund, Channel 7 Children's Research Fund, and the Faculty of Social and Behavioral Sciences.

It was published in the peer-reviewed journal, Pediatrics.

The Mail's reporting was accurate, but took the findings at face value, not discussing any of the study's limitations, such as its small size, and how these could affect the findings.

What kind of research was this?

This randomised control trial (RCT) looked at two approaches to improving an infant's disturbed sleep, compared with a standard control intervention.

Many parents experience trouble getting their infant to have a good night's sleep. Struggles can include settling your infant before bed, helping them fall asleep, or frequent waking in the night.

There are a lot of approaches people suggest to help. The researchers wanted to find which one worked the best:

  • Should you comfort your child each time they cry, or show "tough love" and leave them to cry and sooth themselves?
  • Should you pick up your infant to comfort them, or is it best to only show your face but leave them where they are?
  • Is setting a standard bedtime better, or does it make more sense to be flexible, depending on how tired your baby seems to be?

These questions can leave parents confused, and sometimes feeling guilty about what's best – and they aren't the only ones.

Researchers couldn't find any clear answers from past studies they'd seen, either. They designed this trial to test two behavioural approaches against an educational-only approach to improve infants' disturbed sleep, hoping there would be a clear winner.

What did the research involve?

All families in the study answered yes to the question "Do you think your child has a sleep problem?", so they were a special group of disturbed sleepers.

Infants with mothers with significant postnatal depression scores were excluded. Most parents were graduates and middle- to high-income earners.

A total of 43 infants aged 6 to 16 months – mostly (63%) girls – were randomised to one of three sleep test groups:

  • graduated extinction (14 infants) – gradually delaying parents' responses to their infant's cry each night and each time they wake in the night. Parents were told to put their infant to bed awake, and leave within one minute. When re-entering the room, they were allowed to comfort their child, but couldn't pick them up or turn on the lights.
  • bedtime fading (15 infants) – delaying the infant's bedtime by 30 minutes each time they took more than 15 minutes to fall asleep.
  • sleep education (14 infants) – this was the control group. Parents were given information on reasons for night wakings, settling tips, and sleep cycles in infants. The graduated extinction and bedtime fading groups also received this information.

Parents filled in sleep diaries to document their infant's sleep habits, wore ankle tags to track their night-time movements, and filled in ratings scales assessing the mother's level of depression, mood and stress.

Infants' stress levels were also monitored in the morning and afternoon, testing their saliva for the stress hormone cortisol.

Parental-reported changes in sleep patterns were obtained before the test and one week, one month, three months and one year into the test to monitor change.

A year after the tests, mothers rated their children for emotional or behavioural problems, and a series of separation and reunion tests assessed parent-child attachment.

All mothers and infants who started the test finished it through to a year, but there was data missing for approximately half (seven) of the families by the third month.

The main analysis compared the two active tests – graduated extinction and bedtime fading – with the control group, sleep education given to all, and for changes over time.

The focus was on any changes in the time it took for the infant to fall asleep (sleep latency), how often they woke in the night, and whether they woke up after falling asleep.

What were the basic results?

Three months into the intervention, a lot of sleep measures had improved across all three groups.

However, it wasn't clear whether they were statistically different across the three test conditions, or before and after the study, as they were presented as graphs.

After three months:

  • The time it took infants to go to sleep had fallen from around 18 minutes to less than 10 minutes in both graduated extinction (-12.7 minutes) and bedtime fading groups (-10 minutes). Stayed more or less the same in the control at around 20 minutes (+2 minutes).
  • Average number of times the infant woke in the night appeared to decline in all groups, but it wasn't clear if these were statistically significant compared with the education only group, or over time.
  • The time spent awake after first falling asleep fell across all groups. For graduated extinction, it fell from just under an hour at the start of the study to around 15 minutes (44.4 minutes). The control group and bedtime fading improved a little less, by 31.7 minutes and 24.6 minutes respectively.
  • Total time asleep improved for those trying graduated extinction  (+19.2 minutes) and the control group (+21.6 minutes) but there was little change for bedtime fading (+5.4 minutes).

Over the first month, maternal stress in the control group was largely unchanged, but reduced in both sleep test conditions. Maternal mood improved in all groups, most of all for bedtime fading.

At one year no effects on parent-child bonding or emotional or behavioural problems were found.

How did the researchers interpret the results?

The researchers concluded that their study showed "meaningful effects for both graduated extinction and bedtime fading".

They went on to say that, "Compared with the control group, large reductions in nocturnal wakefulness resulted from each treatment.

"Despite assertions that extinction-based methods may result in elevated cortisol, emotional and behavioural problems, and insecure parent-infant attachment; our data did not support this hypothesis."


This randomised control trial suggests two behavioural approaches to remedy disturbed sleep in infants may work better than a sleep education only control group approach.

This may be true, but may also be a chance finding or affected by bias. For example, the statistical significance of some of the results was hard to interpret, as many were presented as graphs only. This means we can't be sure that some, or even many, of the differences are down to chance.

The study was also very small, with only 14 to 15 people in each of the three test conditions at the start of the study.

There were even less after three months – only seven in each group. This isn't enough to make accurate, reliable or generalisable statements about which method works best.

Small studies like this are also more likely to throw out unusual and unrepresentative results. For these reasons, we can't say anything too solid based on it.

You may wish to experiment with different techniques to see if a specific approach suits your baby better.

If you have persistent problems getting your baby to sleep and it is beginning to have a significant impact on your quality of life and ability to function during the day, speak to your health visitor or GP.

Read more advice about helping your baby (and you) get a good night's sleep.

NHS Attribution