Sleep Training Research

Overview

The methods devised to ‘sleep train’ babies vary widely, however, they can be thought of as falling broadly into two groups:

• Methods designed to treat existing ‘sleep problems’: These include ‘extinction’ methods (more commonly known as Cry-It-Out and Controlled Crying) and are designed to prevent a baby being ‘rewarded’ with parents’ attention for crying whilst settling to sleep, or on waking during the night.
  Other methods which fall into this category include ‘Gradual Withdrawal’ (e.g. The Disappearing Chair) and ‘Camping Out’ (where a caregiver remains with the child but does not respond or interact).

• Methods designed to prevent ‘sleep problems’ from developing: These focus either on development of routines or schedules which are broadly designed to help babies differentiate between night-time as sleep time, and day-time as wake, play and feed time; or they are designed to teach the baby to fall asleep in their cot/crib (e.g. Shush and Pat) until they can eventually do so without parental support.

Reviews of the sleep training literature show that the majority of published studies (over 80%) report positive outcomes, regardless of what method is being tested, with the greatest amount of support (in terms of number and quality of studies) existing for extinction methods.

‘Positive outcomes’ means that the research studies found one or more significant differences between families who used the sleep training method and those who did not. In most cases these involve an increase in the amount of sleep mothers reported for their babies or themselves, a decrease in the number of times mothers reported that their babies woke during the night, or a reduction on the severity of maternal depression.

These studies often find no difference in the sleep behaviour of the baby who has received the training – it is the perception of the parent, or whether the baby has ceased signalling when they wake, that is the successful outcome.

Some studies report that the improvement in mothers’ or babies’ self-reported sleep was only significant for part of the group who took part. For example a large randomised trial of controlled-crying/camping out was conducted with 8-10 months old babies. The researchers found that the overall improvement seen in the number of mothers reporting infant sleep problems two months later was due to significant improvement occurring only in a sub-group of depressed mothers. There was no significant improvement in the ‘non-depressed’ group

(Randomised controlled trial of behavioural infant sleep intervention to improve infant sleep and maternal mood. Hiscock & Wake, 2002.). 

Although the vast majority of studies have reported initial increases in (maternally-reported) infant sleep duration, and/or reduction in night waking, relatively few studies have investigated whether this effect persists long-term, or whether the perception of sleep improvement is supported by objective data (e.g. via actigraphy). In most published studies the initial improvement, or benefit of using the method disappeared over time, and outcomes were the same for the group of babies who had been sleep trained as for those in the control group who had not.

Limitations of the research

Although a high percentage of published studies report ‘positive’ results, these figures alone do not tell the whole story. One of the biggest problems with evaluating this body of research is the huge variety in the way studies were conducted, and what they aimed to achieve. Some of this variation includes:

• Who takes part in the research: babies involved in the studies we reviewed for this summary ranged in age from 0-45 months, and included families with newborns (and hence no ‘sleep problems’), to families needing ‘some help’ with their baby’s sleep, to families reporting severe and ongoing sleep disruption.
• How sleep training methods are implemented, including what methods are used: not only are there a wide range of techniques, from Cry It Out to ‘positive bedtime routines’ and ‘parental education’, often more than one method is used within one study, making it hard to disentangle what works from what has no effect.

In addition to these variations, studies can be very different in terms of design:

• Some are large randomised controlled trials that often have the benefit of large sample sizes, and participants are randomly allocated in an attempt to isolate the effects of the method being used, but also tend to need to use low-cost, easily collected data (such as parental reports) due to the huge numbers of participants;
• Others are very small controlled or non-controlled trials that don’t have the benefits of large-scale RCTs but can often use costly but objective data-collection methods such as video observation or sleep monitoring devices, but because of the small sample sizes these studies cannot be generalised beyond the specific context of the study.
• Some studies attempt to alter the way babies sleep, or the way mothers sleep, or focus on other sleep-related issues such as mother or child mental and physical health, and family relationships.

For these reasons, few studies are directly comparable to others, and those that are comparable tend to be done by the same research team, using similar groups of participants. Drawing conclusions about the effects of using sleep training methods in one group of individuals based on studies done on another – quite different – group, is therefore extremely difficult.

We also need to be aware that for some participants, simply taking part in a research project can change the way they act, or think about their infants’ sleep. One study that looked at parents’ experiences of taking part in a sleep training intervention, found that being in the project made some participants feel more determined to implement aspects of the method – including controlled crying – than they would otherwise have been, thus making the method more successful in changing their babies’ sleep as part of a research trial than they would have been if they were using the method alone without taking part in a research project.

(A qualitative study of parents’ perceptions of a behavioural sleep intervention. Tse & Hall, 2007.)

In another study, parents surveyed by Canadian researchers remarked upon the difficulty they had implementing techniques within the home versus in the clinical or research setting. The reality was that many sleep training techniques were more stressful to implement than parents had initially anticipated. These parents quickly gave up and dismissed sleep training as too challenging or traumatic.

Finally, the fact that the vast majority of publications reporting on tests of sleep training methods produce ‘positive’ results raises the question of whether studies that produce negative results get published at all (known as publication bias). How many studies have examined sleep training methods that haven’t worked? We have no way of knowing.

Overall, there is good evidence that sleep training methods can improve parental mood disorders, or change (parental reports of their) babies’ sleep patterns in the short term. Evidence that sleep training actually helps babies to sleep better or longer is scarce. It is also not clear what other — unintended — consequences of such training might be. We consider these issues in the following page.