New research in the Journal of Pediatrics

[1] has claimed to look at the success of real-world implementations of sleep training, with initial reports claiming it’s successful over 80% of the time.  I know how everyone likes to run with such stories, but I think we need to take a closer look at what exactly “success” means and what we can take from the research.

Who took part?

Participants were parents who were recruited from a Facebook group that supports families doing behavioural extinction methods for infant sleep.  Only parents in this group were offered the chance to take part and there was data from over 650 families in the analyses (though some analyses had less than 500 because of missing information).  Parents completed this when their children were, on average, 12 months of age, but the infants were on average 5.6 months of age when sleep training took place.  Most families started sleep training when infants were between 3 and 5 months of age.

There are a couple issues to discuss here.  First, this is quite potentially a biased sample.  Individuals who are in this group already have a belief about extinction methods that is positive and this can lead to a positive response bias.  Further, these same individuals are seeking strong support for their chosen methods and thus have an investment in it and will likely see it through to whatever they deem a “success”.  Individuals are less likely to report any failures or to even take part if there were (these people would potentially be more likely to leave this group thus making them unavailable for the survey).  Think of it like doing a study on the effectiveness of Weight Watchers and then only recruiting from an active support group instead of a wider population.

Second, there was a lengthy time between the average implementation of the methods and the completion of the survey yet the survey data looks at timing in minutes.  This raises accuracy questions as many parents don’t actually remember how long their babies cried for.  Notably, they removed the time questions for families who completed sleep training greater than 1 year earlier over concerns.  Including only families that were currently sleep training might offer a more accurate assessment.

What kind of training was used?

Most families used extinction methods, though a small percentage used ones with parental presence (15% total).  Around 35% used cry-it-out and almost 50% used controlled crying.  The majority of parents used these methods for both bedtimes and awakenings.  Interestingly, there seem to be other changes that took place too, such as implementing a regular bedtime routine after sleep training.  It’s unclear the effect of this on any sleep training “success”.

Importantly, for all of these methods, parents reported getting support from their medical doctors, sleep trainers, and then family.  This is important as it informs on who is providing this type of information.  It is worth mentioning here too that medical doctors are not trained on infant sleep nor are they trained on elements to do with long-term emotional development.

What did they report/find?

It seems that most families found their methods successful and they were quite happy with them despite reporting high stress on day 1.  The average crying bout was 44 minutes on day one then reduced in subsequent days.  Parents reported “success” after an average of 12.5 days of these methods and close to 90% of parents surveyed say they were pleased with the results.  There is no actual definition of “success” except parents saying they were no longer doing it, but this may or may not be a matter of crying at night or a certain level of crying.

I want to mention the break down the “success” time too.  Thankfully, the authors didn’t just provide a mean for days, but also the percentage of families who took 1, 2, 3, or 4 or more weeks to find success.  While 55% said it took one week for “success”, 24% said it was within 2 weeks, 7% said 3 weeks, and 14% said 4 or more weeks.  Despite these numbers, close to 90% of families were “satisfied” with their experience.  This may speak to their experience but it may also speak to the types of families that were recruited (see the bit on the sample).

There was also data on feedings and wakings.  Obviously parents reported fewer wakings (though this doesn’t mean there were actually fewer, to be discussed below), but they also reported fewer feedings, going from an average of between 3-4 to just over 1 per night.  Given the average age of 3-5 months for the training, this is quite bothersome.  Babies are designed to eat frequently and this is actually important and even more important if you are breastfeeding.  Sadly many doctors have no training or information on infant feeding outside of global numbers of calories and amounts and this doesn’t jive with the massive amount of research by those who study lactation and infant feeding.

What isn’t included in the results or survey data?

  1. Data on actual infant sleep. Although, like in previous studies, parents reported their infant sleep had improved, we also know that when objective measures of sleep are included, there does not seem to be actual improvement (see here for a summary).
  2. Data on any child outcomes. There is nothing on attachment or any social-emotional infant or toddler outcome.
  3. Data on infant temperament or special needs. Again, we have nothing to compare children to who were successful in a shorter or longer period of time.
  4. Later outcomes of sleep. Parents did report if they did it again if they weren’t successful the first time, but it seems there was no option for parents to say it was successful and then have to do it again, something that was addressed in another survey on the same topic (see a review of that here).
  5. Maternal support (or lack thereof) or mental health. These are factors that influence the perceived need to engage in sleep training and can account for views of success.
  6. Infant sleep expectations or Western cultural beliefs. New research has shown us that the engagement of extinction sleep training is better predicted by a Western mindset not by stress or need[2].  I imagine, though can’t say for sure, that the group was one that held this mindset in high regard, but it would be helpful to know that for certain.

I find these to be critical missing factors if we are going to look at sleep training in depth and try to make actual conclusions about the practice.

Overall conclusions?

Overall, I find this to be informative in that we can see the “best case” scenario for extinction methods given the group sampled and the type of information gathered.  I still personally don’t find it convincing as anything we should be promoting and is very clearly lacking lots of relevant information, including alternatives for families.  I worry about the effects on breastfeeding mothers as well.  I’m not quite sure what this adds to the body of research, but I suppose something will come to mind one day.


[1] Honaker SM, Schwichtenberg AJ, Kreps TA, Mindell JA.  Real-world implementation of infant behavioral sleep interventions: results of a parental survey.  Journal of Pediatrics in press.

[2] Maute M, Perren S.  Ignoring children’s bedtime crying: the power of Western-oriented beliefs.  Infant Mental Health Journal 2018; 39: 220-30.