This is a special post for quick takes by Mvolz. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
How much of sleep training is placebo for the parents?
A few years ago I read this BBC Future article on sleep training. It's a well written article, balanced, a bit sceptical, but overall positive.
Recently, though I actually read one of the papers heavily referenced in the article, this 2015 RCT which claims sleep training works. After reading the paper I now think the researchers were just putting a good face on what was originally a null result.
The researchers used sleep actigraphy to tell how often the infant woke up. There was no difference in number of wakings between groups - the sleep trained infants woke up just as often. The researchers fully admit this. But then the paper rescues this null effect by saying that probably the infants are self-soothing and going back to sleep without crying despite waking up, and isn't that what we want? Who cares if the infant wakes if no one hears it?
We weren't saying that the kids wouldn't wake. We were saying that they would wake, but they wouldn't have to signal their parents – Wendy Hall
The problem with this claim is that in the study there is no objective measure of how much the children were signalling at night (crying). There also is no objective measure of how much the parents were actually waking up.
What they do have is a sleep diary filled out by the parents that shows the parents record the intervention arm infants waking less. They also claim parent sleep quality and mood improves, based on a one time survey done 6 weeks after the intervention.
The parents are doing the sleep training themselves and obviously can’t be blinded, so they know they’re in the intervention arm. This introduces a source of bias into both the infant sleep diary and the sleep quality and mood survey.
This means there are several additional possibilities that explains the data without infants crying less at all; we only know they record fewer wakings, but don't know why.
It's possible the parents themselves have been trained to wake up less themselves when their infant cries, which would still mean they're actually getting more sleep. This would be great, but it's not the same as the infants self-soothing and crying less.
The worst case scenario is simply that parents woke up as often, but simply recorded fewer wakings in the diary and reported better sleep 6 weeks later because they personally believe sleep training works- or because they wanted to please the researchers- which is common in un-blinded self-report data.
There was a big missed opportunity here to have the parents wear an actigraphy device. If you’re putting one on the child, you might as well put one on the parent. It would have been nice to see how much the parents were actually waking up. And it would have recorded if parents were waking up but not recording the infant waking in the sleep diary.
Additionally they could have set up an audio recording of the infants to see how much they were crying to see if they actually cried less or not at all during night wakings. At first you'd expect infants being sleep trained to cry more, but later they should have cried less. But we don't have that data.
The pre-registration shows the experiment was originally designed to see if the infants actually woke up less, and also to see if this changed parents' opinion about whether or not the infant had a sleep problem or not. Per the pre-registration, the results were null. There were no objective measures of parental sleep quality because the study wasn't initially designed to test this self-soothing hypothesis. When the hypothesis that infants would wake less didn't pan out, they went with the self-soothing angle after the fact, despite the fact they didn't actually test this.
The data did find the sleep trained infants slept slightly longer - 18 more minutes a night. Lead with that! But this paper simply doesn't distinguish between self-soothing and placebo effect for the parents, and I think the researchers oversold this angle.
As a parent, it is generally difficult to tell the difference between "I brought up my child to behave this way" and "my child has grown up". Both consist of you doing something, and then something happening.
To some degree the second child is a control group, because you are often more tired and pay less attention, and yet they usually turn out to be okay. However, they could have learned from the older sibling, so again you don't know.
As a parent, it is generally difficult to tell the difference between "I brought up my child to behave this way" and "my child has grown up".
As a kid, I would frequently hit my brother (and get hit by my brother). I would get punished every time, and I knew I shouldn't do it, but from the inside it felt impossible to stop—it was an uncontrollable impulse. No amount of punishment could get me to stop.
One time my parents told me, "If you hit people as an adult, you go to jail." I thought, well I guess I'm going to jail, because I can't stop myself from hitting people.
What I did not realize is that as you get older, your impulse control automatically gets better. The problem went away by itself once I was old enough.
It's possible the parents themselves have been trained to wake up less themselves when their infant cries, which would still mean they're actually getting more sleep. This would be great, but it's not the same as the infants self-soothing and crying less
Maybe sleep training isn't training the infant to sleep.