Preface: This blog is dedicated to all the mums caught in a confusing tug-of-war between opposing baby sleep experts.

Several years ago there was a hypothesis that letting a baby cry themselves to sleep caused them harm.

The way it worked was that the baby was so stressed by such an event that elevated levels of the stress hormone - cortisol - were released in the baby’s body.

A well known baby doctor in the USA - Dr Sears - even once said that these elevated levels “attack the baby’s brain”.

This led to damage to the baby’s emotional and behavioural regulation.

And ultimately - it led to the child having attachment issues with their parent.

If there was ever a way to convince a parent to not let their baby cry themselves to sleep - this was it.

But was it True?

One clear way to know if a hypothesis is correct or not is to test it by gathering evidence.

But as you will see in this blog, gathering evidence needs to be as complete as possible.

So let’s first look at the evidence to support this hypothesis.

A hypothesis that we will call the Sleep Training Hypothesis.

Evidence For Harm

It comes primarily from 2 situations:

1. deprived living conditions of orphaned children in Romania

2. rat puppies separated from their mothers.

In these situations, there is chronic stress. And by ‘chronic’ we mean sustained stress to the ‘child’. And these children/puppies have shown elevated levels of the stress hormone cortisol.

Thus proponents of the ‘For’ hypothesis have extrapolated evidence from these situations to a parent leaving their child to cry themselves to sleep at night.

This is what you would call indirect evidence.

And then came along a pivotal study …

Middlemiss and Friends

In 2012, a study was published in the scientific journal - Early Human Development - which directly tested the levels of cortisol in babies - as well as their mums - who enrolled into a 3-day sleep program.

In this study, cortisol was measured on Nights 1 and 3 of the program. And the measurements on these nights were (i) measured before the babies were separated from their mums at the beginning of the night - and - (ii) when the children were separated from their mums and given the opportunity to fall asleep by themselves.

Middlemiss and Friends found that on Night 1 - before the mums and babies were separated - that their cortisol levels were synchronised. Later in the sleep program - when the babies were separated from their mums on Night 3 - the researchers noticed that the synchrony between the babies’ and mums’ cortisol levels were diminished.

Hence the title of their study ….

When this study was published it made it greatest waves amongst parenting groups all over the internet than it did in the scientific world.

Various advocates -against the notion of sleep training a baby - were elated.

Indeed, that was how I found out about the Middlemiss and Friends study.

A well known baby expert in Australia - Pinky McKay - let all of her newsletter subscribers know about this landmark study. And someone sent that email to me.

See if you can spot the part when the description of the study turns into what’s called ‘Research Translation’ (ie, what a scientific study’s findings mean in the real world) …

As I write this blog, Pinky McKay has over 32,000 followers on Instagram - and who knows how many email subscribers since this infamous newsletter was sent 9 years ago.

Pink McKay put her scientific interpretation onto a study … that applied their own scientific interpretation onto their own data. And as you will see by the end of this blog - that is problematic (to say the least).

But to return to the point, Middlemiss and Friends directly tested a part of the ‘For’ hypothesis. Specifically, measuring cortisol levels in babies when they are left to cry themselves to sleep.

Before we move onto the ‘Evidence Against’ side - let’s take a …

Quick Lesson in Science

The second person to alert me to the Middlemiss and Friends study was Professor Harriet Hiscock. Harriet has published some of the leading science when it comes to treating sleep problems in infants.

To cut a long story short - Harriet invited me to co-write a short paper analysing the Middlemiss and Friends study …

In the very same journal that the Middlemiss and Friends study was published, we highlighted a few important points - but one of them was focused on the interpretation of the cortisol data:

In 2012, the world did not have any data for whether cortisol levels were high or not.

To determine this, it would require the measurement of cortisol from a lot of babies.

And that’s what happened in 2015.

For reference, the Middlemiss and Friends study used a sample of babies aged between 4 to 10 months of age, and cortisol levels ranged between 0.41 to 0.58 micrograms per decilitre. When converted to nanomoles per litre (nM/L) - the Middlemiss values range from 11.3 to 16.0 nM/L.

Compare these values on the graph of infants’ cortisol levels that are considered normal:

Looking at the evening cortisol values (dark grey) from 4 to 10 months, the tops of the dark grey boxes are at 5 nM/L. The line coming out of the top of these dark grey boxes denote the maximum value, and these bounce around, ranging from about 9 nM/L to 26 nM/L.

Nevertheless, the cortisol levels in the Middlemiss study are all higher than normal (above the 75th percentile).

But we shouldn’t stop there.

Before the babies undertook sleep training - before they even knew what they were in store for - their cortisol levels were 12.4 nM/L.

And Middlemiss and Friends’ analysis shows that the babies’ cortisol levels did not significantly change during the sleep program.

Said another way, the babies’ cortisol levels were high before they started the sleep program, and their cortisol levels did not change as a result of the sleep program.

So by critically analysing the cortisol data from the Middlemiss study, we can conclude that the first part of the hypothesis was not confirmed - cortisol levels did not increase because baby’s were left to cry themselves to sleep.

Thus far, the evidence to confirm the hypothesis is tenuous - with most of it indirect, and all of it incomplete.

So let’s not beat around the bush. Let’s see a direct test!

Evidence Against Harm

In 2016 we tested babies’ chronic cortisol levels before and 1 week after undertaking ‘sleep training’.

Side note here, the Middlemiss study involved ‘cry-it-out’ (not responding to the infant’s cry all night). Our study allowed parents to check on their infants after a specific amount of time - known colloquially as sleep training.

We’ve written about this study in various ways - and you can download a copy of it here.

Suffice to say, we:

• witnessed no significant increases in the babies’ cortisol levels from before sleep training (6 nM/L) to 1 week after sleep training (4 nM/L). Yep, that’s right. The cortisol levels were in the normal range.

• 1 year later, we saw no significant difference in emotion and behavioural regulation of infants who were sleep trained and those in the control group.

• And we saw no significant difference in the attachment of infants to their mums if they were sleep trained or not.

This is direct evidence.

However, it is only 1 study.

We noted in our 2016 study that a previous study followed up infants 5 years after they were sleep trained. Even though this study did not measure cortisol levels, like us, they did not find any issues with emotion and behavioural regulation, nor attachment.

Seeing multiple studies discovering the same finding is the convergence of evidence. Convergence is a strong scientific principle.

The Conclusions?

Sleep training does not elevate a baby’s chronic levels of cortisol. The first step that is hypothesised to cause harm.

I’m going to go out on a limb here and state that in our study, sleep training resulted in a 33% decrease in babies’ chronic levels of cortisol.

When you ‘hear’ a scientific finding, find the primary source (ie, original study) and critically evaluate it.

We can have more confidence in whether a study finding is ‘true’ when multiple studies find the same thing.

32K followers on Instagram does not immediately qualify someone to make correct scientific interpretations. It just means they’re popular on Instagram.

If you believe - not feel - that the content in this blog is scientifically sound - share it - and tell your friends and colleagues to share it. If this keeps happening, 32K people will eventually read it.

Not because it’s popular.

But because 32K people believe in the science behind a baby’s cortisol levels - when they are provided the opportunity to learn how to independently fall asleep.