The Sleep Training Cortisol Study: What the Middlemiss Research Actually Showed

If you’ve spent any time researching sleep training online, chances are you’ve come across the infamous “cortisol study” often shared as proof that babies who stop crying during sleep training are still experiencing hidden stress or emotional shutdown.

The Middlemiss Cortisol Sleep Training Study:

What the Research Actually Showed — And What It Didn’t

The study most commonly referenced is the 2012 Middlemiss Cortisol Study study.

Online it is frequently presented as definitive evidence that sleep training harms babies emotionally.

But there’s an important problem...

Many of the conclusions circulating online go far beyond what the study itself was actually designed to prove.

And perhaps most importantly, the study had several major limitations that are rarely discussed.

A major limitation:

No baseline cortisol measurements.

One of the biggest methodological concerns is that researchers did not establish any pre-intervention baseline cortisol level before the sleep intervention began.

This matters enormously.

Cortisol levels in infants can already be elevated for many reasons including chronic overtiredness, fragmented sleep, illness, developmental changes, environmental disruption, schedule instability and parental stress within the home.

Without understanding what the infant’s cortisol levels looked like before intervention, we cannot confidently conclude that elevated cortisol was caused specifically by the sleep training process itself.

Chronic sleep deprivation itself elevates cortisol.

This is another critically important factor often missing from online discussions.

Research consistently shows that chronic sleep deprivation increases cortisol and physiological stress responses in both adults and children.

This is also important biologically. When babies (and adults) stay awake beyond a reasonable window, their brains begin releasing excess cortisol and adrenaline to help the body push through fatigue and remain awake. Many babies and their Mothers entering sleep training are already chronically overtired from ongoing fragmented sleep and prolonged wakefulness, meaning cortisol levels are already elevated before sleep support even begins.

So an important question becomes:

Were cortisol levels reflecting temporary adjustment during behavioural change or were they reflecting pre-existing chronic sleep deprivation and overtiredness?

The study cannot fully answer that question.

The study lacked broader environmental controls: And was not a controlled clinical study

Another major limitation is that the study did not fully account for other contributing variables during the intervention process.

For example we do not know:

• whether schedules were adjusted

• whether babies became temporarily more overtired during transition

• whether feeding routines changed

• whether the sleep environment was altered

• whether parental anxiety increased throughout the process

• how much crying occurred overall

• or what level of responsiveness was used

  
  

These factors matter because cortisol is highly sensitive to many forms of physiological and emotional arousal, not simply crying alone.

Small sample size and no long-term follow-up

The study itself was also very small, with limited data. Although 25 mother–infant pairs initially enrolled, only 7 dyads ultimately provided complete cortisol data used within the final analysis.

Importantly, it did not assess attachment security, measure trauma, evaluate long-term emotional outcomes or follow children into later childhood development.

This means the study cannot be used to conclude that sleep training causes attachment damage or long-term psychological harm.

It simply was not designed to answer those questions.

So what did the study actually find?

The study observed a small group of infants participating in a behavioural sleep intervention within a hospital-based sleep program.

Researchers measured cortisol levels in both mothers and babies across several nights.

One of the most quoted findings was that as the nights progressed, some babies cried less while their cortisol levels remained elevated compared to their Mothers’.

This has led to the now-viral claim that:

“The baby stopped signaling distress because they learned nobody was coming.”

Which by anyone's standard sounds deeply upsetting.

But again, this interpretation extends far beyond what the study itself actually demonstrated.

The study measured short-term physiological arousal during a period of behavioural change.

It did not prove emotional shutdown, emotional abandonment, trauma or attachment damage. Were the Mother's responding to their babies during the intervention, or not at all? We don't know.

Crying and cortisol are not automatically signs of trauma

This is where online conversations often become highly emotionally charged.

Cortisol is not inherently harmful.

It is a normal adaptive hormone involved in stress regulation, metabolism, immune function, wakefulness and learning responses. In the last few hours of the night for example, our bodies will gradually and continuously elevate cortisol levels, in order to wake us from sleep. First thing in the morning, our cortisol levels will typically be their highest in a 24 hour period.

Which poses another important question: Was the cortisol of the Mother's and their infants measured first thing in the morning? Was it measured at the same time each day over the three days, or at a different time each day?

We don't know. These vital elements were not documented or controlled within this study.

Temporary increases in cortisol occur during many normal childhood experiences, including starting daycare, vaccinations, learning new skills, frustration, transitions and disrupted routine (and is there anything more disruptive than bringing your baby to sleep in a hospital lab filled with strangers...?)

A temporary rise in cortisol during change does not automatically equal toxic stress trauma or emotional harm.

Those are very different physiological and psychological concepts.

Another important consideration is that cortisol samples were collected by parents themselves, within the home environment, using salivary swabs, rather than within a tightly controlled laboratory setting. While home cortisol sampling is commonly used in research, it also introduces substantially more potential variability, including differences in collection timing, feeding status, crying before sampling, sleep quality, environmental stimulation, and parental stress levels. In a study with an already extremely small final sample size, this level of uncontrolled variability makes it difficult to draw sweeping conclusions about infant emotional states, attachment, or long-term psychological outcomes.

What do long-term sleep training studies show?

This is why looking at the full body of evidence matters.

Long-term clinical studies on behavioural sleep interventions have repeatedly found:

• no evidence of attachment damage

• no evidence of adverse emotional outcomes

• no behavioural differences later in childhood

• and significant improvements in parental mental health and family wellbeing

  
  

One of the best-known follow-up studies published in Pediatrics followed children for five years after behavioural sleep interventions and found no evidence of harm to emotional health attachment or parent-child relationships.

The bigger picture parents deserve

Parents are absolutely right to ask thoughtful questions about research.

But single studies, particularly small studies with significant limitations, should never be used in isolation to make sweeping claims about parenting or child development.

The Middlemiss study raised interesting questions about infant stress physiology during behavioural change.

What it did not prove was that responsive evidence-based sleep training causes emotional shutdown trauma or attachment damage.

Healthy attachment is built across thousands of daily interactions filled with comfort, responsiveness, affection, co-regulation, feeding, temperature regulation and emotional safety. Supporting a child in learning independent sleep skills within a loving, connected, responsive home is not the same thing as emotional neglect.

Parents deserve nuance context and accurate interpretation of the science not fear-based headlines.

References

Middlemiss, W. et al. (2012). Asynchrony of mother–infant hypothalamic–pituitary–adrenal axis activity following extinction of infant crying responses induced during the transition to sleep.

Hiscock, H. et al. (2012). Five-year follow-up of harms and benefits of behavioral infant sleep intervention. Pediatrics.

Mindell, J.A. et al. (2006). Behavioral treatment of bedtime problems and night wakings in infants and young children. American Academy of Sleep Medicine.

Ready to discover attachment led sleep support?