I recently posted information on crying-it-out and received some interesting comments, but one was a particularly interesting question that I felt warranted a fuller explanation for everyone.  You see, the person questioned the conclusions from the Middlemiss study which found high cortisol activity in children undergoing a sleep training program in New Zealand.  The commenter rightly points out that the research itself does not show a cortisol spike pre- to post-bedtime and so instead believes the high cortisol found in the infants on both days one and three is due to general all-around stress and that it might be very different for a child in his or her own home undergoing CIO.

This is where I need to explain.  We all know the limitations of science – the best study is still hardly conclusive.  After all, there hasn’t been (and never will be) a study that actually shows that smoking causes cancer, despite the general knowledge that it, in fact, does.  That’s not to say this one study provides anything near the level of evidence built up in the smoking case, but it does highlight that when examining research, we make logical deductions in order to analyze the results we’re presented with.  I do this each time I write about a study, though I don’t usually include this thought process because a) it’s long, and b) I imagine it’s generally boring.  However, for the sake of clarity, honesty, and an example, I have decided to share this process for anyone interested.  As with anything I write, you are free to disagree with the conclusions I’ve reached, but you will at least realize the process I go through as I read and review the literature cited in my posts.

The study:  The cortisol levels reported in the Middlemiss study were obtained at the beginning of the bedtime routine and 20 minutes after the child has fallen asleep.  On both days 1 and 3, the levels were elevated, but did not significantly increase pre- to post-sleep time.  As the commentator pointed out, someone could say that this is evidence that CIO does not cause an increase in cortisol.

Facts agreed upon:  Both the commentator and I agree that the levels of cortisol displayed by the infants is high.  There is no debate there.  As a reference point, the levels obtained in this study are higher than those obtained in other studies examining infant’s cortisol reactions to a known high-stressor[1][2].  This is echoed by the conclusions I cited from the researchers themselves that the cortisol levels displayed were indicative of “high levels of physiological distress”[3].

The problem:  There are two possibilities accounting for the cortisol data obtained:

(A) That the high-cortisol levels are associated with the sleep routine and the CIO methodology.

(B) That the infants are generally stressed and this high level of cortisol is indicative of all around stress, presumably due to being in a different environment from home (i.e., the sleep training centre).

I landed on the side of (A) which I will explain, but I will first explain why I did not land on (B).

The logical problem of (B):  First, we have the problem of the actual cortisol levels being as high as they are at the end of the day.  The natural circadian rhythm has individuals (children too) showing their highest levels earlier in the day with levels naturally declining over time (the rhythm is diurnal).  While this has been occasionally found to not be the case in certain individuals or infants, for it to be the case for all infants in the program is highly unlikely (for example, in one study using 11 infants, only 2 were found to have higher cortisol levels at night than morning[4]).  And if these results were being driven by four or five extremely high cases with the rest showing low cortisol activation, the researchers would be remiss and unethical not to mention that.  I don’t believe that to be the case.

The other reason to see consistently high cortisol levels is when there is chronic stress.  Day one of a program would not be enough to elicit this consistent level of stress suggested, particularly as the daytime portion of the program solely involves activities with mom.  As stated in the paper, “During infants’ awake times, mothers and infants could spend time in the shared lounge with other program participants, in the mothers’ room, or could run errands or take walks with their infants.”  Therefore, despite the new surroundings, the infants were with their mothers, and as Megan Gunner’s research has shown, parents serve a very important role in attenuating the cortisol stress response[5].  And given that the environment was new, it should not elicit a chronic stress response, but rather an increase which slowly decreases with time, the typical stressor response.  A chronic stress response would suggest that there are parenting issues at hand causing all children to have insecure, neglectful, or abusive parents—the primary reasons to see chronic stress in an infant[6].  This would mean that the stress response was a continuation of chronic stress in the home environment, and while there were reported sleep problems in these families, there is no reason to believe that there was child maltreatment in the form of neglect or abuse strong enough to elicit this type of chronic stress response, particularly as the mothers were not referred for any psychiatric help in addition to sleep training (and in fact, one of the requirements of the study was that mothers were not taking medication for post-partum depression).  One last reason I discount the chronic stress hypothesis is that while chronic stress initially results in the higher cortisol output later in the day, as time passes, chronic stress seems to change its effect on cortisol and results in a lowering of baseline cortisol[7], something that was not demonstrated in the current study if the stress had been ongoing for a period of time at home.

Now, some might be thinking that the new environment should elicit some stress response on the first day, but that then ignores the same high levels found on day three, when the new environment is not so new (children do adapt quickly, especially with a parent present).  So the high levels on both days seem to work to counter the idea that it’s the environment per se that’s causing a stress reaction.  But are there stressors that cause increases in evening cortisol?  Greg Miller and colleagues examined the type of stress by cortisol response[7] and found that afternoon/evening samples with an increase were linked to the following types of consistent stressors: physical threat, social threat, traumatic stressor, uncontrollable stressor, loss, and shame.  I don’t believe that these infants were faced with any of these types of stress on an ongoing basis or even on a regular basis when starting in the program so on top of discounting chronic stress, even the types of stressors known to elicit evening peaks, none of them seem relevant to the entire group of infants on hand.  Daytimes of spending time with mom, going for walks, and running errands or playing with other moms and kids is hardly under the category of these forms of stress, even if it does take place in a new environment.

Finally, there is research on the cortisol effects of beginning a new school year on school-aged children[8] and on starting daycare for infants and toddlers[9][10].  The school year research shows that the stress of the new school year leads to a greater decrease by the end of the day.  That is, the slope from morning until evening is steeper than when the child has adapted to the new school year.  Of course, this is research with older children so it’s questionable as to how pertinent it is, but it is there and to me, serves as just that added extra, not as a study that helped shape the view against (B).  The daycare research on infants and toddlers does find an increase in cortisol across the day in daycare (though in less than half of infants[10]), but – and this is a big ‘but’ – being in the new environment of daycare during an adaptation period with mom did not result in an increase in cortisol for the infants who were securely attached[9] (and remember that synchrony is a sign of secure attachment so it’s fair to assume these infants were at least relatively securely attached).  Mom’s presence in the new environment worked to eliminate the expected rise in cortisol from an infant being placed in a new environment.

Why I sided with (A):  It wasn’t just the lack of logical reasoning with (B) as to why I chose (A), there is the additional fact of mom’s cortisol levels.  Pre-sleep routine mom also showed signs of stress, as demonstrated by heightened cortisol levels.  Why would this be the case?  I do not believe it is because of the daytime environment, and because of the drop in cortisol for moms on day three (post-bedtime), it is clear mom’s cortisol can be reduced.  I believe it’s because she knew the routine was coming.  Now, does this mean baby knew the routine was coming?  No.  But it means there were cues to the nighttime routine beginning which the infant may have picked up on (they’ve been known to pick up on a lot), but it also means that mom was giving off cues signaling distress.  Remember that synchrony works both ways – mom picks up on infant emotions and vice versa.  A calm mom can elicit calmness in her infant.  A stressed mother can elicit stress and so a mother that is in sync with her child and starting to dread the CIO routine that is to come may very well elicit the high cortisol response in her infant.  The notion that because the saliva cortisol sample suggests a stress response 20 minutes prior to the initiation of the bedtime routine does not eliminate the notion that it was the impending bedtime routine that elicited the stress.  It simply requires that the infants either picked up on a cue to the impending bedtime routine (which is not unheard of) or that they picked up on mom’s stress about the impending bedtime routine.  Importantly though, the stress of CIO remained because if CIO did not cause stress, the levels of cortisol would have dropped during the bedtime routine and sleep period, which they did not.  By day 3, mom’s cortisol had dropped, but the infant would now be expecting the CIO routine – remember infants pick up on routines very easily and well – and thus the high cortisol pre- and post- on day 3 is, to me, indicative of the CIO routine.

Other conclusions:  I must add though that this debate has nothing to do with what I perceive to be the most important conclusion gleaned from this research—that infants who are not crying are still experiencing distress.  Parents typically speak of CIO as being a few days of hell but then everything is fine.  What the Middlemiss study showed us, though, is that that may not always be the case.  Opponents of CIO techniques typically state that CIO teaches infants that someone will not always be there to help them when they are distressed, and here is research supporting that.  There are infants who very clearly and quickly learned that their cries were not going to be responded to in any manner and so stopped crying after only two nights.  And yet their distress levels remained very high.  So they didn’t learn to “self-soothe”, they learned to keep quiet and preserve energy (as crying results in a large energy expenditure relative to other activities[11]).  If no one is coming to help you, you simply waste energy by calling out.  It doesn’t make you less scared, alone, or distressed.

There will always be the question of how long this distress continues on for.  No one expects this rise in cortisol to continue forever, but the fact that it happens at all tells me that we’re not doing our infants any service by utilizing CIO techniques.  Why?  Because it seems clear that the first lesson they learned is that they can’t trust that mom will be there to help.  Whatever else comes next, that’s a pretty shitty lesson to learn so early in life.  The fact that one of the “positive” side effects of CIO, as reported by the Sleep review[12] is fewer bouts of crying at all times suggests that these infants are internalizing and extending this message.  Further support for this comes from the same daycare research[10] as infants who were insecurely attached demonstrated consistently high cortisol levels during both the time when mom was there and after she left, despite showing low levels of crying and fussing behaviour.  As to the question of how long the distress continues for, again the daycare research may provide some hint.  Securely attached infants, when separated from their mothers, demonstrated the rise in cortisol that the insecurely attached infants showed during the adaptation period.  These levels did not drop for nine days (the longest they measured cortisol before a 5-month follow up).  Fussing and crying behaviour decreased over the nine days (though for securely attached infants, it was still higher than when mom was there), but cortisol levels remained equally high from day one through day nine.  It’s also worth noting that these infants did have someone to respond to their distress at daycare – they were not left alone – and yet they continued to show physiological distress at the separation from their mother.

I know some people will make different conclusions based on the studies I’ve discussed above and the possibilities surrounding them.  That’s fine.  But I was asked why I stated this (with a hint that I was being dishonest in my reporting) and I have always strived to be transparent so this is a long-winded way of being transparent.  Hopefully this helps provide information for those who want more and further explains why I feel that CIO is not the way to go.


[1] White BP, Gunnar MR, Larson MC, Donzella B, Barr RG.  Behavioral and physiological responsivity, sleep, and patterns of daily cortisol production in infants with and without colic.  Child Development 2000; 71: 862-877.

[2] Goldberg S, Levitan R, Leung E, Masellis M, Basile VS, Nemeroff CB, Atkinson L.  Cortison concentrations in 12- to 18-month-old infants: stability over time, location, and stressor.  Biological Psychiatry 2003; 54: 719-726.

[3] Middlemiss W, Granger DA, Goldberg WA, Nathans L.  Asynchrony of mother-infant hypothalamic-pituitary-adrenal axis activity following extinction of infant crying responses induced during the transition to sleep.  Early Human Development in press.

[4] Kiess W, Meidert A, Dressendorfer RA, Schriever K, Kessler U, Konig A, Schwarz HP, Strasburger CJ.  Salivary cortisol levels throughout childhood and adolescence: relation with age, pubertal stage, and weight.  Pediatric Research 1995; 37: 502-506.

[5] Gunnar, M. R. (2006). Social regulation of stress in early childhood. In K. McCartney & D. Phillips (Eds.), Blackwell Handbook of Early Childhood Development (pp. 106-125). Malden: Blackwell Publishing.

[6] Cichetti D, Rogosh FA, Gunnar MR, Toth SL.  The differential impacts of early physical and sexual abuse and internalizing problems on daytime cortisol rhythm in school-aged children.  Child Development 2010; 81: 252-269.

[7] Miller GE, Chen E, Zhou ES.  If it goes up, must it come down?  Chronic stress and the hypothalamic-pituitary-adrenocortical axis in humans.  Psychological Bulletin 2007; 133: 25-45.

[8] Bruce J, Davis EP, Gunnar MR.  Individual differences in children’s cortisol response to the beginning of a new school year.  Psychoeuroendocrinology 2002; 27: 635-650.

[9] Ahnert L, Gunnar MR, Lamb ME, Barthel M.  Transition to child care: associations with infant-mother attachment, infant negative emotion, and cortisol elevations.  Child Development 2004; 75: 639-650.

[10] Watamura SE, Donzella B, Alwin J, Gunnar MR.  Morning-to-afternoon increases in cortisol concentrations for infants and toddlers at child care: age differences and behavioral correlates.  Child Development 2003; 74: 1006-1020.

[11] Thureen PJ, Phillips RE, Baron KA, DeMarie MP, Hay Jr WW.  Direct measurement of the energy expenditure of physical activity in preterm infants.  Journal of Applied Physiology 1998; 85: 223-230.

[12] Mindell JA, Kuhn B, Lewin DS, Meltzer LJ, Sadeh A.  Behavioral treatment of bedtime problems and night wakings in infants and young children.  Sleep 2006; 29: 1263-1276.