Are babies “stressed out” from sleep training? Part Two

More on sleep training, stress, and Gradisar et al. (2016)

Natural Parenting. (Image via xkcd)

In Part One of this series, I talked about a recent New Yorker article on sleep training that discussed two studies that examined cortisol and stress in infants. I already summarized Middlemiss et al. (2011), a study that examined cortisol in mothers and infants in a residential sleep training program. For Part Two, I’m going to discuss the second study from the New Yorker piece and summarize what these two studies together are able to tell us about stress and sleep training. 

Gradisar et al. (2016)¹

Unlike the Middlemiss article, Gradisar and colleagues were directly invested in whether or not sleep training has long-term consequences for babies and their parents. Specifically, they were interested in whether two types of sleep training methods (graduated extinction and bedtime fading) were associated with elevations in cortisol as well as behavioral problems in children over the course of a year. 

The researchers recruited 43 infants with their parents (mothers and fathers) from Australia in a randomized controlled trial. Generally, randomized controlled trials are the gold standard of research methods. Randomly assigning which parents tried sleep training methods and which did not allowed the researchers more certainty that any differences they found between parents were due to the sleep training interventions themselves rather than on whether or not there were baseline differences between parents who wanted to sleep train their child versus those who didn’t. Randomized controlled trials also involve a control (comparison) group, which is crucial to understand whether any changes we see in the things we are measuring (e.g., cortisol levels) are due to our intervention (sleep training) or not. The researchers assigned families to one of three groups to implement a specific sleep training strategy: graduated extinction (a form of the “cry it out” method), bedtime fading (a non-“cry it out” method involving making a baby’s bedtime earlier and earlier over subsequent nights), and a control group (parents were not trained on either sleep training method and just did what they would normally do)². They collected cortisol from infants twice a day (morning and afternoon) for two days at five different phases of the study: before sleep training began (pre-intervention), one week after the sleep training intervention, one month post-intervention, three months post-intervention, and 12 months post-intervention. During these phases they also assessed maternal stress. During the 12-month post-intervention follow-up, the researchers also asked parents to fill out a self-report checklist of common behavioral/emotional problems in children as well as complete a task that assessed how well children were attached to their mothers to examine other potential, longer-term behavioral consequences of sleep training. 

This is a pretty rigorously, well-designed study. Unlike the Middlemiss article, these were not families with sleep problems that required residential treatment and all infants were deemed to be typically developing with normative/healthy weight gain. Thus, findings from this study may be more applicable to families other than those with more extreme sleep difficulties. While the researchers did not examine maternal cortisol levels (like Middlemiss), they examined infant cortisol levels across multiple timepoints over the course of a year, providing a better window into any potential long-term hormonal side effects of sleep training. They also studied cortisol levels prior to any intervention so that we have an understanding of what might be “normal” cortisol levels for these infants (something Middlemiss didn’t do). Finally, and perhaps most importantly, the whole point of this study was to examine potential long-term consequences of sleep training for infants and parents. 

Conclusions

The researchers found that there were significant, large declines in infant afternoon cortisol levels at 12-month follow-up (compared to pre-intervention levels) in both sleep training conditions and a significant, smaller decline in the control group. In other words, cortisol levels declined in infants across all groups, with a slightly larger decline found in the infants who were sleep trained. They also noted that the cortisol levels among all infants were within normative limits (i.e., there was no evidence of any chronically high levels of cortisol in the children). Mothers in both sleep training conditions reported improvements in self-reported stress after a month, while stress remained unchanged for moms in the control group. However, after 12 months, moms in all groups reported significant improvements in stress. Regardless of whether infants were sleep trained or not, there were no significant differences between groups in parent-child attachment style or behavioral problems in children during the 12-month follow-up. Overall, they found no evidence that sleep training had detrimental effects on either parents or children, and sleep training was not associated with elevations in cortisol levels in infants. They also cited a larger, separate randomized controlled trial that followed hundreds of infants for five years that also did not find any behavioral consequences of sleep training. 

Summary

It should be noted that both studies have limitations, as all studies do. Parents who volunteer for a research study to examine infant sleep problems are going to be different than parents who do not - for example, these parents may have infants with more extreme sleep problems than your average parent. Both studies had pretty small sample sizes (25 families in the Middlemiss; 12-15 families per group in the Gradisar). Both studies examined cortisol through saliva, which is a standard practice in these types of studies but is less accurate than, say, assaying cortisol from blood samples (something that Gradisar and colleagues discuss).

Despite what the New Yorker article suggested, these studies are extremely dissimilar. Middlemiss and colleagues were interested in short-term differences in cortisol responses between mothers and infants in a residential sleep training program, while Gradisar and colleagues were interested in comparing two different sleep training methods with a control group to understand cortisol response and many other outcomes on parents and children over the course of a year. The Middlemiss article does not tell us much about the potential consequences of sleep training for infants, mostly because that’s not what they were interested in studying in the first place - instead, they show that after three days of extinction sleep training, moms and babies have different cortisol levels. On the other hand, Gradisar and colleagues conducted a rigorously designed study that suggests that sleep training (at least the methods they examined) does not appear to have any significant effects on cortisol levels or behavioral problems in children up to a year after sleep training. Importantly, both studies suggest that sleep training is helpful in reducing parental stress, including a significant decline in maternal cortisol levels after just three days of sleep training (Middlemiss) as well as improvements in self-reported stress in moms up to a year after a sleep training intervention (Gradisar). 

In other words: You are not inadvertently stressing your child out by sleep training, and it’s definitely not selfish to want to get some extra sleep.

In the next newsletter I’ll continue the conversation about infant sleep and what we know about its relationship to parental mental health. Let me know what you think in the comments, and feel free to suggest other topics you’d like to see me tackle. Thanks for reading! 

1. Gradisar et al. (2016). Behavioral interventions for infant sleep problems: A randomized controlled trial. Pediatrics, 137: e20151486. https://doi.org/10.1542/peds.2015-1486

2. See the Gradisar et al. (2016) supplemental information for more details about their sleep training methods.