Parents are largely taking this advice. A survey conducted in Canada on the use of graduated extinction for sleep found that more than 70% of parents reported starting before six months (Loutzenhiser et al., 2014); however, the survey also found that 40% of those said that the approach “didn’t work at all.” Because extinction often results in prolonged distress in the absence of soothing, we must also consider whether such an approach is physiologically safe for younger infants, whose neurological capacities for self-calming are more limited.

​

What do we actually know about using extinction (crying it out) with younger infants?

          The overwhelming endorsement of extinction (crying it out) with infants under six months represents a significant departure from what is known from existing research, as well as from knowledge about the neurological and developmental trajectories of both sleep and emotional regulation. A review of the literature on behavioral infant sleep interventions reveals that the advice to start this early is not empirically well-supported (Gordon, 2022).

​

Very little research has been conducted using extinction (pure or graduated) with infants under six months of age. Indeed, one of the most seminal researchers on extinction submitted that the intervention should not be used before six months and, under a year, should only occur under close monitoring (France, 1994). A systematic literature review (Gordon, 2022) found that of the 49 studies listed as empirical support for extinction, only five included any infants under six months (Eckerberg, 2002, 2004; Matthey & Speyer, 2008; Symon et al., 2005; Weir & Dinnick, 1988). Of the studies supporting the “well-established” designation for extinction (Kuhn & Elliot, 2003), only one included any infants younger than six months (Symon et al., 2005).

​

Studies with younger infants include them only as part of a wide age range (e.g., 4-48 months; Eckerberg, 2004; 3 weeks-36 months; Matthey & Speyer, 2008; 4-52 months; Weir & Dinnick, 1988), with only group averages reported. Information about the number of participants in each age bracket or results by age is routinely not reported. As a result, it is unknown how the younger infants fared compared to the older children in the sample.

​

Prevention studies are not employing actual extinction interventions. Some studies with newborns required only extremely limited waiting periods (e.g., five minutes or less; Crncec et al., 2010; Matthey & Speyer, 2008)—not the much longer delay suggested in parent-directed advice books for this age group (e.g., 45 minutes to no maximum, Weissbluth, 2015). As a result, interventions for young infants, which are said to be empirically supported, differ significantly from the real-world instructions given to parents, who are routinely told that such interventions are not only effective but will not be harmful to younger infants. These points have so far not been researched.

Currently, any advice to use pure extinction on infants under six months is without empirical investigation or support.

​​​

No research has been conducted on the need or benefit of intervening before six months.

          Popular advice suggests that waiting until after six months to sleep train will be significantly more difficult (Dubief, 2017; Mindell, 2009; Weissbluth, 2015). Research has suggested that sleep training is necessary because sleep problems do not resolve on their own (Byars et al., 2012; Zuckerman et al., 1987). While it is true that without intervention, sleep problems will persist, no research has examined whether there are any benefits to starting early or whether waiting until six months or later poses any additional problems (Thomas et al., 2014). In fact, the four-month developmental regression may make intervening at this point particularly difficult. The notion that sleep interventions will be more difficult if parents wait until six months or later has never been investigated.

Results of prevention studies can be nonexistent, small, or ephemeral.

          Results of research on prevention strategies have not found significant changes that endure past a month or two. Some studies aimed at preventing sleep problems by intervening early reported differences between groups at the end of intervention, but these differences disappeared by follow-up (Hauck et al., 2012; Pinilla & Birch, 1993; Wolfson et al., 1992). One study comparing a behavioral intervention, an educational intervention, and a control group found statistical differences in the number of “uninterrupted nights,” but this number was only 0.5 more (5.7 versus 5.2; Sleep et al., 2002). Other studies found changes in parental perceptions (e.g., confidence, mood; Adachi et al., 2009) or estimates of the infant’s sleep (Hall et al., 2015), but not in the infant’s objectively measured sleep behavior.

Some studies failed to objectively measure the infant’s sleep and relied on parent report of improvement (Matthey & Speyer, 2008). A meta-analysis found “no evidence showing these interventions reduce mother reported night waking in infants." (Kempler et al., 2016, p. 19) Two of the highest quality studies in this analysis "found no evidence of the efficacy of the sleep-focused interventions on reported total sleep time, infant wakings or maternal mood in the short-term" (Kempler et al., 2016, p. 30). Others likewise found no differences between intervention and control groups (Galland et al., 2017; Hiscock et al., 2014; Stremler et al., 2013). Owens et al. (1999), leading researchers on extinction, suggest:

​

The presumption is that these programmes will prevent sleep disturbance, rather than simply change the sleep patterns of young infants in the short term. While short-term relief may be of benefit to parents, if the effects do not persist, then this does not fit within the definition of primary prevention. (Owens et al., 1999, p. 294)

​

Parents are nevertheless strongly told to start sleep training early and are warned about the consequences of waiting. These points are not supported by any research that exists.​

​

The effect for time is rarely considered.

          Sleep develops in the brain. As a result, sleep patterns naturally improve over the course of the first several months. Several researchers have pointed out that studies of this time period may inflate results by failing to account for normal developmental improvements (Douglas & Hill, 2013; Owens et al., 1999).

​​​

Prevention studies employ a package of strategies in addition to behavioral interventions.

          Prevention research rarely focuses on behavioral strategies alone. Instead, studies offer parents a package of strategies that often include psychoeducation, professional support, individualized consultations, and regular follow-ups with providers (Adachi et al., 2009; Didden et al., 1999; Eckerberg, 2002; Fisher et al., 2004; Matthey & Speyer, 2008; Smart & Hiscock, 2007; Stremler et al., 2013). A few studies were conducted at in-patient parenting centers, where mothers received instrumental support from nurses, professional input from clinical staff, and social support from other mothers in addition to information on sleep (Fisher et al., 2004; Matthey & Speyer, 2008; Stremler et al., 2014). These settings differ dramatically from real-world contexts where parents are at home alone with a baby. 

​

Because multimodal interventions included a variety of strategies, it is impossible to know which part of the program had the most influence on outcomes. Both social support (Weir & Dinnick, 1988) and parental self-confidence (Cutrona & Troutman, 1986) have been shown to improve infants' sleep. It is possible that the support component could have been the key to successful outcomes—and not the intervention. In one study, parents reported that having someone to talk to and learning about normative development was the most influential (Smart & Hiscock, 2007). In another study, parents reported that simply having an established routine improved sleep (Hall et al., 2006). Another study found that the use of a tracking chart improved sleep in nearly 40% of the sample (Largo & Hunziker, 1984). The assessment, support, and follow-up parents receive in these studies are rarely accounted for in results, and as a result, the effectiveness of the intervention is artificially inflated.

​

Exact details on sleep interventions in behavioral packages are often lacking.

          In many prevention studies, it is unclear what the sleep intervention actually was. Details about what parents were told regarding responding to nightwakings are often not provided. For example, one study indicated that parents were given information about “the importance of self-soothing back to sleep” (Paul et al., 2016, p. 3), but failed to articulate what that meant in practical terms for parents’ behavior. Other studies merely report that education included information about “normative sleep,” “appropriate responses to waking,” or “settling methods” without further detail about what those responses should be (e.g., Kerr et al., 1996; Santos et al., 2016).

​

There has been no investigation of negative side effects in infants under six months.

          Though researchers and others continue to maintain that extended crying has never been shown to be harmful, actual investigation of this point is lacking. The majority of studies on extinction failed to gather any data other than sleep. Claims about the lack of negative side effects of extinction can be traced to only a small number of studies—all on children over six months (Eckerberg, 2004; France, 1992; France et al., 1991; Gradisar et al., 2016; Hiscock, Bayer, et al., 2007; Matthey & Crncec, 2010; Price et al., 2012).

​

Given the importance of development on self-calming capacities, it is inappropriate to suggest that because extinction has not been found to be harmful in older infants, toddlers, and preschoolers that it is without effect for younger samples. To date, we know nothing about how much crying at what ages is without harm.

The approach to studying the side effects of extinction in general is flawed.

          While some studies have used validated scales to measure outcomes (Child Behavior Checklist; Hiscock et al., 2008; 8-10 mos.; Reid, Walter, & O'Leary, 1999; 16-48 mos.), objective measures of stress (salivary cortisol; Gradisar et al., 2016; Middlemiss et al., 2012; Price et al., 2012), or attachment (Strange Situation; Gradisar et al., 2016; 6-16 mos.), other studies have relied on non-validated assessments. Some used questions written by the researchers. One study asked parents to rate their child on scales of happy/depressed, alert/tired, and accommodating/obstinate  (Eckerberg, 2004). Another asked the parents to endorse statements such as “My baby is happy,” “My baby is cooperative” (Matthey & Crncec, 2012). The third was a qualitative interview with questions like, “Did you notice any negative change?” (Seymour et al., 1983). One study merely reported that “no negative side effects were observed” without detailing what effects they were looking for (Williams, 1959).

While parent report can be informative, it is not completely reliable. Would a parent be able to say that the intervention they just used produced a negative change in their child? Is it possible that a parent who feels that they now have some control over sleep or that they’re more rested will see everything more positively?

​

The majority of studies assessing the presence of negative impacts utilized the Flint Infant Security Scale (FISS; Flint, 1974). The FISS was originally constructed in the 1970s as an assessment of security—a precursor to the conceptualization of attachment—in institutionalized infants. Given the proximity of the two terms, the scale's results have been frequently misinterpreted by researchers and experts as evidence of a lack of impact on attachment (Didden et al., 2011). This is not the case.

​

In order to measure security, the FISS asks parents to rate their infant in response to statements such as “Accepts new foods readily,” “Enjoys rough play,” “Recovers readily when physically hurt or if feelings are hurt,” or “Enjoys crowds.” It is unclear how these statements relate to what we now understand as attachment. These items appear to measure aspects of temperament more than security. To date, the FISS has never been validated on non-institutionalized or traumatized infants. It also has never been tested for internal validity as a measure of security or concurrent validity as a measure of attachment.

The study of effects on attachment has also been flawed. Only two studies have explicitly measured attachment. One study (Gradisar et al., 2016) utilized the Strange Situation with a sample of children (6-16 mos.) one year post-intervention. The other study by Price et al. (2012) assessed attachment at a five-year follow-up of graduated extinction. In both cases, attachment was not measured at baseline. We don’t know if any of the subjects changed category. The Price et al. (2012) study assessed attachment five years after extinction was deployed. Here, researchers did not assess attachment using a known measure, but instead, utilized the Disinhibited Attachment Interview (Chisolm, 1998), a five-item measure to measure a style of attachment that is a sub-type of Reactive Attachment Disorder (Zeanah et al., 2002) and is marked by indiscriminate friendliness with strangers. It is seen primarily in children who have experienced profound early neglect. Authors appear to suggest that since extinction did not result in the most extreme form of attachment disruption five years post-intervention, it did not impact attachment at all.

The notion that all crying in an extinction context is benign has never been investigated. While it is true that some crying in many contexts is without harm, the same may not hold true when crying gets extreme or within certain family contexts or for certain temperamentally reactive children. No research has yet investigated whether there are differential effects by age, temperament, or duration/intensity of crying.

Typically, the amount of crying is rarely recorded or reported. Results are based on group averages. Analysis of effects by different durations of crying has never been done. Therefore, nothing is known about whether there are amounts of crying that have side effects. France and Blampied (2005) is the study most cited as evidence for the lack of negative side effects. She compared the effects between three groups of infants (6 to 15 months): pure extinction, minimal check, and parental presence (camping out). This is one of the few studies that reported how much crying occurred for each subject. Results showed wide variation in the duration, degree, and persistence of crying across individuals.

The infant who cried the least (130 minutes over the three weeks) was 15 months and never cried more than 15 minutes on any given night. On the other hand, one six-month-old cried a total of 900 minutes, with the longest bout of crying (110 minutes) on the 21st night. A different six-month-old, on the other hand, cried for a total of 220 minutes, with the longest episode occurring on the first night, and subsequently decreased to only five minutes per night. While statistically, this data can be compiled and analyzed as group means, qualitatively, these represent very different experiences for both the infant and the parent. Collapsing the data into an average for a group obscures the wide variation that may exist and that parents rarely hear about.

For younger infants, this variation may be even more important to consider. Given the rapid brain development that occurs in early infancy, unmodulated distress (i.e., when parents let a baby cry for longer than a few minutes) may exceed their immature capacity to down-regulate. The regions of the brain experiencing the most rapid growth are the most vulnerable to disturbance (Webb et al., 2000). During these reorganization periods, very little stress can throw an organism off balance (Schore, 1996). The ability to tolerate and manage distress is highly maturation-dependent (Kopp, 1989; Schore, 1996). Distress must be manageable for an infant to marshal the resources they have. The younger the infant, the fewer strategies are at their disposal. Eight-week-old infants are likely still swaddled and would be unable to even get their hand to their mouth for sucking. So, a similar duration and amplitude of crying would be experienced very differently based on the available self-calming resources at various ages.

​

Further, for temperamentally or neurologically sensitive infants, distress may be experienced even more intensely, further diminishing their capacity to manage it. It is possible that extended distress in some infants, at certain points of development, or in specific family contexts may be concerning. There has been no inquiry into whether there are differential effects for infants based on age, temperament, or duration of crying.

​

It is not appropriate to simply apply strategies that have been tested on older infants, toddlers, and preschoolers to younger infants (Douglas & Hill, 2013). In the first six months, development is rapid, feeding patterns are becoming established, and infants differ widely in their capacity to manage both sleep and distress. It is possible that, for some infants, at certain points in development, there is a degree of distress that may be impactful. To date, sleep book authors continue to contend that even pure extinction is appropriate and benign for newborns despite the fact that these assertions have never been empirically investigated. A more careful investigation of these practices is urgently warranted to determine how parents can intervene safely and whether it is even worthwhile to do so.

​

References