The Costs of Sleep Training

While the high proportion of ‘positive’ outcomes from studies evaluating sleep training measures is impressive, and demonstrates the malleability of human behaviour, there are some important issues relating to the limitations of the research to date, and the potential costs of sleep training, that health professionals and parents need to consider when thinking about their child’s sleep, and the appropriateness of sleep management techniques or behavioural interventions, particularly those involving extinction (Blunden et al 2011). No study in our review of recent work explicitly indicated that parents were informed of potential disadvantages of early sleep consolidation, or of lengthening infant sleep bouts, before consenting to take part in any research. There is a dearth of information relating to negative outcomes associated with sleep training, however such consequences may include effects on breastfeeding, increase in SIDS risk, and alteration in aspects of infant circadian and neurophysiological development and parent-infant relationships.

Mother-Infant Separation

As secondarily altricial primates, human infants rely on close proximity with their mothers in order to support physiological functioning — including breathing, temperature and heart rate — in the early weeks and months of life (Small 1998; Hrdy 1999; Moore et al 2007). While information about the impact of parents’ presence on infant physiology is scarce, our knowledge about the role of proximity during sleep on neurological and emotional state and development is very limited.

A small body of research suggests that mother-infant separation during infant sleep may have detrimental effects on both the immediate well-being and long-term development of infants (Ferber & Makhoul 2004; Bergman et al 2004; Ohgi et al 2001; Feldman et al 2002; Feldman & Eidelman 2003). A recent study compared healthy, full-term, two-day old neonates sleeping in skin-to-skin contact with their mother, and then sleeping separately swaddled in a bassinet close to the mother’s bed (Morgan et al 2011). Although reporting on a small sample (n=16), results showed that newborn infants sleeping separately from their mothers experienced less quiet sleep, and longer sleep latency (time to enter quiet sleep) than infants sleeping in skin-to-skin contact with their mother with an increase in heart-rate variability during separate sleep. The authors suggest these findings are indicative of ‘anxious arousal’ in infants sleeping apart from their mothers.

Another recent study (Middlemiss et al 2012) examined changes in synchrony in mother-infant physiology, whilst 25 4-10 month-old infants were undergoing a hospital-based extinction programme. Physiological and behavioural synchrony is generally considered to be adaptive, and implicated in the management of developmental processes and responses to environmental challenge (ibid.). Two key results emerged: Firstly infants undergoing extinction training ceased crying at bed-time by the third night of the programme. Their cortisol levels, however, remained elevated, at the same level as on the first night of the programme. At that time elevated cortisol had been accompanied by intense crying. This indicates that while the infants’ behaviour had become ‘habituated’ to the environmental change (sleep whilst alone), their physiology had not, and remained indicative of continued stress. Secondly, mothers’ and infants’ physiological stress responses — indicated by cortisol levels — exhibited synchrony on the first night of the programme. By the third night, when infants’ behavioural distress cues were absent but physiological stress was still high, mothers’ cortisol levels were no longer in sync — they were considerably lower. While these findings reflect short-term effects (it is possible that habituation of physiology may follow, and mother-infant synchrony may re-establish), they indicate the importance of behavioural cues for maintenance of mother-infant physiological synchrony, and that there are potentially consequences of sleep training beyond simple cessation of crying. It additionally lends empirical support to the notion that infants who learn to ‘self-sooth’ are in fact learning to ‘give-up’.

These results suggest there are potentially harmful consequences of maternal-infant separation in the immediate postpartum period, and in infancy, and build upon a large body of studies in primates and other non-human animals in which the harmful effects of maternal-infant separation have been observed (e.g. Blum 2002; Pryce et al 2011; Pryce et al 2004; Leventopoulos et al 2009).

SIDS and breastfeeding

According to SIDS-risk reduction advice, until 6 months of age infants should sleep in same room as a caregiver, for naps as well as for night-time sleep. We also know that greater responsiveness and closer proximity of mothers and infants promotes both initiation and longer-term maintenance of breastfeeding (Ball et al 2016; Yamauchi & Yamanouchi 1990; Buxton et al 1991; Baddock et al 2007, Santos et al 2009, Ball et al 2011). The emphasis on lone sleeping that is characteristic of Western parenting culture conflicts both with the SIDS reduction message and with promotion of breastfeeding, in that it inherently focusses on training infants — regardless of their age, developmental stage or biological needs — to self-settle, and to sleep for longer periods.

Encouraging babies to sleep longer and more deeply than is normal for their developmental stage may also put some arousal-deficient infants at further increased risk of SIDS, especially in the absence of parental presence (Mosko et al 1997). Use of extinction methods seems additionally problematic in this regard as continuous or intermittent absence of the parent is an intrinsic element of most regimes. It seems likely that one of the consequences of ‘success’ in sleep training regimes is parents leaving their infants alone for long periods of time, in practice if not in principle, putting younger infants at further increased risk of SIDS. Prevention methods, however, are also focussed on encouraging earlier sleep consolidation in younger infants, including those <6 months, and most at risk of SIDS.

For both younger and older infants, increasing the length of night-time sleep bouts may reduce or eliminate the opportunity to maintain a successful breastfeeding relationship. One study (Nikolopoulou & St Robert-James 2003) reported that while night-time feeds reduced, day-time feeds increased; however in some circumstances — for example where mothers have returned to work during the day — this redistribution of feeding opportunities may not be possible, or adequate for breastfeeding maintenance in the longer-term.

Definition of ‘sleep problem’

In most (treatment orientated studies) studies participants have been selected on the basis of parental report of sleep problems — however the definition of ‘sleep problem’ (and therefore the applicability of the results to the wider population) varies wildly! In addition to variation in the way parents perceive their infant’s sleep prior to participation in a sleep training intervention, it is important to consider the impact that study participation itself might have on perception of sleep problems.

A rare qualitative analysis of one intervention (Tse & Hall 2007) revealed an interesting consequence of the way information about infant sleep was received by parent participants. Parents were given information which indicated that babies (aged 6-12 months at inclusion in this study) could learn to self-sooth. Authors of this study reported that taking part in the programme “helped [parents] realize the need to facilitate behavioural changes in their babies”(p164). In cases where parents are informed that the ‘ability’ to fall asleep alone is a skill that can be taught, parents may interpret this as meaning that children ‘need’ to be actively taught this skill, with the implicit corollary that if such skills are not taught, children will not learn to fall asleep by themselves.

In most cases, however, children will learn to fall asleep by themselves when they are developmentally ready to do so, regardless of ‘training’ received. The few studies that have conducted long-term follow-up of behavioural interventions add support to this assertion. Lam et al (2003) conducted an RCT with 156 8-10 month old participants comparing sleep outcomes in a group largely utilising controlled crying (some parents chose to use ‘camping out’) with a control group. At 2 months post-intervention there was a significant improvement in reported sleep problems in the intervention group, compared to the control group. However at 4 months post-intervention, and at a later follow-up at 3-4 years old, there were no differences between groups. They also found that there was no difference between in maternal coping, stress or family functioning at this stage with regard to presence or absence of sleep problems at this point. A study reporting on the long term effects of a cluster-randomised trial utilising controlled crying/camping out with 7 month old infants (Hiscock et al 2007, 2008; Wake et al 2011) found significantly fewer parental reports of ‘sleep problems’, and fewer reports of poor or insufficient maternal sleep at 10 and 12 months of age in the intervention group, however when infants reached 24 months and 6 years there was no difference in report of sleep problems between groups. In a study (Thunström 2000) which focussed on 6-12 month-old infants with 6 months+ history of long sleep latency and more than 3 wakings/night occurring at least 5 nights/week, a significant improvement in sleep-outcomes was observed at 1 month post-intervention. At 1 year and 2.5 years follow-ups, intervention infants’ sleep characteristics were no different to those of a match-control group with no reported sleep problems at inclusion. However no non-intervention control group reporting sleep problems at inclusion was used, so it is not possible to attribute the lack of difference at these time points to the intervention, rather than to natural resolution via normal/developmental changes in sleep patterns. Overall, findings suggest that even where short-term improvements are reported in intervention-group sleep compared to control-group sleep, long-term no differences in sleep-behaviour between groups persist.

The sense that parents ‘need’ to teach their babies to sleep not only involves the use of behaviour modification strategies which may, or may not, have harmful consequences; it may also impose unnecessary and stressful limitations upon parental and family life. Parents who use extinction methods frequently find the technique stressful and distressing (Loutzenhiser et al 2012, Tse & Hall 2007), and are disappointed to find that the method does not provide an easy or permanent ‘magic bullet’ solution, and that relapses are common. Where parental education and other non-extinction based methods are employed, parents have expressed dissatisfaction at the impact that imposition of strict routines, schedules and other techniques have upon daily life (Tse & Hall 2007).

Alternative approaches to optimising parent-infant sleep

In light of the growing body of evidence highlighting the unintended negative consequences of dominant behavioural sleep interventions, particularly in breastfed infants and infants under six months of age (Douglas and Hill 2013), increasing numbers of health professionals and researchers are arguing that what we tell parents about normal infant sleep, and how we provide support, requires reframing (Ball 2013; Blunden et al 2011).

One such alternative approach is the Possums Sleep Intervention (Whittingham and Douglas 2014). This integrates interdisciplinary knowledge from developmental psychology, medical science, lactation science, evolutionary science, and neuroscience with third-wave contextual behaviourism, acceptance and commitment therapy, to create a unique, new intervention that supports parental flexibility, cued care, and the establishment of healthy biopsychosocial rhythms. Essentially, the main difference between this intervention and interventions from the dominant sleep training paradigm is that this approach does not aim to maximize the duration of the infant’s nocturnal sleep. Instead, this method advocates for responding to infants in a manner that replies to the infant’s communications and meets the need underlying the infant’s cues whilst optimizing the parent’s sleep efficiency and resilience.