Effectiveness of behavioral sleep interventions on children’s and mothers’ sleep quality and maternal depression: a systematic review and meta-analysis

This systematic review and meta-analysis was conducted to investigate the effectiveness of behavioral sleep interventions (BSIs) on the number of child night awakenings, and maternal sleep quality and depression. The search followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (PRISMA) using PubMed, CINAHL, Cochrane, and EMBASE databases and retrieved studies published until April 2021. We calculated the odds ratios (ORs) and 95% confidence intervals (CIs) for child sleep problems, and the mean differences (MD) and 95% CI for the number of child night awakenings, and maternal sleep quality and depression. Ten studies of 1628 initial searched were included in the final analysis. Two of the 10 studies were divided into two subgroups by participants and intervention type; thus, 12 subgroups were included in the meta-analysis. BSIs significantly reduced child sleep problems (OR 0.51; 95% CI 0.37–0.69) and improved maternal sleep quality (MD − 1.30; 95% CI − 1.82 to − 0.77) in the intervention group. There were no significant differences in the number of child night awakenings and maternal depression between the two groups. More RCTs to examine the effect of BSIs considering children’s age, duration of intervention, and outcome measuring time points are needed.

www.nature.com/scientificreports/ and repetitive activities that was carried out before bedtime every night. These interventions have no standard components since they depend on the developmental stage of the children and the cultural situation. Therefore, bedtime routine interventions was defined as a practice that consisted of more than two activities before going to sleep.

Outcome measures. All extracted outcome variables in the included studies are presented in detail in
Supplementary Table 2. Studies that evaluated two specific aspects were chosen. The first aspect was child night awakenings and sleep problems reported by mothers using a sleep diary for their children. The second aspect was maternal sleep quality measured by the Pittsburgh Sleep Quality Index (PSQI), and maternal depression measured by the Edinburgh Postnatal Depression Scale (EPDS) for mothers.
Quality of appraisal. The Risk of Bias 2 (RoB2) tool was used to assess the quality of appraisal in this SR.
The RoB2 tool is a revised version of the original risk of bias tool that aimed to overcome certain limitations of the original version 38 . The RoB2 is structured according to five domains (randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, selection of the reported result), and overall bias. Judgment can be at "low" or "high risk" or can express "some concerns. " After two researchers independently assessed the risk of bias using the RoB2, they cross-checked the results and reached an agreement through discussions with each other about any disagreements or unclear judgments.
Data extraction. The data were extracted and converted into an excel file based on the Cochrane recommendations 39 : publication year, country, study design, sample size of each group, participants' characteristics including mean and range of age, contents of intervention, duration and frequency of intervention, and main outcomes.
The numerical values, such as the mean and standard deviation (SD), of the main outcomes (i.e., child night awakenings, maternal sleep quality, and maternal depression) were determined, as well as the sample size at certain time points after the intervention (within 1 to 4 months). The main outcome of sleep problems in children as recorded by their mothers in a sleep diary was event rate and the sample size at the aforementioned time points after the intervention.
Data synthesis and statistical analysis. Mean and SD were used for continuous variables such as child night awakenings, maternal sleep quality, and depression scores. The odds ratio and 95% confidence interval were calculated as data synthesis statistics for dichotomous data, such as the child sleep problems for dichotomous data, such as the child sleep problems. For three of the studies [40][41][42] , an email was sent to the corresponding author of each in order to request for additional data. Additionally, 95% confidence intervals (CIs) were used in the calculation of the effect size of the overall mean and each effect size. Review Manager (RevMan) version 5.3 was used for meta-analysis. A homogeneity test between studies was performed using chi-square tests, and I 2 tests. Meta-analysis using a random-effect model is appropriate for integrating intervention studies. These include different interventions, different results of ROB assessments and different participants' characteristics [43][44][45] . Studies included in meta-analysis have heterogeneity. Therefore, all subgroup analyses were applied to the random-effect model.

Results
Selection of studies. The selection process of the studies was based on the PRISMA guidelines as illustrated in Fig. 1. A total of 1628 studies were initially searched: 644 studies from PubMed, 470 studies from Embase, 216 studies from CINHAL, and 298 studies from Cochrane. Of these, 82 studies were excluded due to duplication. Among the remaining 1546 studies, 1513 studies were excluded because they were not related to the PICO of this study as determined by the title and abstract of each study. After 33 full-text studies were reviewed, 23 studies were excluded according to specific exclusion criteria: non-English language (n = 1); irrelevant participants (n = 9); irrelevant intervention (n = 4); and non-RCTs design (n = 9), which included one mixed-method study, one second-data analysis, one study protocol, and six quasi-experimental designs using one group pre-post test. Finally, 10 studies were included in this study 24,30,31,[40][41][42][46][47][48][49] . However, two of these studies were divided into two subgroups based on their participants and intervention type 41,47 ; thus, a total of 12 subgroup analyses were included in the meta-analysis.
Characteristics of included studies. The characteristics of the included studies are summarized in Table 1. The studies were published in 2002,2007,2009,2011,2013,2017,2018, and 2021, while two were published in 2016. Four studies were conducted in the United States, three studies in Australia, and one each in Canada, New Zealand, and Iran. The age of the children who participated in these studies ranged from 0 to 36 months. Regarding the intervention, five studies used settling methods, four studies used bedtime routines, and one study used both. In all the studies, the control group received usual care. The intervention periods were 2 weeks (three studies), 4 weeks, 8 weeks, 40 weeks, and from prenatal to 3 weeks after birth; the intervention period was not given in three studies. The time for assessing outcome measurements was up to 12 months after completing the intervention. Regarding the outcome variables, the presence of child sleep problems (two studies), the number of child night awakenings (eight studies), maternal depression measured by EPSD (three studies), and maternal sleep quality measured by PSQI (three studies) were used to evaluate the effect of BSIs on children.  Fig. 2. and Table 2. The overall bias was "high risk" in one study, "some concerns" in three studies, and "low risk" in six studies. In the randomization process item, two studies were judged to have "some concerns" because they randomly assigned study participants and did not mention a specific allocation sequence or allocation concealment. Another study was classified under "some concerns" due to deviations from the intended interventions and because the study described the impossibility of blinding. The risk of bias in the selection of the reported results was assessed as "high risk" in another study because it reported the outcome based on meaning-

Effect of the behavioral sleep interventions on maternal depression. The effect of BSIs on mater-
nal depression is shown in Fig. 3d. Studies reported maternal depression in both intervention and control groups using EPSD after BSIs and were homogeneous with I 2 = 13%. The mean difference in maternal depression was − 0.22 (95% CI − 0.68-0.25, p = 0.33; k = 7), indicating that there was no significant difference between groups.

Discussion
This study was conducted to examine the effects of BSIs on the sleep of children and mothers. As a result of the previous study search, 10 RCTs published from 2002 to 2021 were included, and significant effects of the BSIs on child sleep problems and maternal sleep quality were found.
In children's aspects, when comparing BSIs with usual care, BSIs significantly decreased child sleep problems as reported by their parents. Parent-reported child sleep can provide integrative information on environmental and behavioral dimensions and is the most commonly used method for measuring children's sleep 50,51 . Parentreported problematic sleep is a strong predictor of sleep problems in children 41 . Child sleep problems vary according to developmental stage, are common in early childhood, and decline with age 52 . Williamson et al. examined sleep behaviors from birth to middle childhood using a longitudinal research design over a ten-year period. They found that reports of child sleep problems by caregivers were most prevalent in the infant stage and reduced with age, and the most common problem from infancy to seven years old was night awakening 53    www.nature.com/scientificreports/ and found that infant sleep patterns more clearly develop at 5-6 months. That is, night-time sleep starts to integrate at 5 months and stabilizes more clearly at 6 months 54 . These changes in children, especially in infants' sleep problems and patterns, are due to maturation of the circadian system. Sleep-wake homeostasis is related to multiple parenting and environmental factors, such as feeding, exposure to light, and sound 50 . Therefore, BSIs for infants should be considered at this critical time (i.e., 5-6 months after birth). www.nature.com/scientificreports/ The duration of a behavioral sleep intervention in addition to the child's developmental stage could affect sleep outcomes. That is, a subgroup analysis of the effect of BSIs by children's age and intervention duration may be meaningful for detailed practical implications. However, the number of studies included in the current study was so small that subgroup analyses could not be performed. Therefore, more RCTs on children of varied developmental stages and with varied intervention durations are needed to further examine the effects of BSIs. With the accumulation of such RCT research, more specific and practical applications of BSIs can be expected.
The child night awakenings did not significantly change given the type of behavioral sleep intervention. A similar result was found in a SR and meta-analysis of infants aged under 12 months 37 . The number of night awakenings is one of the most commonly used methods for measuring sleep problems 55 , but night awakenings may not actually be sleep problems in children, especially infants. A possible interpretation of these results may be related to the participants' developmental characteristics. That is, regardless of BSIs, night-time sleep awakenings could normally occur in infants due to feeding and diaper change 56,57 , and their immature circadian system. Moreover, parents of breastfed infants have reported more frequent night awakenings and shorter night-time sleep periods 58 . Among the studies included in the current study, there were eight studies used night awakenings as an outcome variable. Although seven studies of them included infants under 12 months of age, only two studies consider night feeding 49 and feeding type 40 , and most of the studies did not fully consider participants' developmental characteristics. Therefore, future research needs to consider how to measure infants' night awakenings caused by sleep problems,s not by feeding or diaper change.
Although more studies on sleep have measured sleep patterns or problems using actigraphy in recent years, we used parent-reported child sleep as one of the outcome variables because most of the included studies utilized it. It is known that parent-reported sleep problems are easily measured, especially for infants and toddlers, but this method may overestimate sleep duration compared to actigraphy 51,59 . Actigraphy also has its limitations. For instance, it overestimates the number of awakenings since it can measure normal motor activity during sleep as a night awakening 60 . To consider the benefits and limitations of parent-reported data and actigraphy, the concurrent use of subjective and objective methods could be the best way to evaluate child sleep problems more accurately.
In mothers' aspects, the current study found that maternal sleep quality significantly improved in the intervention groups. These results could be interpreted as sleep improvement in children through BSIs, and could also enhance maternal sleep quality. In previous studies, it was found that child sleep patterns have a significant correlation with maternal sleep. Covington et al. 61 found that total maternal sleep time was associated with mothers' perception of their toddlers' sleep problems when sleeping with their toddlers. The number of night awakenings in infants was significantly positively correlated with that of mothers 62 . Goldman et al. 63 examined the association between child and maternal sleep using actigraphy in typically developed children and children diagnosed with autism spectrum disorder. In this study, they found significant associations between time spent in bed, sleep time, and sleep fragmentation in mothers and children 63 .
In contrast, BSIs for children did not have a significant effect on maternal depression in the current study. This is inconsistent with the previous SR of psychosocial sleep interventions for children, which found a significant improvement in maternal depression 37 . Maternal depression could be related to their child's sleep and the postpartum period. In a previous SR study of postpartum depression in healthy mothers, the overall prevalence of postpartum depression was approximately 12% 64 . Most postpartum depression is serious in the early postpartum period but decreases and becomes stable over time 65,66 . That is, because there were more mothers in their early postpartum period who were included in the present study than in the study of Kemplar et al. 35 , it seems that maternal depression might be related to the postpartum period, not the child's sleep. In turn, this resulted in an insignificant effect of behavioral sleep intervention on maternal depression. Maternal postpartum depression could also be affected by various factors, such as history of anxiety or depression, prenatal depressive symptoms, and maternal fatigue [65][66][67][68] . Because of these other factors, maternal depression might not be improved by BSIs for their children.
In addition, the inappropriate use of EPDS for measuring maternal depression has been found in previous RCTs. EPDS was originally developed to measure maternal depression in the postpartum period 69 , which is defined as the period until 6 months after birth 70 . However, several studies included in the current study measured maternal depression using EPDS even if the mothers were not in the postpartum period 24,42,46 . Therefore, maternal depression might not be correctly assessed because of the use of inappropriate measurement tools. Therefore, more research is needed to measure maternal depression using appropriate instruments, and compounders affecting maternal depression should be considered when examining the effect of BSIs for children, on maternal depression. Additionally, future studies are needed to identify the long-term effects of BSIs on maternal depression and changes in maternal depression using a longitudinal study design.
As a result of the current study's meta-analysis, BSIs, including settling and bedtime routine interventions, can reduce sleep issues in children and improve the mothers' sleep quality. The application of age-appropriate planned sleep behavior interventions and sleep culture can contribute to the formation of positive sleep habits for both mother and child. It could contribute to early adaptation to the sleep environment and improvement of sleep quality. Therefore, the application of BSI can be implemented when the circadian rhythm is stabilized. Nurses should prepare the BSI plan and provide the mothers with the information soon after childbirth. An accuracy-evaluation of the number of night awakenings using an actigraph is required in future studies, and more studies measuring maternal depression or fatigue using appropriate tools are needed.
The current study has several limitations. First, only ten RCTs were included in the current study because of the limited number of RCTs on child BSIs. Because of the limited number of included studies, we could not perform subgroup analyses according to the ages of the children and the durations of the interventions, which could have helped provide detailed guidelines for clinical practice. Therefore, it is necessary to conduct more RCTs and accumulate evidence on the effects of child BSIs. Second, the current study could not use the data monitored by www.nature.com/scientificreports/ actigraphy because most studies confirmed sleep patterns using a sleep diary or a sleep-related questionnaire. However, in terms of measuring the sleep quality of children and adults, actigraphy could provide more objective data 71,72 ; thus, future studies need to use actigraphy to achieve and evaluate outcomes of sleep intervention. Third, the current study did not separate participants according to developmental stage (e.g., infants, toddlers) because of the characteristics of the included studies. Since a child's age could be a confounder of their sleep pattern, future RCTs need to consider child sleep patterns by developmental stage. Fourth, because the duration of interventions and follow-up times were different in most of the included studies, this could cause deviations in the meta-analysis results. Therefore, future SR and meta-analysis studies need to identify the effects of BSIs according to the duration of intervention and follow-up times. Fifth, because the sample size of the control group in Galland's (2017) study was considered twice, some caution is needed not to overestimate results.

Conclusion
This study found that BSIs in children significantly improved child sleep problems and maternal sleep quality.
In general, RCTs of BSIs for children are limited, and therefore, more RCTs need to be conducted to provide further evidence for BSIs. Future SR and meta-analysis studies need to examine the effect of BSIs according to the child's age, duration of intervention, and outcome measurement time points.

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.