Night-to-Night Variability in the Bedtime Routine Predicts Sleep in Toddlers

Abstract

The present study examined relations between nightly bedtime routines and sleep outcome measures in a sample of 185 toddlers aged 30 months. Parents reported on their toddler’s sleep duration and the length and activities included in the bedtime routine each night for approximately 2 weeks. Toddlers wore actigraphs to track their sleep during the same time period. Correlation, mean difference, and regression analyses indicated that toddlers experienced different bedtime routines and exhibited differences in parent reported sleep duration between weeknights and weekends. Multi-level models revealed that variability in the bedtime routine on an individual night most consistently affected parent reported sleep duration on that night. Differences in the bedtime routines between weeknights and weekends also affected actigraph recorded sleep duration and sleep efficiency. Results suggest that keeping consistent bedtime routines between weeknights and weekends is important for optimal sleep outcomes.

Introduction

Adequate sleep in young children is associated with healthy mental and physical development, and positive daytime functioning. Toddler and preschool children with poor or short sleep, inconsistent sleep schedules, or other sleep problems such as snoring or difficulty breathing, night waking, difficulty falling asleep, and daytime sleepiness, have been shown to exhibit more behavior problems, such as hyperactivity (Touchette et al., 2009), lower physical and psychosocial health-related quality of life (Hiscock, Canterford, Ukoumunne & Wake, 2006), and smaller gains in pre-kindergarten letter knowledge skills (Molfese, Beswick, Molnar & Jacobi-Vessels, 2009). Sleep problems in young children can also negatively impact parents (Mindell, Telofski, Wiegand & Kurtz, 2009), with mothers of infants and toddlers who experience sleep disturbances displaying elevated levels of depressed mood (Hiscock & Wake, 2001; Lam, Hiscock & Wake, 2003; Meltzer & Mindell, 2007). The National Sleep Foundation recommends that children at ages one to two get 11 to 14 hours of sleep each night and children at ages three to five get 10 to 13 hours of sleep each night (Hirshkowitz et al., 2015). School-age children and adolescents who sleep less than the recommended amount (9 to 11 hours and 8 to 10 hours, respectively) display poorer in school performance, as well as poorer cognitive and behavioral functioning (Dewald, Meijer, Oort, Kerkhof & Bogels, 2010; Geiger, Achermann, & Jenni, 2010; Paavonen et al., 2009). Thus, adequate sleep appears to be important for children’s behavior, school performance, and mothers’ mood.

A consistent sleep schedule, defined as going to bed and waking up at the same time every day, is widely regarded as important for ensuring young children get adequate sleep, yet inconsistent sleep schedules are relatively common in young children. For example, in a study of over 800 German children aged 0–6 years, Randler, Fontius, and Vollmer (2012) found that, in general, children went to bed later and woke up later on the weekends compared to weeknights and that the difference in bedtime and wake time between weeknight and weekend became larger with age. Such inconsistencies in sleep schedules have been linked to lower sleep quality as well as behavior problems. In a study of 48 five-year-old Japanese children, Iwata and colleagues (2012) found that children with more inconsistent sleep schedules took longer to fall asleep, fell asleep later, and slept less on weeknights than children with more consistent sleep schedules. Similarly, in a sample of 1622 Australian school-aged children ages 5–10 years, Biggs and colleagues (2011) found that inconsistent sleep schedules were associated with more hyperactivity and internalizing behavior problems, including worry, nervousness, and somatic complaints. In a study of Head Start preschool-aged children, Bates, Viken, Alexander, Beyers, and Stockton (2002) found that more variability in amount of nightly sleep time and bedtimes was related to children’s difficulties at preschool, as measured by teacher ratings of positive and problem behaviors, even after controlling for family stress and parenting practices.

The research considered here so far has emphasized regularity versus inconsistency in sleep itself. However, sleep is not functionally independent of children’s waking lives. A particularly relevant aspect of waking life involves the events preceding bedtime. The transition to sleep requires reduction of vigilance (Dahl, 1996), and a bedtime routine—a typical sequence of activities supportive of sleep—could foster that. A bedtime routine has been operationally defined as the set of predictable activities that parents engage in with their children in the hour or so before lights out and before the child falls asleep (Mindell, Li, Sadeh, Kwon & Goh, 2015; Mindell & Williamson, 2017). Multiple studies have shown the relevance of a consistent bedtime routine for young children’s sleep. Staples and her colleagues (2015) found that greater adherence to a bedtime routine was concurrently associated with more night time sleep at 36 and 42 months old (r = .30 and r = .29, respectively). In an intervention study of 405 mothers and their children, ages seven to 36 months, Mindell et al. (2009) found that a consistent bedtime routine significantly improved children’s sleep onset latency and sleep consolidation (longest continuous sleep period), decreased the number of night wakings, and improved mother’s self-reported mood as measured by the Profile of Mood States (McNair, Droppleman & Lorr, 1971). A large, cross-cultural study of over 10,000 parents of children ages 0–5 in 14 countries found that bedtime routines had a dose-dependent association with sleep such that better sleep outcomes (including earlier bedtimes, shorter sleep onset latency, decreased wakefulness, and more total sleep time) were associated with the number of nights per week that a bedtime routine was implemented (Mindell et al., 2015).

Conversely, irregular bedtime routines can have a negative impact on children’s sleep. In a study of over 5,000 infants and toddlers ages 0–36 months, Sadeh, Mindell, Luedtke and Wiegand (2009) had parents report how often their child had the same bedtime routine in a typical 7-day week on a five point likert-type scale, ranging from “1 = never” to “5 = every night”. They found that greater irregularity in the bedtime routine was associated with more frequent night wakings. Therefore, the first purpose of this study is to examine overall relationships between bedtime routines and sleep in toddlers, and investigate whether toddlers experienced differences between weeknights and weekends in measures of bedtime routine or sleep.

Bedtime routines can be measured in many ways, and the method used could affect inferences from findings. A majority of studies examining relations between bedtime routines and sleep have relied on a global parental report of sleep (e.g. “What times does your child typically go to bed each night?”) or the bedtime routine (e.g. “Do you have or follow a regular bedtime routine?”). For example, both Mindell et al. (2015) and Sadeh et al. (2009) used the Brief Infant Sleep Questionnaire (BISQ; Sadeh, 2004), which asked parents to report on sleep via the question “how much total time does your child spend sleeping during the night?”, and regularity of the bedtime routine via the question “in a typical 7-day week, how often does your child have the exact same bedtime routine?” with possible responses ranging from “never” to “every night”. Parents in the Biggs et al. (2011) study reported on their child’s usual, earliest, and latest bedtimes during the past week, while both the Molfese et al. (2009) and Touchette et al. (2009) studies asked parents to report how much their child slept on average. Although obtaining global ratings of children’s sleep and/or bedtime routines is common, it may be subject to bias in the estimates of children’s sleep time or consistency of the bedtime routine. Parents may be motivated to report their children going to bed earlier than they actually do on any given night or that they follow a consistent bedtime routine because it is socially desirable. However, daily reports of the bedtime routine and sleep via sleep diaries may help to ameliorate this bias and also minimize the amount of error that occurs in the retrospective reporting of events.

Daily sleep diary reports have been used in a few studies of children’s sleep and/or bedtime routines. In addition to the BISQ, Mindell et al. (2009) also used a daily sleep diary in which parents reported over three weeks on their child’s bedtime, time to fall asleep, and night wakings. Iwata et al. (2012) employed actigraph measures of child sleep and daily sleep diaries collected over seven nights and averaged the nightly recordings separately for weeknights and weekends to obtain their measures of bedtime, sleep latency, and night wakings. Bates et al. (2002) used daily sleep diaries collected and averaged across four weeks to obtain their measures of sleep and bedtime variability.

Researchers using daily reports of sleep or the bedtime routine most often averaged the nightly reports over several nights. Averaging over nights is likely to improve precision of estimates of typical routines compared to a single night’s data or a parent’s global rating, but at the same time, it fails to take advantage of the possibility that there is important information in day-to-day fluctuations. Only two studies have employed daily reports to examine variability in nightly sleep or the bedtime routine in young children (Bates et al., 2002; Staples et al., 2015). Both suggest that variability in the bedtime routine or sleep are more important in relation to different outcomes than are overall measures of sleep or the bedtime routine, but neither study examined the nightly level associations between bedtime routines and sleep.

A further question about implications of bedtime routine variability, is whether toddlers sleep less on specific nights in which they experience a bedtime routine that is discrepant from the usual one. This question is more specific than the previous studies’ questions, and aims at a more detailed understanding of processes involved in children’s sleep than attained so far. The authors are aware of only one study that has used daily reports to examine the nightly level associations between sleep, activities and psychological well-being in 750 14–15 year old adolescents. Using hierarchical linear modeling, Fuligni and Hardway (2006) found that not only was adolescents’ sleep time highly variable across individual nights, but also that activities engaged in during the day, such as studying or experiencing stressful demands, led to reduced sleep on those nights. While the activities in which adolescents were engaged in this study were not a part of a bedtime routine, these results taken together with the previous work linking bedtime routines to sleep in young children lend support for the hypothesis that bedtime routines on individual nights have the potential to impact sleep on those same nights. Therefore, the second purpose of this study is to use daily reports of bedtime routines and daily measures of sleep to investigate whether a bedtime routine or variability in the bedtime routine on a particular night impacts measures of sleep on that same night.

Researchers typically measure children’s sleep in two ways: parent report (either via daily sleep diaries or global ratings as described above) and actigraphy. Actigraphy is a noninvasive method for monitoring periods of rest and activity, and involves wearing a small portable device resembling a watch that records movement. Actigraphy has been established as a valid and reliable method for measuring children’s sleep (Acebo, Sadeh, Seifer, Tzischinsky, Hafer & Carskadon, 2005; Sadaka et al., 2014; Sadeh & Acebo, 2002; Sadeh, Lavie, Scher, Tirosh & Epstein, 1991). When actigraphy is used, it is frequently accompanied by a daily sleep diary, which is used to aid in scoring the actigraph data to provide accurate estimates of children’s sleep.

However, estimates of sleep from parent report and actigraphy are not always in agreement. Several studies have noted that parents consistently overestimate the amount of sleep their child gets each night when compared to actigraph records (Dayyat, Spruyt, Molfese & Gozal, 2011; Molfese et al., 2015; Nelson et al., 2014). A social desirability response set may partly impact parents’ reporting of their children’s sleep, as parents may be motivated to report their children getting sufficient sleep, since lack of sleep is often blamed for behavioral difficulties. However, part of the measured difference can also be attributed to parents being unaware of their child’s night wakings or when their child actually falls asleep or wakes up in the morning. In contrast, actigraph recordings use an algorithm developed and validated for use with pediatric populations (Sadeh, Alster, Urbach & Lavie, 1989) to determine accurate sleep onset and wake times, as well as subtracting out any night wakings, among other sleep measures. Although we do not assume that the method gives perfectly accurate measures, actigraphy is generally considered to be an objective measure of children’s sleep, especially in comparison to parent report. Nevertheless, actigraphy is subject to drawbacks as well including device failure (actigraph unexpectedly breaks or quits collecting data), participant noncompliance (child refusing to wear actigraph), and wrongfully scoring active sleepers or car naps as wake periods (Sadeh & Acebo, 2002). Given the limitations, actigraphy is usually accompanied by parental report of sleep. Understanding how actigraphy and parent report data mesh is of methodological and conceptual value. Therefore, the third purpose of this study is to utilize both parent report and actigraph records of children’s sleep and compare how they each relate to the bedtime routine.

Method

Sample

Participants were 185 (86 female) typically developing toddlers aged 30 months and their primary caregiver who were participating in a larger study on the relations between toddler sleep, temperament, and self-regulation. Participants were recruited from a mid-size city in the Midwest United States. The majority of toddlers were White (83.7%), and also multiracial (8.2%), Hispanic (3.3%), Asian (2.7%), Black (1.6%), and Native American (1.1%), and unreported (1 toddler), which reflect the larger community from which the sample was drawn. The majority of primary caregivers (94.6% mothers; herein referred to as parents) were married (83.8%) and ranged in age from 21 to 46 years old (M = 32.13, SD = 4.67). Parents’ education levels ranged from college degree (83.7%) to some college (13.6%) to high school diploma (2.7%); one parent did not report education level. Family income was reported in $5,000 increments and ranged from less than $10,000 per year to more than $125,000 per year, with a mean of $70,000 - $75,000 per year, and 12.7% of families reporting income of more than $125,000 per year; four parents did not report income.

Procedure

Participants were recruited from local childcare centers and pediatrician offices, through personal contacts, and the distribution of flyers at child-friendly events and locations. Families with toddlers who were within the required age range (not yet 30 months old) and lived within one hour of the testing site were invited to participate. Data were collected at three ages: 30, 36, and 42 months of age; only the data collected at 30 months of age are reported here.

When toddlers turned 30 months old, project staff contacted interested families and an initial home visit was scheduled. At the initial home visit, the parents were given study materials including questionnaires and a daily sleep diary to be completed on their toddler. Toddlers were given a sweatband to wear containing an actigraph used to record their sleep. Parents were instructed to have their toddler wear the actigraph on their non-dominant wrist continuously throughout the day and night, except when the actigraph could get wet (e.g. bath or swimming). If the toddler resisted wearing the actigraph on their wrist, parents were instructed that the actigraph could be placed on the upper arm or ankle. One week after the initial home visit, the parent and toddler visited the lab (to perform tasks considered in future reports), and data from the actigraph were downloaded and daily sleep diary completion was checked. One week after the lab visit, a second home visit was conducted, during which project staff observed the child’s bedtime routine for the one to two hour period leading up to the toddlers’ bedtime (until lights out). At the end of this visit, the actigraph and daily sleep diaries were collected from the parent.

Measures

Sleep.

Actigraphy.

Toddlers’ sleep was measured continuously for the two weeks between the first and second home visits using a Micro-Mini Motion Logger actigraph (Ambulatory Monitoring, Inc., Ardsley, New York), which is a small wristwatch-like device that records motion and activity levels via an accelerometer. Actigraphy is a non-invasive method for tracking rest and activity periods that has been established as a valid and reliable method for measuring children’s sleep (Acebo et al., 2005; Sadaka et al., 2014; Sadeh & Acebo, 2002; Sadeh, et al., 1991). Using the ActionW 2.7 software (Ambulatory Monitoring Inc., Ardsley, New York), actigraph data were scored using the Sadeh algorithm (Sadeh et al., 1989), which has been validated against polysomnography in toddler populations (Sadeh et al., 1991).

Only actigraph data that corresponded with a night for which sleep diary data were available were used. The number of nights available with actigraph data ranged from 0 to 14 (Mean = 11.48, SD = 3.21). Ten toddlers were excluded due to actigraph failure (device unknowingly shut down and did not collect data), loss of the actigraph, or no corresponding sleep diary data.

Five actigraph variables are used in the present study: i.) Actigraph Recorded Sleep Duration– the amount of time spent sleeping during the sleep period (measured in minutes by the actigraph, and converted to and reported in hours to aid in comparison to the parent reported sleep duration), spanning from sleep onset (the time the toddler fell asleep as determined by the actigraph algorithm) to sleep offset (the time the toddler woke up as determined by the actigraph algorithm), ii.) Sleep Efficiency – the percent of time spent sleeping during the sleep period (from sleep onset to sleep offset as determined by the actigraph algorithm), iii.) Wake Minutes – the number of minutes spent awake during the sleep period, iv.) Wake Episodes – the number of wake episodes during the sleep period, and v.) Sleep Latency – the amount of time in minutes from when the parent reported the child in bed to the toddler’s sleep onset as determined by the actigraph.

Sleep Diary.

Parents recorded their toddler’s bedtime each night of the two week protocol, as well as the time their toddler woke up the next morning. Parents also recorded each night in the sleep diary whether the actigraph was worn, when it was temporarily removed (e.g. for bath), and any naps or night wakings the toddler experienced during the 14-day testing period. This information was used to aid in scoring the actigraph data and to compute a parent-reported measure of nightly sleep duration measured in hours. The sleep diary also included space for parents to write in anything unusual about each night’s bedtime or sleep, such as their child being sick or sleeping in an unfamiliar place (e.g. at grandparents house). However, these instances of unusual events in the sleep diaries were rare across all children and days of sleep diary data available. Since the primary focus of the present study was to examine relations between the nightly bedtime routine and nightly sleep, rather than factors influencing the routine or the sleep, data on unusual events were not included in analyses.

Bedtime Routine.

Parents reported on the bedtime routine each night in the sleep diary, including routine start time and the specific activities included in the routine. Parents were provided with six common bedtime routine activities to check off if completed: shower/bath, pajamas/pj’s, story, water, TV, and brush teeth. Four additional spaces were provided for parents to write in their own activities (e.g. pray). The write-ins resulted in 24 additional activities engaged in by families during the bedtime routine. These additional activities could be grouped into nine categories: pray/read scripture, singing, potty/diaper change, snack/treat/meal, snuggle/rocking, play, music/white noise machine, special object (including blanket, stuffed animal or pacifier), and other (mostly family specific, such as trim nails, talk about day, phone calls to family, night walk). Providing drinks of juice, milk, or a bottle were combined with the original “water” category to form a “drink” category; reading was combined with the original “story” category; and YouTube, iPad, and movie watching were combined with the original “TV” category to form a new “technology use” category. This data reduction resulted in 15 categories of bedtime routine activities that parents reported in the sleep diaries: bathe, pj’s, read, drink, technology, brush teeth, pray, sing, potty, snack, snuggles, play, music, special object, and other.

Four variables were calculated for each night from the nightly bedtime routine information. Routine Length (RL), measured in minutes, was calculated as the difference between the parent reported routine start time and time in bed, while Routine Length Variability (RLV), also measured in minutes, was calculated as the absolute difference between the average routine length across the 14 nights and the routine length for each individual night. Routine Activity Total (RAT) was calculated as the total number of bedtime routine activities performed each night. To capture variability in the bedtime routine, a Normal Routine Deviation (NRD) variable was calculated. First a “normal” routine was identified for each child which included all activities (from the list of 15 categories identified above) that occurred on at least 10 of the 14 nights. Next, for each night, deviation from the “normal” routine was calculated such that a score of 1 was added for each bedtime routine activity that occurred and was not part of the normal routine, and for each bedtime routine activity that didn’t occur and was part of the normal routine. Therefore, higher scores indicate more deviation from the normal routine each night.

Statistical Analysis

Three types of analyses were conducted (see Table 1 for research questions and corresponding analyses). First, mean difference and correlation analyses were computed in order to examine overall relationships between the bedtime routine predictors (routine length, routine length variability, routine activity total, and normal routine deviation), and the parent reported and actigraph recorded sleep outcome measures (parent reported sleep duration, actigraph recorded sleep duration, sleep efficiency, sleep latency, wake minutes and wake episodes), averaged across the two-week testing period. These analyses also allowed for examination of differences in the bedtime routine and sleep measures on weeknights versus weekends and for comparison of parent reported sleep duration versus actigraph recorded sleep duration.

Table 1.

Research Questions and Corresponding Analyses

Research Question	Analyses
1. What are the overall relationships between bedtime routines and sleep in toddlers, and do toddlers experiences differences in the bedtime routine or sleep measures between weeknights and weekends?	Mean difference and correlational analyses Aggregate-level regression analyses
2. Does variability in the bedtime routine on a particular night impact sleep measures on that same night?	Multi-level models
3. How do parent report and actigraph records of children’s sleep compare to each other in their relations to the bedtime routine?	Comparisons were embedded within all analyses

Second, all bedtime routine predictors and sleep outcome measures were averaged across the two-week reporting period and multiple regression analyses were run to examine aggregate-level associations between bedtime routines and sleep measures, controlling for gender, ethnicity, and income, based on previous research that has identified differences in the implementation of bedtime routines based on gender, ethnicity and income (Hale, Berger, LeBourgeois, & Brooks-Gunn, 2009; Henderson, & Jordan, 2010; Milan, Snow & Belay, 2007; Mindell et al., 2015). Separate regression models were run for each sleep outcome measure (actigraph recorded sleep duration, parent reported sleep duration, sleep efficiency, sleep latency, wake episodes, and wake minutes) and routine predictor (routine length, routine length variability, routine activity total, and normal routine deviation) pair to examine the individual relations between the bedtime routine variables and sleep outcome measures.

Third, in order to examine associations between bedtime routines and sleep measures within individual nights, and to examine whether these associations differed by weeknight versus weekend, multilevel models were run. In the multilevel models, the bedtime routine variable (routine length, routine length variability, routine activity total or normal routine deviation) and a dummy code indicating if the night was a weeknight were entered as predictors of a given sleep measure at Level 1 and three time-invariant covariates (gender, ethnicity, and income) were entered at Level 2. All Level 1 effects were allowed to be random, resulting in the following model being fit for each combination of the four bedtime routine predictor variables and the six sleep outcome variables:

Level 1:

$$\text{Sleep outcome}={\text{b}}_{0\text{j}}+{\text{b}}_{1\text{j}}\left(\text{bedtime routine variable}\right)+{\text{b}}_{2\text{j}}\left(\text{weeknight}\right)+{\text{e}}_{\text{ij}}$$

Level 2:

$${\text{b}}_{0\text{j}}\left(\text{sleep outcome average}\right)={\text{c}}_{00}+{\text{c}}_{01}\left(\text{gender}\right)+{\text{c}}_{02}\left(\text{ethnicity}\right)+{\text{c}}_{03}\left(\text{income}\right)+{\text{u}}_{0\text{j}}$$

$${\text{b}}_{1\text{j}}\left(\text{bedtime routine variable}\right)={\text{c}}_{10}+{\text{u}}_{1\text{j}}$$

$${\text{b}}_{2\text{j}}\left(\text{weeknight}\right)={\text{c}}_{20}+{\text{u}}_{2\text{j}}$$

Next, the possibility of an interaction between the Level 1 predictors (the bedtime routine variable and weeknight) was tested by adding an interaction term at Level 1, which was also allowed to be random, resulting in the following model:

Level 1:

$$\text{Sleep }{\text{Outcome}}_{\text{ij}}={\text{b}}_{0\text{j}}+{\text{b}}_{1\text{j}}\left(\text{routine variable}\right)+{\text{b}}_{2\text{j}}\left(\text{weeknight}\right)+{\text{b}}_{3\text{j}}\left(\text{routine variable}*\text{weeknight}\right)+{\text{e}}_{\text{ij}}$$

Level 2:

$${\text{b}}_{0\text{j}}\left(\text{sleep outcome average}\right)={\text{c}}_{00}+{\text{c}}_{01}\left(\text{gender}\right)+{\text{c}}_{02}\left(\text{ethnicity}\right)+{\text{c}}_{03}\left(\text{income}\right)+{\text{u}}_{0\text{j}}$$

$${\text{b}}_{1\text{j}}\left(\text{routine variable}\right)={\text{c}}_{10}+{\text{u}}_{1\text{j}}$$

$${\text{b}}_{2\text{j}}\left(\text{weeknight}\right)={\text{c}}_{20}+{\text{u}}_{2\text{j}}$$

$${\text{b}}_{3\text{j}}\left(\text{routine variable*weeknight}\right)={\text{c}}_{30}+{\text{u}}_{3\text{j}}$$

For variable combinations where the Level 1 interaction was not found to be significant, the results from the model that did not contain the interaction term are reported.

Results

Mean Difference and Correlational Analyses

Descriptive statistics for the bedtime routine predictors and sleep outcome measures are found in Table 2. Parents reported their toddlers’ average sleep duration to be 10.53 hours, while the actigraph recorded toddlers’ average sleep duration to be 8.36 hours across the two week reporting period, a statistically significant difference of 2.17 hours (t = 24.01, p < .001). Independent samples t-tests were conducted to test for differences in these variables between gender and ethnicity (Caucasian vs. non-Caucasian). Because income was reported as a categorial variable with 25 levels, it was not included in the mean difference analyses. The results revealed that girls had significantly longer bedtime routines than boys and girls bedtime routine length varied significantly more across the two weeks than did boys. Paired samples t-tests were conducted to test for differences in these variables between weeknights and weekends and found that children completed more bedtime routine activities on the weeknights than the weekends. In addition, parents reported less sleep and the actigraph recorded more wake minutes on the weekends than weeknights. There were no other significant differences.

Table 2.

Descriptive Statistics and Mean Differences on Study Variables by Covariates

Overall	Mean Differences
	Mean (SD)	Gendera	Ethnicityb	Weeknightc
Bedtime Routine Predictors:
Routine Length	43.20 (16.2)	−6.30*	−2.88	0.80
Routine Length Variability	18.60 (9.12)	−3.00*	−0.90	−1.69
Routine Activity Total	4.16 (1.04)	−0.30	0.29	0.19*
Normal Routine Deviation	2.19 (0.88)	−0.04	0.01	−0.16
Sleep Outcome Measures:
Parent Reported Sleep Duration	10.53 (0.64)	−0.02	0.14	−0.20**
Actigraph Recorded Sleep Duration	8.36 (1.01)	0.06	0.17	0.10
Actigraph Recorded Sleep Efficiency	86.91 (8.14)	0.35	0.69	0.83
Actigraph Recorded Sleep Latency	33.21 (16.57)	−0.57	1.67	2.22
Actigraph Recorded Wake Episodes	5.76 (5.66)	0.06	−0.18	−0.14
Actigraph Recorded Wake Minutes	75.48 (46.44)	−0.95	−1.59	−5.94*

Note. Routine length, Routine Length Variability, Sleep Latency, and Wake Minutes are reported in minutes. Actigraph Recorded Sleep and Parent Reported Sleep are reported in hours. Sleep Efficiency is reported as a percentage.

^aPositive values indicate males are higher.

^bPositive values indicate Caucasians are higher.

^cPositive values indicate weeknights are higher.

^*p < .05,

^**p < .01,

^***p < .001

Results of Pearson correlations for the four bedtime routine variables, averaged across the two weeks, and parent reported sleep duration were that longer routines (r = −.15, p = .04) and more variable routine lengths (r = −.18, p = .01) were related to less parent reported sleep duration. More routine activities were related to more parent reported sleep duration (r = .18, p = .02), and more deviation from the normal bedtime routine was related to less parent reported sleep duration (r = −.23, p = .002). None of the bedtime routine variables were related to actigraph-recorded measures of sleep.

In order to determine if there were differences in the relations between variables on weeknights versus weekends, ccorrelations were computed between variables separately for weeknights and weekends. A similar pattern emerged between bedtime routines and parent reported sleep duration on weeknights, with longer routines (r = −.18, p = .01), more variable routine lengths (r = −.23, p = .002), and more deviation from the normal bedtime routine (r = −.21, p = .005) related to less parent reported sleep duration, and more routine activities related to more parent reported sleep duration (r = .19, p = .01). In addition, more deviation from the normal bedtime routine was related to less wake minutes on the weeknights (r = −.16, p = .03). None of the bedtime routine measures were related to the actigraph recorded sleep measures on the weeknights, nor were bedtime routines related to either parent-report or actigraph sleep measures on the weekends.

Aggregate-Level Analyses

The aggregate-level regression analyses were computed on the bedtime routine predictors and sleep outcome measures averaged across the two-week data collection period. Both the tests of mean differences and correlation analyses revealed some differences between weeknights and weekends on the relationships between bedtime routines and sleep measures, indicating a potential moderating effect of weeknight. Since weeknight/weekend is a time-varying variable, however, the interaction between weeknight and the aggregate bedtime routine predictors are impossible to test. Therefore, in order to avoid a potential masking effect of averaging across weeknights and weekends, it was decided to run the aggregate-level regression models for weeknights and weekends separately. Table 3 contains results from the aggregate-level regression models for weeknights only; results for weekends, not reported here, were similar in that coefficients were in the same direction, but smaller and all were non-significant. For completeness and comparison, the aggregate-level regression models were also run aggregating across the two weeks (including weeknights and weekends); results including weeknights and weekends averaged together were similar to the results for weeknights, but again, coefficients were smaller and fewer were significant, as would be expected if the bedtime routine and sleep variables have somewhat different relations in the weeknight versus weekend segments of the week.

Table 3.

Regression Analyses on Aggregate Models- Weeknight

Sleep Outcomes B (SE)
	Actigraph Recorded Sleep Duration	Parent Reported Sleep Duration	Actigraph Recorded Sleep Efficiency	Actigraph Recorded Sleep Latency	Actirgraph Recorded Wake Episodes	Actigraph Recorded Wake Minutes
Predictors
Routine Length	−0.21 (.28)	−0.41 (.18)*	1.24 (2.29)	−3.54 (4.83)	1.28 (1.56)	−7.34 (13.00)
Gender	−0.12 (.16)	0.07 (.10)	−1.30 (1.30)	0.40 (2.74)	0.08 (.89)	5.78 (7.37)
Ethnicity	0.09 (.22)	0.09 (.14)	−0.27 (1.77)	0.05 (3.74)	0.17 (1.21)	2.97 (10.07)
Income	0.00 (.01)	−0.02 (.01)*	0.10 (.09)	−0.43 (.19)*	−0.05 (.06)	−0.66 (0.50)
Routine Length Variability	−0.23 (.55)	−1.00 (.34)**	5.83 (4.54)	−15.93 (9.55)	−1.94 (3.11)	−39.51 (25.73)
Gender	−0.12 (.16)	0.09 (.10)	−1.47 (1.30)	1.00 (2.73)	0.34 (.89)	7.43 (7.37)
Ethnicity	0.08 (.22)	0.07 (.13)	−0.27 (1.76)	0.06 (3.71)	0.22 (1.21)	3.03 (10.00)
Income	0.00 (.01)	−0.02 (.01)*	0.11 (.09)	−0.44 (.18)*	−0.06 (.06)	−0.69 (0.50)
Routine Activity Total	0.05 (.07)	0.11 (.05)*	0.01 (.61)	1.17 (1.28)	−0.10 (.42)	1.10 (3.45)
Gender	−0.15 (.16)	−0.01 (.10)	−1.11 (1.28)	−0.23 (2.70)	0.24 (.88)	4.76 (7.28)
Ethnicity	0.08 (.22)	0.08 (.14)	−0.24 (1.77)	0.04 (3.73)	0.20 (1.21)	2.86 (10.07)
Income	0.00 (.01)	−0.02 (.01)*	0.10 (.09)	−0.44 (.19)*	−0.05 (.06)	−0.66 (0.50)
Normal Routine Deviation	0.05 (.09)	−0.13 (.06)*	1.44 (.71)*	−2.05 (1.50)	−0.52 (.49)	−9.26 (3.99)*
Gender	−0.14 (.15)	0.03 (.10)	−1.18 (1.26)	0.13 (2.68)	0.25 (.87)	5.49 (7.13)
Ethnicity	0.07 (.14)	0.09 (.14)	−0.36 (1.75)	0.13 (3.72)	0.25 (1.21)	3.59 (9.92)
Income	0.00 (.01)	−0.01 (.01)*	0.09 (.09)	−0.41 (.18)*	−0.05 (.06)	−0.59 (0.49)

^*p < .05,

^**p < .01

Results from the aggregate-level regression analyses revealed that on weeknights parent reported sleep duration was most consistently associated with bedtime routines, with the overall models predicting parent reported sleep duration from the four bedtime routine variables all being significant and explaining 5% - 8% of the variance in parent reported sleep. Specifically, longer routines, more variable routine lengths, fewer bedtime routine activities, and greater deviation from the normal routine each predicted less parent reported sleep duration on the weeknights. Income was a significant covariate in all four models with higher incomes predicting less parent reported sleep duration. Income was also a significant predictor of actigraph recorded sleep latency, with higher incomes predicting lower sleep latency, although overall percentage of variance explained in these models were not significant. Finally, greater deviation from the normal routine predicted greater actigraph recorded sleep efficiency and fewer actigraph recorded wake minutes, although again, the overall variance explained in these models were not significant.

Daily-Level Analyses

The mean difference, correlation, and aggregate regression analyses reported above were used to examine overall relations between bedtime routines and sleep outcomes, based on averages of daily reports that were recorded across the two week reporting period. The daily-level analyses were used to examine associations between bedtime routines and sleep outcomes within individual nights, in order to determine if the bedtime routine on an individual night impacted any sleep measures on that night. Table 4 contains the results from the multilevel models that were run for the daily level analyses and are discussed below.

Table 4.

Results from Multilevel Models

Sleep Outcomes B (SE)
	Actigraph Recorded Sleep Duration	Parent Reported Sleep Duration	Actigraph Recorded Sleep Efficiency	Actigraph Recorded Sleep Latency
Routine Length (RL)
Intercept	8.46 (0.26)***	11.06 (0.17)***	86.24 (2.09)***	37.28 (4.42)***
RL	−0.07 (0.06)	−0.34 (0.06)***	0.38 (0.45)	−2.03 (1.43)
Weeknight	0.11 (0.06)	−0.20 (0.06)**	0.71 (0.45)	2.21 (1.43)
Gender	−0.18 (0.15)	0.01 (0.09)	−1.34 (1.23)	0.94 (2.57)
Ethnicity	0.02 (0.21)	0.09 (0.13)	−0.37 (1.71)	1.30 (3.51)
Income	−0.00 (0.01)	−0.01 (0.007)*	0.07 (0.08)	−0.31 (0.17)
RL*Weeknight	---	---	---	---
Routine Length Variability (RLV)
Intercept	8.46 (0.26)***	10.96 (0.16)***	86.52 (2.07)***	36.62 (4.39)***
RLV	−0.15 (0.09)	−0.46 (0.08)***	−2.08 (1.40)	−1.90 (2.27)
Weeknight	0.11 (0.06)	−0.22 (0.06)***	−0.26 (0.66)	1.96 (1.39)
Gender	−0.17 (0.15)	0.01 (0.09)	−1.14 (1.22)	0.88 (2.52)
Ethnicity	0.02 (0.21)	0.09 (0.13)	−0.19 (1.68)	1.32 (3.49)
Income	−0.00 (0.01)	−0.01 (0.01)*	0.10 (0.08)	−0.31 (0.17)
RLV*Weeknight	---	---	3.29 (1.59)*	---
Routine Activity Total (RAT)
Intercept	8.42 (0.26)***	10.55 (0.19)***	86.07 (2.17)***	31.03 (4.42)***
RAT	−0.00 (0.02)	0.07 (0.02)**	0.03 (0.13)	1.27 (0.41)**
Weeknight	0.13 (0.06)*	−0.21 (0.06)***	0.79 (0.43)	1.06 (1.33)
Gender	−0.15 (0.15)	−0.04 (0.09)	−0.93 (1.25)	−0.08 (2.45)
Ethnicity	−0.02 (0.21)	−0.09 (0.13)	−0.54 (1.73)	1.71 (3.42)
Income	−0.00 (0.01)	−0.01 (0.01)*	0.08 (0.09)	−0.33 (0.17)*
RAT*Weeknight	---	---	---	---
Normal Routine Deviation (NRD)
Intercept	8.42 (0.26)***	10.66 (0.18)***	86.19 (2.11)***	35.54 (4.37)***
NRD	−0.01 (0.02)	0.09 (0.04)*	−0.04 (0.16)	0.35 (0.51)
Weeknight	0.12 (0.06)*	−0.001 (0.09)	0.80 (0.42)	2.46 (1.40)
Gender	−0.16 (0.15)	−0.03 (0.09)	−1.17 (1.22)	0.23 (2.49)
Ethnicity	0.02 (0.21)	0.03 (0.13)	−0.07 (1.71)	1.70 (3.43)
Income	−0.00 (0.01)	−0.01 (0.01)*	0.08 (0.08)	−0.32 (0.17)
NRD*Weeknight	---	−0.09 (0.03)**	---	---

Note. None of the models with number of wake episodes or number of wake minutes as outcome variables were significant, and are thus not reported here

^*p < .05,

^**p < .01,

^***p < .001

Bedtime routine length.

In the multilevel models containing bedtime routine length as a predictor, parent reported sleep duration was the only sleep outcome that was significantly related to any of the predictors (See Table 4). Specifically, routine length, weeknight, and income were all significant predictors of parent reported sleep duration such that increases in routine length on an individual night and higher incomes, as well as the night being a weeknight, predicted less parent reported sleep duration that night, controlling for all other variables. The Level 1 interaction between routine length and weeknight was not significant for any of the sleep outcomes.

Routine length variability.

In the models containing routine length variability as a predictor, significant effects were found for both parent reported sleep duration and actigraph recorded sleep efficiency (Table 4). Specifically, routine length variability, weeknight, and income were all significant predictors of parent reported sleep duration such that the more the routine length varied on an individual night from the average routine length, the higher the family income, as well as the night being a weeknight, all predicted less parent reported sleep that night, controlling for all other variables. For sleep efficiency, the Level 1 interaction between routine length variability and weeknight was statistically significant, indicating that the effect of routine length variability on sleep efficiency on a weekend (B = −2.08, p > .05), was not the same as on a weeknight (B = 1.21, p > .05), controlling for all of the other variables, even though the effect was not significant for either weekends or weeknights.

Routine activity total.

In the models containing the number of routine activities as a predictor, there were significant effects on actigraph recorded sleep duration, parent reported sleep duration, and actigraph recorded sleep latency. Specifically, there were significant effects of weeknight versus weekend on actigraph recorded sleep duration, such that children were predicted to have longer actigraph recorded sleep duration on weeknights compared to weekends, controlling for the other variables. There were significant effects of routine activity total and income on actigraph recorded sleep latency, such that higher number of total routine activities on an individual night was related to longer sleep latency, but higher income predicted shorter sleep latency, controlling for all other variables. Finally, there were significant effects of routine activity total, weeknight, and income on parent reported sleep duration, such that higher total number of routine activities on an individual night predicted higher parent reported sleep duration that night, while lower parent reported sleep duration was predicted on weeknights and as the parent’s income increased. None of the Level 1 interactions were significant for total number of routine activities.

Normal routine deviation.

In the models containing normal routine deviation as a predictor, there were significant effects on actigraph recorded sleep duration and parent reported sleep duration. Specifically, weeknight was a significant predictor of actigraph recorded sleep duration, such that more actigraph recorded sleep duration was predicted on weeknights compared to weekends, controlling for the other variables. Normal routine deviation, income, and the Level 1 interaction between normal routine deviation and weeknight were all significant predictors of parent reported sleep duration, such that higher income predicted lower parent reported sleep duration and a higher routine deviation on an individual night predicted higher parent reported sleep duration on that night, if that night were a weekend (B = 0.09, p < .05), but there was no relationship on weeknights (B = 0.00, p > .05).

Wake episodes, wake minutes, and cross-level interactions.

None of the models for number of wake episodes or number of wake minutes had significant effects and thus are not reported in Table 4. In addition, the moderating effect of the Level 2 covariates (gender, ethnicity, and income) on the Level 1 predictors of interest (routine length, routine length variability, routine activity total, normal routine deviation, and weeknight) were tested for each sleep outcome measure. The only model that had significant effects including these cross-level interactions had severe convergence problems that could be not remedied and therefore will not be interpreted here.

Discussion

This study investigated what role nightly bedtime routines play in toddler sleep by examining aggregate and daily relations between bedtime routines and sleep measures in toddlers. Daily reports of the nightly bedtime routines and several sleep measures were averaged across a two week reporting period to explore overall relations between bedtime routines and sleep outcomes, and then daily reports were analyzed individually to examine within-night relations between bedtime routines and sleep outcomes. Three main findings emerged: Parents’ estimates of their toddlers’ nighttime sleep were more than two hours per night greater than actigraph estimates sleep duration. Across all levels of analyses bedtime routines were most consistently related to parent reported sleep duration, in contrast to actigraph recorded sleep measures; and differences between bedtime routines on weeknights and weekends are associated with differences in toddler sleep between weeknights and weekends. Each finding will be discussed in further detail below.

The different levels of analyses used in the present study yielded largely concordant results, although there were a few notable differences suggesting that relationships between bedtime routines and toddler sleep may be more nuanced than can be captured by employing a single method of analysis. All levels of analysis were consistent in identifying differences in parent reported versus actigraph recorded sleep duration and relations between bedtime routine variables, weeknight, and parent reported sleep duration, but were less consistent in showing relations between bedtime routines and actigraph recorded measures of sleep. Mean difference and correlation analyses indicated that parents’ estimates of the amount of sleep their child was getting were significantly greater than actigraph estimates. This is consistent with prior research indicating parents overreporting on their child’s sleep when compared to more objective measures such as actigraphy (e.g. Molfese et al., 2015; Nelson et al., 2014). To the extent that this represents children actually sleeping less than what their parents believe, at least in certain cases, this could have health or behavioral implications. Given the National Sleep Foundation’s recommendations for sleep (11 to 14 hours for toddlers, 10 to 13 hours for preschoolers), parents may (wrongly) assume their child is getting “enough” sleep because their sleep falls within these guidelines. Given that average nighttime sleep duration as measured via actigraphy was a little over 8 hours, this means the majority of toddlers in the present study aren’t meeting the National Sleep Foundation’s recommendations. As a result, parents of children displaying behavioral difficulties may not even consider inadequate sleep as a potential area for intervention in their children’s difficult behavior, and may not talk with their pediatricians about addressing possible sleep problems.

The only significant models from the aggregate regression analyses involved bedtime routine variables predicting parent reported sleep duration. The daily-level analyses utilizing multilevel modeling revealed that the nightly bedtime routine and nightly variability in the bedtime routine were most consistently related to parent reported sleep duration as compared to actigraph recorded sleep duration. Specifically, longer bedtime routines and more variable bedtime routine lengths on weeknights led to less parent reported sleep duration. This may be because longer routines naturally push the actual bedtime later, leading to reduction in time in bed available for sleep. For parents who work full time outside the home and send their children to childcare, morning rise times are less flexible than nightly bedtimes because parents need to arrive at work the same time every day, meaning their children need to wake up at around the same time every morning. So, on nights when a longer bedtime routine pushes the actual bedtime later, parents may assume their toddlers are getting less sleep on that night. The fact that longer routines and more variable routine lengths did not relate to objective measures of sleep assessed via actigraphy, however, suggest that toddlers may not be missing out on sleep to the extent their parents think they are. This finding is consistent with those from Staples et al. (2015) who found that adherence to a nightly bedtime routine was not associated with actigraph recorded sleep duration at 30 months, but was at 36 and 42 months. That study did not include a measure of parent reported sleep duration, so comparisons between parent reported and actigraph recorded sleep duration can not be made.

In contrast, more routine activities completed each night and more deviation from the normal routine each night predicted more parent reported sleep duration on weeknights. At first glance, this seems paradoxical given that the more routine activities completed would be assumed to increase the length of the bedtime routine, which, as discussed, is actually associated with less parent reported sleep. However, it may be that parents who complete more bedtime routine activities each night believe that more activities or “steps” in the routine help to settle their child better so they fall asleep quicker, leading to less time awake in bed after lights out. Or it could also reflect a child who’s workng more cooperatively with the routine, allowing more steps in the allocated time, and greater sense of agency and security in the child. This does not explain why greater deviation from the normal routine predicts more parent reported sleep, unless it is the case that parents try out multiple bedtime routine activities when their children resist bedtime.

Only one of the actigraph sleep variables was predicted by the bedtime routine- sleep efficiency. More routine activities performed predicted higher sleep efficiency, which may indicate that the more steps in a bedtime routine do indeed help children fall asleep faster, leading to less time awake in bed, and thus higher sleep efficiency. However, to our knowledge no other research has been conducted examining the relations between the bedtime routine and sleep efficiency, and more research is needed to understand how bedtime routine activities impact sleep efficiency, including whether it is the total number of steps or if it’s the content of those steps (e.g. reading and snuggling versus bathing and brushing teeth) or affective qualities of the parent-child interaction that matter more.

Differences in the bedtime routine between weeknights and weekends also seemed to play a role in the prediction of nightly sleep. Weeknight was a significant predictor in all four bedtime routine models predicting parent reported sleep duration, as well as in the prediction of actigraph recorded sleep duration for the models including routine activity total and normal routine deviation. In addition, weeknight interacted with routine length variability to predict sleep efficiency, and normal routine deviation to predict parent reported sleep duration. Even at this young age, 30 month old toddlers seem to be experiencing differences in bedtime routines and sleep between weeknights and weekends, and part of this may be attributed to their parents’ work schedules. Parents who work a traditional 40-hour work week may be more regimented in bedtime routines during the week, and more lax in implementing the normal routine or a consistent bedtime on the weekends. This reasoning is supported by findings from Randler et al. (2012) who found that parents try to enforce regular bedtimes on weekdays, but would prefer their children to go to bed later and wake up later on the weekends.

Alternatively, parents may view the weekend as a “catch-up” period, where they allow their toddler to sleep in to catch up on sleep they may have missed during the week. Indeed, the mean differences seem to confirm this, showing that parents report their toddlers getting more sleep on the weekends. Interestingly, the objective measures of sleep via actigraphy showed no differences in total amount of sleep between weeknights and weekends; the only difference was for actigraph recorded wake minutes, indicating children had more wake minutes on weekends, consistent with findings from Iwata et al. (2012). So, toddlers may be spending more time in bed, albeit awake, leading to their parents reporting more sleep on the weekends, even though actual sleep is not increasing.

Some ancillary findings also warrant mention. While not hypothesized, we did explore whether gender, ethnicity, and income characteristics influenced the findings. We did find that girls were reported to have longer and more variable routine lengths than boys, and toddlers in families with higher income was related to lower parent reported sleep duration. Both findings were somewhat surprising given what is currently known from the literature base on sleep and bedtime routines in young children. Few studies have examined relations between gender and sleep outcomes in young children, and at this point, results are mixed. In a population-based longitudinal study of over 14,000 live births in England, Blair et al., (2012) found that while infant and toddler girls went to bed at the same time as boys, they consistently got five to ten more minutes of sleep at night than boys, a statistically significant difference. In contrast, an online survey of over 3,000 US and Canadian parents found no differences in sleep duration between boys and girls aged 0 to three (Sadeh et al., 2009).

However, the authors know of only one study to examining relations between child-level characteristics including gender and bedtime routines, the focus of the present study. In a sample of over 3,000 three-year-old children, Hale et al. (2009) found that in comparison to parent-level factors such as maternal race, age, and education, as well as household level factors including poverty status and number of adults and bedrooms in the house, child-level factors including age, birth weight, and gender were unrelated to bedtime routines. However, because that sample was comprised of primarily low-income and ethnic minorities, those findings may not generalize to the present study’s higher income, and primarily White sample.

We can offer a few suggestions as to why might girls have longer and more variable routine lengths than do boys. It may be that or parents perceive girls as needing or wanting more cuddle time during the bedtime routine. Perhaps the girls are a little more verbally adept, which allows them to manage parents more effectively. Or, parents perceive boys as being more difficult to settle, forcing parents to stick to a more regimented bedtime routine, thus reducing routine length variability. Alternatively, parents may perceive boys as needing more sleep than girls to protect against behavior problems during the day, and so parents may use a shorter routine with boys to ensure earlier sleep onset times. To better understand gender differences in bedtime routine lengths more research is needed, including possible qualitative examinations into parent perceptions of gender differences, sleep, and bedtime routines.

The finding that higher incomes were related to less parent reported sleep duration was unexpected, because in general, lower incomes are associated with shorter sleep duration. For example, in a study of over 2,000 Canadian children ages 1.5 to 5 years, Touchette et al. (2009) found that shorter sleep duration measured via parent report was associated with “insufficient income”, defined as a family spending 20% more of their income on food, shelter, and clothing than the average family. Another study of 493 toddlers aged 30 months found that SES, as measured by the Hollingshead Four-Factor Index (Hollingshead, 1975), was weakly but positively correlated (r = 0.16) with an actigraph sleep duration composite consisting of average sleep period, average duration of time in bed, and average minutes asleep in bed (Hoyniak, Bates, Staples, Rudasill, Molfese, D., & Molfese, V., 2018). Results may not be directly comparable however, because the present study used raw actigraph measures as opposed to an actigraph sleep composite, and a direct measure of household income as a proxy for SES, instead of a composite Hollingshead index, which takes occupational prestige and educational attainment into consideration. In the present study, which consists of primarily high income families, both parents are likely working full time jobs, and parental work schedules naturally put contrainsts on children’s sleep schedules, particularly what time children wake up in the mornings. Parents may perceive needing to wake their toddlers earlier than desired in order for parents to get to work at a specific time. Alternatively, in higher income families, there are more resources available for extra activities such as swim lessons or trips to children’s museums or zoos, ortechnology such as ipads or tablets, both of which may push bedtimes later if attended or used during the evening.

Implications

There are some important implications from the present study to consider. Some parents may not be adequately aware of or concerned about their toddler’s sleep, because their perception is that their toddler is sleeping more than they actually are. Given the importance of adequate and high quality sleep for the developing child, information about objectively-measured versus parent-reported sleep amounts, the importance of consistent nightly bedtime routines, and differences in toddler sleep between weeknights and weekends should all be shared with parents, pediatricians and other professionals working with children and families. Encouraging consistent bedtime routines and more awareness of their toddler’s sleep in general is an easy point of entry for pediatricians and other professionals into discussions with parents about their children’s development.

Limitations and Future Directions

Several limitations to the present study warrant mention. First, parents reported on both bedtime routines and sleep duration on the same sleep diary, which may have inflated relations between the two due to shared method variance. This is a concern given the consistent relations found between bedtime routines and parent reported sleep in comparison to actigraph recorded sleep measures in this study. However, we believe this limitation is attenuated by the fact that daily reports were taken and averaged across a two week reporting period, rather than using global reports of the bedtime routine and sleep duration (e.g. “how much sleep does your child get at night?”). In addition, if a parent perceives their child as not getting enough sleep, they may be inclined to alter the bedtime routine in some way to improve or increase their child’s sleep. Thus, when considering how parents implement the bedtime routine on a nightly basis, parent perception of their child’s sleep duration may be just as important as actual, objectively measured sleep duration.

Second, the way that variability in bedtime routines were aggregated in this study did not allow examination of individual or specific bedtime routine activities, such as reading or technology use, as they relate to sleep outcomes. Future research should examine individual routine activities to determine whether and how they may relate to sleep outcomes. Finally, information on whether children in this study went to childcare full time or part time during the week or stayed at home with a parent was not available for all children, and thus this important covariate could not be included when examining weeknight versus weekend differences in bedtime routines and sleep. It may be that children who do not attend childcare full time due to their parents work schedule do not have as variable bedtime routines or sleep patterns, because there’s no need for a “catch-up” period on the weekend. Conversely, it is plausible that children who do not attend childcare full time during the week may have more variable bedtime routines and sleep patterns because they are not constrained to wake up at a certain time each morning so that their parents can get to work on time. Future research on bedtime routines and sleep should include information on toddlers who attend childcare full time versus toddlers who stay at home with a parent.

Conclusion

This study adds important information about the nature of relations between bedtime routines and sleep measures in toddler children by utilizing daily reports and examining the night-to-night variability in bedtime routines. Bedtime routines and sleep measures in toddlers vary from night to night, and are influenced by a host of factors including day of the week and source of information (i.e. parent report versus objective actigraphy). Toddlers as young as 30 months old experience variability in their sleep between weeknights and weekends. For parents who are concernd about their toddlers sleep, keeping consistent bedtime routines even on the weekends is essential for optimal sleep outcomes.

Highlights.

• Toddlers experience inconsistent bedtime routines between weeknights and weekends

• Nightly differences in bedtime routines impact sleep on those nights

• Bedtime routines are most consistently associated with parent reported sleep