Time for bed! Earlier sleep onset is associated with longer nighttime sleep duration during infancy

Abstract

Objective/background:

Clinical recommendations include putting infants to bed using a consistent bedtime routine at an appropriate hour to promote longer nighttime sleep. Actigraphy was used in this exploratory study to examine how bedtime routines and nighttime sleep onset were associated with nighttime total sleep time (TST) and efficiency from 6 to 24 weeks of age.

Patients/methods:

Infants (n = 24) wore sleep actigraphs for three, one-week periods at 6, 15, and 24 weeks of age. Nighttime TST, sleep efficiency, sleep onset and offset were quantified. Mothers reported on infant bedtime routines using the Brief Infant Sleep Questionnaire at each age. Multilevel models examined between- and within-person associations.

Results:

As infants aged, sleep onset was earlier, and bedtime routines became shorter (p’s < 0.05). Infants fell asleep between 7 and 8:00PM on 24% of the nights. Most mothers (70%) reported that they often fed infants to sleep for the night. For every 1 h earlier in infants’ usual sleep onset, nighttime TST was 34.4 min longer that night (p < 0.01). Infants with earlier than usual sleep onset had slightly earlier sleep offset the next morning (8.4 min for every 1 h earlier in onset; p = 0.02). Between-person analyses showed similar patterns. Infants with a more consistent bedtime routine and who were not typically fed to sleep at bedtime had longer nighttime TST at 6 weeks, with a trend or no association at later ages.

Conclusion:

Infants who fell asleep earlier also slept longer at night. Keeping infants up later in hopes of them sleeping in longer may be counterproductive.

1. Introduction

Shorter sleep duration during infancy (<12 months of age) [1] is associated with adverse health outcomes, including greater adiposity, poorer emotion regulation, impaired growth, and later-life obesity [1–3]. In the first year of life, it is important to establish healthy sleep patterns including appropriate sleep duration, timing, and quality [4]. Infancy is a unique period of sleep development where circadian rhythms are being formed and sleep patterns are rapidly developing [5,6]. During this time, parents have a key role in shaping infants’ sleep patterns [7], and the parenting behaviors that occur at bedtime appear particularly important for influencing nighttime sleep patterns [8–10].

Previous studies have identified different aspects of sleep hygiene that contribute to improved nighttime sleep in young children [11–14]. This includes the use of consistent bedtime routines, which involves predictable, low-stimulus activities (eg, reading, bath) before lights go out and children fall asleep [11,12,15]. Implementing a consistent bedtime routine has been associated with numerous health outcomes that are well-described in a review by Mindell et al., [10] including longer nighttime sleep duration, shorter sleep onset latency, reduced nighttime wakings, and greater sleep consolidation at night [9,10,14–19]. For example, in the 2004 National Sleep Foundation Sleep in America Poll, caregivers who implemented a consistent bedtime routine reported their toddler slept ~1 h longer, compared to toddlers without a consistent bedtime routine [20]. Furthermore, bedtime routines act in a dose-dependent manner over time, such that children with a consistent bedtime routine during infancy and toddlerhood, rather than just toddlerhood, have better sleep outcomes [15], thus highlighting the need for bedtime routines to start in the first year of life.

Other factors that have been shown to promote longer, more consolidated sleep in young children include earlier nighttime sleep onset [21–25] and allowing children to fall asleep on their own without parental presence at bedtime [8,14]. For example, toddlers who went to bed before 9PM slept 78 min longer than those who went to bed later, and those who fell asleep without a parent present slept ~1 h longer than toddlers with a parent present [20].

Few studies have examined these associations during infancy, and those that do use parent-reported questionnaires. These subjective data show similar patterns to those described above, in that consistent bedtime routines during infancy are associated reduced night wakings, greater sleep continuity, and longer nighttime sleep duration [15,16,26]. Clinical recommendations suggest putting infants to bed at an age-appropriate hour, usually between 7 and 8:00PM [27,28]; however, a recent systematic review found that infants ≤6 months of age are often put to bed much later, usually between 7 and 11:00PM [29]. Similar to toddlers, the Sleep in America Poll showed that later sleep onset was associated with shorter nighttime sleep duration for infants, with 1.3 h less sleep at night when infants were put to bed after 9:00PM [20]. There are many reasons why infants may be put to bed later at night, including parent-driven factors (eg, setting later bedtimes, prolonging bedtime routines), as well as infant-driven factors (eg, difficulty in falling asleep). Further, some parents may be keeping their infants up later at night, in hopes that they sleep in later the next morning [30].

Previous research shows that infants should be put to bed while drowsy and awake, rather than being held or fed to sleep [28,30,31]. This is thought to foster the development of self-regulation during the night (or “self-soothing”), where infants fall back asleep on their own during nighttime wakings, rather than rely on their parents to soothe them back to sleep [32–34]. Previous studies support that putting infants to bed while still awake results in longer, more consolidated nighttime sleep [27,32–34].

As mentioned above, the existing data on bedtime practices and sleep outcomes during infancy are based on parent-report [35,36]. It is unclear if objective measures of infant sleep support parents’ perception. In recent years, actigraphy has been widely used in older children [37], with an increasing use of sleep actigraphy among infants that now allows these relationships to be explored more objectively [37,38]. Actigraphy can be used during infancy to better understand how parenting practices around bedtime are associated with objective measures of nighttime sleep duration and efficiency (ie, quality), as well as next-day sleep outcomes, including morning sleep offset and/or next-day daytime sleep. It is important to develop this work within the first months of life when parenting practices and infant behaviors are first established.

In this study, we used sleep actigraphy when infants were 6, 15, and 24 weeks of age to address three research aims. First, we describe infants’ bedtime routines, how infants are put to bed for the night, infants’ sleep schedule, and nighttime sleep duration over time. Second, we examine between- and within-person associations of infants’ nighttime sleep onset with nighttime total sleep time (TST) and sleep maintenance efficiency, as well as the next-day’s sleep offset and daytime sleep. We hypothesize earlier sleep onset will be associated with longer, more efficient nighttime sleep and will not be associated with next-day’s daytime sleep. Third, we explore if characteristics of bedtime routines and how infants are put to bed for the night are associated with nighttime TST and maintenance efficiency. We hypothesize shorter, more consistent bedtime routines, not feeding infants to sleep, and putting infants to bed while awake will be associated with longer, more efficient nighttime sleep.

2. Materials and methods

2.1. Participants and study design

Mothers were recruited during pregnancy in Obstetrics and Gynecology clinics, where nurses distributed study flyers to women in their third trimester. Additional recruitment took place in Labor and Delivery at a local hospital in Central Pennsylvania, where study flyers were given to mothers just before delivery. Interested mothers contacted the study team and were emailed a screening questionnaire to assess study eligibility using REDCap [39]. Both mothers and infants had to be eligible for enrollment. Inclusion criteria for mothers included being 18–35 years of age, with no major pre-existing health conditions (eg, cancer) or plans to move far (>50 miles) from the study area within the following six months. Infants had to be singleton, full-term at delivery (≥37 weeks’ gestation), birth weight ≥2500 g, <6 weeks of age at enrollment, and free from major pre-existing health conditions (eg, congenital heart disease). Prior to enrollment, informed consent was obtained and approved by the Penn State Institutional Review Board. In total, N = 24 mother-infant dyads were enrolled.

This study used a micro-longitudinal burst design, where data were collected each day for three, one-week periods (ie, “bursts”) when infants were 6, 15, and 24 weeks of age. This design allowed for developmental patterns to be characterized longitudinally over time, as well as on a daily level within each measurement burst. At each timepoint, mother-infant dyads attended an in-person lab visit; after this, sleep actigraphy was collected for one-week, and mothers completed online questionnaires using REDCap. At the six-week timepoint only, mothers also completed a demographic questionnaire.

2.2. Measures

2.2.1. Infant sleep actigraphy methods

During the in-person lab visit at the start of each measurement burst, infants were fitted with a sleep monitoring device (Actiwatch Spectrum; Philips-Respironics, Murrysville, PA) placed on their right ankle. This placement site was chosen to be consistent with previous research using actigraphs in infants [38,40]. Sleep devices were placed overtop a thin cotton/polyester sock to prevent skin irritation, while allowing the device to detect “on-ankle” time. Mothers were instructed to leave the device on their infants’ ankle throughout the day and night for one full week. At the end of the week, the devices were returned to the study team.

The sleep device was set to a medium sensitivity (wake threshold value = 40 activity counts for one-min epochs), and collected activity counts in 30-s epochs to be consistent with previous research in pediatric populations [37,38]. Our research team’s process of scoring these data is described in detail in Adams et al. [38] In brief, a SAS-based hierarchical algorithm methodology was applied to identify periods of sleep and wake [41,42]. The nighttime sleep interval was identified as the longest sleep interval that at least partially overlapped the hours of 10:00PM and 8:00AM. Two trained scorers (ELA, LM) then visually assessed periods of sleep and wake using a graphical interface. Scorers independently inserted, removed, or adjusted the sleep intervals to improve the scoring accuracy. The research team created an ordered set of decision rules to be applied when making adjustments [38]. For example, a valid day had to consist of >20 h of wear time, and a valid week had to have three or more valid days. After adjustments were made, a comparison between the two scorers records was conducted, and disagreements were discussed and resolved. A final data set was then established and approved by OMB.

2.2.2. Quantifying actigraphy variables

At each infant age, mothers reported when their infant typically fell asleep for the night and woke up for the day as part of the Brief Infant Sleep Questionnaire (BISQ; described in more detail below) [43]. This information was then used to guide actigraphy scoring. The longest sleep interval, starting closest to when infants typically fell asleep for the night, was identified as infants’ nighttime sleep interval [38]. The end of the nighttime sleep interval was informed by mother-reported typical infant wake time. Sleep onset was defined as the clock time in which infants’ nighttime sleep interval began, while sleep offset was defined as the clock time in which infants’ nighttime sleep interval ended. Nighttime TST was the amount of time infants spent asleep during the nighttime sleep interval (sleep onset to sleep offset). Nighttime TST was calculated as (nighttime sleep interval duration – wake after sleep onset [WASO]) during the nighttime sleep interval. The outcome variable of nighttime TST was used in order to quantify the time infants spent in a sleep state, after subtracting out periods of wakefulness, during nighttime sleep. Sleep maintenance efficiency (ie, quality) was calculated as the percentage of the nighttime sleep interval spent asleep, after accounting for WASO [44]. This was calculated as [(nighttime sleep interval duration – WASO during nighttime sleep)/nighttime sleep interval duration] x 100. Higher values indicate greater sleep maintenance efficiency. For example, sleep maintenance efficiency of 88% indicates that infants were in a sleep state for 88% of their nighttime sleep interval. Daytime sleep duration was calculated as the sum of all nap durations in a given day.

2.2.3. Infant bedtime routines and how infants were put to bed

Once each timepoint, mothers completed the validated BISQ to assess infants’ bedtime routines [43]. Open-ended questions on the BISQ asked about the start time of infants’ bedtime routines and the time infants were usually put to bed at night. Routine duration was calculated as (time infants were usually put to bed at night – start time of bedtime routines) and is reported in minutes. Mothers selected all activities that were usually done during infants’ bedtime routine from a list of developmentally appropriate response options (see Table 2). The sum of each mothers’ responses was used to represent the number of activities typically done during bedtime routines. Mothers also reported the single activity done last during infants’ bedtime routine and all activities done as infants fell asleep for the night, from a list of possible options (Table 2). Another question asked how often infants had a consistent bedtime routine, and similar to other studies [45,46], the response options were dichotomized as <5 nights/week and ≥5 nights/week. Last, mothers were asked how often infants were put to bed while awake, and response options were dichotomized as Usually/Always vs. Sometimes/Rarely/Never based on response distributions.

Table 2.

Percent of mothers (n = 24) who reported that each activity was part of their infants’ usual bedtime routine and typically occurred as infants fell asleep at 6, 15, and 24 weeks of age.

	All activities in bedtime routine (%)			Last activity in bedtime routine (%)			Occurred as infant fell asleep (%)
Infant age	6 wks	15 wks	24 wks	6 wks	15 wks	24 wks	6 wks	15 wks	24 wks
Gave bottle or nursed	91.7	91.7	87.0	20.8	20.8	34.8	79.1	70.9	73.9
Swaddled	83.3	66.7	26.1	12.5	12.5	4.4	62.5	50.0	17.4
Rocked	79.2	62.5	56.5	16.7	16.7	8.7	45.8	37.5	30.4
Cuddled/held	79.2	87.5	78.3	4.2	25.0	17.4	62.5	37.5	39.1
Gave pacifier	66.7	58.3	56.5	29.2	12.5	13.0	54.2	50.0	43.5
White noise	58.3	50.0	52.2	8.3	8.3	13.0	29.2	33.3	34.8
Bathed	41.7	54.2	56.5	0	0	0	–	–	–
Sang songs	37.5	33.3	30.4	0	0	4.4	–	–	–
Read books	25.0	50.0	52.2	4.2	0	0	–	–	–
Massaged	20.8	20.8	17.4	0	0	0	–	–	–
Said prayers	16.7	16.7	26.1	0	0	0	–	–	–
Played	12.5	45.8	47.8	0	0	0	–	–	–
Watched television	16.7	16.7	13.0	0	4.2	4.4	4.2	4.2	4.4
Listened to music	16.7	20.8	13.0	4.2	0	0	–	–	–
Put in swing/stroller	8.3	8.3	8.7	0	0	0	12.5	16.7	0
Had dinner/snack	0	4.2	8.7	0	0	0	–	–	–
In crib/bed alone	–	–	–	–	–	–	20.8	29.2	34.8
In own bed, parent in room	–	–	–	–	–	–	29.2	25.0	21.7
In parents’ bed, parent in room	–	–	–	–	–	–	12.5	8.3	8.7
Sucked thumb/finger	–	–	–	–	–	–	4.2	8.3	13.0
In other room of the house	–	–	–	–	–	–	4.2	0	0
In parents bed alone	–	–	–	–	–	–	0	4.2	0
Other	8.3	0	4.4	0	0	0	0	0	0

Response options differed for these 3 questions. Dashes indicate non-applicability, where that activity was not a possible response option for that question. Responses for “Other” included put infant in vibrating chair (n = 1; 6 weeks) and changed diaper (n = 2; 6 and 24 weeks). Response options for “all activities in bedtime routine” and “occurred as infant fell asleep” were check all that apply, so values sum to >100%. Response options for “last activity in bedtime routine” were select one, so values sum to 100%.

2.3. Statistical analysis

Aim 1 described infants’ bedtime routines and sleep patterns from 6 to 24 weeks of age. Separate mixed models were used to test the main effect of infant age on continuous dependent variables, including bedtime routine start time, duration, number of activities, sleep onset and offset, nighttime TST, sleep maintenance efficiency, and daytime sleep duration. The unit of analysis for actigraphy variables was an individual night, given that actigraphy measures were collected on multiple days, for each individual, at each timepoint. These models also accounted for the lack of independence between observations made repeatedly on the same individual by nesting day within timepoint. The unit of analysis for all BISQ variables was an individual infant, given this questionnaire was administered once at each infant age. Frequencies were used to describe each categorical variable including bedtime routine activities, consistency of bedtime routines, and whether infants were fed to sleep or put to bed while awake. Chi square tests of independence were used to evaluate if patterns of categorical variables differed at each infant age (6 vs. 15 vs. 24 weeks). These categorical variables were dichotomized as: consistency of bedtime routines (≥5 nights/week vs. <5 nights/week), typical routine duration >30 min (yes vs. no), put to bed while awake (Rarely/Never vs. Sometimes vs. Usually/Always), and usually feeding infants to sleep (yes vs. no).

Aim 2 used mixed models with lagged effects to account for the nested structure of these data [47], where 494 total daily observations were nested within three different timepoints for 24 infants. Intraclass correlations were assessed to verify model selection, and the magnitude of these correlations (ICC = 0.39) indicated high correlation within clusters (ie, individuals), thereby justifying our use of multilevel models with random intercepts for these analyses. These models were used to test the temporal associations between nighttime sleep onset (independent variable) and nighttime TST, sleep maintenance efficiency, next-day sleep offset, and daytime sleep duration (dependent variables). Variances for key measures were modeled as within-person (level 1) and between-person (level 2) levels. For within-person effects, each infants’ nighttime sleep onset was compared to his/her personal average. Person-specific averages were tested as the predictor on the dependent variables described above. Between-person effects were derived by comparing infants’ personal mean across days at each timepoint to the sample mean at that timepoint. All models tested for an interaction between nighttime sleep onset and infant age. When not significant, the interaction term was removed from the model and infant age was retained as a covariate. All models also used a Restricted Maximum Likelihood estimation [48] method and a Kenwood-Roger adjustment [49–51] to correct for potential biased level-2 variance estimates and alleviate inflated type I error rates, given the small sample size of individuals used in these models. Studentized residuals were used to examine potential outliers, as indicated by values greater than three (absolute value). Models were run with and without these data points, which did not appear to have high influence on the regression coefficients and significance values; therefore, no outliers were removed in these analyses.

For Aim 3, a similar mixed model framework was used to explore how aspects of bedtime routines were associated with nighttime TST and sleep maintenance efficiency. Bedtime routine consistency, length, feeding infants to sleep, and putting infants to bed awake were independent variables, while nighttime TST and sleep maintenance efficiency were dependent variables. All models accounted for day nested within timepoint for each infant. Interactions between the independent variables and infant age were tested and removed when not significant. All analyses were performed in SAS Version 9.4, and significance was defined a priori as p < 0.05. Data are presented as mean ± standard deviation or mean [95% confidence interval].

3. Results

Infants were mostly non-Hispanic/Latino (100%; n = 23), White (79.2%; n = 19), and just over half of the sample was male (58.3%; n = 14). Average gestational age at birth was 39.6 ± 0.9 weeks, and birth weight averaged 3679.5 ± 418.6 g. At six weeks of age, 82.6% of infants were predominantly breastfed (ie, ≥80% of feeds were breastmilk, either at the breast or by bottle). Infants’ mothers were mostly married (95.7%; n = 22), college educated (95.8%; n = 23), working full- or part-time (66.7%; n = 16), and in households earning ≥$50,000/year (75%; n = 18).

3.1. Description of infant bedtime routines, how infants were put to bed, and nighttime sleep

3.1.1. Bedtime routines

Infants’ bedtime routine at each age is described in Table 1. Bedtime routines started between ~7:35–8:25PM and got earlier with age (p < 0.01). At six weeks, 50% of infants had a consistent bedtime routine on ≥5 nights/week, which increased to 87% of infants at 24 weeks (p < 0.01). Routines lasted between ~35 and 55 min and got shorter over time (p < 0.01). Routine duration exceeded 30 min on most nights at 6 weeks of age and decreased to about half of all nights at 15 and 24 weeks of age (Table 1). Bedtime routines consisted of 6.7 ± 2.1 (range 2–11) different activities; all activities, and the activity done last, are listed in Table 2. Some common activities included feeding, swaddling, rocking, and cuddling. The least common activities included putting infants in a swing and having dinner or a snack. About 30% of infants were usually/always put to bed while awake, which did not differ by infant age (p = 0.17). As infants fell asleep, they were most often being breastfed, swaddled, held, and/or given a pacifier (Table 2). Most mothers (~70%) reported that infants were often fed to sleep (either by breast or bottle), which did not differ by infant age (p = 0.47).

Table 1.

Description of infants’ bedtime routine and nighttime sleep interval at 6, 15, and 24 weeks of age.

Infant age	6 wks	15 wks	24 wks	p value
Bedtime routines
Start time	8:25PM [8:01–8:49]	7:51PM [7:27–8:15]	7:34PM [7:10–7:58]	<0.01
Duration (min)	55.9 [46.0–65.8]	45.8 [36.0–55.6]	35.2 [25.4–45.0]	<0.01
Nights where duration was >30 min (%)	76.5	51.2	46.4	<0.01
Activities (n)	6.6 [5.8–7.4]	6.9 [6.1–7.7]	6.4 [5.6–7.2]	<0.01
Usually/always put to bed while awake (%)	16.7	33.3	39.1	0.17
Consistent routine ≥5 nights/wk (%)	50	83.3	87.0	<0.01
Nighttime sleep interval
Sleep onset	9:35PM [9:14–9:56]	8:41PM [8:22–9:01]	8:19PM [7:59–8:39]	<0.01
Sleep offset	7:53AM [7:34–8:12]	7:29AM [7:11–7:47]	7:12AM [6:54–7:31]	0.01
Nighttime TST (hrs)	8.6 [8.2–8.9]	8.9 [8.6–9.2]	9.0 [8.7–9.3]	0.14
Sleep maintenance efficiency (%)a	83.2 [81.3–85.2]	82.5 [80.8–84.3]	83.1 [81.3–84.9]	0.80

Values are presented as mean [95% confidence interval].

^aSleep maintenance efficiency was calculated as ((nighttime sleep interval duration – wake after sleep onset during nighttime sleep)/nighttime sleep interval duration) x 100. Higher values indicate greater sleep efficiency (ie, quality). TST = total sleep time. P-values for continuous variables (start time, duration, activities, sleep onset, sleep offset, nighttime TST, sleep maintenance efficiency) are from regression models testing the main effect of infant age on these dependent variables. P-values for categorical variables (nights where duration was >30 min, usually/always put to bed while awake, consistent routine ≥5 nights/week) are from chi square tests of independence testing differences by infant age.

3.1.2. Nighttime sleep

Infants’ nighttime TST and sleep maintenance efficiency are described in Table 1. At six weeks of age, sleep onset occurred at ~9:35PM and got earlier with age (p = 0.01). On about one-fourth (24%) of the nights where actigraphy data were recorded, infants fell asleep for the night between 7 and 8:00PM, as recommended. At six weeks of age, sleep offset occurred at ~7:55AM and also got earlier with age (p = 0.01). Nighttime TST averaged 8.8 ± 1.3 h/night, and sleep maintenance efficiency averaged 82.7 ± 7.2%, both of which did not vary with age (p’s > 0.05).

3.2. Between- and within-person associations of sleep onset and nighttime TST

3.2.1. Between-person associations

Infants with an earlier nighttime sleep onset tended to have longer nighttime TST [24.6 min (16.1–33.1) for each 1 h earlier, p < 0.01; Table 3], with no difference in nighttime sleep efficiency [0.7% (−0.2–1.7); p = 0.14; Table 3]. The relationship between nighttime sleep onset and nighttime TST is illustrated in Fig. 1. Further, infants with an earlier nighttime sleep onset had earlier sleep offset the next morning; for every 1 h earlier in sleep onset, sleep offset was 25.2 min (15.6–34.8) earlier the next morning (p < 0.01; Table 3). Earlier sleep onset was not associated with daytime sleep duration the next day (p = 0.28; Table 3).

Table 3.

Results for multilevel models testing the temporal relationships between nighttime sleep onset predicting nighttime sleep variables and next-day daytime sleep duration.

	TST (min)		Nighttime sleep				Daytime sleep
			Sleep maintenanceefficiency (%)		Sleep offset (decimal hours)		Next-day sleepduration (min)
	B	(SE)	B	(SE)	B	(SE)	B	(SE)
Intercept	529.69	8.05	82.77	0.87	7.55	0.14	164.05	8.38
Nighttimesleep onset
Between-person	−24.61*	4.30	0.71	0.48	0.42*	0.08	5.95	5.44
Within-person	−34.40*	3.17	1.21*	0.36	0.14*	0.06	−3.48	4.39

Data were from 494 total daily observations clustered within 3 timepoints for 24 infants. Results express the between- and within–person relationship of nighttime sleep onset with nighttime sleep variables and next-day daytime sleep duration. Nighttime sleep onset was centered at midnight (0 = 00:00). TST = Total Sleep Time.

^*p < 0.05.

Fig. 1.

Scatterplot of raw data for each night of sleep actigraphy collected in n = 24 infants for 1-week at 6. 15, and 24 weeks of age (resulting in 494 total observations). When infants’ nighttime sleep onset was earlier, then average nighttime total sleep time was longer (p < 0.01). Sleep onset was the clock time in which infants’ nighttime sleep interval began (ie, fell asleep for the night). Nighttime total sleep time was the time spent asleep during the nighttime sleep interval, calculated as (nighttime sleep interval duration – wake after sleep onset).

3.2.2. Within-person associations

On nights when sleep onset was earlier than usual, nighttime TST was 34.4 min (28.2–40.6) longer and 1.2% (0.5–1.9) less efficient that night (p’s < 0.01; Table 3). Nighttime sleep onset was associated with morning sleep offset, such that for every 1 h earlier than usual sleep onset, infants woke 8.4 min (1.8–15.6) earlier the next morning (p = 0.02; Table 3). Nighttime sleep onset was not associated with next-day daytime sleep duration (p = 0.43; Table 3).

3.3. Associations between bedtime routines, how infants were put to bed, and nighttime sleep

The associations between bedtime routine consistency and feeding infants to sleep on nighttime TST differed by infant age (p’s < 0.05; Fig. 2). At 6 weeks, infants with a consistent bedtime routine ≥5 nights/week, and infants not usually fed to sleep, had a longer nighttime TST [37.0 min (13.8–60.1) and 34.8 min (9.4–60.1) longer, respectively] compared to infants with a consistent bedtime routines <5 nights/week and infants usually fed to sleep (p ≥ 0.01). There was a trend for this same pattern at 15 weeks of age, with no association at 24 weeks of age (Fig. 2). Bedtime routine consistency and feeding infants to sleep were not associated with sleep maintenance efficiency at any age (p’s > 0.05). Further, bedtime routine length and putting infants to bed still awake were not associated with nighttime TST or sleep maintenance efficiency at any age (p’s > 0.05).

Fig. 2.

At 6 weeks, infants A) with a consistent bedtime routine ≥5 nights/week and B) not usually fed to sleep, had greater nighttime total sleep time. There was a trend for these associations at 15 weeks, and no associations at 24 weeks. *p < 0.05; #0.05 ≤ p ≤ 0.06. Data are from a sample of n = 24 infants ages 6–24 weeks. Nighttime total sleep time is time spent asleep during the nighttime sleep interval (nighttime sleep interval duration minus wake after sleep onset) collected using actigraphy. Routine consistency and feeding to sleep were mother reported on the Brief Infant Sleep Questionnaire.

4. Discussion

This study explored associations between aspects of bedtime routines and actigraph-obtained sleep patterns during infancy. Results showed that infants’ nighttime sleep onset was often late and occurred between 7 and 8:00PM on only one-fourth of the nights where actigraphy data were recorded. Infants who fell asleep earlier at night had a longer nighttime sleep duration, without influences on the next-day’s daytime sleep. Earlier sleep onset was also associated with a slightly earlier wake time the next morning, and these patterns were seen when observing both between- and within-person associations. Infants’ bedtime routines lasted 35–55 min on average. Only about one-third of infants were typically put to bed while still awake and feeding infants to sleep was common at all ages. Infants with more consistent bedtime routines and not typically fed to sleep had longer nighttime sleep durations at 6 and 15 weeks of age.

Infants in our sample fell asleep for the night between 7 and 8:00PM, as recommended, on about one-fourth of the nights where actigraphy data were collected. Instead, infants’ sleep onset was typically later, both in our sample (8:20–9:35PM) and in previous research such as the nationwide Sleep in America Poll (9:11PM average infant bedtime) [52]. Given this, it is important to discuss our findings in the direction of later sleep onset and subsequent sleep. Infants who fell asleep later also slept less at night, while daytime sleep the next day was unaffected. Our findings support similar patterns observed in toddlers and preschool-aged children [21–25], as well as a study with infants where parent-reported measures of sleep were obtained [29]. Future research is needed to fully understand infant-driven factors versus parent-driven factors that influence infants’ sleep schedule. This should include why parents put their infants to bed when they do, and their reasons and intentions for later bedtimes. One common myth is that some parents put their infant to bed later in hopes of them sleeping in later the next morning [53]; and while our data support this relationship, we found that for every 1 h later in infants’ usual sleep onset, infants woke only 8 min later the next morning. Thus, it does not appear advantageous to keep infants up later for this reason alone. Other plausible reasons for a later bedtime include difficulty getting infants to sleep, bedtime resistance, and overtiredness [20,31]. In any case, the later infants are up at night, the less sleep they will likely get that night, without compensating by sleeping longer the next day. Recent evidence using actigraphy also suggests that more frequent exposure to late sleep is associated with increased adiposity in 2 to 6-year old children [54]; thus, future studies should examine nighttime sleep onset and growth measures, starting in infancy. Another important area for future research is to examine these associations among infants from different cultures, given that sleep patterns differ substantially across cultures [45,46]. For example, our sample was predominantly Caucasian, and children from predominantly Asian countries tend to have later bedtimes, shorter total sleep times, and are less likely to fall asleep independently at night [45,46]. It should also be noted that in these data, nighttime sleep maintenance efficiency was greater when infants’ sleep onset was later than usual, suggesting higher sleep quality; yet, the magnitude of change was minimal (~1% per hour later). Collectively, these results support an earlier bedtime, which appears advantageous for infants’ overall sleep duration.

Another predictor of longer nighttime sleep duration included infants who had a more consistent bedtime routine at 6 and 15 weeks of age. A large body of evidence shows the benefits of consistent bedtime routines on child sleep outcomes [13–17], as well as emotional and behavioral regulation, family stress and functioning, language development and cognition, and parent-child bonding [13]. Our findings demonstrate these associations during early infancy using objective measures of nighttime sleep duration and justify the importance of establishing a consistent bedtime routine within in the first months of life. Forming a bedtime routine this early may give infants the structure and predictability around bedtime that is needed to reduce nighttime struggles and bedtime resistance in later childhood; yet, larger longitudinal studies are needed to examine these relationships across development and explore the potential long-term benefits. It is unclear why an association between consistent bedtime routines and nighttime sleep duration was not observed at 24 weeks in our sample; yet, perhaps the small sample size of this study, and the fact that almost all infants had a consistent bedtime routine at this age, limited the variability to detect differences. We suggest future work examine this association in a larger sample before firm conclusions are drawn.

Bedtime routines may be maladaptive if they are too long or consist of high-stimulus activities (eg, television viewing, running around). Clinical recommendations suggest engaging in low-stimulus activities that last no more than 30 min total in duration [11,12,15]. Fortunately, in this sample, low stimulus activities such as feeding, rocking, cuddling, and swaddling were common, while high-stimulus activities, such as putting infant in a swing and playing, were rarely done. Different patterns may be seen for families of lower socioeconomic status and racial/ethnic minority infants, given the evidence that older children in these populations exhibit more maladaptive bedtime activities [55]. Infants’ average bedtime routines lasted just under an hour at 6 weeks of age and decreased to ~35 min by 24 weeks of age. An important area for future research is to better understand the ideal length of bedtime routines; yet, given the previous suggestions to keep bedtime routines at no more than 30-min, our data show that infants routines exceed this duration at all ages. It may be possible that longer bedtime routines were due to more infant-driven behaviors (eg, length of feeding, high amount of fussiness), and future research is needed to disentangle how parent versus infant-driven behaviors influence the length of bedtime routines.

Putting infants to bed while awake is recommended so that infants to learn to fall asleep on their own [28,30,31]. This includes not feeding infants as they fall asleep for the night. Previous research has shown that parental presence at bedtime and feeding infants to sleep is associated with sleep disturbances during the night [26,32–34], while infants who fall asleep independently may have a greater ability to self-soothe during nighttime wakings. Our results showed that infants who were not typically fed to sleep at bedtime ended up sleeping longer at 6 and 15 weeks of age, but not at 24 weeks of age. Further, putting infants to bed while awake was not associated with nighttime sleep duration. There are a few possible reasons for these non-significant associations that are contrary to our hypotheses as well as the past literature. As described above, the small sample size of this study may have limited the power and variability needed to identify differences in these behaviors. The second reason may be due to difference in objective versus subjective measures of infant sleep. When measured objectively, infants’ nighttime sleep duration may not differ in relation to them being put to bed asleep or awake; yet, infants put to bed awake may have more occurrences of self-soothing during the night. Parents are often unaware of these self-soothing episodes and therefore over-estimate infants nighttime sleep [56–58]. To further explore these associations, objective measures of self-soothing during nighttime wakings are needed. Future work from this study will examine if feeding infants to sleep and putting infants to bed while awake are associated with proxy measures of infant self-soothing using actigraphy. Third, not feeding infants to sleep has been associated with increased nighttime sleep consolidation [33,34], and while these actigraphy data were not scored in such a way to quantify nighttime sleep consolidation, future studies should consider doing so to explore associations between bedtime parenting practices, such as feeding infants to sleep, and actigraphy-derived nighttime sleep consolidation. Lastly, parents’ intentions for how they put their infants to sleep (eg, while feeding or while awake) are needed, in order to better understand their motivations around these bedtime practices. With a high percentage of infants typically fed to sleep in this study, it would be intriguing to know parents’ reasons for doing so.

The largest strength of this study includes the objective data collected using sleep actigraphy in the first months of life. Given the novelty of using sleep actigraphy during infancy, and the intensive nature of establishing data scoring procedures for this population [38], it was important to first pilot these methods in a small sample. The use of actigraphy data on the daily level allowed for almost 500 observations across the 24 participants, thus reducing the effects of this potential limitation. However, given the small number of participants, our findings should be interpreted as initial exploratory patterns, to be later tested with a larger, more diverse sample, particularly when using multilevel modeling. Mothers in this study were highly educated and mostly White, thus limiting the generalizability of our findings to more racially diverse and less educated families. Another limitation of this study includes the specificity of the BISQ questionnaire. The BISQ is a widely used, validated measure [43]; yet, the questions and response options limit the type of data that can be collected. For example, mothers could select multiple activities that were done as infants fell asleep for the night, in a “check all that apply” format, and putting infants to bed while awake was rated on a Likert scale ranging “Never” to “Always”. These response options do not get at which singular activity was done as infants fell asleep or identify specific nights when infants were/were not put to bed awake. On the other hand, mothers had to select the one activity that was usually done last during infants’ bedtime routine, when multiple activities could have been done last simultaneously (eg, feeding while rocking). The questions also do not distinguish between bedtime routines on weekdays versus weekends. Future studies should consider methods such as a daily diary, where mothers record bedtime practices each night. With this, associations can be tested on a daily level and weekday versus weekend distinctions can be made.

5. Conclusion

This study is among the few to use actigraphy during early infancy to describe patterns of infant sleep and associations with bedtime routines. Our findings demonstrate that infants who fell asleep earlier at night also slept longer and keeping infants up later resulted in only minimally later wake-times the next morning. As such, infants should be put to bed at an age-appropriate hour to promote longer nighttime sleep and keeping infants up in hopes they will sleep in appears ineffective. Further, our findings support the benefits of a consistent bedtime routine and not feeding infants to sleep, particularly in the earliest months of life. Future research that uses actigraphy during infancy in larger, more diverse samples is encouraged to further explore these associations and the many possible influential factors.

Abbreviations:

TST

total sleep time

WASO

wake after sleep onset