Sleep arrangements, parent-infant sleep during the first year, and family functioning

Abstract

The present longitudinal study addressed the ongoing debate regarding the benefits and risks of infant-parent co-sleeping by examining associations between sleep arrangement patterns across the first year of life and infant and parent sleep, marital and family functioning, and quality of mothers’ behavior with infants at bedtime. Patterns of infant sleep arrangements across the infants’ first year were derived from information obtained from 139 families at 1, 3, 6, 9, and 12 months of infant age in a U.S., central Pennsylvanian sample. Linkages between these patterns and parent-infant sleep, marital and coparenting stress, and maternal behavior at bedtime (from video-recordings) were assessed. Compared to families whose infants were solitary sleepers by 6 months, persistent co-sleeping was associated with sleep disruption in mothers but not in infants, although mothers in persistent co-sleeping arrangements reported that their infants had more frequent night awakenings. Persistent co-sleeping was also associated with mother reports of marital and coparenting distress, and lower maternal emotional availability with infants at bedtime (from home observations). Persistent co-sleeping appeared to be a marker of, though not necessarily a cause of, heightened family stress, although the present design did not enable strong tests of causal processes, and results may be particular to cultures that are not supportive of co-sleeping. Findings are discussed in terms of cultural contexts of infant sleep and the need for further investigations into the role of the health of the family system in influencing how parents structure infant sleep.

The manner in which sleep is structured for infants across the first year, the stability vs. fluidity of sleep arrangements, and whether particular sleep arrangements place infants or parents at risk, are topics of considerable controversy and debate. Part of this debate is focused on the benefits vs. risks of solitary vs. co-sleeping, particularly bedsharing, for infant survival. Although bedsharing may be practiced for many reasons (Salm Ward, 2015), the American Academy of Pediatrics (2005) has warned against bedsharing because of its link with increased rates of Sudden Infant Death Syndrome (SIDS) and other sleep-related infant deaths, such as death due to suffocation (Moon, 2011; Schnitzer, Covington, & Dykstra, 2012). This work is controversial. Some (Blair et al., 2009; Blair, Sidebotham, Pease, & Fleming, 2014) have argued that the bedsharing-SIDS link is reduced and even eliminated in the absence of specific infant and environmental risk factors, which include preterm or ill infants, parents who smoke or use alcohol, infant swaddling, co-sleeping with infants on a sofa, placing infants in prone (rather than supine) sleeping positions, and surrounding infants with bedding material (e.g., pillows) or other soft objects that can possibly occlude the airway. Others, by contrast, argue that bedsharing poses SIDS risks even with all risk factors controlled (Carpenter et al., 2013). Still others have argued that early bedsharing, if safely practiced, may confer developmental advantages for infants. McKenna and colleagues (McKenna, Ball, & Gettler, 2007; McKenna & Dade, 2005) and Blair, Heron, & Fleming (2010) demonstrated that co-sleeping is linked with breastfeeding, which has important health benefits to infants (Ball, 2003; Godfrey & Meyers, 2009). In addition, co-sleeping may protect against Sudden Infant Death Syndrome (SIDS) by increasing the time infants and mothers spend in lighter sleep states, reducing time spent in deeper sleep states, and increasing the frequency of synchronized mother-infant arousals, which promotes maternal monitoring the infant during the night (McKenna et al., 2007).

At the same time, children who co-sleep are reported by parents to experience increased night waking, compared to children who sleep in separate rooms (DeLeon & Karraker, 2007; Ramos, Youngclarke, & Anderson, 2007). This appears to be true even in cultures (e.g., China) that are supportive of co-sleeping (Tan, Marfo, & Dedrick, 2009; Wang et al., 2013) and has raised concerns because of associations between night waking and socio-emotional, behavioral, and attentional problems (Cortesi, Fiannotti, Sebastiani, Vagnoni, & Marioni, 2008; Wang et al., 2013). Further, parents who co-sleep with infants beyond the first few months of life appear to be at risk for marital discord and personal distress (Cortesi et al., 2008; Countermine & Teti, 2010), which themselves are known risk factors for socio-emotional and behavioral problems in young children (Teti & Gelfand, 1990).

Thus, whereas early co-sleeping, and in particular bedsharing, carries health benefits to the infant if safely practiced, persistent co-sleeping in cultures in which solitary sleep is the norm by the end of the infants’ first year has been associated with family dysfunction and infant sleep disruption, both of which are predictive of cognitive and social developmental problems in early childhood. Importantly, it is unclear, from this work, whether persistent co-sleeping plays a causal role in any of this, or whether it is symptomatic of pre-existing family problems. There is some evidence for the latter perspective. For example, Teti et al. (2015) found that coparenting distress when infants were one month of age was associated with co-sleeping that persisted across the infants’ first six months of life, whereas positive coparenting at one month predicted a move out of co-sleeping arrangements and into solitary sleeping arrangements by six months. Further, Teti and Crosby (2012) reported that elevated depressive symptoms and worries about infant night waking in mothers appeared to lead mothers to seek out and spend more time with their infants at night, independent of any sign of infant distress, and that this in turn was predicted of mothers’ reports of increased infant night waking.

The present longitudinal study attempted to shed further light on linkages between sleep arrangement use across the first year post-partum, infant-parent sleep, and family functioning, in a U.S., central Pennsylvania sample. It expands upon earlier work in several ways. First, unlike previous studies, which have mostly been single-point-in-time or short-term longitudinal studies involving small samples, the present study examines linkages between infant sleep arrangements, sleep quality, and family functioning across five different age points in infants’ first year of life in a large, community-based sample. Second, we capitalized on the fluidity of sleeping arrangements (McCoy et al., 2004) to identify subgroups of families with different patterns of sleep arrangements used across the first year [e.g., consistent solitary sleep, early co-sleeping that switched to solitary sleep by 6 months (early switchers), co-sleeping persisting beyond 6 months but switched to solitary sleep between 9 and 12 months (late switchers), consistent co-sleeping across the first year, and inconsistent sleep arrangements]. This enabled us to examine how different first-year sleep arrangement patterns were linked with stability and change in parent and infant sleep quality and marital and coparental functioning across this same time. Third, whereas prior work focused primarily on relations between sleep arrangements and mothers’ reports of infant sleep, the present study assessed linkages between sleep arrangement patterns and mothers’ and fathers’ sleep as well, using both maternal report and objective assessments of sleep (actigraphy). Studies examining relations between infant sleep arrangements and both infant and parent sleep using multiple methods to assess sleep are rare. In a recent study that used both maternal self-report and actigraphy assessments of sleep, Volkovich, Ben-Zion, Karny, Meiri, and Tikotzky (2015) found that whereas mothers in cosleeping arrangements reported more night awakenings in their infants than mothers in solitary sleeping arrangements, actigraph recordings indicated that it was actually the mothers, but not infants, in cosleeping arrangements who experienced more sleep disruption compared to mothers in solitary sleep arrangements. It was of interest to determine if Volkovich et al.’s (2015) findings would be replicated in the present study, and if the additional information we collected on marital and coparenting distress and bedtime parenting quality might provide some insight about why different results are obtained about links between sleep arrangements and infant sleep when using maternal reports vs. actigraph reports of infant sleep.

The following hypotheses were assessed:

1. Consistent with earlier reports, co-sleeping would be associated with mothers’ reports of increased infant night waking. Relevant to this question, however, was whether increases in mother-reported infant night waking would be corroborated by actigraphy, and whether co-sleeping would also be associated with increased wakefulness and sleep fragmentation among parents. Volkovich et al.’s (2015) findings suggest that mothers’ reports of infant sleep will not be corroborated by actigraph reports, and that actigraph data would indicate that it will be mothers’, not infants’ night wakefulness, that will be increased in co-sleeping arrangements. We will determine if these patterns are replicated in the present study, and whether paternal sleep, which was not examined by Volkovich et al., is linked with sleep arrangements.

2. Co-sleeping will be associated with marital and coparenting distress, but particularly so when co-sleeping persists beyond 6 months of infant age. This hypothesis is based on data suggesting that co-sleeping in the U.S. is more accepted during the first six months (McCoy et al., 2004).

3. To be explored as well was the extent to which co-sleeping was associated with the emotional quality of parenting at bedtime across the infant’s first year. Directional hypotheses concerning these associations were withheld because no prior data exist on this question. We nevertheless examined this question in light of expectations that persistent co-sleeping would be associated with sleep loss and marital and family distress, which could reduce one’s capacity for effective parenting with the infant.

The present study was not experimental. It was not possible to determine if any relations obtained between sleep arrangements and the outcomes examined in this study were caused by sleep arrangement differences. Causality attributed to sleep arrangements may be most clear for linkages between sleep arrangements and parent-infant sleep, because sleep arrangements are the proximal contexts in which sleep quality is assessed. Beyond that, however, there is no theory, nor empirical data from prior work, to support the expectation that sleep arrangement differences would cause differences in marital adjustment, coparenting quality, and bedtime parenting. For those outcomes collected at multiple points across the first year, however, support for causality would be suggested if changes in these variables over time co-occurred in expected ways with changes over time in sleep arrangements. For example, if, in early switching families, coparenting improved between 1 and 6 months of infant age, but did not do so in late switching families until after 6 months, that could be taken as imperfect evidence of a causal link between sleep arrangements and coparenting quality. We will be examining such change patterns in all analyses.

Methods

One hundred sixty-seven families and their 1-month-old healthy infants were recruited to participate in a larger longitudinal NICHD-funded study (SIESTA – Study of Infants’ Emergent Sleep TrAjectories) (R01 HD052809, awarded to the first author) of parenting, infant sleep, and infant development across the infants’ first two years. In the present study, we focused on information collected from all first-year home visits, which took place when infants were 1, 3, 6, 9, and 12 months. Mothers were first approached in two central Pennsylvania hospitals within 24–48 hours after delivery by a project staff member, who described the study and provided contact information. Interested mothers were called 2-to-3 weeks after infant discharge, and an initial home visit was scheduled when infants were 4-to-6 weeks of age.

Of the 167 families recruited, 18 withdrew from the study between 1 and 12 months of infant age, and thus there were 149 families in the final study sample. The 18 families who dropped were compared with the 149 participants who remained in the study on all socio-demographic measures and study variables collected at 1 month, using oneway analysis of variance (ANOVA), chi-square and Fisher Exact Probability tests. No differences were found between dropouts and completers on socio-demographic measures, with two exceptions: Dropouts were more likely to be non-White [35% vs. 14%, Χ² (1) = 4.96, p = .026] and less likely to breast feed at 1 month [50% vs. 80%, Χ² (1) = 8.40, p = .004] than completers. Dropouts did not differ from completers in marital adjustment, positive and negative coparenting, and bedtime parenting quality. In addition, no differences were found between dropouts and completers in the sleep arrangements used with infants at 1 month.

There were 80 female and 69 male infants in the 149 families who completed the study through 12 months. Ninety-five percent of parents were married or living with a partner. Thirty-seven percent of mothers were primiparous. Eighty-six percent of mothers and 85% of fathers were White, with the remaining evenly split between African American, Asian American, Latino, or “Other”. The sample was fairly well-educated. Ninety-nine percent of mothers had completed high school, and 60% of mothers had a bachelor’s degree or higher. Eighty-six percent of fathers had completed high school, with 61% completing a bachelor’s degree or higher. Ninety-five percent of fathers and 62% of mothers were employed full or part-time at 1 month; 98% of fathers and 65% of mothers were employed full or part-time at 12 months. Mothers who were employed at 1 months were, by and large, the same mothers who were employed at 12 months (89%). Median yearly family income was $60,000. Eighty percent of mothers were breastfeeding their infants, either full or part-time, at 1 month of age. That dropped to 33% by 12 months.

At 1 month, 62% of employed mothers were on maternity leave. This dropped to 18% at 3 months, to 4% by 6 months, and to 2% by 12 months. To test whether returning to work may have negatively impacted mothers’ bedtime parenting quality or reports of positive and negative coparenting, oneway ANOVAs were conducted comparing employed mothers who had returned to work with employed mothers who had remained on maternity leave at 1 and 3 months, the age points at which sample sizes for mothers who were on maternity leave were large enough for meaningful group comparisons. No differences between mothers who were and who were not on maternity leave at 1 and 3 months were found for mothers’ emotional availability at bedtime and for mothers’ reports of positive and negative coparenting. These analyses suggested that, in the present sample, the return to work did not compromise mothers’ quality of parenting with their infants or their coparenting relationships.

Overall Procedure

Home visits were conducted when infants were 1, 3, 6, 9, and 12 months of age. At each age point, family assessments were done across seven consecutive days, with some assessments occurring once and others across multiple days at each age point. Once at each occasion, mothers and fathers completed questionnaire measures pertaining to their infants’ sleep arrangements, marital adjustment, and coparenting quality. In addition, an observational assessment of the quality of parental behavior during infant bedtimes was made from a video-recordings made during one night at each age point. Infant and parent sleep quality was assessed daily across seven consecutive days in the assessment week using actigraphy for mothers, fathers, and infants, and a mother-reported sleep diary for infants.

Measures

Socio-demographics

Parents at recruitment completed a socio-demographic questionnaire that assessed educational attainment, yearly family income, parental age, partner status (live-in vs. no live-in partner) and family size. In addition, mothers provided information in Keller and Goldberg’s (2004) Sleep Practices Questionnaire (SPQ) about whether they thought they had adequate sleeping space for everyone in their household. That specific SPQ item was, “We have limited sleeping space”, to which mothers responded on a five-point scale (1 = “Not at all, 5 = “Definitely”). At each age point, mothers also indicated whether they were breast feeding their infant, and if so, whether it was part- or full-time.

Infant sleep arrangements

One item in the Sleep Practices Questionnaire (SPQ) (Goldberg & Keller, 2007) was used at each age point to obtain information from parents about their infants’ sleep arrangements. Mothers responded to the SPQ item: “Where does your baby usually sleep at night?” At each age point, four sleep arrangement categories were scored: Solitary sleep (infant slept in a separate room), room sharing (infant slept in the same room as parents, but on a separate sleeping surface), bed sharing (infant slept in the same bed as the parents), and combination (infant’s sleep arrangement varied across the night between solitary sleep, room sharing and/or bed sharing). Complete sleep arrangement data were available on 139 families through 12 months, and these data were aggregated across all five age points to create five sleeping arrangement categories across the first year: Consistent solitary sleep (infant slept in a separate room all night from 1 through 12 months, n = 34), early switch to solitary sleep (infant switched to solitary sleep arrangement by 3 or 6 months, n = 52), late switch to solitary sleep (infant switched to solitary sleep arrangement by 9 or 12 months, n = 13), consistent co-sleeping (room-sharing or bed-sharing from 1 through 12 months, n = 21), and inconsistent (no discernible infant sleep arrangement pattern across the first year, n = 19). These maternal reports of sleep arrangements were corroborated by nighttime video recordings of infant sleep location throughout the night at each age point (Teti, 2014). Infants in consistent solitary sleep were observed to spend significantly more time in their own rooms than infants in consistent co-sleeping, late switching, and combination sleep arrangement groups. In addition, infants in early-switching families were observed to spend more time in own room after 3 months of age, and infants in late-switching families began to spend more time in own room after 9 months.

Table 1 presents the frequency of sleep arrangements used at 1, 3, 6, 9, and 12 months in the five aggregate groups. Seventy-three percent of families were engaged in some form of co-sleeping at 1 month. This percentage dropped across the year: 45% at 3 months, 32% at 6 months, 27% at 9 months, and 23% at 12 months. We note that in the consistent co-sleeping group, only three families engaged in consistent bed sharing through 12 months, and only one family engaged in consistent room sharing through 12 months. Thus, the majority of these families switched between room sharing and bed sharing throughout the infants’ first year.

Table 1.

Sleep arrangements in the five sleep arrangement groups.

	Infant age
	1	3	6	9	12
Consistent solitary sleep
Solitary sleep	34	34	34	34	34
Room sharing	0	0	0	0	0
Bed sharing	0	0	0	0	0
Combinationa	0	0	0	0	0
Early switchers
Solitary sleep	0	28	52	52	52
Room sharing	39	16	0	0	0
Bed sharing	4	2	0	0	0
Combinationa	9	6	0	0	0
Late switchers
Solitary sleep	0	0	0	7	13
Room sharing	9	9	9	4	0
Bed sharing	1	2	1	0	0
Combinationa	3	2	3	2	0
Consistent cosleeping
Solitary sleep	0	0	0	0	0
Room sharing	7	8	4	6	6
Bed sharingb	10	6	11	9	8
Combinationa	4	7	6	6	7
Inconsistent
Solitary sleep	3	14	9	9	7
Room sharing	12	3	3	4	4
Bed sharing	1	0	0	1	1
Combinationa	3	2	7	5	7

^aInfant’s sleep during the night is divided between bed sharing, room sharing, and solitary sleep.

^bOnly three families engaged in consistent bed sharing throughout the full year.

Marital adjustment

When infants were 1 and 12 months old, mothers completed the Locke-Wallace Marital Adjustment Test (MAT) (Locke & Wallace, 1959) with four additional items (religious matters; aims, goals, and things believed to be important; making major decisions; and household tasks) incorporated from Spanier’s (1976) Dyadic Adjustment Scale (DAS). The MAT and DAS are well-established measures and have strong psychometric credentials (Haque & Davenport, 2009; Prouty, Markowski, & Barnes, 2000). In the present study, internal reliability of marital adjustment was .80 and .71 for mothers and fathers at 1 month, and .83 and .77 for mothers and fathers at 12 months.

Quality of coparenting

A second assessment of marital quality, the Coparenting Relationship Scale (CRS) (Feinberg, Brown & Kan, 2012) was also obtained. Unlike the marital adjustment questionnaire, the CRS was specific to a mother’s perception of how well she and her partner worked together as a child-rearing team. The CRS taps parental perceptions of the quality of their coparenting relationship in the dimensions of interparental agreement, closeness, exposure of child to conflict, coparenting support, undermining, endorsement of partner’s parenting, and division of labor. Feinberg et al. (2012) demonstrated adequate internal and test-retest reliability and construct validity of the CRS. In the present study, positive coparenting dimensions (agreement, closeness, support, endorsement, and division of labor) were significantly inter-correlated at all age points, and items in these five dimensions were summed to create a positive coparenting composite at all age points (alphas = .91 to .94 for mothers’ positive coparenting, and .87 to .93 for fathers’ positive coparenting). The two dimensions of negative coparenting (competition-undermining and exposure to conflict) were also significantly intercorrelated at all age points, and items in these two dimensions were summed to create a negative coparenting composite at each age point (alphas = .71 to .80 for mothers’ negative coparenting, and .78 to .85 for fathers’ negative coparenting).

Parent emotional availability at bedtime

Parents’ emotional availability (EA) with infants during bedtimes was scored from the digital video recordings of bedtime, using the Emotional Availability Scales (EAS) (Biringen, Robinson, & Emde, 1998). The recording system used for this purpose was a Bosch Divar XF digital video recorder (DVR), Infrared Color CCD night-vision cameras, Channel Vision 5104 microphones, and a portable DVD player. This system enabled up to four cameras and four microphones to input into the DVR, all of which were time-synched. A typical bedtime setup included one camera-microphone overhanging the crib or bed that contained the infant; a second camera-mic on the opposite corner of the room facing the door and providing a wide-angle view of the room; a third camera-mic trained on an area outside the infant’s crib or bed (e.g., a rocking chair, changing table) that the parent typically used during bedtime; and, if needed, a fourth wireless camera-mic in a separate room if the parent indicated that s/he spent some time in that room during bedtime. Parents were asked to start the recording approximately one hour before they began bedtime with their infant, and to end it when the infant woke up in the morning.

The construct validity of the EAS is well-established (Biringen et al., 2005; Moehler, Biringen, & Poustka, 2007; Trapolini, Ungerer, & McMahon, 2008). The EAS uses four scales to assess parental EA: Sensitivity (9-point scale), assessing the parent’s ability to read accurately and respond contingently to child signals with warmth and emotional connectedness; structuring (5-point scale), measuring parent’s capacity for appropriate scaffolding of child activities and setting appropriate limits; non-intrusiveness (5-point scale, reverse-scored), reflecting parent’s capacity to respect the child’s autonomy and personal space; and non-hostility (5-point scale, reverse-scored), assessing parent’s ability to interact with the child without signs of covert or overt irritability/anger. Two additional child scales (responsiveness and involvement) were not used in the present study because of the limited behavioral repertoire of infants in the first year and because they did not appear to lend themselves readily to a bedtime/sleep context.

EA was coded by the 4^th author, who was trained and certified in the EAS system by Zeynep Biringen in Spring, 2007 and who was blind to all other observational data. EA could not be scored if parents turned the video system on after the infant was already asleep, or if the amount of interaction between parents and infants was less than 5 minutes. From the videos, it was clear that mothers were much more likely to be putting infants to bed than fathers. Bedtime EA on fathers could only be obtained from 44, 37, 37, 31, and 37 families at 1, 3, 6, 9, and 12 months. These numbers, when further subdivided by sleep arrangement groupings, yielded cell sizes that were too small (ns ranged from 2 to 16) for meaningful analyses. Thus, analyses of paternal EA and sleep arrangements were not conducted.

At each age point, the four EA scales for mothers were standardized and summed to create a maternal EA composite. Internal reliability of the maternal EA composite was adequate (standardized item alphas = .81, .77, .78, .88, and .82 at 1, 3, 6, 9, and 12 months, respectively). Inter-rater reliability [intraclass correlation (ICC), absolute agreement] on the maternal EA composite was established between the 4th and 1st author, who was also trained and certified on the EAS system. The ICC on the maternal EA composite was .98, based on 42 mother-infant dyads evenly distributed across all five age points (eight at 1, 3, and 12 months, nine at 6 and 9 months). Stability of maternal EA across infant age was moderate-to-strong (rs = .28 to .63).

Parent-Infant Sleep Quality

Actigraphy

For seven consecutive days at each age point, infants, mothers, and fathers wore a Respironics/Mini Mitter actiwatch (model AW-64) to assess sleep-wake activity across each night. Infants wore the actiwatch on their upper ankle (affixed with a soft elastic band), and mothers and fathers wore it on their wrists. Sampling epoch length for all actiwatches was 1 minute, and a medium (40 activity counts) wake threshold value was used. Using Actiware (Version 5.0) software, actigraphy data, in actogram format, were uploaded onto a personal computer. The Sadeh algorithm (Sadeh, Sharkey, & Carskadon, 1994), which identifies sleep onset as the first of at least 3 continuous minutes of sleep and sleep offset as the last of at least 5 continuous minutes of sleep, was used to identify wake bouts throughout the night. It has been found to accurately identify sleep-wake activity in both adults and infants (Galland, Kennedy, Mitchell, & Taylor, 2012; Sadeh et al., 1994). Actiware^® (version 5.0) software (Respironics/Mini Mitter, 2005) was used to determine two indices for each infant and parent for each of the seven nights. Sleep fragmentation (percent of active bouts + percent of inactive bouts less than 1 minute in duration) is a measure that combines the percentage of higher activity (wake) bouts with the percentage of inactive bouts whose duration is so brief that they are unlikely to be deep sleep. Sleep fragmentation is believed to be an overall measure (percent) of restless, non-recuperative sleep (Levine, Roehrs, Stepanski, & Zorick, F., 1987). Wake after sleep onset (WASO) is the number of minutes spent awake between sleep onset and morning wake time. The mean of each of these measures was obtained for the full week of data collection. In addition, a daily sleep diary (The 24-Hour Sleep Patterns Interview; 24-HSPI) (Meltzer, Mindell, & Levandoski, 2007) was used at each age point to confirm when each parent went to bed and when s/he fell asleep the previous night. This information was used to cross-check the beginning of sleep time for mothers and fathers each night from the actigraphy record.

Infant sleep diary

Mothers also completed an infant sleep diary (adapted from Burnham, Goodlin-Jones, Gaylor, & Anders, 2002) every morning across the full week of data collection at each age that asked the mother to record when the infant was put down to sleep and when s/he fell asleep the previous night, and also the frequency of infant night waking during the previous night. This diary was used to cross-check each infant’s actigraphy record to confirm the onset of sleep time and morning wake time for the infant. In addition, this diary provided information from mothers directly about infant night waking during the week. This index was obtained for each infant by summing across the full seven nights of data collection at each age.

Results

Power analyses

A series of power analyses were conducted that took into account overall sample size, number of sleep arrangement groups, and the number of infant age points, and the expected moderate effect size for most analyses. These analyses indicated power levels that ranged from .63 to .99 across all analyses.

Hypothesis 1, that co-sleeping would be associated with maternal reports of increased night waking, and whether this finding would be corroborated by actigraph records of infant sleep, was addressed with a covariance pattern analytical approach, with sleep arrangements and infant age as between- and within-subjects factors, respectively, using PROC MIXED in Statistical Analysis Software (SAS), v. 9.3. The advantages of this approach over standard repeated measures ANOVA in longitudinal research is its ability to adjust for missing values, using maximum likelihood procedures, and to provide investigators the ability to choose the best-fitting error structure for the data, using AIC and BIC fit indices (Hedeker & Gibbons, 2006). Using a Toeplitz error structure, analyses revealed a main effect of infant age on infant nighttime sleep fragmentation, F(4, 488) = 154.88, p < .0001, but the effects of sleep arrangements, F(4,134) = 0.73, p = .57, and the interaction of sleep arrangements × infant age, F(16,488) = 0.88, p = .59, were not significant. Figure 1 presents mean values of infant fragmentation for the five sleep arrangement groups at all infant age points. Fragmented sleep in infants decreased sharply across the first year, particularly between 1 and 6 months of age, reflecting the expected rapid consolidation of infant nighttime sleep with maturation during the first year of life. These analyses were also conducted on infant WASO (using a Toeplitz error structure), which yielded a main effect of infant age, F (4,488) = 40.33, p < .0001 (again showing sharp decreases in infant wake time across age), no effects of sleep arrangements, F(4,134) = 1.61, p = .18, and no sleep arrangement × infant age interaction, F(4, 488) = 0.93, p = .54. That the main effects of sleep arrangements and the sleep arrangements × infant age interactions were not significant for both infant sleep fragmentation and WASO indicated that, contrary to expectations, infant sleep quality did not appear compromised in co-sleeping relative to solitary sleeping arrangements when using actigraphy-derived indices of infant sleep.

Figure 1.

Actigraph records of infant sleep fragmentation during the infants’ first year, broken down by sleep arrangement pattern.

Similar covariance pattern analyses were conducted on mothers’ sleep fragmentation and WASO. Consistent with results obtained on infant sleep fragmentation, mothers sleep fragmentation (with unstructured error) was found to decrease significantly across the infants’ first year, F(4,134) = 19.05, p < .0001, indicating improved sleep across the year, although this decrease was more gradual and linear compared to that observed for infant sleep fragmentation (see Figure 2). However, in contrast to results obtained for infants, mothers sleep fragmentation was significantly associated with sleep arrangements, F(4,134) = 6.56, p < .0001. Post-hoc pairwise comparisons, using an adjusted per comparison error rate so that family wise error rate did not exceed p = .05, revealed that mothers who used consistent solitary sleeping arrangements across the first year had consistently less fragmented sleep compared to mothers of infants in consistent co-sleeping arrangements (adjusted p = .0004), mothers in late switching arrangements (mothers who did not switch their infants into solitary sleep until 9–12 months) (adjusted p = .02), and mothers of infants in inconsistent sleep arrangements across the first year (adjusted p = .007). In addition, mothers whose infants switched into solitary sleep arrangements between 3 and 6 months had less fragmented sleep than mothers of infants in consistent co-sleeping sleep arrangements (adjusted p = .02). The interaction of sleep arrangements × infant age on mothers’ sleep fragmentation was not significant, F (16,134) = 1.39, p = .16. Additional analyses on mothers’ WASO (using a Toeplitz error structure) yielded a similar main effect of infant age, F(4,481) = 41.10, p < .0001), a main effect of sleep arrangements, F(4, 134) = 4.01, p = .004, but no sleep arrangements × infant age interaction, F(16, 481) = 1.13, p = .32. Consistent with results for sleep fragmentation, mothers’ WASO decreased over time (indicating improved sleep throughout the infants’ first year). In addition, post-hoc pairwise comparisons, again adjusting the error rate per comparison so that family wise error rate ≤ .05, revealed that mothers of infants in consistent solitary sleep arrangements spent significantly fewer minutes awake during the night (M = 62.13) than mothers of infants in consistent co-sleeping arrangements (M = 77.19) (adjusted p = .04) and mothers of infants in inconsistent sleep arrangements (M = 80.81) (adjusted p = .01). In addition, mothers who were early switchers spent significantly fewer minutes awake at night (M = 65.77) compared to mothers of infants in inconsistent sleep arrangements (M = 80.81) (adjusted p = .04)

Figure 2.

Actigraph records of mothers’ sleep fragmentation during the infants’ first year, broken down by sleep arrangement pattern.

Collectively, these results revealed that, with actigraph measures of sleep quality, mothers of infants in co-sleeping arrangements that persisted beyond 6 months of age were more likely to experience sleep disruptions, compared to mothers of infants in consistent solitary arrangements and mothers whose infants switched into solitary sleep before 6 months. These results contrasted markedly with infant actigraphy data, which indicated that infant sleep was not associated with sleep arrangements. Subsequent covariance pattern analyses were conducted on frequency of infant night waking as reported by mothers on the infant sleep diary, summed across the full data collection week. Interestingly, these analyses (using unstructured error) revealed a pattern of results that was more consistent with those obtained with mothers’ fragmented sleep and mothers’ WASO than with infant fragmented sleep and WASO (see Figure 3). Specifically, infant night waking frequency decreased across the first year, F(4,134) = 41.26, p < .0001 (as expected), and was significantly associated with sleep arrangements, F(4,134) = 5.00, p = .0009, with no effect of the sleep arrangements × infant age interaction, F(16, 134) = 1.18, p = .29. Post-hoc pairwise comparisons, again adjusting for a family wise error rate of p ≤ .05, revealed that mothers of infants in consistent co-sleeping arrangements reported significantly more infant night awakenings than mothers of infants in consistent solitary sleep arrangements, adjusted p = .0005, and mothers of infants who switched into solitary sleep before 6 months, p = .007). Thus, mothers’ perceptions of their infants’ sleep quality coincided much more closely with actigraphy measures of mothers’ own sleep quality than with actigraphy measures of infants’ sleep quality.

Figure 3.

Mothers’ diary reports of infant nighttime awakenings during the infants’ first year, broken down by sleep arrangement pattern.

Post-hoc correlational analyses were conducted to explore the nature of the relation between mothers’ reports of infant night waking and sleep arrangements. It was possible that mothers’ daily diary reports of increased infant night waking in the consistent co-sleeping group was because these mothers slept in closer proximity to their infants and thus more likely to detect infant nighttime arousals than mothers in the other sleep arrangement groups. If such was the case, we would expect that mothers’ reports of the frequency of infant night awakenings would be more highly correlated with mothers’ sleep fragmentation and WASO, from actigraphy, in consistent co-sleeping arrangements than in consistent solitary sleeping arrangements. As Table 2 indicates, Pearson correlations did not support this expectation. Instead, stronger and consistently positive associations were found between mothers’ reports of infant night awakenings and infant sleep fragmentation and WASO in the consistent solitary sleeping group than in the consistent co-sleeping group at all infant age points. By contrast, these correlations were inconsistent across infant age in the consistent co-sleeping group, and at 9 months the correlation was unexpectedly negative. These results indicate that objective assessments of mothers’ sleep quality were more closely linked with mothers’ perceptions of their infants’ night awakenings in the consistent solitary sleeping group than in the consistent co-sleeping group.

Table 2.

Correlations between mothers’ perceptions of infant night awakenings, from daily sleep diaries, and actigraph recordings of mothers’ sleep fragmentation and WASO^a in consistent solitary and consistent co-sleeping groups.

	Consistent Solitary		Consistent Co-Sleeping
	Mothers’ sleep fragmentation	Mothers’ WASOa	Mothers’ sleep fragmentation	Mothers’ WASOa
1 month	.48**	.55***	.11	.30
3 months	.26	,32	.32	.18
6 months	.31	.48**	−.27	−.17
9 months	.19	.36*	−,67**	−.16
12 months	.47**	.51**	.28	.45+

^aWake after Sleep Onset – number of minutes awake during the night after first going to sleep

^*p < .05

^**p < .01

^***p < .001

A final set of covariance pattern analyses was performed on fathers’ sleep fragmentation and WASO data, again looking at main effects of sleep arrangement, infant age, and sleep arrangement × infant age interactions. For sleep fragmentation (using an unstructured error covariance pattern), the effect of infant age was significant, F(4,125) = 3.24, p = .01, but neither the effect of sleep arrangements nor the interaction of sleep arrangement with infant age was significant For fathers’ WASO data, using a Toeplitz error structure, the effect of infant age approached significance, F(4, 431) = 2.52, p = .07, but like fathers’ sleep fragmentation, the effects of sleep arrangements and the sleep arrangements × infant age interaction were not significant. These findings indicate that, like mothers’ and infants’ sleep, fathers’ sleep quality generally improved across the infants’ first year, likely the result of improvements in their infants’ sleep across the same time. However, unlike mothers, fathers’ sleep was not associated with their infants’ sleep arrangement patterns across the first year.

Hypothesis 2, that persistent co-sleeping (beyond 6 months of age) would be associated with marital and coparenting distress, was addressed with covariance pattern analyses, with sleep arrangements and infant age as independent variables. Mothers’ marital adjustment, using a compound symmetry error structure, was significantly associated with sleep arrangements, F(4,126) = 5.30, p = .0005. Post-hoc pairwise comparisions, maintaining family wise error rate at p ≤ .05, revealed that mothers of consistently co-sleeping infants had significantly lower marital adjustment scores (M = 114.21) than mothers of infants in consistent solitary sleep (M = 138.32) (adjusted p = .0008), early switching mothers (M = 132.83) (adjusted p = .007), late switching mothers (M = 137.27) (adjusted p = .004), and mothers of infants in inconsistent sleep arrangements in the first year (M = 140.51) (adjusted p = .002). Mothers’ marital adjustment also decreased from 1 (M = 134.2) to 12 months (M = 131.05), F (1,126) = 4.30, p = .04. The sleep arrangements × infant age interaction was not significant, however, indicating that the sleep arrangement differences in marital adjustment were evident as early as 1 month post-partum and were not differentially related to sleep arrangements across time. Similar analyses, using a compound symmetry error structure, conducted on fathers’ reports of marital adjustment revealed no associations with infant age, sleep arrangements, or the sleep arrangements × infant age interaction.

Covariance pattern analyses, using unstructured error, were then performed on mothers’ and fathers’ reports of coparenting quality. Mothers’ reports of negative coparenting were significantly associated with sleep arrangements, F (4,131) = 5.91, p = .0002, but not with infant age, F(4,131) = 0.47, p = .76, or the sleep arrangements × infant age interaction, F(16, 131) = 0.84, p = .64 (see Figure 4). Post-hoc pairwise comparisons, maintaining family wise error at p ≤ .05, revealed that mothers of infants in consistent co-sleeping arrangements reported higher levels of negative coparenting than mothers of infants in mothers of infants in consistent solitary sleep (adjusted p < .0001), mothers of early switching infants (adjusted p = .008), mothers of late switching infants (adjusted p = .03), and mothers of infants in inconsistent sleep arrangements throughout the first year (adjusted p = .01). The non-significant sleep arrangements × infant age interaction indicated that these differences were evident from 1 month of infant age onward and did not show differential change with different sleep arrangements over time.

Figure 4.

Mothers’ reports of negative coparenting during the infants’ first year, broken down by sleep arrangements pattern.

Analyses of fathers’ reports of negative coparenting revealed no effects of infant age, F(4,126) = 1.22, p = .31, no effects of sleep arrangements, F(4,126) = 1.63, p = .17, but a significant sleep arrangement × infant age interaction, F(16,126) = 2.73, p = .0009. Analyses of simple effects, maintaining family wise error at p ≤ .05, revealed a significant effect of infant age on fathers’ negative coparenting in families whose infants were consistent co-sleepers, F (4.126) = 4.20, p = .003. Fathers’ negative coparenting in the consistent co-sleeping group decreased from 1 month (M = 22.75) 6 months (M = 19.19), but then increased back to 1-month levels from 6 to 12 months (M = 22.33).

Mothers’ positive coparenting was associated with both infant age, F(4,131) = 2.76, p = .03 and sleep arrangements, F(4,131) = 3.40, p = .01, but not with the sleep arrangements × infant age interaction, F(16,131) = .84, p = .64, which again did not support differential change with sleep arrangements over time. The cubic contrast for infant age was significant, F(1,131) = 7.52, p = .007 and fit an inverted U function, with an increase from 1 (M = 141.03) to 6 months (M = 143.00), followed by a decrease from 6 to 12 months (M = 137.21). Post-hoc pairwise comparisons, holding family wise error at p ≤ .05, revealed that mothers of infants in consistent co-sleeping arrangements reported less positive coparenting than mothers of infants in consistent solitary sleeping arrangements (adjusted p = .005). No other pairwise comparison was significant. Similar analyses on fathers’ positive coparenting yielded no significant effects.

Hypothesis 3, addressing whether persistent co-sleeping was associated with poorer quality of parenting at bedtime, was examined with a covariance pattern analysis conducted on mothers’ emotional availability with infants at bedtime, using unstructured error. Mothers’ emotional availability was significantly associated with infant sleep arrangements, F(4, 130) = 6.00, p = .0002 (see Figure 5). Post-hoc pairwise comparisons, maintaining family wise error at p = .05, revealed that mothers of infants in consistent solitary sleep were more emotionally available with their infants at bedtime than mothers of infants in consistent co-sleeping arrangements (adjusted p = .0001) and mothers in late switching arrangements (adjusted p = .03). In addition, early switching mothers were more emotionally available to their infants at bedtime than mothers of infants in consistent co-sleeping arrangements (adjusted p = .004). Mothers’ EA was not associated with infant age or with the interaction of sleep arrangements × infant age, indicating that these group differences were evident from 1 month of age onward and were not moderated by sleep arrangements over time.

Figure 5.

Mothers’ emotional availability with infants at bedtime, from home observations, during the infants’ first year, broken down by sleep arrangements pattern.

The fact that sleep arrangement patterns were associated with maternal bedtime EA prompted a final set of analyses to determine the extent to which maternal EA was associated with other correlates of infant sleep arrangements, identified above. Mothers’ sleep fragmentation, marital adjustment, positive coparenting perceptions, negative coparenting perceptions, and bedtime EA were each averaged across the full year, and the full correlation matrix of these averaged scores is presented in Table 3. Maternal bedtime EA was inversely associated with mothers’ sleep fragmentation and negative coparenting, and positively associated with mothers’ marital adjustment and positive coparenting.

Table 3.

Inter-correlations among variables that were significantly associated with infant sleep arrangement patterns.

	Mothers’ fragmented sleepa	Mothers’ marital adjustmenta	Mothers’ positive coparentinga	Mothers’ negative coparentinga	Mothers’ EA at bedtimea
Mothers fragmented sleep	------	−.06	−.13	.12	−.18*
Mothers’ marital adjustment		------	.79***	−.67***	.29**
Mothers’ positive coparenting			------	−.68***	.23**
Mothers’ negative coparenting				------	−.33***
Mothers’ EA at Bedtime					------

^aMeaned across the full year.

^*p < .05

^**p < .01

^***p < .001

A closer look at first-year sleep arrangement patterns and socio-demographics

A final series of analyses were conducted to identify socio-demographic and family characteristics associated with specific sleep arrangement patterns across the infants’ first year. These analyses were prompted by earlier work reporting associations between co-sleeping and younger and more socio-economically stressed parents (McCoy et al., 2004), limited sleeping spaces (Jenni & O’Connor, 2005), and breastfeeding (Ball, 2003). One way analyses of variance, followed by Student Newman-Keuls post-hoc comparisons, revealed no associations between maternal education, paternal education, mothers’ age, fathers’ age, yearly family income, or family size and sleep arrangement patterns. However, at 1, 6, and 9 months, late switching mothers more strongly indicated limited sleeping space as a reason for their sleep arrangement choice than mothers in all other sleeping arrangements. Chi-square analyses revealed no associations between breastfeeding, partner status, and sleep arrangement patterns across the first year, but did reveal an association between race (White, non-White) and sleep arrangements, Χ² (4) = 23.47, p < .001. Percentages of non-White families was low and evenly distributed across consistent solitary, early switchers, late switchers, and inconsistent sleep arrangement groups (< 15%), but higher in the consistent co-sleeping group [43%, n = 9, (1 African American, 4 Asian, 3 Latino, 1 mixed-race]. Race, however, was not associated with parent age, education, family income, family size, or reports of limited sleep space. In addition, one way analyses of variance revealed no differences between white and non-white families in the consistent co-sleeping group in mothers’ and fathers’ sleep fragmentation, marital adjustment, positive and negative coparenting, and mothers’ bedtime EA.

Discussion

The present study of infant sleep arrangements and family functioning extended prior work by identifying different sleep arrangement patterns across the infant’s first year and examining relations between these patterns and longitudinal patterns of sleep, marital and coparenting distress, and parenting. Results indeed indicated that sleep quality, marital and coparenting quality, and mothers’ emotional availability at bedtime were less optimal in persistent co-sleeping families, but not in families who stopped co-sleeping by 6 months of infant age, although these findings pertained almost exclusively to mothers.

Sleep Arrangements and Infant-Parent Sleep

Actigraphy data revealed that it was mothers’ but not infant sleep, that appeared to be disrupted in persistent (i.e., beyond six months post-partum) co-sleeping arrangements, the latter which included consistent co-sleeping mothers, late-switching mothers, and mothers who showed inconsistency in their sleep arrangements across the first year. In addition, mothers of infants in consistent co-sleeping arrangements reported more infant night awakenings compared to mothers of infants in consistent solitary sleep and early switching mothers. These findings are consistent with earlier studies using maternal reports of infant sleep (DeLeon & Karraker, 2007). They are also consistent with Volkovich et al. (2015), who found that, whereas mothers in co-sleeping arrangements reported higher levels of infant night waking, this link was not borne out by actigraphy, and that it was actually mothers in co-sleeping arrangements, but not infants, whose sleep appeared to be disrupted in co-sleeping arrangements.

It is possible that in the present study, actigraphy was simply not sensitive enough to detect subtle infant waking states that would have been detected by polysomnography (Yoshida, Shinohara, & Kodama, 2015). Actigraphy, however, has been found to be a valid indicator of sleep-wake activity in infants and adults (Galland et al., 2012), and in the present study was more than adequate to detect the expected increase in sleep consolidation in infants across the first year (Henderson et al., 2011). Actigraphy is still generally regarded as superior to maternal report in estimating child sleep-wake activity (Asaka & Takada, 2011; Simard, Bernier, Bélanger, & Carrier, 2013) because mothers tend to under-estimate child waking frequencies and durations (Acebo et al., 2005). We propose that the tendency of mothers in persistent co-sleeping arrangements to report higher frequencies of infant night awakenings may be related to other, family-based and personal factors. We return to this point later in our discussion.

Analyses of actigraphy data indicated that both mothers’ and fathers’ sleep fragmentation and WASO decreased significantly over time, likely reflecting the fact that both parents’ sleep improved with the consolidation of infant sleep across the first year. However, in contrast to mothers, fathers’ sleep was remarkably unrelated to sleep arrangement patterns. Mothers in this sample were almost exclusively the primary caregivers for their infants, and much more likely than fathers to take primary responsibility for putting their infants to bed. This was corroborated by our bedtime video data, which almost always involved mothers but sporadically involved fathers in bedtime activities. We suspect that, whereas mothers’ involvement with their infants at bedtime and at night was a constant, fathers’ involvement in these activities varied widely. We would expect stronger linkages between infant sleep arrangements and fathers’ sleep quality under conditions of high father involvement. This is an important question for further research.

Sleep Arrangements and Marital and Coparenting Distress

It was hypothesized that marital and coparenting distress would be higher in families engaging in co-sleeping, but especially so when co-sleeping persisted beyond 6 months of infant age. This hypothesis was, in the main, supported, but primarily for mothers, in that (a) mothers’ reports of marital adjustment and positive coparenting were lower and negative coparenting higher among mothers of infants in consistent co-sleeping arrangements across the first year. These “main effect” differences were apparent as early as 1 month of infant age and persisted throughout the full year. In contrast, fathers’ marital adjustment and positive coparenting bore no main effect relations to sleep arrangements whatsoever, although among fathers of infants in consistent co-sleeping arrangements, paternal reports of negative coparenting decreased from 1 to 6 months, but then increased back to 1-month levels from 6 to 12 months. The genesis of this U-shaped pattern is unclear and requires replication with additional data on fathers’ reactions to and attributions about co-sleeping at various points throughout the infant’s first year.

Collectively, these findings do not support the premise that sleeping arrangements were causal to mothers’ marital and coparenting quality, given that (a) these differences were evident very early in the infants’ life (1 month) and (b) no interactive effects between sleep arrangements patterns and infant age were found for any infant and parent outcome. Indeed, parents’ reasons for using specific sleep arrangements are quite varied, and decisions to co-sleep, particularly in the first few months of life, could be based on convenience, ease of breastfeeding, the belief that doing so promotes infant security, facilitating a good night’s sleep (for infant and parent), etc. (Ball, Hooker, & Kelly, 1999). This was borne out by the fact that 73% of families engaged in some form of co-sleeping with their infants at 1 month. This percentage dropped precipitously across the year, however (45%, 32%, 27%, and 23% at 3, 6, 9, and 12 months, respectively), and it was the persistent co-sleepers in the present study (i.e., co-sleeping beyond 6 months of age) who reported heightened marital and coparenting distress compared to families who never co-slept and families who stopped co-sleeping by 6 months. These data suggest that persistent co-sleeping throughout the infant’s first year, particularly in a culture in which persistent co-sleeping is not supported, may be symptomatic of pre-existing heightened marital and family stress that is evident very early in the life of the infant. This heightened distress, which appears to be particularly felt by mothers in the marital and coparenting domains, may predispose mothers to structure infant sleeping arrangements to allow them to spend more time with their infants at night, perhaps as a way of compensating for a perceived lack of closeness and intimacy in their marriages, and to maintain these arrangements in the face of continuing marital and coparenting distress. These findings support and extend earlier work (Teti, Crosby, McDaniel, Shimizu, & Whitesell, 2015), who reported that maritally distressed mothers who co-slept with their infants at 1 month were more likely to maintain these co-sleeping arrangements through 6 months, compared to non-distressed co-sleeping mothers at 1 month, who by contrast were more likely to move their infants into solitary sleep by 6 months. The present findings also extend the findings of Teti and Crosby (2012), who found that mothers with elevated depressive symptoms and excessive worries about their infants’ nighttime sleep behavior were more likely to seek out and spend time with their infants during the night than mothers with low symptom levels, and that this propensity for doing so was largely unrelated to whether or not their infants were distressed.

The question remains, however, why mothers of infants in consistent co-sleeping arrangements reported higher levels of night awakenings in their infants relative to the other sleep arrangement groups, when such differences were not supported by actigraphy? One hypothesis, that mothers who sleep in close proximity with their infants notice their infant nighttime arousals more than mothers who sleep separately from their infants, and thus more likely to report them than mothers in solitary sleeping arrangements, was not supported by correlational analyses. Indeed, stronger, more consistent positive associations were found between mothers’ reports of infant night waking and actigraph reports of mothers’ sleep disruption in consistent solitary sleeping arrangements than in consistent co-sleeping arrangements (Table 2). We offer the following alternative hypothesis: If persistent co-sleeping in a culture that does not support it is a marker of family and maternal distress (as it appeared to be in this study), one manifestation of maternal distress may be a hypersensitivity to and hyper-vigilance about infant night awakenings, leading distressed mothers to count even very brief infant arousals and postural shifts as night awakenings. These very brief infant arousals may go unnoticed by non-distressed mothers, and they may be too brief and below threshold to be identified as wake activity by actigraphy. The question of whether distressed mothers and/or co-sleeping mothers over-report infant night awakenings has, to date, rarely been addressed in the mother-infant sleep literature, and emphasizes the need to corroborate parent reports of infant/child sleep with objective assessments, such as actigraphy and polysomnography.

Sleep Arrangements and Mothers’ Bedtime EA

Although there was no theoretical or empirical basis for expecting linkages between sleep arrangements and mothers’ bedtime EA, finding such linkages was not surprising in light of the fact that mothers in persistent co-sleeping arrangements had more fragmented sleep (and thus at risk for cumulative sleep debt) and experienced more marital and coparental distress. Sleep deprivation and fatigue is known to disrupt the quality of social relationships, including the parent-child relationship (Tikotzky, 2016; Tikotzky, Chambers, Kent, Gaylor, & Manber, 2012), and marital maladjustment and poor coparenting quality are established predictors of parenting difficulties (Cabrera, Shannon, & La Taillade, 2009; Coln, Jordan, & Mercer, 2013; McCoy, George, Cummings, & Davies, 2013). Again, sleep arrangement differences did not moderate longitudinal patterns of EA across the year, and thus there was no evidence that sleep arrangement differences caused differences in mothers’ bedtime EA. The sleep arrangement differences in mothers’ bedtime EA raises additional questions about whether persistent co-sleeping in a culture that does not endorse it is not just a marker for maternal sleep loss and family stress but also a risk marker for early socio-emotional problems in children. We emphasize that if this was the case, we would expect that such risks would be realized only to the extent that persistent co-sleeping was associated with persistent family and parenting stress across the first year. Clearly, this is a topic for further research.

Limitations and Future Directions

Although power analyses suggested there was sufficient power to detect differences of moderate effect, this study was nevertheless limited by relatively small sample sizes in the persistent co-sleeping subgroups, compared to the solitary and early switching groups. In addition, similar to Ball’s (2002) identification of “habitual” vs. “combination” bedsharers, persistent co-sleeping families typically made use of both bed sharing and room sharing, such that it was not possible to compare consistent room sharers with consistent bed sharers across the year. It may be the case that maternal sleep was especially disrupted when mothers and infants bedshared, compared to when they roomshared. A future study that examines differences in sleep and family-based variables between a “pure” group of consistent bed sharing families and a “pure” group of consistent room sharing families across the infants first year (if such could be obtained) would shed light on whether persistent bed sharing alone is associated with maternal sleep disruption and elevated family risk, or if persistent room sharing is also associated with risk. The present findings nevertheless told a coherent story, that infant-parent co-sleeping that persists beyond 6 months of infant age was associated with maternal sleep disruption and marital and coparenting distress as reported by mothers.

Another limitation of the present study is the lack of clarity regarding the nature and risk status of the inconsistent group, so defined in terms of the lack of consistency in sleep arrangement use across the infants’ first year. They, along with consistent co-sleepers and late-switching mothers, experienced increased sleep fragmentation and WASO, compared to mothers in consistent solitary sleep, but better marital and coparenting quality than consistent co-sleeping mothers, and they were not different from consistent solitary sleeping mothers in bedtime EA. Unfortunately, the circumstances and reasons underlying the inconsistent use of sleep arrangements in this group were not assessed in this study, and thus it was unclear on what this inconsistency was based. It behooves future researchers to attempt to assess more closely and systematically the circumstances and (if they can be accessed) the reasons behind parents’ use of sleep arrangements, and change in sleep arrangements, as they evolve across time.

A third limitation is that, despite the lack of evidence that sleep arrangements caused the differences observed in family stress and parenting, the present study’s design and measures were insufficient to address issues of causality. Sleep arrangement patterns and measures of marital and family stress were first measured at the end of the infants’ first month, and thus we are not privy to parenting and family dynamics that may have unfolded during the infants’ first month and what possible organizational impact they may have had on sleep arrangement choices, parents’ perceptions of marital adjustment and coparenting, and the emotional quality of bedtime parenting over time. It is also possible that the measures of marital and coparenting distress used in the present study, which were broad-based and independent of context, were simply not sensitive enough to tap into possible subtle mechanisms of influence between sleep arrangements and family stress. More sensitive, time-intensive approaches to the study of marital and coparenting adjustment, obtaining parent reports and observational measurements across micro-longitudinal bursts of time, beginning during infants’ first few weeks of life (i.e., a measurement-burst design approach; Sliwinski, 2008) may better address questions about any causal mechanisms and how they unfold between sleep arrangements and family stress.

It is important to note that the present findings are likely to be applicable to cultures in which persistent parent-infant co-sleeping is not normative. In a co-sleeping culture, we would not expect persistent co-sleeping during the infant’s first year to be a marker of risk. It is tempting to speculate about whether parents in traditionally co-sleeping cultures (e.g., Japan) who decide not to co-sleep with their infants are more likely to experience criticism from others and, in turn, elevations in marital, personal, and parenting distress. There are no published data on this point, but we note that Shimizu, Park, and Greenfield (2014) found that mothers in contemporary Japan had co-sleeping rates in 2008 that were largely unchanged from co-sleeping rates assessed in the 1960s (about 90%), despite Japan’s evolution from a post- World War II agrarian, rural society into a technologically sophisticated, highly educated society, one that would be expected to support more independent sleeping arrangements for infants (Greenfield, 2009). Interestingly, Shimizu et al. found that many mothers in their 2008 sample indeed endorsed ethnotheories of individualism and independence, reflecting Japan’s emergence as an industrialized, technologically advanced society, yet they nevertheless co-slept with their infants. The apparent conflict between these mothers’ ethnotheories and their sleep arrangement practices leads one to speculate about whether such conflicts would be associated with heightened personal or coparenting distress, compared to mothers whose ethnotheories matched their parenting practices. The cultural backdrop regarding infant sleep arrangements clearly plays a very large role in interpretations about whether sleep arrangement choices affect and are affected by marital and family stress and is, in our view, a very important area for future study.

If mothers who engage in persistent co-sleeping do so because they are distressed, and if doing so places mothers at risk for sleep disruption, cumulative sleep debt, and relationship difficulties with their infants, such mothers comprise a high-risk group in need of intervention. The primary goal of such an intervention should not be, in our view, to advise against co-sleeping, but instead to improve the marital and coparenting relationship, particularly around but not limited to decisions parents make about infant sleep arrangements, and to make certain that parents who wish to co-sleep think carefully about and discuss with each other the reasons they wish to do it, and ideally to be in full agreement if the decision to co-sleep is made. We believe such an intervention should aim to make parents fully aware of the perks and pitfalls of co-sleeping and help parents understand that whereas co-sleeping if practiced safely can facilitate breastfeeding and parents’ ability to monitor the infant during the night, persistent co-sleeping could lead to parent sleep disruption and cumulative sleep debt, which could detrimentally impact the marital and parent-infant relationships over the long term. Co-sleeping, in other words, should be done with full knowledge and support of both parenting partners. It may be problematic if co-sleeping is practiced as a way of compensating for a lack of marital intimacy or because parents harbor unrealistic anxieties about their infants’ nighttime sleep-wake behavior.

We are cognizant of the fact that this study addresses a topic that is highly personal. Our goal was not to take a unilateral stand for or against any particular sleep arrangement, but to address the topic empirically and present data that, in conjunction with earlier work (e.g., Ball, 2003; McKenna et al., 2007), Ball’s (2003), can be used to make more informed choices about structuring infant sleep across the first year of life. What seems clear from this study is sleep arrangements patterns across the first year are associated with the health of the family system, and we are just beginning to understand the very complex dynamics that underlie these links. We hope this investigation stimulates more work on this topic.