A Longitudinal Study of Postpartum Maternal Sleep and Sensitivity: Examining Depressive Symptoms and Social Support as Moderators

ABSTRACT

This longitudinal study examined whether links between postpartum maternal sleep and the quality of the mother–infant relationship are moderated by depressive symptoms and social support. Assessments were conducted at 4 months (N = 130) and 8 months postpartum (N = 108). Maternal sleep (i.e., sleep percent, minutes and variability) was measured over seven nights using actigraphy and sleep diaries. Questionnaires captured socio‐demographic information, feeding methods, social support and maternal depressive symptoms. At the 8‐month assessment, a 20‐min free‐play interaction between mother and infant was video recorded and later maternal sensitivity was coded using the Emotional Availability Scales (EAS). The findings showed no significant Pearson correlations between maternal sleep and maternal sensitivity when controlling for infant feeding methods. A series of two‐step hierarchical multiple regression models was used to examine depression and social support as moderators of the links between maternal sleep and maternal sensitivity. While depressive symptoms did not moderate these relationships, social support did. Specifically, higher actigraphic and reported sleep variability (i.e., less stable sleep) was associated with lower maternal sensitivity during play interactions, but only among mothers with high levels of social support. Evidence emerged for both concurrent and prospective links. The results suggest that maternal sleep is associated with sensitivity only in mothers with high social support. Possible explanations for this unexpected finding are discussed. Moreover, in this group, sleep variability is more strongly linked to maternal sensitivity than other sleep characteristics, highlighting the importance of sleep stability.

1. Introduction

A positive mother–infant relationship is consistently linked to better child emotional, social and cognitive development (Humphreys et al. 2024; Lubotzky‐Gete et al. 2021). Given the growing body of evidence connecting the quality of the mother–infant relationship to child outcomes, it is crucial to investigate factors that may undermine the development of this bond. While maternal psychological risk factors such as depression have been extensively studied in relation to parenting and the quality of the mother–infant relationship (Goodman et al. 2020; Martins and Gaffan 2000; Lovejoy et al. 2000), the impact of maternal sleep disturbances on these aspects has received comparatively less attention. This is surprising, considering that maternal sleep in the postpartum period is frequently disrupted due to factors like nighttime infant caregiving and hormonal changes, which put mothers at risk for negative emotional and psychological consequences associated with poor sleep (Hunter et al. 2009; Horwitz et al. 2023).

Several studies examined the links between maternal sleep and subjective reports of the mother–infant relationship and found that shorter maternal sleep duration and lower sleep quality were associated with more negative feelings and perceptions of mothers toward their infants (Kalmbach et al. 2022; Tikotzky 2016). These studies provided initial support for the importance of maternal sleep as a factor contributing to the evolving mother–infant relationship, but they primarily relied on subjective reports. While self‐reported sleep measures offer valuable insights into participants' subjective sleep quality, they may be limited by recall biases and shared method variance (Sadeh 2015). Only a few studies have used objective sleep assessment methods, such as actigraphy, to investigate the links between maternal sleep and the mother–infant relationship. In a recent study that combined both objective and subjective sleep measures and employed a diary‐study design, lower actigraphic and diary‐based sleep quality on a given night predicted more negative maternal feelings toward her infant the following day (Ran‐Peled et al. 2022), thus supporting previous research on the links between poorer maternal sleep and less favourable caregiving.

The mother–infant relationship is a complex construct that can be measured through self‐reports and observational methods. Few studies have explored the links between maternal sleep disturbances and the observed mother–infant relationship quality. King et al. (2020) found that poorer objective sleep quality was related to observers' ratings of a reduction in maternal sensitivity throughout a mother–infant play interaction. In a longitudinal study conducted during the first year postpartum, mothers' sleep characteristics (e.g., sleep variability, quality) were more strongly linked with bedtime emotional availability (EA) than with daytime EA assessed during a free play interaction (Bai et al. 2020, 2023). Another study found no support for direct associations between self‐reported maternal sleep quality and observed maternal sensitivity during a mother–infant play interaction (Leerkes et al. 2024). Thus, findings regarding the links between maternal sleep and the observed mother–infant relationship are inconsistent, particularly when mother–infant daytime interactions are considered.

Moreover, research in this field has traditionally focused on average sleep measures over time, but daily variations in sleep are common and the emotional and behavioural correlates of sleep variability (i.e., sleep instability) are still not well understood (Bei et al. 2016). The few studies that explored changes in maternal sleep variability show that sleep becomes less stable in the postpartum compared to pregnancy (Signal et al. 2007; Yamazaki et al. 2005) and that lower sleep timing regularity is associated with worse functioning on a psychomotor vigilance test (McBean and Montgomery‐Downs 2013). Research on sleep variability beyond the early postpartum weeks is limited (Kalogeropoulos et al. 2022) and, to the best of our knowledge, only two studies to date examined sleep variability in mothers of infants in relation to their parenting quality (Ran‐Peled et al. 2022; Bai et al. 2023). Thus, in the present study, we examined both average and variability sleep variables.

Another limitation of previous research on maternal sleep and parenting quality is its focus on the direct links between these constructs, overlooking potential moderators. From a developmental psychopathology perspective, individual risk factors explain only a small part of outcome variance; instead, it is the interaction between risk factors that likely explains maladaptation (Goodman and Gotlib 1999). In considering the role of disturbed maternal sleep in parenting, it is important to take into account other risk factors, such as mothers' emotional functioning and the broader context in which the parent–child relationship is embedded (Bronfenbrenner 1986), for instance, the amount of support that mothers receive from significant others (Taraban et al. 2017).

Maternal depression has been associated with poorer maternal sleep (Dorheim et al. 2009; Okun and Lac 2023) and with a more negative mother–infant relationship (Lovejoy et al. 2000). We therefore aimed at examining whether, under higher levels of depressive symptoms, the association between maternal sleep disturbances and the mother–infant relationships would be stronger.

Research on social support suggests that it buffers against stress and negative events and promotes psychological well‐being (Zell and Stockus 2025). Mothers' reports of social support have been linked with reduced depressive and anxiety symptoms (Arnold and Kalibatseva 2021; Cairney et al. 2003) and with greater maternal self‐efficacy and positive parenting behaviors (Suzuki et al. 2009; MacMillan et al. 2021). Social support has also been linked to better sleep quality (de Grey et al. 2018), possibly by protecting against stress and feelings of insecurity, which have been linked to disturbed sleep (Dahl and El‐Sheikh 2007; Pow et al. 2017). Based on this evidence and the fact that little research has moved beyond exploring direct associations between maternal sleep and parenting quality, the present study examined depressive symptoms and social support as moderators of the links between maternal sleep disturbances and maternal sensitivity.

1.1. The Present Study

The high prevalence of maternal sleep disruptions during the postpartum period, along with growing evidence on the adverse effects of poor sleep on emotional functioning—both in the general population and in mothers of infants—raises concerns about the challenges mothers may face in maintaining positive parenting when sleep is chronically disturbed. Therefore, the present study aimed to examine whether maternal objective and subjective sleep disturbances, as well as sleep instability, are associated with lower maternal sensitivity toward her infant. Additionally, it explored whether these links are moderated by depressive symptoms and perceived social support.

The specific aims and hypotheses of this study were as follows:

1. To assess the concomitant and predictive associations between maternal sleep (quality, quanitiy and stability) and maternal sensitivity. We hypothesized that poorer sleep quality, shorter sleep duration and higher sleep variability (i.e., lower stability) assessed objectively and subjectively would be associated with lower maternal sensitivity during a free play interaction.

2. To examine whether the links between maternal sleep and maternal sensitivity are moderated by maternal depressive symptoms and social support. We hypothesised that the links between poorer maternal sleep and lower maternal sensitivity would be stronger for mothers with higher maternal depressive symptoms and lower social support.

2. Methods

2.1. Participants

One‐hundred and fifty‐one married Israeli women expecting their first (n = 135) or second (n = 16) child were recruited during pregnancy through announcements on internet forums for expectant parents and through prenatal courses. Inclusion criteria were singleton pregnancy and full mastery of Hebrew. Exclusion criteria included (1) chronic use of any sleep or psychiatric medication that started prior to pregnancy (based on self‐report); (2) maternal breathing‐related sleep problems or circadian rhythm disorders, diagnosed prior to pregnancy. Post hoc exclusion criteria were (1) Pre‐term pregnancy (i.e., < 37 weeks); (2) Infant chronic health conditions. Two infants were born prematurely, and one infant was born with a major health condition and therefore were excluded.

Because of dropout, the N at 4 months was 130 and at 8 months, N was 108. The main reasons for discontinuation were overload or change of residence. The families who withdrew from the study were compared to the retained families on socio‐demographic variables, and those who withdrew at 8 months were also compared to participating families on the sleep variables at 4 months. No differences were found for any of the variables.

The first phase of the study occurred during the third trimester of pregnancy. At this time, we collected demographic data along with other measures, including parental sleep (not reported in this study). Demographic characteristics are described in Table 1. At 4 months postpartum, the sample included 76 female infants (58.5%).

TABLE 1.

Demographic characteristics of the study sample (N = 151 at pregnancy, 130 at 4 months, 108 at 8 months).

	%	Mean	SD	Min	Max
Maternal age		28.85	2.77	22	37
Maternal education
High school education	11.2
Bachelor's degree	56.5
Master's degree	29.6
Doctoral degree	2.7
Household income
Below average	27.9
Average	23.8
Above average	48.3
Employment status
Pregnancy
75%–100%	70.7
25%–50%	17.7
Bed rest	2.0
Unemployment	9.5
4 months
75%–100%	19.2
25%–50%	22.4
Maternity leave	48.0
Unemployment	10.4
8 months
75%–100%	67
25%–50%	16.5
Leave of absence	5.2
Unemployment	11.3
Feeding method
4 months
Full breastfeeding	65.9
Partial breastfeeding	19.7
Bottle feeding	14.4
8 months
Full breastfeeding	30.6
Partial breastfeeding	30.6
Bottle feeding	38.8

2.2. Procedure

The study was approved by the Helsinki Committee of Soroka Medical Center. All participants signed informed consent during the third trimester of pregnancy after receiving a full explanation of the study procedures. Assessments were conducted at 4 and 8 months postpartum. Participants were visited by a research assistant who provided instructions on the study procedures. Maternal sleep was assessed over seven nights (excluding weekends) using actigraphy and sleep diaries, with mothers completing the diaries each morning. Assessments were conducted only on nights when all family members slept at home and were healthy. Participants completed questionnaires online (via Qualtrics) to assess sleep patterns, socio‐demographic measures, feeding methods, social support and maternal depressive symptoms. Questionnaires were completed during the sleep assessment period. At the end of the 8‐month assessment, participants visited the lab for a 20‐min free‐play dyadic interaction between mother and infant, which was video recorded and later coded using the sensitivity subscale of the Emotional Availability Scales (EAS) (Biringen, 2008). After completing each assessment, participants received a small gift (valued at about $20) and a graphic report of their sleep.

2.3. Measures

2.3.1. Actigraphy

Actigraphy, a motility‐based watch‐like device, is a reliable and valid method to continuously and objectively assess sleep–wake patterns in the participant's natural sleep environment (Sadeh et al. 1994) We used the micromotion logger sleep watch (Ambulatory Monitoring Inc. Ardsley, NY) with a 1‐min epoch interval, according to the standard working mode for sleep–wake scoring.

The Actigraphic Sleep Analysis (ASA) was used to score the data based on Sadeh's validated scoring algorithm for adults (Sadeh et al. 1994). Sleep onset was defined as the first minute after reported bedtime that was identified as sleep by the ASA sleep–wake algorithm and was followed by at least 15 min of uninterrupted sleep. The end of the night was defined as the last minute that was identified as sleep and was preceded by at least 15 min of uninterrupted sleep. Mothers were asked to attach the actigraph to their non‐dominant wrist 15 min before they went to sleep. Most mothers (~93%) had valid actigraphy data for at least 5 assessment nights per assessment period—the minimum number of nights recommended for obtaining reliable measures (Acebo et al. 2005). However, some data were missing because of technical problems (e.g., the actigraph fell off in the middle of the night). Because fewer than 5 days have been reported before with reasonable reliability (Sadeh 2015) and to maximise available data, we retained participants with at least 3 nights of data, after ensuring that the mean scores of actigraphic measures for mothers with fewer than five nights were not significantly different than those who were assessed for five or more nights. Thus, the minimum number of nights included in the analyses was 3. Three‐night assessments were available for two mothers at 4 months and four mothers at 8 months; four‐night assessments were obtained for seven mothers at 4 months and three mothers at 8 months; Seven‐night assessments were available for 89 mothers at 4 months and 71 at 8 months. The actigraphic measures included in the present study were (1) sleep percent—percentage of sleep minutes from total sleep period (from sleep onset to morning wake‐up time) and (2) sleep minutes—true sleep time, excluding night‐time wakefulness. These measures were selected to represent both sleep quality and sleep duration. Variability in maternal actigraphic sleep patterns was examined using the standard deviation for sleep percent and sleep minutes across the measurement period. All measures were averaged across the monitoring period.

2.3.2. Sleep Diaries

Sleep diaries are commonly used in sleep research and have been validated relative to objective sleep measures such as polysomnography (Monk et al. 1994). In the present study, mothers were instructed to complete a sleep diary based on the validated Pittsburgh sleep diary (Monk et al. 1994). The diaries were completed during the same monitoring period as the actigraphic assessment. Most mothers (~96%) had valid diary data for at least five assessment nights per assessment period. Diary data was obtained for at least three nights for all mothers. Three‐night assessments were available for one mother at 4 months and two mothers at 8 months; four‐night assessments were obtained for five mothers at 4 months and two mothers at 8 months; Seven‐night assessments were available for 95 mothers at 4 months and 76 at 8 months. The sleep measures included in the study were: (1) sleep percent—the ratio of maternal‐reported sleep minutes to the total sleep period and (2) sleep minutes. Additionally, variability in maternal‐reported sleep was examined using the standard deviation for both sleep percent and sleep minutes. All measures were averaged across the monitoring period.

2.3.3. Emotional Availability Scales (EAS)

Dyadic mother–infant interactions were videotaped in the lab, and maternal sensitivity was coded using Biringen's 4th edition of EAS (Biringen 2008). This system is designed to assess the capacity of a dyad to share and express an emotionally healthy relationship through observations of mother–child interactions. The EAS provides a sensitive and valid measure of the mother–child relationship quality (Biringen et al. 2014).

The EAS includes six mother–infant relationship dimensions: four scales measure maternal behaviour, and two assess child behaviour. In the present study, we focused on the sensitivity subscale which evaluates how attuned the mother is to the child's emotional needs and assesses the mother's ability to respond to the infant's cues in a warm, affectionate, flexible, accurate and appropriate manner. This focus was chosen because maternal sensitivity is considered the central behavioural construct in observational studies of the parent–child relationship and has been associated with the mother–infant attachment relationship (Leerkes et al. 2024; Mesman and Emmen 2013).

Four coders were trained by one of the authors (N.A.P), who is certified in the EAS. To ensure proper inter‐rater reliability and prevent coder drift, 20% of the interactions were randomly selected and coded by two trained coders. Scores ranged on a scale from 1 to 7. If the coders differed by more than 1 point, they discussed the coding until a consensus score was reached. Intraclass coefficients for absolute agreement were calculated for the sensitivity scale, which demonstrated good inter‐rater reliability (Intraclass correlation coefficient = 0.89).

2.3.4. The Edinburgh Postnatal Depression Scale (EPDS)

The Edinburgh Postnatal Depression Scale (EPDS) is a widely used and validated screening tool for postpartum depression (Cox et al. 1987). It consists of 10 short statements asking about depressive symptoms during the past week. Each item is rated on a four‐point scale (0–3), and the total score ranges from 0 to 30, with a higher score representing greater depressive symptom severity. In the present study, Cronbach's alpha was 0.77 at 4 months and 0.73 at 8 months.

2.3.5. The Social Support Questionnaire 6 (SSQ6)

The SSQ6 is a validated questionnaire that assesses the amount of social support and satisfaction with the received support (Sarason et al. 1987). In the present study, we focused on the amount of social support. The questionnaire consists of 6 items, and for each item, participants are asked to indicate the initials of all the people who provide support for that specific item (with a range of 0–9 people per item). Specific items include: ‘Whom can you really count on to distract you from your worries when you feel under stress; Whom can you count on to care about you, regardless of what is happening to you?’ Total scores were calculated by adding the number of people listed across all items and dividing the sum by the number of items (i.e., six). Cronbach's α was 0.93 at both 4 months and 8 months. A previous validation study (Furukawa et al. 1999) found that the mean score for individuals with psychiatric disorders in a similar age range to our sample was 3.03, while the mean score in a control sample of the same age range was 4.47.

2.3.6. Background Variables

During the third trimester of pregnancy, questionnaires were used to collect socio‐demographic data, including the mother's age, education level and household income. Mother's employment status was assessed at each assessment point. Information on the number of children and the infant's gender was collected at 4 months. Based on prior research indicating significant associations between breastfeeding and both infant and maternal sleep (Volkovich et al. 2015), we assessed feeding method at both 4 and 8 months. We used a three‐point scale: (1) exclusive breastfeeding, (2) partial breastfeeding or (3) formula‐fed.

2.4. Data Analysis

Pearson product–moment correlation coefficients were used to examine the associations between maternal sleep measures and the EA sensitivity subscale. The significance threshold was set at p < 0.05. Two‐step hierarchical multiple regressions were conducted to test whether EPDS and social support scores moderate the relationship between maternal sleep and sensitivity. We followed the process recommended by Aiken and West (1991). Missing data were handled using listwise deletion, and all analyses were conducted using SPSS version 24.0. Moderation analyses were performed using Hayes' PROCESS macro for SPSS (Hayes 2017).

3. Results

3.1. Preliminary Analysis

3.1.1. Descriptive Statistics

Descriptive statistics of the main study variables are presented in Table 2.

TABLE 2.

Descriptive statistics of study variables.

	4 months				8 months
	Mean	SD	Min	Max	Mean	SD	Min	Max
Sleep diaries
Sleep percent	88.36	7.15	63.05	99.17	90.40	6.95	63.73	99.78
Sleep percent SD	6.26	4.79	0.54	24.91	6.19	4.39	0.30	19.85
Sleep minutes	403.45	53.40	276.38	527.83	391.79	51.57	276.00	493.60
Sleep minutes SD	58.07	25.54	13.86	152.81	57.48	25.13	10.68	161.24
Actigraphy
Sleep percent	87.11	6.98	68.58	99.43	89.32	6.21	70.15	98.81
Sleep percent SD	5.34	2.63	0.75	13.31	5.26	2.78	1.11	14.32
Sleep minutes	391.58	51.31	268.50	504.0	384.91	48.57	261.0	503.14
Sleep minutes SD	48.15	18.27	8.38	94.78	46.34	17.34	7.26	113.16
EA maternal sensitivity	NA	NA	NA	NA	5.27	0.40	4.0	6.0
EPDS	4.34	3.35	0	13.0	4.58	3.07	0	13
Social support (SSQ6)	3.60	1.85	0.6	9.0	3.56	1.76	1.0	8.67

Abbreviations: EPDS, The Edinburgh Postnatal Depression Scale; SD, standard deviation; SSQ, Social Support Questionnaire—score range from 0 to 9 indicating the number of people providing support.

3.1.2. Correlations Between Background Variables and Maternal Sleep and Sensitivity

We examined the correlations between demographic variables (i.e., maternal age, education, income and employment hours) and between maternal sleep measures and the EA sensitivity subscale. Because no consistent pattern emerged, we did not control for these variables in subsequent analyses. We also conducted Spearman correlations between feeding method (full breastfeeding, partial breastfeeding or formula feeding) and the main variables. Breastfeeding was associated with lower actigraphic sleep percent at 4 and 8 months (rho = 0.23, p < 0.05 at both time points). Due to these statistically significant correlations, feeding method was controlled for in subsequent analyses.

3.2. Main Analyses

3.2.1. Correlations Between Maternal Sleep and the EA Maternal Sensitivity Subscale

Pearson correlations between maternal sleep at 4 and 8 months and maternal sensitivity at 8 months are displayed in Table 3. The p‐values were adjusted using Benjamini and Hochberg's False Discovery Rate (FDR) correction for multiple comparisons (Benjamini and Hochberg 1995). No statistically significant correlations were found for any of the variables.

TABLE 3.

Concurrent and predictive Pearson correlations between maternal sleep and maternal sensitivity (controlling for breastfeeding).

Maternal sleep	Maternal sensitivity
4 months
Sleep diaries
Sleep percent	−0.20
Sleep percent SD	0.09
Sleep minutes	−0.16
Sleep minutes SD	−0.14
Actigraphy
Sleep percent	−0.15
Sleep percent SD	−0.01
Sleep minutes	−0.06
Sleep minutes SD	−0.10
8 months
Sleep diaries
Sleep percent	−0.06
Sleep percent SD	−0.16
Sleep minutes	0.00
Sleep minutes SD	−0.21
Actigraphy
Sleep percent	0.08
Sleep percent SD	0.05
Sleep minutes	0.05
Sleep minutes SD	0.02

Abbreviation: SD, standard deviation.

3.2.2. Moderation Analysis

The mean EPDS scores are presented in Table 2. Only a few mothers met the clinical cutoff for depression according to the EPDS. At the 4‐month assessment, four women scored in the 10–12 ‘minor depression’ range, and three exceeded the clinical cut‐off score of > 12. At the 8‐month assessment, three women scored in the 10–12 range, and one met the clinical cut‐off score of > 12.

Regarding social support, the mean SSQ6 score was 3.6 (±1.85) at 4 months and 3.56 (±1.76) at 8 months, indicating that, on average, mothers reported having about 3.6 people to rely on at both assessment points.

To examine whether the links between maternal sleep and maternal sensitivity at 8 months were moderated by EPDS and social support (both at 8 months), we conducted a series of two‐step hierarchical multiple regression models. In step 1, we entered the predictor (i.e., one of the maternal sleep variables), the moderating variable (EPDS or social support) and breastfeeding (as a covariate). In step 2, we added the interaction term between the sleep variable and the moderator. This procedure was repeated for each sleep variable (actigraphic and reported sleep percent, sleep minutes and the SD scores). Significant models are presented in Table 4.

TABLE 4.

Hierarchical multiple regression models testing the moderating effects of social support on the associations between maternal sleep and maternal sensitivity.

Dependent variable	Step	Predictor	Total adj R 2	Δ R 2	β	t	F (4,78)
Concomitant model—8 months
EA Sensitivity	Step 1	Breastfeeding	0.19	0.05*	−0.26	−2.37*	5.21**
		Diary S. minutes SD			−0.25	−2.30*
		Social support			0.20	1.86
	Step 2	Breastfeeding			−0.27	−2.52*
		Diary S. minutes SD			−0.31	−2.81**
		Social support			0.22	2.06*
		Diary S. minutes SD × Social support			−0.23	−2.12*
Predictive models—4–8 months
EA Sensitivity	Step 1	Breastfeeding	0.13	0.10**	−0.19	−1.72	3.95**
		Diary S. minutes SD			−0.18	−1.59
		Social support			0.12	1.07
	Step 2	Breastfeeding			−0.12	−1.11
		Diary S. minutes SD			−0.11	−1.03
		Social support			0.22	1.95
		Diary S. minutes SD × Social support			−0.35	−3.06***
EA Sensitivity	Step 1	Breastfeeding	0.11	0.09**	−0.20	−1.80	3.27*
		Actigraphy S. minutes SD			−0.15	−1.29
		Social support			0.09	0.83
	Step 2	Breastfeeding			−0.21	−1.91
		Actigraphy S. minutes SD			−0.12	−1.06
		Social support			0.19	1.64
		Actigraphy S. minutes SD × Social support			−0.31	−2.72**

Note: *p < 0.05; **p < 0.01; ***p < 0.001.

Abbreviations: S, sleep; S. minutes, true sleep time; SD, standard deviation.

For significant interactions, simple slopes were plotted at three levels of the moderator: high (one standard deviation above the mean), average and low (one standard deviation below the mean). Slopes were examined to determine whether they were significantly different from zero (Aiken and West 1991). A similar procedure was applied to test the predictive association between maternal sleep at 4 months (interacting with the moderators at 4 months) and the EA scales at 8 months. Due to the multiple moderation analyses conducted, we applied the Benjamini–Hochberg procedure to control the false discovery rate (FDR) at 0.05. The effects described below remained significant after FDR correction.

3.2.2.1. Moderation by EPDS

There were no significant interactions between the sleep variables and EPDS in the prediction of maternal EA.

3.2.2.2. Moderation by Social Support—Concomitant Links

For diary‐based maternal sleep minutes variability (but not for the other sleep variables), the interaction term significantly explained additional variance beyond the previous step of the model. This indicates that the link between diary‐based maternal sleep minutes variability and maternal sensitivity at 8 months varied depending on the level of social support at that time. Simple slopes analysis revealed that for mothers with high (β = −0.50, SE = 0.16, p < 0.01) and average (β = −0.27, SE = 0.10, p < 0.01) social support, greater sleep minutes variability (indicating lower stability) was associated with lower maternal sensitivity. However, no significant association was found for mothers with low social support (β = −0.05, SE = 0.14, p = 0.73) (see Figure 1a).

FIGURE 1.

(a–c) Social support moderates the association between maternal sleep minutes variability and maternal sensitivity.

3.2.2.3. Moderation by Social Support—Predictive Links

For diary‐based sleep minutes variability and actigraphic sleep minutes variability at 4 months, the interaction terms significantly explained additional variance beyond the previous step of the model. This suggests that the associations between both diary and actigraphic sleep minutes variability at 4 months and maternal sensitivity at 8 months were moderated by the level of social support at 4 months.

A simple slopes analysis revealed that for mothers with high social support, greater diary sleep minutes variability at 4 months predicted lower maternal sensitivity at 8 months (β = −0.41, SE = 0.13, p < 0.01). No significant associations were found for mothers with low (β = 0.20, SE = 0.16, p = 0.21) or average (β = 0.10, SE = 0.10, p = 0.32) levels of social support (see Figure 1b). Similarly, for actigraphic sleep minutes variability, the simple slopes analysis indicated that higher variability was associated with lower maternal sensitivity for mothers with high social support (β = −0.38, SE = 0.14, p < 0.01). No significant associations were found for mothers with low (β = 0.14, SE = 0.16, p = 0.38) or average (β = −0.12, SE = 0.12, p = 0.32) social support (see Figure 1c).

4. Discussion

To the best of our knowledge, this study was the first to examine the moderating role of maternal depressive symptoms and social support on the links between maternal sleep and the mother–infant observed relationship in the context of a longitudinal design utilising objective and observational measures.

Contrary to expectations, we did not find support for significant simple associations between any of the sleep variables and maternal sensitivity. Although growing evidence suggests that poorer maternal sleep is linked to maternal reports of more negative feelings toward the infant, findings regarding the association with observed parenting quality have been sparse and inconclusive (King et al. 2020; Bai et al. 2023). Our findings align with a few studies that did not identify significant links (Leerkes et al. 2024; Kim and Teti 2014). One possible explanation for this lack of association could be the specific nature of the mother–infant interactions observed in this study, which focused on free play. During these brief interactions, and knowing they are being filmed, mothers may still succeed in being sensitive and responsive, even when experiencing poor sleep. Indeed, overall, the EA sensitivity scale scores in our sample were on the higher end. It could also be, as has been suggested before, that maternal sleep is more strongly associated with mother–infant interactions that occur around bedtime than during the day. This is because mothers with disturbed sleep may be particularly vulnerable to the effects of poor sleep and heightened fatigue at the end of the day (Bai et al. 2023).

Another possible explanation for the lack of significant links is that disturbed maternal sleep may not function as a standalone risk factor, but rather interacts with other factors to explain the variance in the observed mother–infant relationship (Leerkes et al. 2024). To explore this, we conducted moderation analyses to examine whether maternal sleep interacts with depressive symptoms and social support in predicting maternal sensitivity. We hypothesized that the links would be stronger for mothers with higher depressive symptoms and lower social support.

Overall, our findings did not support the role of depressive symptoms as a moderator. Notably, in our study, only a few mothers scored above the clinical cutoff of depression (> 12) based on the EPDS (three mothers at 4 months and one at 8 months). Perhaps in a sample with greater variance and severity, depressive symptoms would significantly interact with sleep problems in explaining maternal sensitivity.

Regarding social support, we hypothesised that the relationship between poorer maternal sleep and lower sensitivity would be stronger for mothers with lower social support, while this relationship would be attenuated for mothers with higher social support. This hypothesis was based on previous findings suggesting that social support buffers against the harmful effects of stress in the general population (Zell and Stockus 2025) and is associated with better adjustment of mothers during the postpartum period (Arnold and Kalibatseva 2021; Racine et al. 2019). However, our findings revealed significant links between poorer maternal sleep and lower maternal sensitivity only for mothers who reported higher social support. For mothers with lower social support, no significant associations between sleep and maternal sensitivity were found, suggesting that these mothers might be less vulnerable to the negative effects of poor sleep, and maybe for them, other risk factors interacting with social support, such as mental health, are more critical in explaining parenting quality.

The results showing stronger links in mothers with high social support are consistent with another study of mother–toddler dyads that tested the moderating role of social support satisfaction on the association between maternal depressive symptoms and parenting (Taraban et al. 2017). This study reported that the association between maternal depressive symptoms and harsh parenting was strongest for mothers with higher levels of social support satisfaction. The explanation offered by the authors for this surprising finding was that mothers with high social support, when facing depressive symptoms, may turn to their spouses, family or friends to assume parenting duties. Consequently, these mothers may have fewer experiences in caregiving, which limits their opportunities to develop and sustain high‐quality parenting skills in the context of depression compared to mothers with depressive symptoms who do not feel they can depend on others for support (Taraban et al. 2017).

Similarly, in the context of this study's findings, it is possible that when mothers with higher social support experience sleep disturbances, particularly irregular and unpredictable sleep, they are more likely to rely on others for help with infant care. This reliance may lessen their opportunities for practising sensitive and attuned interactions with their infants during free play. In contrast, mothers with lower support may have developed greater resilience to maintaining sensitivity while experiencing disturbed sleep, as they are less likely to rely on others. Another possible explanation is that mothers with high social support may set higher expectations and standards for their parenting due to their greater resources. Perhaps, sleep disturbances hinder their ability to meet these standards, thereby increasing stress and reducing sensitivity. However, as these explanations are speculative, and causality cannot be inferred from our correlational data, our suggestions should be interpreted with caution and investigated in future studies.

Notably, the only sleep variables that interacted with social support in explaining maternal sensitivity were the sleep minutes variability measures, and not the average sleep minutes and sleep percent variables. Specifically, greater variability in diary‐based sleep minutes was concurrently associated with lower maternal sensitivity in these mothers at 8 months postpartum. Additionally, greater variability in both actigraphic and diary‐based sleep minutes at 4 months predicted lower sensitivity at 8 months. Our findings are in line with one previous research study which found that greater maternal variability in sleep period was associated with lower maternal emotional availability at bedtime during the first 6 months (Bai et al. 2020). The prospective links found in our study suggest that sleep variability interacts with social support to predict later maternal sensitivity.

Sleep variability has received limited attention in research so far, though accumulating evidence suggests that greater sleep variability is associated with higher physical and mental health problems (Bei et al. 2016). In postpartum women, less stable sleep timing was associated with worse neurobehavioral functioning (McBean and Montgomery‐Downs 2013). Based on this previous body of literature, there are several possible explanations for our findings. One possibility is that experiencing unpredictable sleep may heighten maternal concerns and anxiety about sleep quality and daytime functioning (Bei et al. 2016), which may reduce the mother's capacity to maintain sensitive and responsive care (Ran‐Peled et al. 2022; Bai et al. 2020). It could also be speculated that unstable sleep may impair mothers' emotion regulation (Palmer and Alfano 2017), and lead to reduced vigilance and energy levels and increased fatigue (McBean and Montgomery‐Downs 2013; Zohar et al. 2005), making it harder to cope with the demands of caregiving, which could manifest as reduced sensitivity during interactions with the infant. Nevertheless, given the lack of research on the mechanisms through which sleep variability is associated with functioning (Bei et al. 2016) and as none of the suggested potential mediators were examined in this study, these theoretical explanations should be interpreted cautiously and explored in subsequent research.

To summarise, the results suggest that maternal higher sleep variability is more strongly linked with maternal lower sensitivity than other maternal sleep characteristics. However, contrary to our hypothesis, these effects were only apparent in mothers with higher levels of social support.

4.1. Study Limitations

The findings of this study must be considered in light of several limitations. First, although the longitudinal design is a strength, the study remains correlational, and no causal conclusions can be drawn. Additionally, the sample was relatively homogeneous, consisting only of married mothers from middle to upper socioeconomic backgrounds in Israel. Moreover, most participants scored in the non‐clinical range for depression, limiting the generalisability of the findings to mothers with more severe depression and reducing the likelihood of detecting a moderating effect. Therefore, further studies should examine the potential moderating role of depression in a sample with higher and more variable depressive symptom scores. Another limitation is that the study focused exclusively on mothers; further research should explore sleep disturbances and parent‐infant relationships among both mothers and their partners.

4.2. Conclusions and Clinical Implications

This study adds to the growing body of knowledge on the role of maternal sleep in the mother–infant relationship by showing, for the first time, that both objective and self‐reported variability in maternal sleep interact with social support to predict observed maternal sensitivity. The findings, which reveal significant links between sleep variability and maternal sensitivity only in mothers with high levels of social support, are intriguing. One possible explanation is that these mothers may rely more on others for help with infant care, potentially reducing their opportunities to practise sensitive caregiving. However, further research is needed to replicate these findings before drawing firm conclusions. Future research could also explore potential mediators of the links between maternal sleep variability and maternal sensitivity, such as sleep‐related anxiety, emotion regulation, energy depletion and fatigue. Additionally, future intervention research should examine whether efforts to stabilise maternal sleep could lead to enhanced maternal sensitivity.

Author Contributions

Dar Ran‐Peled: conceptualization, writing – original draft, methodology, writing – review and editing, formal analysis, investigation, project administration, data curation, visualization. Avel Horwitz: investigation, writing – review and editing, project administration, data curation. Omer Finkelstein: investigation, writing – review and editing, project administration, data curation. Naama Atzaba‐Poria: methodology, writing – review and editing, supervision. Liat Tikotzky: conceptualization, investigation, funding acquisition, writing – original draft, writing – review and editing, methodology, supervision, validation.

Conflicts of Interest

The authors declare no conflicts of interest.

Ran‐Peled, D. , Horwitz A., Finkelstein O., Atzaba‐Poria N., and Tikotzky L.. 2025. “A Longitudinal Study of Postpartum Maternal Sleep and Sensitivity: Examining Depressive Symptoms and Social Support as Moderators.” Journal of Sleep Research 34, no. 5: e70056. 10.1111/jsr.70056.

Data Availability Statement

The data underlying this article will be shared on reasonable request to the corresponding author.