Objectives. To describe infant sleep patterns and investigate relationships between infant sleep problems and maternal well-being in the community setting.
Design. Cross-sectional community survey.
Setting. Maternal and Child Health Centers in 3 middle-class local government areas in Melbourne, Australia.
Participants. Mothers of infants 6 to 12 months of age.
Main Outcome Measures. Maternal well-being (Edinburgh Postnatal Depression Scale) and infant sleep problems (standardized maternal questionnaire).
Results. The survey was completed by 738 mothers (94% response rate), of whom 46% reported their infant's sleep as a problem. In the univariate analyses, sleep patterns characterizing a sleep problem included the infant sleeping in the parent's bed, being nursed to sleep, taking longer to fall asleep, waking more often and for longer periods overnight, and taking shorter naps. The same sleep patterns were associated with high depression scores and tended to increase as depression scores increased.
Because of positive skew, the Edinburgh Postnatal Depression Score was analyzed in 3 categories (<10, 10–12, and >12) using validated cutoff scores from community and clinical studies. Fifteen percent of mothers scored above 12 on the depression scale, indicating probable clinical depression, and 18% scored between 10 and 12, indicating possible clinical depression. After adjusting for potential confounders and factors significant in the univariate analyses, maternal report of an infant sleep problem remained a significant predictor of a depression score >12 (odds ratio: 2.13; 95% confidence interval: 1.27,3.56) and >10 (odds ratio: 2.88; 95% confidence interval: 1.93,4.31). However, mothers reporting good sleep quality, despite an infant sleep problem, were not more likely to suffer depression.
Conclusions. Maternal report of infant sleep problems and depression symptoms are common in middle-class Australian communities. There is a strong association between the 2, even when known depression risk factors are taken into account. Maternal report of good sleep quality attenuates this relationship. Appropriate anticipatory guidance addressing infant sleep could potentially decrease maternal report of depressive symptoms.
Infant sleep problems and postnatal depression (PND) are important, not only because of their high prevalence, but also because of their adverse sequelae. In Australia, more than one third of parents report problems with their infant's sleep.1 Estimates of the prevalence of PND range between 10% and 15% in the general population,2 with an even higher proportion reporting problems that, although less severe, still significantly reduce maternal well-being.3
Infant sleep problems and PND can both impact adversely and in apparently similar ways on the mother, the infant, and the mother–infant relationship.2,,4 Infant sleep disruption can lead to maternal sleep disruption, resulting in adverse effects on mood, cognition, and motor function,5 similar symptoms to those reported by mothers suffering from PND.2 Infants with sleep problems are more likely to be tired, inattentive, and irritable and have difficulty modulating their impulses and emotions.6 Children of depressed mothers experience similar problems as well as attachment difficulties and poorer cognitive and behavioral outcomes.7,,8
How infant sleep problems and PND are related (and, in particular, whether 1 causes the other) is yet to be established. Case–control studies consistently report more depressive symptoms in mothers of poor sleepers than in mothers of good sleepers,9,,10 regardless of social class.11 Maternal depressed mood predicts both the presence of sleep problems9 and their persistence.12 In 1 sample of consecutive mothers attending a pediatric sleep outpatient clinic, 40% reported symptoms of PND.13 Uncontrolled sleep intervention studies have demonstrated significant reductions in maternal depression scores after treatment of the child's sleep problem.13,,14
We investigated the relationship between maternal depression and sleep problems for 3 reasons. First, important questions remain unanswered. What sleep behaviors do mothers consider problematic? Do depressed mothers differ from nondepressed mothers in reporting their infant's sleep behaviors? Are depressed mothers characterized by a cluster of maternal, infant, and demographic variables, possibly including perceived infant sleep problems? Second, there are few good data describing the relationship between maternal depression symptoms and infant sleep problems in the community setting. Third, previous studies have tended to use nonvalidated depression measures or measures containing items about somatic symptoms that are normal changes in the postnatal setting. In addition, small sample sizes, poor response rates, and researcher-imposed definitions of sleep problems have reduced the validity and generalizability of the data.
This survey addressed these issues. We examined the relationship between PND and infant sleep problems in a large Australian community sample. We used a well-validated PND screening scale and standard definitions of parent-reported infant sleep problems.
Between May 1998 and April 1999, mothers attending routine hearing testing sessions at participating Maternal and Child Health Centers in 3 statistical local government areas in suburban Melbourne, Australia, were invited to take part. The areas chosen represented predominantly middle-class mothers who were deemed likely to participate in a second study involving a community-based treatment of infant sleep problems. The Maternal and Child Health service is provided free to all children in the state of Victoria 0 to 6 years of age, and ∼90% of children are known to attend the hearing sessions offered to all infants 7 to 9 months of age.
After giving verbal consent to the principal investigator, mothers completed a 48-item survey while waiting for their infant's appointment or were provided with a reply paid envelope if they chose to take the survey home. The survey asked for details of the infant's sleep over the previous 2 weeks and whether the mother considered her infant's sleep to be a problem. If the infant's sleep was a problem, mothers were asked to rate its severity on a 7-point visual analog scale designed for the study. Maternal well-being was assessed by the Edinburgh Postnatal Depression Scale (EPDS),15 a 10-item scale with scores ranging from 0 to 30. Scores >10 and 12 indicate probable depression in community and clinical samples, respectively.16,,17 Mothers were also asked whether they had been previously diagnosed with depression. Infant factors (sex, birth weight, gestation, delivery type, health, ordinal number, breastfeeding status, and use of childcare) and sociodemographic factors (maternal and paternal education, employment status, occupational prestige for current or most recent occupation [ANU3 Occupational Prestige Scale],18 country of birth, maternal age, and marital status) were sought.
An item regarding maternal sleep quality19 was added for the final 305 questionnaires because maternal comments indicated that this might be an important intermediary between infant sleep and maternal well-being.
Ethics approval was obtained from the Ethics in Human Research Committee of the Royal Children's Hospital, Melbourne, Australia.
Because of positive skew, the EPDS score was analyzed in 3 categories using cutoff scores recommended in clinical and community studies (<10, 10–12, and >12). Univariate analyses were conducted to characterize: 1) sleep problems and 2) the relationship among infant, maternal, and sociodemographic variables and the 3 EPDS categories. Means and standard deviations (SDs) are reported for continuous variables that were approximately normally distributed, and medians and ranges for those that were not. Pearson's χ2analyses were used for categorical variables. Two-sample ttests and analysis of variance were used to test differences in means of continuous parametric variables, while Wilcoxon rank-sum tests were used for nonparametric variables.
For the multivariate analyses, the EPDS score was dichotomized in 2 ways: 1) using a cutoff score of 10 or more16 and 2) using a cutoff score of 12.17 The resulting binary variables were analyzed with logistic regression models to explore the association with infant sleep problems. The final models included risk factors that held an association with the dichotomized EPDS scores and variables that confounded the EPDS–sleep problem association. Data were analyzed with Stata, Version 6 (Stata Corp, College Station, TX).
Of 815 eligible mothers, 674 (86%) completed the survey at the session; 110 (13%) took the survey home and 67 (8%) returned it; 1 declined to complete the survey and 30 (4%) left before being approached. The response rate from those approached was 94% (n = 738).
Table 1 shows the characteristics of the sample. As expected from the geographical areas sampled, mothers were in general well-educated (62% tertiary degree), Australian born (75%), and married to employed partners (96%).
Sleep disruption was prevalent. Fifty-nine percent of infants were reported to wake 4 or more nights per week (with 37% of the total sample waking every night); only 13% were not reported to wake overnight. The average child in the sample went to bed at 8 pm, woke once overnight, rose at 7:30 am, and had 2 naps per day lasting 75 minutes.
Forty-two percent of mothers reported unusual events that they perceived had affected their child's sleep over the previous 2 weeks. These included illness (reported by 26% of all mothers), teething (15%), travel (9%), guests (7%), and other (9%).
Factors Associated With Report of Sleep Problems
Forty-six percent of mothers reported their infant's sleep as a problem, of whom >40% rated the severity of the problem as ≥5 of a possible 7. Sleep patterns were significantly different for those reporting a sleep problem (Table 2). In particular, infants with a sleep problem were more likely to sleep in the parent's bed, be nursed to sleep, take longer to fall asleep, wake more often and for longer periods overnight, and take shorter naps (allP < .01). Mothers who were exclusively breastfeeding were more likely to report a sleep problem (χ2= 16.12; df = 1; P < .001), compared with nonbreastfeeding mothers or those who gave their infant both formula and breast milk. Mean maternal occupational prestige scores were lower for those reporting sleep problems (55.2 vs 28.6;t = 2.6; P = .02), but other infant and sociodemographic factors did not differ.
Factors Associated With Maternal Well-Being
The median EPDS score was 7 (mean: 7.6; SD: 4.2; range: 0–25).Table 3 outlines sleep and nonsleep characteristics by EPDS category. Of the sleep characteristics, maternal report of an infant sleep problem was most strongly associated with EPDS category (P < .001). EPDS scores increased with sleep problem severity (P < .001). Mothers with a higher EPDS score were more likely to sleep with their child, nurse them to sleep, and report more nighttime wakings more nights per week (all P < .02).
A number of nonsleep characteristics were associated with a high EPDS score (Table 3). These included a past history of maternal depression, being divorced, and having a partner born overseas.
Logistic Regression Analyses
Many individual sleep patterns were strongly associated with maternal report of a sleep problem. Because of this high collinearity, only 2 sleep patterns continued to be independently associated with high EPDS scores (duration of night waking and infant sleeping in the parents' bed) over and above maternal report of a sleep problem. The final model, therefore, included maternal report of a sleep problem, duration of night waking, infant sleeping in the parents' bed, history of maternal depression, and marital status/partner born overseas (both predictive of depression in the univariate analyses). In this model, maternal report of a sleep problem continued to be strongly associated with EPDS scores greater than both 10 and 12 (Table 4). Mothers with a history of depression were also more likely to have high EPDS scores. Mothers whose partners were born overseas were more likely to score >10 on the EPDS (odds ratio [OR]: 1.91; P = .003) and a similar effect was observed for those scoring over 12 (OR: 1.59; P = .08).
For the subgroup of mothers (42%) with sleep quality data available, logistic regression analyses were repeated with this predictor variable included. These mothers did not differ from other mothers in terms of marital status, past history of depression, education level, partner born overseas, or report of sleep problem. After controlling for sleep quality as good or very good, the presence of an infant sleep problem no longer predicted an EPDS score above 10 (OR: 1.22; 95% confidence interval [CI]: 0.64–2.35) or above 12 (OR: 1. 27; 95% CI: 0.59–2.72). Infants whose mothers reported good sleep quality were more likely to sleep in their own bed, wake less often, and settle themselves to sleep.
Infant sleep problems were reported by almost one half of the mothers in our study. Infants reported to have a sleep problem were significantly more likely to wake frequently and for longer periods, sleep in their parent's bed, and need an adult to settle them to sleep. These behaviors are usually learnt and, as such, are amenable to change through behavior modification.20 If parents could be helped to teach their infant to settle independently, mothers would not need to attend night wakings. Because these night wakings were also associated with a high EPDS score, this might decrease maternal report of depression symptoms.
Depressed mothers were more likely to report the cluster of sleep patterns that characterized an infant sleep problem than nondepressed mothers. There could be several reasons for this. Depressed mothers may sleep poorly and may be more aware of their infant's night wakings than nondepressed mothers who sleep through the night. Infants of depressed mothers may be more difficult to settle and may wake more frequently. Depressed mothers may be biased in their report of sleep problems, an issue that has not been addressed in the literature. Alternatively, infants who have not learnt to settle independently may disrupt their mother's sleep resulting in maternal sleep deprivation and report of symptoms similar to depression. Because this study was cross-sectional, we cannot define causality. However, it is clear that depressed mothers perceive their infant's sleep patterns quite differently from nondepressed mothers.
Our results confirm previous findings that child sleep problems and maternal report of depression symptoms are common in middle-class Australian communities.1,,21 A strong relationship exists between the 2, even when established risk factors for PND, such as a past history of depression, are taken into account. Maternal sleep quality may be an important mediator in the relationship between depression and infant sleep problems because mothers who reported their sleep quality as good or very good were less likely to report depressive symptoms, even when they perceived their infant's sleep to be a problem. This suggests that other factors are operating to promote good maternal sleep quality and protect mothers from depressive symptoms in the presence of an infant sleep problem.
The high response rate (94%) in our study suggests that our results are likely to be representative of middle-class Australian populations. Prevalences of sleep problems and PND in this study were similar to Australian studies using the same measures across broader sociodemographic strata,1,,21 suggesting that our results may, in fact, generalize beyond a middle-class sample. Given that both PND and sleep problems occur across all socioeconomic classes, it is unlikely that mothers from lower classes will be suffering less than are mothers in our sample.
Subjective report of an infant sleep problem, and its severity if present, was the principal variable against which we assessed reports of depression symptoms. Several more objective parent report measures have been proposed, usually requiring a certain number of wakings for a certain duration occurring a certain number of nights per week.10,,22 Such definitions may severely underestimate the impact of sleep disruption experienced by families. For example, in the study by Morrell23 only 17.3% of infants met Richman's criteria, but 35% had a sleep problem defined by the mother. From a clinical point of view, parent report may provide the most relevant and useful measure of a sleep problem. Furthermore, there is good evidence that mothers of poor sleepers report their child's sleep patterns more accurately than mothers of good sleepers, compared with overnight video recording,24 and in this study, objective indicators of disturbed infant sleep were strongly related to maternal reporting of a sleep problem.
Infant sleep problems are common and should be taken seriously. Many mothers reporting sleep problems are experiencing symptoms of depression, even if they do not reach a cutoff score for clinical depression. Many of the sleep patterns perceived by mothers as sleep problems are amenable to change, and much of this change could potentially be implemented in the primary care setting. Controlled trials are required to establish whether treatment of infant sleep problems can improve maternal well-being.
Randomized, controlled trials have demonstrated that with appropriate anticipatory guidance, problem sleep patterns are preventable.25,,26 Future research needs to examine whether such guidance can reduce the incidence of maternal report of depression symptoms. In the interim, practitioners should be aware that a mother who reports infant sleep problems is likely to be experiencing emotional difficulties. Attending to the sleep problem may not only improve the infant's sleep, but may also have significant ramifications for the well-being of the mother, her infant, and her family.
This project was funded by the Research Institute, Royal Children's Hospital, Melbourne, and a Public Health Postgraduate National Health and Medical Research Council Scholarship.
- Received January 3, 2000.
- Accepted September 26, 2000.
Reprint requests to (H.H.) Centre for Community Child Health, Royal Children's Hospital, Parkville VIC 3052, Melbourne, Australia. E-mail:
- PND =
- postnatal depression •
- EPDS =
- Edinburgh Postnatal Depression Scale •
- SD =
- standard deviation •
- OR =
- odds ratio •
- CI =
- confidence interval
- Lozoff B,
- Wolf AW,
- Davis N
- Richman N
- Zuckerman B,
- Stevenson J,
- Bailey V
- Cox JL,
- Holden JM,
- Sagovsky R
- Murray L,
- Carothers AD
- Boyce P,
- Stubbs J,
- Todd A
- Jones FL
- ↵Ferber R. Sleeplessness in children. In: Ferber R, Kryger M, eds.Principles and Practice of Sleep Medicine in the Child. Philadelphia, PA: WB Saunders Company; 1995:79–89
- Moore T,
- Ucko LE
- Adair R,
- Zuckerman B,
- Bauchner H,
- Philipp B,
- Levenson S
- Copyright © 2001 American Academy of Pediatrics