Determinants of postpartum sleep duration and sleep efficiency in minority women

Abstract

Study Objectives

To examine demographic, psychosocial, and behavioral determinants of postpartum sleep duration and sleep efficiency among a cohort of black and Latina women.

Methods

Data were from 148 women (67% black, 32% Latina) at 5 months postpartum, recruited from an academic medical center in Philadelphia. Relevant demographic, psychosocial and behavioral predictors were assessed via questionnaire. Nocturnal sleep was objectively measured for 1 week using wrist actigraphy. Sleep duration was examined as a continuous variable and in categories (<7 versus ≥7 h per night); sleep efficiency was examined as a continuous variable. Independent multiple linear regression models were built to evaluate significant determinants of sleep.

Results

Adjusted models revealed that breastfeeding, having a bedtime after midnight, and being employed were associated with shorter sleep duration (–25–33 min, all p < 0.05). Multiparity, being unmarried, being employed, breastfeeding, having a bedtime after midnight, bedsharing, and responding to infant awakenings by getting up immediately rather than waiting a few minutes to see if the infant fell back asleep, were all significant determinants of sleeping <7 h per night (OR varying: 2.29–4.59, all p < 0.05). Bedsharing was the only variable identified from the multiple regression model that associated with poorer sleep efficiency (–3.8%, p < 0.05).

Conclusions

Findings may inform interventions for improving postpartum sleep in socioeconomically disadvantaged, racial/ethnic minority postpartum women.

Statement of Significance.

Sleep remains significantly disrupted at 5 months postpartum among socioeconomically disadvantaged, racial/ethnic minority mothers. The majority (62%) of the women in this study were sleeping less than the recommended 7–9 h per night and many women experienced a lot of wakefulness during their nocturnal sleep period. Several modifiable behaviors, such as maintaining an early bedtime, refraining from bedsharing and waiting a few minutes for the infant to self-soothe during nighttime awakenings, were identified as predictors of short sleep duration and/or poor sleep efficiency. These results can inform future interventions aimed at improving sleep in this population and support policy changes to optimize the health of new mothers, including paid family leave and increased support for breastfeeding.

Introduction

Postpartum sleep undergoes dynamic changes as the woman and infant adapt and develop feeding, activity, and sleep–wake rhythms. During the postpartum period, women report more nighttime awakenings, and lower sleep efficiency (the amount of time actually asleep while in bed) than during the late pregnancy period [1, 2]. Excessive daytime sleepiness (Epworth Sleepiness Score ≥ 12) is also common among postpartum women [1, 2]. Predictors of insufficient sleep duration and low sleep efficiency during the postpartum period include hormonal changes, anxiety and mood disturbance, childcare demands, the sleep and feeding schedules of children, and the sleep quality of children [3–9]. During the postpartum period, women also experience major life changes related to work and home responsibilities, relationships with family and friends, and personal finances that have also been associated with increased risk of sleep disorders [10–12]. Poor postpartum sleep associates with adverse health outcomes for the mother and child, including stress level, mental health, interpersonal relationships, postpartum weight retention, birth outcomes, and child development [4].

Data suggest that racial/ethnic minority women and those who are socioeconomically disadvantaged experience greater sleep disruption during the postpartum period than white, socioeconomically advantaged women [4, 13, 14]. For example, in a sample of majority white (93%) and married (91%) mothers who were well-educated, with higher than average annual income, sleep duration during the first 4 months postpartum was relatively stable at approximately 7.2 h per night, and sleep efficiency increased from 80% at 2 weeks postpartum to approximately 91% at 4 months postpartum [15]. The authors posited that sleep efficiency improved as sleep became more consolidated and less interrupted by infant awakenings [15]. Conversely, we observed that in a sample of black and Latina socioeconomically disadvantaged mothers, with the majority not married (75%), sleep duration during the first 5 months postpartum was also stable but was only approximately 6.5 h per night. In comparison, sleep efficiency was also much lower and did not show the same magnitude of improvement over time (6 weeks postpartum: 74%, 5 months postpartum: 78%) in this group [16]. These disparities in postpartum sleep are consistent with studies of the general population demonstrating that socioeconomically-disadvantaged, racial-minority adults in the United States are at increased risk for habitual short sleep duration and poor sleep quality [17–20].

There are likely unique demographic, psychosocial, and behavioral determinants of postpartum sleep in racial/ethnic minority women who are socioeconomically disadvantaged. Thus, we aimed to identify important predictors of objectively measured sleep duration and sleep efficiency at 5 months postpartum in this vulnerable population. We were most interested in exploring predictors of sleep at this time point due to previously reported associations of 5- or 6-month postpartum sleep duration with weight change over the first postpartum year [16, 21].

Methods

Study design and participants

Study subjects were participants in a longitudinal cohort study of maternal cardiometabolic risk, sleep, and offspring health over the first year after childbirth [16]. Cohort participants were recruited between 2012 and 2013 in late pregnancy or at delivery from Temple University Hospital, an urban academic medical center in Philadelphia that serves predominately black and Latina Medicaid-insured patients. Eligibility criteria included age ≥16 years, <1 month postpartum at recruitment, fluency in English, and plans to stay in Philadelphia for the first postpartum year. Women were excluded if they delivered multiples (e.g. twins, triplets), had pre-existing cardiometabolic disease (e.g. cardiovascular disease, diabetes mellitus, and hypertension), or carried a diagnosis of obstructive sleep apnea. All participants provided written informed consent, and the institutional review board of Temple University approved the study.

Assessment of outcomes: objectively measured postpartum sleep duration and sleep efficiency

Sleep duration and sleep efficiency were assessed using an actigraph wristwatch (AW-2, Philips Respironics) that participants were asked to wear on their non-dominant wrists continuously for 7 days and nights. Actigraphs use highly sensitive accelerometers to measure gross motor activity, analyzed to identify sleep periods [22]. Wrist actigraphy has been compared to polysomnography (PSG)—considered the gold standard for measuring sleep, demonstrating high accuracy (87%), and sensitivity (97%) for detecting nighttime sleep, but low specificity (33%) for detecting wakefulness [23]. However, actigraphy is preferable for measuring nocturnal sleep duration and efficiency, because it is unobtrusive, records for multiple nights at a lower cost than PSG and eliminates participant burden or laboratory effects.

The actigraph device had an event marker that could be pressed to indicate specific times; participants were asked to press this marker when they got in and out of bed at night for sleep. Participants were also asked to keep a daily sleep log (adapted from our prior research) [24], recording their nighttime sleep onset time, morning wake time, and any periods of device removal. An actigraphy recording was considered successful if there was a minimum of three 24-h periods recorded and if there was no off-wrist time within these periods noted in sleep log records.

Actigraphy data were analyzed by two trained research staff with the use of Actiware Software version 6.0.0 (Phillips Respironics). We first used the manufacturer default setting of 10 min immobile for determination of nocturnal sleep onset and end times, followed by verification or manual adjustment of the sleep period by trained staff after review of each participant’s event marker data. Because naps may differ from overnight sleep in length and amount of deep sleep and rapid-eye-movement sleep [25], scoring algorithms for nighttime sleep estimation may not be equivalent for naps and have not been validated [26]. Therefore, even though participants were wearing the device, we were unable to use actigraphy to accurately measure daytime sleep. Sleep logs aided scoring nocturnal sleep when participants forgot to push event markers. Large discrepancies in sleep onset/end times were brought to biweekly meetings for adjudication. For each 30-s epoch, a medium sensitivity threshold (40 activity counts/min) was used to calculate sleep/wake variables within the interval between sleep onset time and sleep end time [27].

For each participant, all valid nights were included when calculating mean sleep duration and sleep efficiency. The nocturnal sleep period is the number of minutes between sleep onset and sleep offset (defined above). Sleep duration reflects time spent asleep during the nocturnal sleep period and is defined as the number of minutes scored as sleep during the sleep period. Sleep efficiency reflects the percentage of time spent asleep during the sleep period and is as defined as sleep duration (minutes)/sleep period (minutes) × 100. Sleep duration and sleep efficiency were used as continuous variables. We also examined sleep duration by category (<7 h per night and ≥7 h per night), as this cut-point is consistent with American Academy of Sleep Medicine (AASM) recommendations for adequate sleep duration in adults and has been shown to be clinically relevant [28], including in cohorts of postpartum women [16]. We were not able to examine sleep efficiency by category due to the relatively small number of women (n = 17) who were above the AASM recommended clinical cut-point (85%) [29].

We chose to focus on determinants of sleep at 5-month postpartum because at 5 months, most typically developing infants have established the ability to sleep for 8 h, which leads to less disruption for the mother. In addition, we [16], and others [21], have observed that sleep at 5- or 6-month postpartum predicts subsequent postpartum weight gain in this population.

Assessment of predictors

Sociodemographic and medical variables

We collected information on maternal race/ethnicity, age, education, employment, parity, marital status, and eligibility to receive benefits from the Special Supplemental Nutrition Program for Women, Infants, and Children (income proxy), all reported by the participants at enrollment in the early postpartum period. We measured weight in light clothing, without shoes, using calibrated SECA digital scales at 5 months postpartum and categorized participants as having normal/overweight or obesity based on their body mass index (BMI). To assess postpartum weight retention at 5 months postpartum we calculated the difference between 5-month postpartum and early pregnancy weights (early pregnancy weight was abstracted from participant medical records).

Psychosocial and behavioral variables

At 5 months postpartum, participants reported smoking habits, infant milk feeding, and childcare. Specifically, participants were asked “Are you a current smoker?” and responded Yes or No. If they responded Yes, a follow-up question asked “How many cigarettes do you smoke on a normal day?” with the following responses: <5, 5–10, 10–20, >20. Regarding feeding modality, participants were asked “Since your last visit with us (at 1 month), what kind of milk have you been feeding your baby?” and responded: Breastmilk only, Formula and Breastmilk, Formula only, Other _____. We then categorized these responses as any breastfeeding (exclusive breastfeeding or breastfeeding/formula mix) compared with exclusive formula feeding. Regarding childcare, participants were asked “Was your baby cared for by someone other than you on a regular schedule during the past 4 weeks? That is, did someone else usually keep your baby at least once a week for three or more hours at a time?” and responded Yes or No. Our measure of childcare comes from The Infant Feeding Practices Study II (IFPS II, conducted by the Food and Drug Administration (FDA) and the Centers for Disease Control and Prevention (CDC) in 2005–2007) [30] and is inclusive of the baby’s father, grandparent(s), other family member or someone not in the family.

Symptoms of postpartum depression were evaluated using the 10-item Edinburgh Postnatal Depression Scale (EPDS) [31, 32], a widely used and validated self-report screening measure. The EPDS asks women to report how often they have experienced different feelings (e.g. enjoyment, self-blame, worry, fear) over the past 7 days by selecting one of four options (e.g. most of the time, some of the time, not very often, never). Each item is scored 0–3 (with some items being reverse scored); therefore, scores range from 0 to 30 with higher scores reflecting greater depressive symptoms. We used a cut-point of ≥9 to define possible depressive symptoms, as this has been validated in a large sample of low-income, urban women [33].

The Brief Infant Sleep Questionnaire was used to assess sleep patterns, parent perception, and sleep-related behaviors in young children (0–36 months) [34]. This widely-used and validated questionnaire asks respondents to report about their child’s sleep by selecting from pre-determined answers or writing-in a response [35, 36]. Items inquire about the child’s sleeping arrangement, sleeping position, nocturnal sleep duration, daytime sleep duration, nighttime awakenings, snoring, sleep latency, bedtime routine, bedtime, and sleep problems. We dichotomized responses to the “sleep arrangement” item to determine if bedsharing was practiced (i.e. if participant responded that infant slept in parents’ bed vs. responded that infant slept somewhere besides parents’ bed). One of the questions related to nighttime awakenings included a follow-up question asking “If your child wakes up during the night, what do you do? Check all that apply.” Responses were: Pick up my child and hold/rock him/her until child falls asleep, Pick up my child and put him/her back down while child is still awake, Rub or pat my child but do not pick up or take out of crib/bed, Bottle feed child back to sleep, Breastfeed/nurse child back to sleep, Give my child a pacifier, Change diaper, Comfort my child verbally but don’t pick child up or take child out of crib/bed, Bring my child into my bed, Let my child cry and fall back to sleep by himself/herself, Give my child a few minutes to see if he/she falls back to sleep, Play with my child until child is ready to go back to sleep, Watch television/a video with my child until he/she falls asleep, Sing to child, Other ______.

Participants were also asked about their bedtime routine. Specifically, they were asked “Which of the following usually occurs on most nights for you in the hour before bedtime? Check all that apply” and then were provided with the following behaviors: Bath, Massage, Read books, Talk on the phone, Use the internet, Watch television, Have dinner or a snack, Brush teeth, Shower, Cuddle, play or feed your baby, Say prayers, Listen to music, Other __________. A second question asked “In a typical 7-day week, how often do you have the exact same bedtime routine?” with the following responses: Never, 1–2 nights per week, 3–4 night per week, 5–6 nights per week, Every night. A third question asked “What time do you usually start your bedtime routine?” which participants then answered by filling in: Time: ____:_____ pm.

Physical activity was assessed at 5 months postpartum using questions adapted from the 2001 Behavioral Risk Factor Surveillance System survey [37]. We asked participants to recall overall frequency and duration of time spent in activities of light intensity (e.g. walking), moderate intensity (e.g. vacuuming), and vigorous intensity (e.g. running) in a usual week over the past month. Because so few women engaged in moderate and/or vigorous activity (<10%), only data for walking were analyzed. Total minutes of walking during a usual week was examined as well as a categorical variable using cut-points adapted from Bentley et al. [38]: (1) not walking at least 10 min at a time in a usual week; (2) walking between 10 and 60 min at a time in a usual week; or (3) walking >60 min at a time in a usual week. The categorical walking measure has moderate reliability across gender and race (kappa coefficient = 0.4), and fair validity compared to physical activity logs (kappa coefficient = 0.2) [37].

Self-reported sleep quality was measured using the Pittsburgh Sleep Quality Index Questionnaire (PSQI) [39]. Respondents answer 10 questions which are grouped into subscales (subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medications, and daytime dysfunction), and yield scores ranging from 0 to 3, with 3 indicating greatest dysfunction. PSQI subscales are then summed to create a Global Score, with higher scores reflecting poorer sleep quality. Participants with a global score >5 are considered poor sleepers. At the end of the PSQI, we added an additional question: “During the past month, how many hours per day did you spend napping?” and participants responded by filling in: _____ hours of sleep napping per day. Responses were used as a continuous variable and dichotomized as napping vs. no napping (a response of “0”).

Sleep behaviors predictive of sleep duration and sleep efficiency were measured using actigraphy data and averaged across all of each participants’ valid nights. These variables included bedtime (sleep onset), wake time (sleep offset), sleep period (minutes from bedtime to wake time), and wake after sleep onset during the sleep period (WASO, minutes scored as wake during the sleep period). In order to capture variability of bedtime and wake time, we also calculated the standard deviation of these measures across nights for each participant. In addition to examining bedtime as a continuous variable, we dichotomized the variable to before vs. after midnight.

Statistical analyses

Preliminary analyses involved summary statistics using means, standard deviations, medians, ranges or interquartile range (IQR), and frequencies and percentages for participant characteristics at 5 months postpartum. Associations between outcome variables (sleep duration, sleep duration categories, and sleep efficiency) and covariates (demographic, medical, psychosocial, and behavioral variables) were assessed using Fisher’s exact or chi-square tests for all categorical variables, Wilcoxon or Kruskal–Wallis tests for continuous variables versus categorical variables, and Spearman correlation coefficients for relationships between continuous variables. We then built independent multiple linear or logistic regression main-effects models to evaluate relationships of sleep duration, considered as both a continuous and categorical outcome, and sleep efficiency, considered as a continuous outcome, with all demographic, medical, psychosocial, and behavioral variables, first by starting with the full model of exposures and covariates and then iteratively eliminating the least significant predictor (p ≥ 0.10) followed by adding back a potentially significant variable (p < 0.25), one at a time, until no more covariates could be added or dropped from the final regression models. Post-hoc multiple comparison adjustments were made via the Tukey–Kramer method. We explored effect modification by including biologically plausible interaction terms into the multivariable regression models. SAS version 9.4 (SAS Institute, Carey, NC) was utilized to carry out all analyses. The data underlying this article will be shared on reasonable request to the corresponding author.

Results

Of the 289 mothers who enrolled, we restricted our analysis to include the 191 women who had sleep duration and efficiency measured objectively via wrist actigraphy at 5 months postpartum. Women were excluded from analyses if they had missing surveyed information for multiple demographic, psychosocial, and behavioral predictors (n = 43), yielding a final analytic sample of 148 women. Participants excluded from analyses were more likely to practice bedsharing compared to those who were included (26.3% vs. 11.5%), however; the two groups did not differ on other variables in these analyses.

Table 1 presents the demographic characteristics of the 148 mothers who were included in this study. The mean (standard deviation [SD]) age at baseline was 25.1 (5.8) years and two-thirds (66.9%) were black or African American. Approximately half (51.4%) of the participants were employed or students and about a quarter (26.4%) had more than high school education. Average (SD) BMI at 5 months postpartum was 29.6 (7.1), with approximately 44% being obese and 29% overweight. Only about a quarter (26.4%) of participants were married or living as married and approximately half (49.3%) were living with family rather than in their own apartment or home.

Table 1.

Participant characteristics (N = 148)

	n (%) or mean (SD)
Age, years	25.1 (5.8)
Race/ethnicity‡
Black or African American	99 (66.9%)
Latina	48 (32.4%)
BMI category‡
Normal weight/overweight (≤30 kg/m2)	89 (62.7%)
Obese (>30 kg/m2)	53 (37.3%)
5-month postpartum weight retention (kg)	3.0 (7.8)
WIC eligible	148 (100%)
Employment
Unemployed	72 (48.6%)
Employed/student	76 (51.4%)
Education
High School or less than High School	109 (73.6%)
More than High School	39 (26.4%)
Living situation
In own apartment or home	73 (49.3%)
With family	73 (49.3%)
Other	2 (1.4%)
Parity
Primiparous	52 (35.1%)
Multiparous	96 (64.9%)
Marital status
Married or living as married	39 (26.4%)
Single, never married, separated, or divorced	109 (73.6%)
Breastfeeding status
Exclusively breastfeeding and/or mixed feeding	62 (41.9%)
Exclusively formula feeding	86 (58.1%)
Edinburgh Postpartum Depression Scale
Score	2.52 (4.5)
Depressive symptoms (EPDS ≥ 9)	14 (9.5%)
Currently smoking	47 (31.8%)
Walk for at least 10 min a day	91 (61.5%)
PSQI total score	6.8 (2.5)
Perceived poor sleep quality (PSQI score > 5)	96 (64.9%)
Actigraph-measured sleep
Bedtime (h:mm (min))	00:36 (97.2)
Waketime (h:mm (min))	08:29 (100.6)
Sleep period (h)	7.9 (1.4)
Total sleep time (h)	6.6 (1.1)
Short sleep duration (<7 h per night)	92 (62.2%)
Wake after sleep onset (min)	78.5 (30.7)
Sleep efficiency (%)	77.5 (6.5)

^‡  n = 1 participant did not report race/ethnicity and n = 6 participants did not provide BMI data, causing the sum of % not equaling to 100% for these variables.

When asked how many hours of sleep would be needed per night to feel refreshed, 14.2% of participants reported fewer than 7 h, 77.7% reported 7–9 h and 8.1% reported greater than 9 h. Despite the fact that the majority reported needing at least 7 h per night, the average sleep duration (measured using the wrist actigraph) was 6.6 (1.1) h and 62% (n = 92) of women averaged less than 7 h of sleep per night (“short sleepers”). On average, participants wore the actigraph wristwatch for 6 nights (range 3–13), with the majority (>70%) wearing the device for six or seven nights. For each participant, all valid nights of data were included in the calculation of actigraphy-derived sleep variables. See Table 1 for sleep characteristics of the study participants.

In unadjusted univariable analyses, short sleepers were more likely to be black although this finding was not statistically significant (73% vs. 57%; p = 0.07). Latina ethnicity was not significantly associated with sleep duration (p = 0.37) or efficiency (p = 0.66). Employment associated with shorter sleep duration (median (IQR) = 6.74 (6.05–7.65) vs. 6.37 (5.52–7.31) h, p = 0.04). No other demographic variables were associated with sleep duration, sleep efficiency, or being a short vs. sufficient sleeper in unadjusted univariable analyses.

Psychosocial and behavioral risk factors associated with shorter sleep duration in unadjusted analyses included: breastfeeding (median (IQR)= 6.20 (5.31–7.17) vs. 6.94 (6.06–7.61) h, p = 0.002), waiting and giving their infant a chance to self-soothe during nighttime awakenings (median (IQR) = 6.46 (5.77–7.35) vs. 7.34 (6.05–7.82) h; p = 0.06), fewer self-reported minutes spent walking during the day (ρ = 0.20, p = 0.06), and more self-reported minutes spent napping during the day (ρ = –0.17, p = 0.04). Bedsharing was associated with poorer sleep efficiency (median (IQR) = 74.38 (70.37–78.44) vs. 78.41 (74.29–82.26), p = 0.01). Women who were short sleepers (<7 h per night) were more likely to report napping during the day (median (IQR) = 0 (0–1) vs. 1 (0–2) times; p = 0.05), were less likely to wait and give their infant a chance to self-soothe during nighttime awakenings (92% vs. 80%; p = 0.04), and were more likely to breastfeed (70% vs. 51%; p =0.04) compared to short sleepers.

In regards to bedtime routine, participants sleeping <7 h per night were less likely to report talking on the phone (54% vs. 37%; p = 0.06), listening to music (38% vs. 23%; p = 0.06), brushing their teeth (63% vs. 45%; p = 0.04), using the internet (71% vs. 49%; p = 0.01) and having a dinner or snack (70% vs. 52%, p = 0.04) during the hour before bed compared to those sleeping ≥7 h per night. Smoking, childcare, depressive symptoms, and global PSQI score were not associated with sleep duration or sleep efficiency, and did not differ between short and sufficient sleepers.

When examining sleep behaviors, later bedtime (Spearman correlation coefficient ρ = –0.27, p < 0.001); earlier waketime (ρ = 0.43, p < 0.001); and greater variability in bedtime (ρ = –0.24, p = 0.003) associated with shorter sleep duration. Bedtime variability was also associated with lower sleep efficiency (ρ = –0.18, p = 0.03). Average bedtime was later (12:55 am vs. 12:04 am; p = 0.002) and mean waketime was earlier (8:00 am vs. 9:14 am; p < 0.001) in participants sleeping <7 h per night compared to those sleeping ≥7h per night. Therefore, those sleeping ≥7h per night spent more time in bed for sleep (the period from bedtime to waketime) compared to short sleepers (9.2 h vs. 7.1 h; p < 0.001). During this longer sleep opportunity, sufficient sleepers experienced more sleep (7.7 h vs. 5.9 h; p < 0.001) and slightly more wakefulness (WASO, 1.5 h vs. 1.2 h; p = 0.01) than short sleepers. Because the amount of additional sleep exceeded the amount of additional wakefulness during the longer sleep opportunity, the mean overall sleep efficiency was higher among sufficient sleepers compared to short sleepers (79.5% vs. 76.3%; p = 0.01).

The multivariable regression models for predicting sleep duration as a continuous variable and as a categorical variable (≥7 vs. <7 h sleep per night) are presented in Tables 2 and 3. Independent predictors of sleep duration as a continuous variable (Table 2) included infant feeding modality, bedtime, and employment such that breastfeeding, having a bedtime after midnight, and being employed were each associated with significantly shorter sleep duration (mean (standard error [SE]) reduction in sleep ≥25 (11) min, p < 0.05). Multiparity, being unmarried, being employed, breastfeeding, having a bedtime after midnight, bedsharing with infant, and responding to infant night awakenings by getting up immediately rather than waiting a few minutes to see if he/she fell back asleep were significantly associated with 130%–360% higher odds of having <7 h per night sleep in a fully adjusted multiple logistic regression model (ORs varying from 2.3 to 4.6; all p < 0.05). The multiple regression model for sleep efficiency (Table 4) revealed that bedsharing was the only variable associated with lower sleep efficiency in this cohort (mean (SE) reduction in sleep efficiency = 3.8% (1.6%); p < 0.05).

Table 2.

Multiple regression results: determinants of continuous sleep duration (N = 148)

	β ± SE (min)	p-value
Infant feeding		0.003
Exclusively breastfeeding/mixed feeding	–32.9 ± 10.7
Exclusively formula feeding	Reference
Bedtime		0.005
After midnight	–31.4 ± 11.1
Before midnight	Reference
Employment		0.02
Employed or student	–24.5 ± 10.5
Unemployed	Reference

Table 3.

Multiple regression results: determinants of categorical sleep duration (<7 vs. ≥7 hours/night, N = 148)

	OR (95% CI) of sleeping <7 h	p-value
Parity		0.01
Multiparous	2.91 (1.26, 6.71)
Primiparous	Reference
Marital status		0.005
Single	3.59 (1.48, 8.72)
Married/living as married	Reference
Employment		0.02
Employed or student	2.58 (1.18, 5.61)
Unemployed	Reference
Infant feeding		0.04
Exclusively breastfeeding/ mixed feeding	2.29 (1.05, 5.01)
Exclusively formula feeding	Reference
Bedtime		0.02
After midnight	2.53 (1.14, 5.64)
Before midnight	Reference
Sleeping arrangement		0.03
Bedsharing	4.30 (1.13, 16.44)
Infant does not sleep in bed with parent(s)	Reference
Response to infant nocturnal awakenings		0.01
Got my child immediately	4.59 (1.38, 15.21)
Gave my child a few minutes to see if he/she fell back asleep	Reference

Table 4.

Multiple regression results: determinants of mean sleep efficiency^a (N = 148)

	β ± SE (%)	p-value
Sleeping arrangement		0.02
Bedsharing	–3.8 ± 1.6
Infant does not sleep in bed with parent(s)	Reference

^aSleep efficiency calculated as total sleep time (min)/sleep period (min) × 100.

Discussion

Despite reporting that they needed 7–9 h of sleep per night to feel refreshed, the majority of disadvantaged, racial/ethnic minority women in our study averaged less than 7 h of sleep per night and sleep efficiency remained quite low (77%) at 5 months postpartum, well below the current AASM recommended clinical threshold (85%). We also note that the sleep efficiency observed in these women during the postpartum period is lower compared to those with a similar background who are not in the postpartum period. In one study of predominantly African American women with an average annual household income of $20,900, sleep efficiency was approximately 81% [40] and in another study of Hispanic adults with 44% of the sample having a household year income less than $20,000, sleep efficiency was approximately 87% [41]. Determinants of insufficient sleep included several modifiable factors that could be addressed at the individual level (e.g. bedsharing, better maternal sleep hygiene, infant sleep training) and promoted at the policy level (e.g. via paid family leave or postpartum support programs). The findings of the current study contribute to clinical research working to promote enhanced sleep in disadvantaged populations and inform future studies related to disease prevention and treatment.

Regarding sociodemographic variables, we found that women who reported being employed, were not married, and had more than one child, were at increased risk for habitually obtaining insufficient sleep. These findings are consistent with previous studies in adults showing relationships between employment, marital status, and parity and sleep duration [42]. Data from the American Time Use Survey [42, 43] and the Behavioral Risk Factor Surveillance System [44], have demonstrated that adults who are not married, have more children, and have lower income and less education, are at increased risk for short sleep and sleep disturbances. A recent study in the United Kingdom [45] found that shorter maternal sleep predicted less participation in the labor market, fewer hours worked and lower household income, suggesting a bidirectional relationship between sleep and employment. More research is needed in understanding the relationship between sleep and employment during the postpartum period, particularly in socioeconomically disadvantaged women, especially given ongoing discussions about paid family leave in the United States.

Regarding marital status, it has been posited that unmarried mothers have less time compared with married mothers since they do not have a partner to split responsibilities with [46] and our finding, that unmarried women spent less time sleeping, supports this idea. However, other studies have found that marital status does not impact sleep time. Pepin et al. [47] found that even though married mothers theoretically can share household labor with their partners, living with a heterosexual male partner actually associated with greater time spent on housework and less time asleep. Studies are needed to better understand the dynamics of social support and contributions of household members in multi-generational families during the postpartum period and how this impacts maternal sleep. Future studies should also examine if the relationship between marital status and sleep is moderated by sociodemographic factors such as race, culture, employment and socioeconomic status.

Parity is an important factor to consider when understanding and improving postpartum sleep. Women with multiple children experience increased childcare demands, varying child sleep schedules, and increased nighttime awakenings, which impact maternal sleep [48]. In a German study examining changes in subjective measures of sleep satisfaction and duration in parents across pregnancy and the postpartum period, women exhibited marked declines in sleep satisfaction and duration after childbirth with ratings reaching a nadir during the first 3 months postpartum [49]. Furthermore, sleep satisfaction and duration did not fully recover even up to 6 years following the birth of their first child and became worse with the birth of each additional child [49]. Studies measuring sleep objectively have found that primiparous women exhibit shorter sleep duration and poorer sleep efficiency compared to multiparous women during the early postpartum period (1–3 months), whereas multiparous women report less time-in-bed for sleep than primiparous women during the later postpartum period (6–7 months+) [50, 51]. Thus, it is important to distinguish early from late postpartum when examining the relationship between parity and sleep. Future research is also needed to assess the sleep habits of children other than the infant to better understand the overall construct of how the sleep of older children influences the mother’s sleep. Collectively, these findings suggest that providing socioeconomically disadvantaged women with social support, especially those who are employed, unmarried and with multiple children, and/or policies that expand paid family leave may be critical for improving sleep and consequent health outcomes during the postpartum period.

We also identified several mother-infant behaviors associated with sleep duration and sleep efficiency. Mothers who responded to infant night awakenings immediately (rather than waiting a few minutes to see if the infant fell back asleep on their own), breastfed, or practiced bedsharing were more likely to exhibit short sleep duration and poor sleep efficiency. Although most infants in the United States develop long nocturnal sleep periods 4–5 months of age, a significant number of infants still sleep for short periods and then “signal” (cry) upon waking in the night. Four randomized controlled trials (RCTs) found that “limit-setting” parenting - employing routines and delaying response time in order to encourage infants to develop autonomous settling—increased the number of infants with long nocturnal sleep periods [52, 53]. In contrast, “infant-cued” parenting—maintaining high proximity and rapid responses—has been associated with persistent infant waking and signaling at night [52]. Our finding supports the use of limit-setting parenting for improving maternal sleep.

The majority of studies examining bedsharing have focused on infant sleep and health outcomes where results have been mixed [54]. We found that bedsharing associated with poorer sleep efficiency and increased risk for short sleep duration in the mother. This is consistent with a study by Volkovich et al. that found that women practicing bedsharing exhibited more nocturnal awakenings, longer periods of wakefulness at night, and shorter continuous sleep periods, even after controlling for breastfeeding [55]. Studies demonstrating a relationship between bedsharing and poorer sleep have posited that parents may be awakened by infant movement or vocalizations more easily; however, other studies report perceived improvements in sleep quality as a benefit of bedsharing. Although there are reported benefits to bedsharing (e.g. breastfeeding, mother–infant bonding), bedsharing, combined with certain factors, has been associated with increased risk of sudden infant death syndrome. These include maternal smoking, alcohol or drug consumption before bed, maternal excessive weight, overtiredness, household overcrowding, excessive/soft bedding, bed-sharers other than parents, unsafe sleep surfaces such as sofas [54]. It is important to note that many of these factors are more prevalent among socioeconomically disadvantaged, racial/ethnic minority women. Thus, it may be helpful for clinicians to provide simple suggestions to new mothers that are in line with recent guidelines from the American Academy of Pediatrics recommending that infants sleep in the same room as parents but in their own crib or bassinet [56]. In addition to reducing sudden infant death syndrome, clinicians can inform parents that their own sleep may be improved if they follow these guidelines. This is likely to be particularly helpful for employed mothers, among whom increased sleep efficiency is critical given that their sleep opportunity is limited by work obligations.

We observed that in this cohort, formula feeding was associated with increased sleep duration whereas any breastfeeding was associated with a decreased sleep duration of approximately 33 min on average. Studies examining the relationship between breastfeeding and maternal sleep have been mixed; however, several studies have shown that breastfeeding promotes longer nocturnal sleep durations in mother and infant compared to formula feeding [57–61]. Understanding why we observed less sleep in women who were breastfeeding is an important topic to pursue and may be due to a lack of lactation resources in socioeconomically disadvantaged mothers [62]. Future studies should also examine how additional support for breastfeeding mothers, through maternity leave policies, community programs, and lactation consultants, affects nocturnal sleep duration. Given the important role for sleep in improving mood and overall health, improving sleep in breastfeeding mothers would likely contribute to continued practice and improved milk production.

Increased physical activity has been associated with improved sleep quality and daytime alertness in the general population [63, 64]. Findings from studies examining the relationship between physical activity and sleep during the postpartum period have demonstrated mixed/conflicting results, with two studies showing no association between physical activity levels and self-reported sleep measures [65, 66], and another showing that self-reported physical activity is positively associated with self-reported sleep quality [67]. We observed that self-reported minutes spent walking was associated with increased sleep duration; however this correlation did not reach statistical significance. It should be noted that our ability to examine the relationship between physical activity and sleep duration was limited to only minutes spent walking because of the extremely low levels of physical activity reported by this population. Future research is needed to objectively measure physical activity levels (e.g. hip-worn accelerometers) in postpartum women to gain a clearer understanding of how physical activity and sleep behaviors are related in this population.

Surprisingly, we did not observe an association between depressive symptoms (measured using self-report) and sleep outcomes. Evidence suggests that although rates of postpartum depression do not differ by race and ethnicity, detection and treatment rates are lower in racial/ethnic minority women compared to white women and there are racial disparities in the proportion of women who seek treatment [68]. In our sample, 5.4% of mothers scored higher than 12 on the EPDS (the standard cut point) and 9.5% scored higher than 8 (the adjusted cut point for this population). This prevalence rate is lower than what was observed in a previous study which found that 56% of low-income, black young mothers in an urban pediatric clinic met the criteria for major depressive disorder using a diagnostic interview [33]. Therefore, it is possible that the screening tool we used was insufficient in accurately capturing depressive symptoms in this population. Future studies are needed to more carefully evaluate depression in these women and determine how symptoms relate to sleep during the postpartum period. Understanding the role of postpartum sleep patterns in the etiology of postpartum depression remains an important area of study. There may also be a difference how subjective measures of sleep (e.g. perceived sleep quality and sleep duration) and objective measures of sleep (e.g. sleep efficiency and sleep duration) relate to depressive symptoms. Our finding that objective measures of sleep did not associate with depressive symptoms is consistent with prior research [69–72]. Thus, there may be a disconnect for women with postpartum depression between their actual sleep versus their perceptions of sleep. Future studies are needed to disentangle these constructs.

We identified several modifiable sleep behaviors that determined postpartum sleep duration. Later bedtime was a consistently significant predictor of short sleep duration; therefore, it may be prudent to educate mothers about the importance of sleep for health and instructed to prioritize an early (e.g. before midnight) bedtime. Using a qualitative approach, we previously found that low-income African American mothers perceived their own sleep quality to be very poor at 3–6 months postpartum despite the fact that they believed their babies’ sleep had improved substantially. They cited work/school commitments, household chores, the home environment, and television viewing [73] as significant barriers to sleep. Coaching new mothers on executive skills like time management and planning strategies may be helpful for promoting an earlier bedtime. Given the low sleep efficiency observed in this study, it may be necessary for these women to allow themselves a longer sleep opportunity (time in bed) each night compared to other adult populations. We found that women who were able to obtain at least 7 h of sleep per night averaged 9 h time in bed for sleep. Thus, disadvantaged women should be educated that there is a difference between these two constructs (time in bed vs. time asleep) and encouraged to schedule a sufficient amount of time in bed for sleep in order to achieve a sufficient amount of sleep. Furthermore, future research is needed to better characterize and understand what goes on during maternal nighttime awakenings in this population.

Finally, consistent with a prior study [74], we observed that self-reported daytime napping was associated with shorter nocturnal sleep duration. Women may need to nap during the early postpartum period in order to supplement the major nocturnal sleep period since infants’ circadian rhythms have not yet matured and they are more to wake through the night and nocturnal feeding. However, by 5 months postpartum, maternal napping could be done more strategically. Women could consider duration and time of day (e.g. napping later in the day may have a more adverse impact on nocturnal sleep than napping earlier in the day), and prioritize an earlier bedtime rather than daytime naps.

Collectively, our findings indicate that many of the predictors of insufficient sleep observed in this sample of racial/ethnic minority women who are socioeconomically disadvantaged are similar to those observed in other studies of insufficient sleep (e.g. demographic characteristics [42–44] and bedsharing [55]). It remains critical to study sleep, using objective measures, in this population in order to understand the nuance that will be valuable for framing future interventions. As an example, bedsharing may be a predictor of insufficient sleep in both racial minority and white postpartum women; however, the way in which we create interventions to address this behavior may be very different. Future studies, particularly those using qualitative methodologies, can build upon our findings in black and Latina postpartum women to better understand cultural factors that may be important to address as we develop meaningful interventions that effectively modify the predictors we have identified in this study. In addition, we found that the measures used to assess physical activity and postpartum depression may not be as useful in this population. We encourage the continued efforts of those who are developing and validating new measures that are more appropriate for racial/ethnic minority women who are socioeconomically disadvantaged [75, 76]. Finally, we found that sleep efficiency is particularly low in this population, which may inform future interventions specifically targeting maternal sleep behaviors.

Strengths of our study include objective assessments of sleep in a relatively large sample of socioeconomically disadvantaged black and Latina women during the postpartum period and detailed information on a large number of potential demographic and psychosocial determinants. We did not collect environmental and physiological information that also likely affect sleep duration and sleep efficiency (e.g. presence of light/noise in the room, dim-light melatonin onset). Unmeasured environmental factors, like light and noise pollution [77, 78] or neighborhood safety [79] may have played an important role increasing nighttime awakenings, and underlying circadian differences likely contribute to individual differences in sleep efficiency during the postpartum period. Finally, although the use of actigraphy to measure sleep is a strength of the study (compared to relying only on self-reported assessments), actigraphy provides an indirect behavioral measure of sleep based on activity patterns, rather than a direct neurophysiological measure and has only been validated to measure nocturnal sleep.

In summary, sleep remains significantly disrupted at 5 months postpartum among socioeconomically disadvantaged, racial/ethnic minority mothers, with the majority (62%) obtaining less than the recommended 7 or more h per night and the average sleep efficiency quite poor (77%). Several modifiable behaviors, such as maintaining an early bedtime, refraining from bedsharing and waiting a few minutes for the infant to self-soothe during nighttime awakenings, were identified that can inform future interventions aimed at improving sleep in this population. Data also support policy changes to optimize the health of new mothers, including paid family leave and increased support for breastfeeding.

Funding

This study was supported by grants from the Doris Duke Charitable Foundation (2012065) and the National Institutes of Health (R01 HL130816). The information or content and conclusions of this study are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by DDCF, NIH, HHS, or the US Government.

Author contribution

Sharon Herring performed the research. Sharon Herring and Grace Pien, designed the study. Andrea Spaeth led and designed the analyses for the current paper. Risha Khetarpal wrote an initial draft of the paper, while Andrea Spaeth wrote the final paper. Daohai Yu performed the analyses. All authors made scientific contributions to the interpretation of the data. All authors edited and approved the final paper for submission.

Disclosure statement

Financial disclosure: The authors do not have any financial disclosures.

Non-financial disclosure: The authors do not have any potential conflicts of interest.