The preliminary efficacy of a sleep self-management intervention using a personalized health monitoring device during pregnancy

Abstract

Background:

Sleep disturbances are common during pregnancy and are associated with the development of adverse pregnancy outcomes. Personal health monitors (PHM) can facilitate change in health behaviors, though few studies have examined their use in improving sleep during pregnancy. This pilot study aimed to characterize sleep changes during pregnancy in women participating in a self-management intervention using a PHM.

Participants/Methods:

Participants with low risk, singleton pregnancies from Western Massachusetts were randomized at 24 weeks gestation to receive sleep education only (n=12) or sleep education, and PHM intervention (n=12). The single-session sleep education was given at baseline by a registered nurse. Sleep quality, duration, efficiency, disturbances, daytime sleepiness, and fatigue were assessed at baseline and 12 weeks follow-up using questionnaires. We described mean ± standard deviation within and between-group changes in each sleep outcome from baseline to 12 weeks follow-up.

Results:

The PHM arm experienced larger sleep quality improvements and daytime sleepiness than the sleep-education only arm, but the differences were not statistically significant. In the PHM arm, the Pittsburgh Sleep Quality Index (PSQI) score decreased (i.e., sleep quality increased) 1.22 ±2.39 (p=0.16), and the Epworth Sleepiness Scale (ESS) score decreased (i.e., daytime sleepiness decreased) 1.11 ±2.08 (p=0.15). In the sleep-education arm PSQI decreased 0.57±2.37 (p=0.55) and ESS decreased 1.29±2.93 (p=0.29). Neither group experienced statistically significant changes in sleep duration, efficiency, disturbances, or fatigue.

Conclusion:

Sleep education with PHM may improve or prevent decreases in sleep outcomes during pregnancy. Further investigation in larger trials is warranted.

Introduction

Women often experience decreases in sleep quality throughout pregnancy.(Hedman et al., 2002; M’Saad et al., 2015; Mindell & Jacobson, 2000; Sedov et al., 2017) For example, a meta-analysis of four prospective studies examined the change in sleep quality, assessed with the Pittsburgh Sleep Quality Index (PSQI), from the second to the third trimester of pregnancy.(Sedov et al., 2017) The authors found that the mean PSQI global score increased 1.68 (0.42 to 2.94) points from the second to the third trimester, indicating decreasing sleep quality.(Sedov et al., 2017) Poor sleep during pregnancy is a concern because it is associated with increased inflammatory response and subsequent increased risk for adverse pregnancy (e.g., preeclampsia, gestational diabetes) and birth (e.g., preterm birth, birth weight, Apgar score) outcomes. (Micheli et al., 2011; Nodine & Matthews, 2013; Okun et al., 2009; Qiu et al., 2010; Zafarghandi et al., 2011)

Personal health monitors (PHM) remotely track health indices, including physical activity, diet, and sleep. PHMs include interactive behavior change techniques, such as self-monitoring, goal setting, review of behavior goals, rewards, and facilitation of social support and peer comparison.(Lyons et al., 2014) PHMs have been used to promote changes in diet and physical activity. (Ferreira et al., 2008; Pellegrini et al., 2012; Shuger et al., 2011) A meta-regression of healthy eating and physical activity interventions among adults by Michie et al. found that physical activity interventions that used behavioral self-monitoring were more effective than interventions that did not. (Michie et al., 2009) Moreover, behavioral self-monitoring explained more of the heterogeneity in study effectiveness than any other behavioral change technique. PHMs have been used to promote physical activity during pregnancy (Kim et al., 2015); however, there is a lack of research using PHMs to intervene on sleep during pregnancy.

There is limited evidence that PHMs are effective stand-alone intervention strategies (Jakicic et al., 2016); however, PHMs are effective facilitators of change. Therefore, combining PHMs with a sleep education intervention that addresses sleep hygiene has the potential to promote sleep health during pregnancy and can be easily applied in clinical or community settings. There is limited evidence that poor sleep hygiene exacerbates sleep disruptions in pregnant women. Sleep hygiene is defined as modifiable behavioral (e.g., bedtime routine, limiting arousing activities at night) and environmental (e.g., TV in the bedroom, unpleasant temperature) factors that influence sleep. (Hutchison et al., 2012) For example, in a cross-sectional study of 197 pregnant women in their third trimester, those with poor sleep quality, defined as having a PSQI score >5, had worse sleep hygiene than women with good sleep quality (PSQI≤5). (Tsai et al., 2015) Specifically, poor sleepers (vs. good sleepers) reported more arousal-related behaviors (e.g., TV in bed, worry about not being able to sleep), having irregular sleep schedules (e.g., sleeping in on the weekends, limited outdoor exposure), and having poor sleep environments (e.g., noisy or quiet bedrooms, uncomfortable bedding). Interventions that addressed sleep hygiene during pregnancy effectively reduced depressive symptoms, which is a predictor of sleep quality. However, they did not determine the impact of the interventions on sleep outcomes.(Rezaei et al., 2015; Rezaei et al., 2014)

This pilot study’s primary goal was to characterize within and between-group changes in sleep-related outcomes (i.e., sleep duration, sleep efficiency, sleep quality, fatigue, daytime sleepiness) among pregnant women randomized to receive a sleep education with or without a PHM device intervention. This paper’s ultimate goal of this paper was to produce point estimates and measures of variability for changes in multiple domains of sleep from mid- to late-pregnancy to inform the design of a larger intervention.

Methods

Study Population

We recruited participants living in Western Massachusetts within 50 miles of the University of Massachusetts Amherst campus. We included nulliparous and multiparous women who (1) were at least 18 years of age, (2) were 14 to <24 completed weeks gestation, (3) had a low-risk pregnancy (i.e., no known maternal or fetal complications) during the current pregnancy, (4) had a mobile phone (iOS or Android operating systems) compatible with the PHM device, (5) had internet access with cell phone data plan, (6) were English speaking, and (7) were receiving prenatal care. We excluded women with pre-existing diabetes mellitus, hypertension, or diagnosed sleep disorder, as some previous studies indicated that the associations between pregnancy complications and a pre-existing sleep disorder may bias the results. (August et al., 2013) We primarily recruited women through targeted advertisements on Facebook. The University of Massachusetts Amherst Institutional Review Board approved this study. (Trial ID: NCT03783663) Additional information on the design of this trial can be found elsewhere. (Hawkins et al., 2019)

Study Design

This was a 12-week, parallel-arm, pilot randomized controlled trial (RCT). All participants were randomized to receive in-person sleep education only, or in-person sleep education plus a PHM device. We used randomized blocks (n=6) to ensure a participant balance between the intervention arms (n=12 in each arm). We used open-source randomization software to carry out the randomization scheme. (Haahr, 2019)

We identified eligible participants between 14 and <24 weeks gestation. Participants completed a screening survey through phone calls to determine their eligibility. The intervention started at 24 weeks of gestation. We chose to start the intervention at 24 weeks gestation to provide sleep education before the third trimester when sleep quality is known to decline. (Sedov et al., 2017) Additionally, we sought to test a 12-week intervention with an end-point of 36 weeks to reduce the risk of participants delivering their infants before the end of the intervention time frame. At baseline, the participants completed baseline surveys (see measures below) and demographics assessments. The participants also received a sleep hygiene brochure and an individual 30-minute education delivered by a registered nurse. The sleep education intervention and brochure discussed the importance of sleep for maternal health, factors that can affect sleep (e.g., diet, physical activity, sleep environment), and strategies to improve sleep during pregnancy (e.g., limiting arousal behaviors). The sleep education provided tips on achieving a better night’s sleep during pregnancy through behavioral and environmental factors, including managing pregnancy weight gain, physical activity, and keeping a consistent sleep schedule. The nurse revealed the intervention group assignment allocation after the sleep education session. The participants completed in-person follow-up surveys post-intervention, followed by a qualitative interview to contextualize study findings further.

Participants receive a $20 gift card for completing the baseline assessment at 24 weeks gestation and a $30 gift card for completing the post-intervention assessment at 36 weeks gestation. Participants from both groups received the same amount of incentives for completing the assessment. Participants from the PHM group kept their devices, and participants from the comparison group received a PHM device at the end of the study.

Intervention

The PHM intervention group received a MisFit Shine 2 for self-monitoring of sleep, goal setting, and tracking progress towards meeting their goals. We chose the MisFit Shine 2 because it can be worn continuously, is water-proof, is battery-powered, can last six months without a battery change, and has acceptable accuracy for estimating total sleep time. (El-Amrawy & Nounou, 2015) We instructed participants to download the following apps to their mobile phone: MisFit, which was the platform for health monitoring comparable to the device; and IFTTT, which was used for data sharing from the PHM device to the research team. (IFTTT, n.d.) We created a password-protected, study-specific Google account for each participant accessible only by the research team. Sleep data from participants’ phones were then automatically downloaded into a spreadsheet in Google Drive. The accounts did not store personal identifying information. After the trial ended, we instructed participants to change their MisFit accounts to their email addresses, and participants were allowed to keep the device. We deleted the study-specific Google accounts at the study end.

Primary Outcomes

Sleep quality and duration

The PSQI was used to measure sleep quality and duration. The PSQI is a 19-item, self-reported scale that measures the frequency of sleep disturbances in the past month in seven domains (i.e., subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, sleep medication use, and daytime dysfunction). (Buysse et al., 1989) The PSQI has high validity and reliability (Cronbach’s α = 0.83) among pregnant women. (Qiu et al., 2016) The PSQI was used in this study to assess sleep quality and duration over the past week. The PSQI scores range from 0 to 21, with higher scores indicating worse sleep quality. Poor sleep quality was defined as a PSQI score > 5. A threshold of 5 has high sensitivity (89.6%) and specificity (86.5%) for identifying poor sleep quality. (Buysse et al., 1989)

Participants were also asked to report what time they go to bed, what time they wake up, and their actual sleep duration at night. The total time in bed was calculated as the difference between bedtime and wake time. We calculated sleep efficiency by dividing actual sleep time by total time in bed. We defined poor sleep efficiency as less than 85% of the total time in bed sleeping. (Ohayon et al., 2017) Short sleep duration was defined as sleeping <6 hours per night.

Sleep disturbances

The PROMIS Short Form v1.0-Sleep Disturbance 6a was used to measure sleep disturbance. (Buysse et al., 2010) The Sleep Disturbance scale is a 6-item, 5-point Likert scale measuring individual sleep disturbance over the past week. The overall score was calculated by summing all item responses, with a higher score indicating higher sleep disturbances. The sleep disturbance subscale has demonstrated acceptable reliability in diverse populations, with Cronbach α > 0.7 among pregnant women. (Jensen et al., 2016)

Excessive daytime sleepiness

The Epworth Sleepiness Scale (ESS) was used for measuring daytime sleepiness. (Johns, 1991) The ESS is an 8-item scale with adequate validity and reliability (Cronbach’s α =0.75) for measuring daytime sleepiness among pregnant women.(Baumgartel et al., 2012) Higher scores indicate higher daytime sleepiness of the individual. Excessive daytime sleepiness was defined as an ESS score >10. (Baumgartel et al., 2012)

Fatigue

The PROMIS Fatigue Short Form 4a is a 4-item scale that measures symptoms of fatigue in the past week. (Fatigue: A brief guide to the PROMISE fatigue instrument, 2019) The PROMIS Fatigue Short Form 4a has adequate validity and reliability among pregnant women (Cronbach’s α =0.72). (Ameringer et al., 2016) Each item is scored using a 5-point Likert scale. The overall fatigue score was calculated by summing each item, with higher scores indicate more severe fatigue symptoms.

Statistical analysis

We presented the change in each sleep outcome continuously. We calculated mean and standard deviation for within-group change for each sleep measure from baseline to post-intervention. Paired t-tests were used to compare within-group change for each sleep measure. Next, we calculated means and standard deviations for differences in change between the intervention arm. We used two-sample t-tests for the baseline and follow-up differences in sleep outcomes between the PHM and sleep-education only arms.

We additionally described categorical changes in sleep quality and duration using thresholds for change deemed clinically meaningful. (Buysse et al., 2011) Specifically, a decrease in PSQI global score of three points or greater, or the resolution of poor sleep quality (i.e., decreased below a threshold of five points), is considered a clinically significant improvement in sleep quality. There are no specific guidelines on clinically meaningful increases in sleep duration, but prior researchers have used 30-minute benchmarks for change in sleep duration. (Edinger & Fins, 1995) We calculated within-group relative frequencies of making these respective changes in sleep quality and duration.

We conducted multivariable linear regression to compare mean change in each sleep outcome between the intervention arms, adjusting for potential measured confounders. We compared the distribution of demographic factors between the two arms to identify potential measured confounders. We presented the adjusted mean changes in sleep outcomes between the two groups.

Our primary goal was to calculate point estimates and measures of variability to inform power calculations for a larger, definitive RCT. Therefore, hypothesis testing was not the primary goal. As Schoenfeld recommended, we accepted a higher type 1 error (α = 0.25) during the pilot test. (Schoenfeld, 1980) SAS version 9.4 (SAS Inc. Cary, NC) was used for all analyses.

Results

The total sample included 24 participants, 12 in each group at baseline. All participants had data on at least one sleep measure at baseline. The missing sleep data at the baseline were due to internet issues during data collection. Four participants were lost to follow-up at 36 weeks gestation. One participant moved to a different state and was unable to complete the follow-up survey. A total of 19 participants completed follow-up assessments (Figure 1). We did not exclude participants if they developed complications after enrolling in the study. The attrition rate of this study at 12 weeks follow-up was 12.5%. (Paterno et al., 2020)

Figure 1 -.

Participant Flow Diagram

The PHM group has one and the comparison has two missings in PSQI and ESS assessment due to technical issues

Among the participants in the final sample, the average age was 31.8±5.1 years. The majority of the participants were non-Hispanic white, married, college-educated, and worked full-time. There were no differences in demographic characteristics between the study groups except for employment status. Participants in the PHM group were more likely to report full-time employment than the women in the comparison group (91.67% vs. 50.00%, p=0.09) (Table 1).

Table 1.

Demographic characteristics of the study population

	Total (N=24)		Intervention (n=12)		Control (n=12)
	mean±SD	n (%)	mean±SD	n (%)	mean±SD	n (%)
Age, years	30.83±5.10		31.17±5.11		30.50±5.28
non-Hispanic White		23 (95.83)		12 (100.00)		11 (91.67)
Married, yes		21 (87.50)		11 (91.67)		10 (83.33)
College educated, yes		22 (91.67)		12 (100.00)		10 (83.33)
Employment, full time		17 (70.83)		11 (91.67)		6 (50.00)

For our primary analysis, we first characterized baseline, post-intervention, and within-subjects change in sleep outcomes in the PHM arm (Table 2). The mean PSQI score at baseline was 6.27±2.2, with 45.5% reporting poor sleep quality (PSQI>5). At follow-up, the mean PSQI score was 4.8±2.04, with 40% reporting poor sleep quality (p=0.16). The majority of participants in the PHM (55.56%) reported reductions in PSQI of at least three points or the resolution of poor sleep quality. The fatigue score was 10.92±2.71 in the PHM group at baseline and virtually unchanged at follow-up (10.2±2.66). Sleep efficiency was 84.22±11.49% at baseline, with 58.33% reporting poor sleep efficiency (<85%). The mean sleep efficiency score was 91.1%±12.04 at follow-up, with 40% reporting poor sleep efficiency. The average sleep duration was 7.54±1.1 hours at baseline and 7.7±0.75 after the intervention. Four participants (40%) in the PHM arm reported 30-minute or greater increases in sleep duration from baseline to post-intervention. No participant reported short sleep duration (<6 hours/night) at baseline or post-intervention follow-up. The mean ESS score was 7.18±4.26 at baseline and 6.5±4.81 post-intervention (p=0.15). The percentage of participants reporting excess daytime sleepiness decreased from 36.36% to 20% after the intervention. The mean score of sleep disturbance was 14.83±5.37 at baseline, and 13.7±5.44 post-intervention (Table 2).

Table 2.

Baseline, post-intervention, and within subjects change in sleep outcomes in the intervention and comparison group

	PHM Intervention										Comparison group
	Baseline			Post-Intervention			Mean changea				Baseline			Post-Intervention			Mean Changea
	n	Mean	SD	n	Mean	SD	n	Mean	SD	p-value	n	Mean	SD	n	Mean	SD	n	Mean	SD	p-value
PSQI Global Scoreb	11	6.27	2.24	10	4.8	2.04	9	−1.22	2.39	0.16	10	8.20	3.19	9	6.11	1.45	7	−0.57	2.37	0.55
Sleep efficiency, %	12	84.22	11.49	10	91.1	12.04	10	4.77	13.49	0.29	10	76.71	11.56	9	85.54	8.23	7	4.91	12.57	0.34
Sleep duration, hrs	12	7.54	1.1	10	7.7	0.75	10	0.05	0.76	0.84	10	6.80	1.09	9	7.44	1.29	7	0.07	0.67	0.79
Fatigueb	12	10.92	2.71	10	10.2	2.66	10	−0.20	2.49	0.80	12	12.33	3.31	9	12.00	3.16	9	−0.11	3.86	0.93
Daytime sleepinessb	11	7.18	4.26	10	6.5	4.81	9	−1.11	2.08	0.15	10	8.80	3.94	9	6.56	3.64	7	−1.29	2.93	0.29
Sleep disturbanceb	12	14.83	5.37	10	13.7	5.44	10	−0.3	5.38	0.86	10	18.00	5.12	9	16.22	3.23	7	−1.29	6.52	0.62

^aMean change= the mean of post-pre estimates;

^bPSQI: Pittsburgh sleep quality index; Fatigue score is based on the PROMIS Fatigue short form 4; Daytime sleepiness is measured by the Epworth Sleepiness Scale (ESS); Sleep disturbance is measured by PROMIS Short Form v1.0-Sleep Disturbance 6a; p-values <0.25 are highlighted in bold font

Next, we characterized baseline, post-intervention, and within-subjects change in sleep outcomes in the sleep-education only arm (Table 2). The mean PSQI score was 8.2±3.2 and 6.11±1.45 at baseline and post-intervention, respectively. Eighty percent and 66.67% of sleep-education only participants reported poor sleep quality at baseline and post-intervention, respectively. Only one participant reported reductions in PSQI of at least three points or the resolution of poor sleep quality. The average sleep efficiency decreased from 76.71±11.56% at baseline to 85.54±8.23% post-intervention. The percentage of participants reporting poor sleep efficiency was 70% at baseline and 55.56% at post-intervention. Participants reported an average of 6.8±1.09 hours of sleep at night at baseline and 7.44±1.29 hours after the intervention. Three participants (42.86%) reported 30-minute or greater increases in sleep duration from baseline to post-intervention. Only one participant reported short sleep duration at baseline and follow-up. The mean ESS score was 8.8±3.94 and 6.56±3.64 at baseline and post-intervention, respectively. The percentage of participants reporting excessive daytime sleepiness was slightly changed from 30% at baseline to 22.22% post-intervention. The mean score for sleep disturbance was 18±5.12 at baseline and 16.22±3.23 at follow-up.

Finally, we compared the differences in change for each sleep outcome between the intervention arms (Table 3). We reported mean changes (post-pre intervention) in each sleep outcome, adjusted for employment status to account for differences in employment status at baseline between the intervention arms. Overall, the PHM arm reported slightly greater improvements in sleep outcomes, but the differences failed to reach statistical significance. Specifically, the PHM arm had a 0.59 point greater decrease in PSQI score than the sleep education only arm (95% CI: −2.03, 3.21; p=0.66). The PHM arm also reported a 4.04% (95% CI: −9.14, 17.21, p=0.56) greater increase in sleep efficiency compared to the sleep education only arm. The sleep education-only arm had a 0.65 point greater decrease in sleep disturbance than the PHM arm (95% CI: −3.53, 2.23; p=0.84).

Table 3.

Difference between the intervention groups in change in sleep outcomes from baseline to postintervention

	Differences in meansa Comparison-PHM	95% CI		p-value	Adjusted differences in meansa,b Comparison-PHM	95% CI		p-value
PSQI Global Scorec	0.65	−1.92	3.22	0.60	0.59	−2.03	3.21	0.66
Sleep efficiency, %	0.14	−13.65	13.93	0.98	4.04	−9.14	17.21	0.56
Sleep duration, hrs	0.02	−0.74	0.79	0.95	0.08	−0.70	0.48	0.84
Fatiguec	0.09	−3.02	3.19	0.95	0.61	−2.52	3.75	0.71
Daytime sleepinessc	−0.17	−2.86	2.51	0.89	0.18	−2.50	2.86	0.9
Sleep disturbancec	−0.99	−7.14	5.17	0.74	−0.65	−3.53	2.23	0.84

^aDifferences in means=the mean difference of the pre-post estimates between the intervention and the control group, control-intervention group;

^bAdjusted for employment status and group allocation;

^cPSQI: Pittsburgh sleep quality index; Fatigue score is based on the PROMIS Fatigue short form 4; Daytime sleepiness is measured by the Epworth Sleepiness Scale (ESS); Sleep disturbance is measured by PROMIS Short Form v1.0-Sleep Disturbance 6a

Discussion

In this pilot RCT of a sleep intervention with or without a PHM during pregnancy, we characterized within and between-group changes in sleep-related outcomes. We found increases in sleep quality and decreases in daytime sleepiness in the PHM arm. There was a mean increase in sleep quality and sleep efficiency and a mean decrease in fatigue in both intervention arms. There was no change in other sleep outcomes in either group. Although the observed differences were not statistically significant, our findings warrant further investigation in a larger, definitive clinical trial.

Our findings are in contrast with the majority of observational studies of sleep during pregnancy, which mainly reported decreases in sleep quality from mid to late pregnancy. For example, in an observational study of 100 pregnant women, 50% of participants reported decreased sleep duration during the third trimester. (Schweiger, 1972) Likewise, Facco et al. (2010) found that among 189 nulliparous women, the prevalence of poor sleep quality (>5 PSQI), increased from the first (39%) to third trimester (53.5%) (p=0.001). In contrast, participants in the current study in both intervention arms reported an increase in sleep quality, with slightly higher improvements in the PHM arm. However, our findings should be interpreted with caution as we did not have a control group with no intervention. The primary aim of the current study was to evaluate the feasibility and preliminary efficacy of PHM in facilitating changes in sleep outcomes. Additional intervention studies are needed to further clarify the effectiveness of sleep education with or without PHM for improving sleep in pregnancy.

The findings of the current study are consistent with findings from prior behavioral interventions during pregnancy. Specifically, the majority of prior behavioral interventions had positive effects on one or more dimensions of sleep. (Bacaro et al., 2019; Hollenbach et al., 2013) For example, a pilot trial of 15 nulliparous women practicing yoga for seven weeks during the second trimester found significantly fewer awakening and sleep disturbances in the third trimester, assessed through actigraphy and the General Sleep Disturbance Scale (p=0.03). (Beddoe et al., 2010) Another prospective longitudinal study with 65 nulliparous women found a weak association between physical activity and sleep quality and duration during late pregnancy, assessed by the General Health Questionnaire (p=0.04). (Goodwin et al., 2000) An RCT with 26 pregnant women randomized to either massage therapy or relaxation therapy only found significant improvement in sleep disturbance measured by the Verran and Snyder-Halpern Sleep Scale (Shahid et al., 2011) in the massage therapy group. The sleep disturbance score in the massage group changed from 49.4 at the start of the intervention to 38.0 at the end of the intervention, with no change observed in the relaxation therapy only group. (Field et al., 1999) However, it is difficult to make direct comparisons between these studies and ours because of differences in the starting time of the intervention (ranged from 14–30 weeks of gestation) and outcome measures. Additional studies are required with comparable sleep measures and timing of intervention initiation in order to better examine the effect of behavioral interventions on sleep improvement during pregnancy.

This study has some limitations worth noting. First, our small sample limits our ability to make definitive conclusions about the intervention’s efficacy. However, this pilot trial’s primary goal was to inform the development of a larger, more definitive trial. This study’s findings served our main aim of providing point estimates and measures of variability to inform sample size estimation for a larger scale trial in the future. Second, we did not collect data on some lifestyle factors, including physical activity and third-shift jobs, which may introduce bias if imbalances were present between the study arms. Third, we did not collect data on clinical sleep disorders (e.g., insomnia, restless leg syndrome, or sleep-disordered breathing) or maternal complications occurring after randomization. However, our intervention was designed to address sleep health dimensions that are more common; thus, the intervention likely would not have made a significant impact on clinical sleep disorders. Moreover, missing data on maternal complications occurring after randomization is unlikely to bias results. The loss to follow-up was similar in both groups and mainly unrelated to maternal complications. Fourth, our study may have limited generalizability due to the sample characteristics and mobile phone access requirement to participate in the study. Our sample was predominantly non-Hispanic White, college-educated women. Therefore, it is unclear whether there are differences in the feasibility or magnitude of the intervention’s effect across racial and ethnic groups. Moreover, racial and ethnic minority women experience a disproportionate burden of sleep disturbances and adverse maternal health outcomes compared to non-Hispanic White women. Therefore, it is important to engage these populations in interventions to address sleep behaviors. (Patel et al., 2010) Furthermore, access to mobile phone compatibility with the study monitor was required for this study. However, access to mobile phones has increased from 35% to 77% from 2010 to 2016 and is predicted to increase in the future. (Smith & Pew Researcher Center, 2017) Fifth, we used self-reported questionnaires to identify maternal and fetal complications. One study has shown some level of discordance between self-reported recall and medical records of maternal and fetal complications at 5 to 8 years postpartum. (Keenan et al., 2017) However, our participants were asked to recall in a short period of time; therefore the risk of inaccurate reporting of pregnancy complications is likely low. Lastly, we did not exclude participants using their own PHM. Therefore, the comparison group participants may have had access to a PHM during the study. However, none of the participants in the sleep-education-only arm reported using a sleep monitor during the intervention.

Conclusion

In this pilot study, we found that a single session, nurse-administered sleep education intervention increased sleep quality and decreased daytime sleepiness in pregnant women with or without the use of a PHM. The additional use of a PHM provided slightly larger improvements in sleep quality. Larger, fully powered studies are needed to determine the efficacy of sleep education, paired with PHM for making clinically meaningful sleep improvements among pregnant women.