Abstract
Study Objectives
We investigated daily associations between step count and sleep quality across trimesters using wearable devices.
Methods
Participants (N = 243; pre-pregnancy body mass index≥ 25 kg/m2) from a mobile health randomized clinical trial intervention arm, wore Fitbits day and night from ~8 weeks’ gestation- delivery. Devices tracked daily step count (primary exposure), moderate to vigorous physical activity (MVPA) and light physical activity (LPA) (secondary exposures), and sleep measures (duration, stage length, efficiency, awakenings, midpoint and multidimensional sleep score). Covariate-adjusted mixed effects models estimated daily associations between movement and sleep outcomes, stratified by trimester.
Results
Participants averaged 5795 steps/day. In trimester 1, step count (per 1000) was associated with shorter sleep duration (−23 min, odds ratio [OR] = 0.84, 95% confidence interval [CI] = −38.7 to −8.6). In the trimester 2, step count was associated with shorter sleep duration (−22 min, CI = −27.6 to −16.4), shorter light sleep (−10 min, CI = −13.3 to −6.6), longer deep (+4 min, CI = 2.7 to 6.1), and rapid eye movement (REM) sleep (+6 min, CI = 3.5 to 7.7). In trimester 3, step count was associated with lower odds of poor sleep (OR = 0.84, CI = 0.70 to 1.00), shorter light sleep (−14 min, CI = −18.4 to −9.3), longer deep (+6 min, CI = 3.3 to 7.8), and REM sleep (+8 min, CI = 5.5 to 11.4), and more awakenings (+0.9, CI = 0.4 to 1.4). Associations of MVPA and LPA with sleep were smaller in magnitude but relatively consistent with step count.
Conclusions
Higher daily step count was associated with higher quality sleep in the following night during second and third trimesters. These findings highlight step count as a potential target to support prenatal sleep quality.
Keywords: pregnancy, wearables, step-count, physical activity, sleep quality
Statement of Significance.
High quality sleep is a vital aspect of whole-body and mind health for pregnant individuals. In our study, we used device-based longitudinal data to illustrate the relationship of wearable device-recorded daily step-count and physical activity minutes on sleep quality in pregnant individuals across trimesters. We showed that engaging in more steps led to improvements in overall sleep quality and sleep architecture, with the strongest effects observed in the third trimester. We also found shorter sleep duration on days with higher step count or moderate to vigorous physical activity, potentially due to reductions in light sleep. This paper highlights the importance of daily steps and activity, regardless of intensity, and acknowledges the complexity of sleep hygiene by adjusting for key confounders. Future directions include exploring potential mediators of the physical activity-sleep relationship in pregnancy.
Introduction
Pregnancy represents a critical period where various physiological changes can influence health outcomes for both the mother and the developing fetus. Among the many factors that affect maternal health, physical activity (PA), and sleep quality play pivotal roles. PA during pregnancy has been shown to have a range of health benefits, including reducing the risk of gestational diabetes, preeclampsia, and excessive weight gain [1, 2]. On the other hand, sleep disturbances, which are common during pregnancy, can lead to negative outcomes such as poor maternal glycemic control [3], increased risk for preterm birth, and poorer fetal development [4].
While both PA and sleep are individually important for maternal and fetal health, their interrelationship remains poorly understood. Recent advancements in technology have made it easier to objectively assess PA and sleep through wearable devices, providing an opportunity to analyze these two physical states in a continuous manner. Device-assessed PA offers a more precise measurement of activity levels, including step-count, intensity, frequency, and duration, compared to self-reported data [5]. Similarly, wearable devices can track sleep patterns, offering insights into sleep quality, including sleep duration, fragmentation, and overall restfulness [6].
The relationship between PA and sleep quality during pregnancy is complex and may vary across trimesters of pregnancy. Each trimester brings its own set of physiological changes, such as hormonal fluctuations, body composition alterations, and increasing fetal growth, all of which could impact both activity levels and sleep quality [7, 8]. In the first trimester, fatigue and morning sickness may reduce PA levels [9], while the third trimester may be associated with physical discomfort, back pain, and increased difficulty sleeping [10]. Thus, it is crucial to investigate how device-assessed PA correlates with sleep quality during different trimesters of pregnancy to provide a clearer understanding of how these factors interact and evolve throughout the trimesters.
Step count, as a simple yet informative metric of PA, may be particularly relevant for pregnant populations. Unlike structured exercise, which may be difficult to maintain due to pregnancy-related fatigue, nausea, or physical discomfort, walking is a low-impact, accessible form of movement that can be integrated into daily routines and sustained across trimesters, and is the most common form of exercise in pregnancy [11]. Step count provides a cumulative measure of overall movement throughout the day, making it a practical and non-invasive indicator of lifestyle activity. Prior research in non-pregnant populations has shown that higher daily step counts are associated with better sleep outcomes, including longer sleep duration and improved sleep efficiency [12, 13]. In pregnancy, when both sleep and exercise patterns fluctuate substantially, step count offers a feasible behavioral target for intervention and self-monitoring. Furthermore, because wearable devices can passively capture this data without relying on participant recall, step count is a reliable and scalable metric to explore the nuanced relationship between habitual PA and sleep health during pregnancy.
This analysis aims to explore the relationship between device-assessed PA and sleep quality across the three trimesters of pregnancy. Daily step count, a cumulative and interpretable indicator of total movement, was the primary exposure. Moderate to vigorous physical activity (MVPA) and light physical activity (LPA) minutes were examined as secondary exposures to contextualize the intensity-specific contributions to sleep outcomes. By using objective data, this study can help elucidate how changes in PA may influence sleep patterns, and how these dynamics vary across the different trimesters of pregnancy. Understanding this relationship could inform interventions aimed at improving both PA and sleep quality during pregnancy, ultimately contributing to better health outcomes for both mothers and their infants.
Materials and Methods
Participants
This secondary analysis utilizes data from a cluster randomized clinical trial with randomization at the clinician level (ClinicalTrials.gov NCT03880461) and the full study protocol has been previously published [14]. Briefly, the aim of the study was to evaluate a mobile health (mHealth) intervention promoting appropriate gestational weight gain in an integrated health system and enrolling individuals with a pre-pregnancy BMI of 25–40 kg/m2 and a singleton pregnancy into standard care or standard care and mHealth intervention. Participants in the intervention arm of LEAP received a structured, adaptive mHealth lifestyle intervention that included behavioral goals related to gestational weight gain, diet, stress management, and gradual progression toward 150 min per week of moderate-intensity PA. Intervention components included Fitbit-based self-monitoring, goal-setting with a lifestyle coach, automated feedback, and tailored calorie and activity targets. Sleep was not a behavioral target, and no sleep-related counseling or behavior change strategies were delivered; sleep measures were collected passively for monitoring and research purposes. Data specific to the present analysis come from patients of clinicians randomized to the mHealth intervention who wore a Fitbit device continuously (i.e. most or all days) from enrollment ~8 weeks until the end of the pregnancy. All participants were included regardless of pregnancy outcome (i.e. livebirth, pregnancy loss, and stillbirth).
Device-based measures of physical activity
Participants were provided with either a Charge 4 or Charge 5 Fitbit model or were able to use their personal Fitbit of any model if preferred. Fitbit Charge 4 and 5 devices used in this study have demonstrated high validity for step count [intraclass correlation coefficient (ICC) ~0.95–0.97] and heart rate (ICC ~0.90 for running) under most laboratory and free-living conditions [15, 16].
Fitbit devices tracked daily step count (primary exposure) as well as secondary exposures of moderate to vigorous “exercise minutes” (henceforth, MVPA) and lightly active minutes (henceforth, LPA). Daily step count was derived as the total number of steps accumulated within each 24-h period, as automatically recorded by Fitbit’s continuous tracking algorithm.
Device-based sleep measures
Nightly minutes of light sleep, deep sleep, and REM sleep, minutes awake in bed, wake-up count, bedtime, and waketime were generated by Fitbit internal algorithms. Fitbit sleep stage data have demonstrated moderate-to-good agreement with polysomnography, with particularly strong sensitivity for detecting deep and REM sleep stages [6, 17]. Additionally, total sleep duration (in h), sleep midpoint (the halfway point between bedtime and waketime), and a modified multidimensional sleep score were calculated. The multidimensional sleep score was conceptually adapted from the American Heart Association sleep health framework [18], which identifies sleep duration, timing, and continuity as key dimensions of cardiometabolic-relevant sleep. Three nightly indicators were included: sleep duration ≥7 h, sleep midpoint between 2:00 and 4:00 am, and sleep efficiency ≥85 per cent (score range: 0–3). These thresholds were selected based on adult sleep health recommendations [18] and pregnancy sleep literature [19] and represent commonly used cutpoints for sufficient duration, aligned timing, and consolidated sleep. Because no validated multidimensional device-based sleep index exists for pregnant populations, this score should be interpreted as a theory-informed summary indicator rather than a clinical diagnostic tool. For analysis, scores were categorized as 3 vs 0–2 to distinguish nights meeting all sleep health dimensions from those with at least one suboptimal domain, improving interpretability and model stability.
Prior to analysis, all days with bedtime onset outside of an 8 pm–4 am time window were removed, to exclude potential shift work, non-wear nights, and device errors which could bias our data. Daily activity exposures (step count, MVPA, LPA) were aligned with the nocturnal sleep period beginning the same evening (i.e. activity from 12:00 am to 11:59 pm was paired with sleep onset that night).
Covariates
We adjusted for a priori identified covariates: maternal age at enrollment, gestational day, pre-pregnancy BMI, total sleep duration, marital status, parity, race and ethnicity. Maternal age and pre-pregnancy BMI were extracted from electronic health records and presented continuously in our models. Gestational day (continuous) was defined as the number of days from the estimated last menstrual period to the specific observation day and used as a time-varying exposure in our models. Marital status was dichotomized as “living with partner/spouse” and “not living with partner/spouse.” Parity (continuous) was self-reported and defined numerically as the number of previous live births. We also dichotomized by “nulliparous” and “multiparous” to for descriptive purposes. Race and ethnicity were self-reported and categorized as “American Indian,” “Asian/Pacific Islander,” “Black,” “Hispanic,” “Islander,” and “Non-Hispanic White.”
Statistical analyses
All analyses were performed in SAS version 9.4. Descriptive statistics included measures of central tendency and variance for continuous variables and counts and proportions for categorical variables. To assess how similar the primary exposure (step count) was to the other secondary exposures (MVPA and LPA minutes), we ran Pearson correlations.
Analyses were conducted among participants in the intervention arm of LEAP and therefore do not estimate intervention effects but instead examine longitudinal associations within this cohort. Repeated nightly observations were nested within individuals, and mixed-effects models were used to account for within-person correlation. Random intercepts modeled between-person heterogeneity, and an autoregressive (AR [1]) covariance structure was specified for residuals, as behavioral measures collected closer in time are expected to be more strongly correlated.
Continuous sleep outcomes were analyzed using linear mixed-effects models, and the ordinal multidimensional sleep score was modeled using cumulative logit mixed-effects models. Models were estimated using maximum likelihood and adjusted for maternal age, pre-pregnancy BMI, race and ethnicity, parity, and marital status.
Additionally, differences in associations by trimesters were assessed by including multiplicative interaction terms in each model (alpha = 0.05). Results were presented for the overall sample and by trimester. Only complete cases were used for each model. Outliers/implausible data were removed.
Results
Participants and device data are described in Tables 1 and 2, respectively. There were 138 median total number of observation days per person with complete data (14 first trimester, 83 second trimester, and 66.5 third trimester). Step-count was moderately correlated with MVPA minutes (Pearson’s r = 0.57) and LPA minutes (Pearson’s r = 0.64).
Table 1.
Characteristics of pregnant participants who wore Fitbit devices (N = 243)
| Age, mean (SD) | 34.3 (4.2) |
| Gestational age at enrollment (weeks), mean (SD) | 13.9 (2.9) |
| Pre-pregnancy BMI, mean (SD) | 28.9 (3.2) |
| Nulliparous, n (%) | 101 (41.4) |
| Married/Partnered, n (%) | 219 (89.7) |
| Self-reported race and ethnicity | |
| American Indian | 7 (2.9) |
| Asian/Pacific Islander | 55 (22.5) |
| Black | 14 (6.1) |
| Hispanic | 58 (23.8) |
| Islander | 3 (1.2) |
| Non-Hispanic White | 86 (35.2) |
Table 2.
Descriptive characteristics of device data
| Overall | First trimester | Second trimester | Third trimester | |
|---|---|---|---|---|
| Total days with activity and sleep data * | 18,149 | 1,171 | 10,334 | 6,644 |
| Number of wear days per participant, median (IQR) | 138 (76, 200) | 14 (6, 24) | 83 (48, 97) | 66.5 (38, 83) |
| Daily activity characteristics | ||||
| Daily step count, mean (SD) | 6,761.5 (3606.5) | 7,089.1 (4011.4) | 7,053.8 (6674.5) | 6,256.0 (3376.2) |
| Daily MVPA minutes, mean (SD) | 19.6 (28.1) | 26.3 (34.0) | 21.0 (28.8) | 16.3 (25.3) |
| Daily LPA minutes, mean (SD) | 238.6 (104.0) | 234.7 (128.6) | 241.3 (102.3) | 235.1 (101.7) |
| Nightly sleep characteristics | ||||
| Sleep duration hours, mean (SD) | 7.25 (1.6) | 7.5 (1.5) | 7.3 (1.6) | 7.1 (1.6) |
| Bedtime HH:MM:SS, median (Q1, Q3) | 23:07:00 (22:15:30, 0:09:00) |
22:58:30 (21:58:30, 23:56:30) |
23:03:30 (22:11:00, 0:05:30) |
23:14:30 (22:23:00, 0:16:00) |
| Wake Time HH:MM:SS, median (Q1, Q3) | 7:13:30 (6:16:30, 8:04:30) |
7:11:30 (6:18:30, 8:03:30) |
7:12:30 (6:17:30, 8:05:30) |
7:15:00 (6:15:00, 8:03:30) |
| Light sleep minutes, mean (SD) | 291.2 (86.1) | 303.9 (89.1) | 289.3 (86.3) | 291.1 (85.1) |
| Deep sleep minutes, mean (SD) | 59.1 (29.1) | 63.1 (29.7) | 62.3 (28.9) | 53.5 (28.5) |
| REM sleep minutes, mean (SD) | 85.1 (38.6) | 82.8 (39.0) | 87.4 (38.5) | 82.0 (38.6) |
| Awake in bed minutes, mean (SD) | 64.1 (28.1) | 65.2 (30.3) | 63.6 (28.0) | 64.7 (27.8) |
| Sleep efficiency %, mean (SD) | 87.3 (4.7) | 87.5 (4.7) | 87.5 (4.4) | 86.9 (5.1) |
| Sleep midpoint HH:MM:SS, median (Q1, Q3) | 3:13:20 (2:27:30, 4:05:00) |
3:04:10 (2:20:50, 3:58:20) |
3:12:30 (2:26:40, 4:04:10) |
3:18:20 (2:30:50, 4:06:40) |
| Multidimensional sleep score | ||||
| 0, n (%) | 989 (5.45) | 51 (4.36) | 529 (5.12) | 409 (6.16) |
| 1, n (%) | 4,891 (26.96) | 263 (22.48) | 2,748 (26.60) | 1,880 (28.30) |
| 2, n (%) | 7,234 (39.87) | 508 (43.42) | 4,092 (39.61) | 2,634 (39.65) |
| 3, n (%) | 5,031 (27.73) | 348 (29.74) | 2,963 (28.68) | 1,720 (25.89) |
*Only days with a sleep start time between 8 pm and 4 am included. IQR= interquartile range
Several activity and sleep metrics were most favorable in the first trimester and least favorable in the third. Specifically, daily step count and MVPA were highest in the first trimester (7089 steps and 26 MVPA minutes) and lowest in the third (6256 steps and 16 MVPA minutes). Total sleep duration was longest in the first trimester (7.5 h) and shortest in the third (7.1 h). Bedtime and waketime were earliest in the first trimester (22:58 and 7:11) and latest in the third (23:14 and 7:15). Light and deep sleep were also highest in the first trimester (304 and 63 min) and lowest in the third (291 and 53 min). Sleep midpoint was earliest in the first trimester (3:04) and latest in the third (3:18). The proportion of nights with the best sleep score (score of 3) was also highest in the first trimester (30 per cent) and lowest in the third (26 per cent), while scores of 0 were least common in the first (4 per cent) and most common in the third (6 per cent).
Other metrics showed different patterns. Minutes of LPA were highest in the second trimester (241 min) and roughly equal in the first and third (235 min). REM sleep peaked in the second trimester (87 min) and was lowest in the third (82 min). Time awake in bed and sleep efficiency were relatively stable across trimesters, averaging 64 min awake and 87 per cent efficiency. Associations of step-count, MVPA, and LPA with sleep measures varied by trimester (p interaction <.05).
Associations of step count with sleep measures
Modified multidimensional sleep score
Each additional 1000 daily steps was associated with 16 per cent lower odds of having a low sleep score (defined as scoring 0–2 compared to 3) [OR: 0.84, confidence interval (CI) = 0.70 to 1.00, p = .049] in the third trimester (Figure 1). No associations were found between step count and multidimensional sleep score in the first and second trimesters.
Figure 1.
Associations of step count (per 1000 steps) with device-based sleep measures by trimester. The odds ratio in the top panel was estimated from a cumulative logit repeated measures model. Estimates from bottom six panels represent fixed effects results and standard errors from a mixed effects model with AR covariance structure. Models were adjusted for age, gestational age, pre-pregnancy BMI, total sleep duration (except for when it is modelled as an outcome), marital status, parity, race and ethnicity. CIs are 95%. Dashed lines represent the null.
Sleep duration
For every 1000 steps, sleep duration was shorter by 23 min in the first trimester (CI = −38.7 to −8.6), and 22 min in the second trimester (CI = −27.6 to −16.4), with no association found in the third trimester.
Sleep stages
For every 1000 steps taken, light sleep was shorter by 14 min overall (CI = −16.5 to −11.4), 10 min in the second trimester (CI = −13.3 to −6.6), and 14 min in the third trimester (CI = −18.4 to −9.3), with no association found in the first trimester. For every 1000 steps taken deep sleep was 4 min longer in the second trimester (CI = 2.7 to 6.1) and 6 min longer in the third trimester (CI = 3.3 to 7.8), with no association found in the first trimester. For every 1000 steps taken, REM sleep was 6 min longer in the second trimester (CI = 3.5 to 7.7) and 8 min longer in the third trimester (CI = 5.5 to 11.4), with no association found in the first trimester.
Sleep midpoint
There were no associations found between step-count and sleep midpoint.
Sleep efficiency
For every 1000 steps taken, sleep efficiency was 0.4 per cent greater overall (CI = 0.2 to 0.6), but no significant effect modification was found by trimester.
Sleep interruptions
For every 1000 steps taken, the number of awakenings was greater by 0.9 in the third trimester (CI = 0.4 to 1.4), but no associations were found in the first and second trimesters. For every 1000 steps taken, time awake in bed was 2 min shorter overall (CI = −3.6 to −1.1), but no significant effect modification was found by trimester.
Associations of moderate to vigorous and light physical activity on sleep measures
Multidimensional sleep score
Results on mixed effects models of the association between MVPA and LPA with sleep measures, stratified by trimester, are presented in Figures 2 and 3, respectively. Overall, for every additional 10 min of MVPA, the odds of having a score of 0–2 (compared to 3) was 1 per cent lower (OR: 0.99, CI = 0.97 to 0.997, p = .0149) but no associations between MVPA and sleep score were found by trimester. In the first trimester, for every additional 10 min of LPA, the odds of having a score of 0–2 (compared to 3) was 1 per cent higher (OR: 1.014, CI = 1.003 to 1.025, p = .0316), but no associations between LPA and sleep score were found in the second or third trimesters.
Figure 2.
Associations of moderate to vigorous physical activity with device-based sleep measures by trimester. The odds ratio in panel a was estimated from a cumulative logit repeated measures model. Estimates from panels b-i represent fixed effects results and standard errors from a mixed effects model with AR covariance structure. Models were adjusted for age, gestational age, pre-pregnancy BMI, total sleep duration (except for when it is modelled as an outcome), marital status, parity, race and ethnicity. CIs are 95%. Dashed lines represent the null.
Figure 3.
Associations of light physical activity with device-based sleep measures by trimester. The odds ratio in panel a was estimated from a cumulative logit repeated measures model. Estimates from panels b-i represent fixed effects results and standard errors from a mixed effects model with AR covariance structure. Models were adjusted for age, gestational age, pre-pregnancy BMI, total sleep duration (except for when it is modelled as an outcome), marital status, parity, race and ethnicity. CIs are 95%. Dashed lines represent the null.
Sleep duration
For every 10 min of LPA, sleep duration was shorter by 1 minute in the first and second trimesters (CI = −1.4 to −0.5 and −1.3 to −0.7, respectively), but no association in the third trimester. There were no associations between minutes of MVPA and sleep duration.
Sleep stages
For every 10 min of MVPA, light sleep was 0.6 min shorter in the second and third trimesters (CIs = −1.0 to −0.2 and −1.2 to −0.05, respectively). For every 10 min of MVPA, deep sleep was 0.2 min longer in the second trimester (CI = 0.01 to 0.4). For every 10 min of MVPA, REM sleep 0.4 min (CI = 0.1 to 0.6) longer in the second and third trimesters (CIs = 0.1 to 0.6 and 0.04 to 0.8, respectively). There were no associations between MVPA and minutes of light, deep, or REM sleep in the first trimester or deep sleep in the second trimester.
For every 10 min of LPA, light sleep was 0.3 min (CI = −0.5 to −0.2) shorter in the second and third trimesters (CIs = 0.4 to 0.1 and 0.4 to 0.1, respectively). For every 10 min of LPA, deep sleep was 0.2 min (CI = −0.3 to −0.03) shorter in the first trimester, 0.2 min longer in the second trimester (CI = 0.1 to 0.2), and 0.1 min (CI = 0.06 to 0.2) longer in the third trimester. For every 10 min of LPA, REM sleep was 0.17 min (0.1, 0.25) longer in the second trimester, 0.15 min (0.06, 0.26) longer in the third trimester. There were no associations between LPA and minutes of light or REM sleep in the first trimester.
Sleep midpoint
For every 10 min of LPA, sleep midpoint was 1 minute later (CI = 0.2 to 1.8) in the first trimester but no associations were found in the second and third trimesters. MVPA was not associated with sleep midpoint at any timepoint.
Sleep efficiency
For every 10 min of LPA, sleep efficiency was statistically, but not clinically significantly higher in the first trimester (+0.044% CI = 0.02% to 0.07%). There were no associations between LPA and sleep efficiency in the second and third trimesters and between MVPA and sleep efficiency at any timepoint.
Sleep interruptions
For every 10 min of MVPA, the number of awakenings was higher by 0.05 (CI = 0.01 to 0.09) overall, but there were no associations between MVPA and awakenings by trimester. LPA was not associated with number of awakenings at any timepoint. For every 10 min of LPA, time awake in bed was 0.23 min (CI = −0.37 to −0.08) shorter in the first trimester. There were no associations between LPA and time awake in bed in the second or third trimesters and between MVPA and time awake in bed at any timepoint.
Discussion
Step-count and sleep quality
In our sample of 243 pregnant people with 18 149 days of device data (median of 138 days per person), we found that greater step-count was generally associated with more favorable sleep measures, with few exceptions, and variation in associations by trimester.
Overall, a greater step count was associated with lower odds of poor sleep as measured by a modified multidimensional sleep score. This association was strongest in the third trimester, suggesting that maintaining or increasing steps in late pregnancy may have a particularly beneficial effect on composite measures of sleep health. These results support prior research indicating that PA promotes better sleep during pregnancy, particularly in reducing sleep complaints and improving subjective sleep quality [20, 21]. These trimester-specific findings support the hypothesis that PA may play a compensatory role in late pregnancy by enhancing restorative sleep stages and offsetting sleep disruptions due to fetal movement, discomfort, or frequent urination [22]. (Although, number of awakenings was greater with more steps in the third trimester—potentially due to sleep disturbances due to pelvic discomfort on days with more steps).
Conversely, in early pregnancy, the association of more steps or LPA with shorter sleep may indicate that that some of the potential sleep time is replaced by time in light stepping movement. This may be more of an indicator of movement at night from getting up due to restlessness, finishing up household or work duties, or other forms of movement that may be more obligatory (i.e. necessity vs choice movement [23]). There may also be unknown confounders in this relationship, but little data exists on determinants of sleep duration in pregnancy.
However, higher step counts were associated with shorter sleep duration, by approximately 14 min per 1000 steps overall, and up to 23 min shorter in the first trimester. While counterintuitive, this may reflect a tradeoff between total sleep time and more restorative sleep stages. Supporting this interpretation, step count was associated with longer deep and REM sleep, both critical for physical restoration and cognitive processing [24]. Specifically, deep sleep was longer by 6 min and REM by 7 min per 1000 steps overall, suggesting that PA may enhance sleep architecture even as it slightly shortens overall sleep time. These findings align with studies in non-pregnant populations where PA improves sleep efficiency and increases time spent in restorative stages, even when total sleep duration remains unchanged or slightly reduced [25]. Importantly, sleep duration during pregnancy is often shorter due to physiological and hormonal changes [10], so increases in deep and REM sleep may compensate for decreased sleep length.
To aid interpretation of the observed activity levels in our cohort, it is useful to consider these values relative to existing normative data. Large population studies in U.S. adults have reported average daily step counts of approximately 6000–7000 steps per day, with many adults not meeting commonly cited thresholds for “active” status (e.g. ≥7000 steps/day) [26]. Although normative data specific to pregnancy are limited, prior studies using accelerometer data in pregnant populations have generally observed mean daily steps in the range of ~5000–8000 steps/day across trimesters [27]. In this context, the mean step counts in our cohort (7089 in the first trimester and 6256 in the third trimester) are similar to general adult norms and within the range reported for pregnant populations, suggesting that our participants maintained activity levels that approximate typical adult ambulatory behavior, with a modest decline across pregnancy.
MVPA and LPA: diverging influences on sleep
When examining MVPA, we observed a slightly lower (1 per cent) odds of poor sleep, overall, suggesting that although MVPA may be helpful for sleep in pregnancy, it may be better suited as adjuvant therapy paired with an intervention such as cognitive behavioral therapy [28]. MVPA was also associated with longer deep and REM sleep, echoing the step count results. These effects may reflect the role of more intense physical exertion in regulating circadian rhythms, increasing homeostatic sleep pressure, and reducing stress, all of which are implicated in improved sleep outcomes [29].
Several associations appeared trimester-specific, emphasizing the dynamic physiological and behavioral shifts across pregnancy. The stronger association between step count and sleep quality in the third trimester may reflect the increasing burden of physical discomfort and sleep disruption later in gestation, which activity may partially offset.
In contrast, LPA exhibited more nuanced associations. While it was associated with shorter sleep duration and light sleep, it was also linked to modestly greater sleep efficiency and shorter time awake in bed, particularly in the first trimester. Interestingly, more LPA was associated with slightly worse (but not clinically significant) sleep scores in the first trimester (OR: 1.014), raising the possibility that very low-intensity activity without sufficient exertion may not provide the same sleep-promoting benefits as MVPA. LPA may capture other types of movement (such as household chores or occupational activity) that may not reap the same sleep benefits [30]. This finding warrants further exploration but may reflect confounding from nausea or fatigue-related activity limitations early in pregnancy [31].
Taken together, these findings suggest that while step count and MVPA were consistently linked to deeper, more restorative sleep, LPA showed more variable associations, likely due to differences in intensity and context. MVPA may reflect intentional exertion that promotes sleep, whereas LPA may capture low-effort movements that occur even during fatigue or discomfort, particularly in early pregnancy.
Strengths, limitations, and future directions
This study’s strengths include the use of device-based measures of both activity and sleep, a large range of sleep outcomes, and trimester-stratified analyses. By using timestamped, device assessments of PA and sleep, we were able to assess the prospective temporal association between PA and sleep. Furthermore, continuous recording of movement and sleep data, allowed us to maximize our cohort to obtain a large amount of longitudinal data during pregnancy.
Limitations should be noted: observational design which limits causal inference and residual confounding (e.g. stress, diet, or sleep environment) may persist. Additionally, the generalizability of our findings may be limited by the characteristics of the study sample. Participants were enrolled and participated in the intervention arm of a randomized controlled trial and were required to continuously wear a Fitbit device throughout pregnancy. This level of engagement may select for individuals who are more health-conscious, technologically comfortable, or motivated, which may not reflect the broader pregnant population. Thus, findings may be most applicable to individuals with similar characteristics and access to digital health tools. Although PA was an intervention target in LEAP, the present analyses do not estimate intervention effects; rather, they examine associations between variation in activity and sleep measures within the cohort of participants in the intervention arm only. Therefore, findings should be interpreted as observational associations, not causal effects of the LEAP program. Furthermore, the overall intervention did not have a significant effect on PA. Nonetheless, because the intervention also targeted BMI, it is possible that participation influenced PA behaviors, which could in turn affect the associations observed with sleep. This potential influence should be considered when interpreting results, as the observed activity–sleep relationships may partly reflect behavioral changes related to the intervention rather than purely naturally occurring variation. Furthermore, although associations were statistically significant, most effect sizes were small (e.g. 6–8 minute increases in deep/REM sleep per 1000 steps) and may not translate into clinically meaningful changes on their own. However, previous studies on sleep health generally report summary-level measures of sleep from the past month or week, whereas we report daily associations, which may reflect small, daily improvements in sleep health during pregnancy that accumulate to larger benefits in cognitive and metabolic outcomes [3, 32, 33]. Notably, the 1 per cent lower odds of having a worse modified multidimensional sleep score may not lead to substantial health changes, but a 1-point change in a 22-point composite sleep score led to an increased risk of mortality in the Sleep Heart Health Study [34], suggesting that daily engagement in MVPA may cumulatively lead to improved sleep health and subsequent mortality prevention. While this composite score is not formally validated in pregnancy, it provides an interpretable multidimensional representation of nightly sleep health grounded in established sleep domains relevant to cardiometabolic risk [18, 19].
It is also important to acknowledge the limitations of commercial wearable devices, such as Fitbit, in detecting movement and sleep. Device-derived activity and sleep measures may share some measurement variance. Wearable devices can register low-intensity movement during nocturnal awakenings as steps, which could artifactually link higher step counts or light activity with shorter sleep duration. The absence of an association with MVPA is consistent with this possibility, as higher-intensity movement is unlikely during awakenings. Thus, part of the observed association may reflect nocturnal movement misclassification rather than a true behavioral trade-off. However, the magnitude of the effect suggests that daytime activity differences likely also contribute. These findings should therefore be interpreted with consideration of potential device measurement error. Furthermore, we are limited in our ability to exclude nocturnal activity due to the proprietary nature of commercial wearable data. Nevertheless, Fitbit has been previously validated against the gold standard [6], polysomnography (SG), and found to have moderate accuracy in assessing sleep stages and total sleep time compared to polysomnography (PSG) [17]. Specifically, the Fitbit has a mere 4-minute disagreement in REM sleep relative to PSG and has higher sensitivities to deep sleep (75 per cent) and REM sleep (86.5 per cent) compared to other popular commercial brand wearables on the market. It should also be noted that although PSG is the gold standard, there is currently no direct method of measuring sleep [18]. Although estimates of MVPA are moderately valid in Fitbit devices, they tend to be overestimated relative to research-grade accelerometers due to Fitbit’s reliance on proprietary heart rate-based zone algorithms [35, 36].
Conclusion
Higher daily step count was associated with more optimal sleep quality during mid- to late-pregnancy. Despite shorter total sleep duration on higher PA days, sleep architecture, specifically longer deep and REM sleep, was more favorable on nights following more steps or MVPA. These findings highlight step count as a potentially accessible behavioral target to support better sleep quality during pregnancy, particularly in the third trimester.
Contributor Information
Bethany R Hallenbeck, Division of Research, Kaiser Permanente Northern California, Pleasanton, CA, United States.
Sylvia E Badon, Division of Research, Kaiser Permanente Northern California, Pleasanton, CA, United States.
Fei Xu, Division of Research, Kaiser Permanente Northern California, Pleasanton, CA, United States.
Charles P Quesenberry, Division of Research, Kaiser Permanente Northern California, Pleasanton, CA, United States.
Monique M Hedderson, Division of Research, Kaiser Permanente Northern California, Pleasanton, CA, United States.
Author contributions
Bethany R. Hallenbeck (Conceptualization [lead], Formal analysis [lead], Investigation [lead], Methodology [lead], Project administration [lead], Visualization [lead], Writing—original draft [lead], Writing—review & editing [lead]), Sylvia E. Badon (Conceptualization [supporting], Formal analysis [supporting], Funding acquisition [supporting], Investigation [supporting], Methodology [supporting], Project administration [supporting], Supervision [equal], Visualization [supporting], Writing—review & editing [supporting]), Fei Xu (Data curation [lead], Formal analysis [supporting], Methodology [supporting], Resources [supporting], Writing—review & editing [supporting]), Charles P. Quesenberry (Data curation [supporting], Formal analysis [supporting], Funding acquisition [supporting], Investigation [supporting], Methodology [supporting], Project administration [supporting], Resources [supporting], Supervision [supporting], Validation [supporting], Writing—review & editing [supporting]), and Monique M. Hedderson (Conceptualization [supporting], Data curation [equal], Formal analysis [supporting], Funding acquisition [lead], Investigation [supporting], Methodology [supporting], Project administration [supporting], Resources [lead], Software [lead], Supervision [equal], Validation [supporting], Visualization [supporting], Writing—original draft [supporting], Writing—review & editing [supporting])
Funding
Dr. Hallenbeck received funding from the Translational Research Fellowship Program at the Kaiser Permanente Northern California Division of Research. This project was also supported by the NIDDK R01DK118455 (PI: Monique Hedderson).
Disclosure statement
Financial disclosure: None declared.
Non-financial disclosure: None declared.
Data availability
Individual level data may not be made publicly available due to IRB and privacy concerns. The data used for this study contain protected health information (PHI) and access is protected by the Kaiser Permanente Northern California Institutional Review Board (IRB). Data are available from the Kaiser Permanente Division of Research for researchers who meet the criteria for access to confidential data. For more information about data access and criteria for access to confidential data, please contact Kaiser Permanente Division of Research: DOR.IRB.Submissions@kp.org. All other relevant data are within the paper.
References
- 1. Mottola MF, Artal R. Fetal and maternal metabolic responses to exercise during pregnancy. Early Hum Dev. 2016;94:33–41. 10.1016/j.earlhumdev.2016.01.008 [DOI] [PubMed] [Google Scholar]
- 2. Physical activity and exercise during pregnancy and the postpartum period: ACOG Committee opinion, number 804. Obstet Gynecol. 2020;135(4):e178–e188. 10.1097/AOG.0000000000003772 [DOI] [PubMed] [Google Scholar]
- 3. Jiang M, Sui R, Wu X. Association between sleep quality and duration during pregnancy and risk of gestational diabetes: a systematic review and meta-analysis. Gynecol Endocrinol. 2024;40(1):2391925. 10.1080/09513590.2024.2391925 [DOI] [PubMed] [Google Scholar]
- 4. Lu Q, Zhang X, Wang Y, et al. Sleep disturbances during pregnancy and adverse maternal and fetal outcomes: a systematic review and meta-analysis. Sleep Med Rev. 2021;58:101436. 10.1016/j.smrv.2021.101436 [DOI] [PubMed] [Google Scholar]
- 5. Ehrlich SF, Casteel AJ, Crouter SE, et al. Alternative wear-time estimation methods compared to traditional diary logs for wrist-worn ActiGraph accelerometers in pregnant women. J Meas Phys Behav. 2020;3(2):110–117. 10.1123/jmpb.2019-0049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Haghayegh S, Khoshnevis S, Smolensky MH, Diller KR, Castriotta RJ. Accuracy of wristband Fitbit models in assessing sleep: systematic review and meta-analysis. J Med Internet Res. 2019;21(11):e16273. 10.2196/16273 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Pien GW, Schwab RJ. Sleep disorders during pregnancy. Sleep. 2004;27(7):1405–1417. 10.1093/sleep/27.7.1405 [DOI] [PubMed] [Google Scholar]
- 8. Sedov ID, Cameron EE, Madigan S, Tomfohr-Madsen LM. Sleep quality during pregnancy: a meta-analysis. Sleep Med Rev. 2018;38:168–176. 10.1016/j.smrv.2017.06.005 [DOI] [PubMed] [Google Scholar]
- 9. Barakat R, Pelaez M, Cordero Y, et al. Exercise during pregnancy protects against hypertension and macrosomia: randomized clinical trial. Am J Obstet Gynecol. 2016;214(5):649.e1–649.e8. 10.1016/j.ajog.2015.11.039 [DOI] [PubMed] [Google Scholar]
- 10. Mindell JA, Cook RA, Nikolovski J. Sleep patterns and sleep disturbances across pregnancy. Sleep Med. 2015;16(4):483–488. 10.1016/j.sleep.2014.12.006 [DOI] [PubMed] [Google Scholar]
- 11. Evenson KR, Barakat R, Brown WJ, et al. Guidelines for physical activity during pregnancy: comparisons from around the world. Am J Lifestyle Med. 2014;8(2):102–121. 10.1177/1559827613498204 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sullivan Bisson AN, Robinson SA, Lachman ME. Walk to a better night of sleep: testing the relationship between physical activity and sleep. Sleep Health. 2019;5(5):487–494. 10.1016/j.sleh.2019.06.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Buman MP, Hekler EB, Bliwise DL, King AC. Moderators and mediators of exercise-induced objective sleep improvements in midlife and older adults with sleep complaints. Health Psychol. 2011;30(5):579–587. 10.1037/a0024293 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Sridhar SB, Ferrara A, Brown SD, et al. Protocol of an adaptive mobile health intervention for the management of gestational weight gain: the LEAP cluster randomized controlled trial. Contemp Clin Trials. 2025;149:107781. 10.1016/j.cct.2024.107781 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Ceugniez M, Devanne H, Hermand E. Reliability and accuracy of the Fitbit charge 4 photoplethysmography heart rate sensor in ecological conditions: validation study. JMIR Mhealth Uhealth. 2025;13:e54871. 10.2196/54871 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Feehan LM, Geldman J, Sayre EC, et al. Accuracy of Fitbit devices: systematic review and narrative syntheses of quantitative data. JMIR Mhealth Uhealth. 2018;6(8):e10527. 10.2196/10527 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Schyvens AM, Van Oost NC, Aerts JM, et al. Accuracy of Fitbit charge 4, Garmin Vivosmart 4, and WHOOP versus polysomnography: systematic review. JMIR Mhealth Uhealth. 2024;12:e52192. 10.2196/52192 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. St-Onge M-P, Aggarwal B, Fernandez-Mendoza J, et al. Multidimensional sleep health: definitions and implications for cardiometabolic health: a scientific statement from the American Heart Association. Circ Cardiovasc Qual Outcomes. 2025;18(5):e000139. 10.1161/HCQ.0000000000000139 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Hawkins MS, Pokutnaya DY, Bodnar LM, et al. The association between multidimensional sleep health and gestational weight gain. Paediatr Perinat Epidemiol. 2023;37(7):586–595. 10.1111/ppe.13004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Yang SY, Lan SJ, Yen YY, Hsieh YP, Kung PT, Lan SH. Effects of exercise on sleep quality in pregnant women: a systematic review and meta-analysis of randomized controlled trials. Asian Nurs Res (Korean Soc Nurs Sci). 2020;14(1):1–10. 10.1016/j.anr.2020.01.003 [DOI] [PubMed] [Google Scholar]
- 21. Banno M, Harada Y, Taniguchi M, et al. Exercise can improve sleep quality: a systematic review and meta-analysis. PeerJ. 2018;6:e5172. 10.7717/peerj.5172 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Facco FL, Kramer J, Ho KH, Zee PC, Grobman WA. Sleep disturbances in pregnancy. Obstet Gynecol. 2010;115(1):77–83. 10.1097/AOG.0b013e3181c4f8ec [DOI] [PubMed] [Google Scholar]
- 23. Salvo D, Jáuregui A, Adlakha D, Sarmiento OL, Reis RS. When moving is the only option: the role of necessity versus choice for understanding and promoting physical activity in low- and middle-income countries. Annu Rev Public Health. 2023;44(1):151–169. 10.1146/annurev-publhealth-071321-042211 [DOI] [PubMed] [Google Scholar]
- 24. Hirshkowitz M, Whiton K, Albert SM, et al. National Sleep Foundation's updated sleep duration recommendations: final report. Sleep Health. 2015;1(4):233–243. 10.1016/j.sleh.2015.10.004 [DOI] [PubMed] [Google Scholar]
- 25. Dolezal BA, Neufeld EV, Boland DM, Martin JL, Cooper CB. Interrelationship between sleep and exercise: a systematic review. Adv Prev Med. 2017;2017:1364387. 10.1155/2017/1364387 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? For adults. Int J Behav Nutr Phys Act. 2011;8(1):79. 10.1186/1479-5868-8-79 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Kim Y, Chung E. Descriptive epidemiology of objectively measured walking among US pregnant women: National Health and Nutrition Examination Survey, 2005-2006. Prev Chronic Dis. 2015;12:E217. 10.5888/pcd12.150437 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Bei B, Pinnington DM, Quin N, et al. Improving perinatal sleep via a scalable cognitive behavioural intervention: findings from a randomised controlled trial from pregnancy to 2 years postpartum. Psychol Med. 2023;53(2):513–523. 10.1017/S0033291721001860 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Kredlow MA, Capozzoli MC, Hearon BA, Calkins AW, Otto MW. The effects of physical activity on sleep: a meta-analytic review. J Behav Med. 2015;38(3):427–449. 10.1007/s10865-015-9617-6 [DOI] [PubMed] [Google Scholar]
- 30. Borodulin K, Evenson KR, Monda K, Wen F, Herring AH, Dole N. Physical activity and sleep among pregnant women. Paediatr Perinat Epidemiol. 2010;24(1):45–52. 10.1111/j.1365-3016.2009.01081.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Liu D, Liu K, Xie L, He M, Liu X. The effect of physical activity on sleep disorders in pregnant people: a meta-analysis of randomized controlled trials. BMC Pregnancy Childbirth. 2025;25(1):139. 10.1186/s12884-024-07129-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Nevarez-Brewster M, Demers CH, Mejia A, et al. Longitudinal and prospective assessment of prenatal maternal sleep quality and associations with newborn hippocampal and amygdala volume. Dev Cogn Neurosci. 2022;58:101174. 10.1016/j.dcn.2022.101174 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Zhang L, Wang HX, Li WX, et al. Association of maternal short sleep duration with neurodevelopmental delay in offspring: a prospective cohort study. J Clin Endocrinol Metab. 2025;110(7):2016–2023. 10.1210/clinem/dgae569 [DOI] [PubMed] [Google Scholar]
- 34. Xing M, Zhang L, Li J, Li Z, Yu Q, Li W. Development and validation of a novel sleep health score in the sleep heart health study. Eur J Intern Med. 2024;127:112–118. 10.1016/j.ejim.2024.05.002 [DOI] [PubMed] [Google Scholar]
- 35. Ferguson T, Rowlands AV, Olds T, Maher C. The validity of consumer-level, activity monitors in healthy adults worn in free-living conditions: a cross-sectional study. Int J Behav Nutr Phys Act. 2015;12(1):42. 10.1186/s12966-015-0201-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Dominick GM, Winfree KN, Pohlig RT, Papas MA. Physical activity assessment between consumer- and research-grade accelerometers: a comparative study in free-living conditions. JMIR Mhealth Uhealth. 2016;4(3):e110. 10.2196/mhealth.6281 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Individual level data may not be made publicly available due to IRB and privacy concerns. The data used for this study contain protected health information (PHI) and access is protected by the Kaiser Permanente Northern California Institutional Review Board (IRB). Data are available from the Kaiser Permanente Division of Research for researchers who meet the criteria for access to confidential data. For more information about data access and criteria for access to confidential data, please contact Kaiser Permanente Division of Research: DOR.IRB.Submissions@kp.org. All other relevant data are within the paper.