Sleep Quality in Individuals with and without Persistent Postpartum Hypertension

Abstract

Objective

This study aimed to compare sleep quality at 1 year postpartum following a hypertensive disorder of pregnancy (HDP) among individuals with persistent postpartum hypertension (HTN) compared with those with normal blood pressures (BPs).

Study Design

We combined data from the Heart Health 4 New Moms pilot randomized trial (n = 118) and the Pathways prospective cohort study (n = 36). Individuals with a singleton pregnancy complicated by gestational HTN or preeclampsia underwent a research study visit at a mean 48.7 ± 9.5 weeks postpartum with standardized BP measurement and assessment of subjective sleep quality with the Pittsburgh Sleep Quality Index (PSQI). Persistent postpartum HTN was defined as Stage 1 HTN or greater (mean systolic BP ≥ 130 mm Hg or mean diastolic BP ≥ 80 mm Hg over three measurements at rest) or requiring antihypertensive medication. Statistical analysis was performed using univariate and multivariable logistic regression analyses.

Results

Of 154 individuals with an HDP included in the analysis, 84 (55%) were normotensive at 1 year postpartum and 70 (45%) had persistent postpartum HTN. Individuals with persistent postpartum HTN were more likely to be older, self-identify as Black race, have higher prepregnancy and 1-year postpartum body mass index (BMI), be multiparous, and deliver at an earlier gestational age. The mean global PSQI score was 8.7 ± 3.7, with 81% reporting poor sleep (PSQI > 5), and scores were higher among individuals who were persistently hypertensive (9.6 ± 3.5) compared with those who were normotensive at 1 year postpartum (7.9 ± 3.6), p < 0.01. Findings were unchanged in a multivariable model adjusting for age, self-reported race, prepregnancy BMI, and parity.

Conclusion

Following an HDP, individuals reported poor sleep quality at 1 year postpartum. Individuals with persistent postpartum HTN reported lower sleep quality, suggesting that sleep behavior may be a target for intervention to improve maternal cardiovascular health following an HDP.

Hypertensive disorders of pregnancy (HDP) including gestational hypertension (HTN) and preeclampsia have been shown to be risk factors for the future development of cardiovascular disease. The underlying mechanisms contributing to this increased risk are not well understood but thought to be related to endothelial dysfunction and progression to chronic HTN in the years following delivery.^1–7

Outside of pregnancy and the postpartum period, sleep disturbances are associated with an increased risk of cardiovascular disease by what is thought to be a similar mechanism involving endothelial dysfunction and vascular inflammation.^8,9 In the first few months postpartum, sleep disturbances are widespread and uniquely characterized by frequent nocturnal awakenings, resultant sleep fragmentation, and reduced quality of sleep.^10–14 It seems plausible that in individuals with an HDP, who already have endothelial dysfunction,⁷ poor sleep in the postpartum period may further compound hypertensive and cardiovascular risks in this susceptible population.

Given that disrupted and inadequate sleep are nearly universal in the postpartum period,¹⁵ there is an urgent need to study its impact on cardiovascular health in high-risk postpartum individuals. As such, we sought to compare sleep quality at 1 year postpartum following an HDP among individuals with persistent postpartum HTN compared with those with normal blood pressures (BPs).

Methods

This was a secondary analysis of combined data from the Heart Health 4 New Moms (HH4NM) pilot randomized clinical trial and the Pathways prospective cohort study. Institutional Review Board approval was granted for the two studies.

The HH4NM study was a prospective pilot randomized trial consisting of 148 overweight and obese individuals (body mass index [BMI] ≥25 kg/m²) with an HDP who delivered at either Magee-Womens Hospital in Pittsburgh or Erie, PA, or were recruited through the institution’s partner community organization, Healthy Start, between January 2019 and May 2021. Individuals were randomized 1:1:1 to either the control group, home BP monitoring program alone, or home BP monitoring plus an internet-based lifestyle program, HH4NM. There were no statistical differences in systolic BP, diastolic BP, or sleep characteristics across intervention groups; thus, all study arms were included for this analysis.¹⁶ Participants were followed through 1 year postpartum with research study visits consisting of standardized BP measurements, dietary surveys, mood assessments, and sleep quality assessment including the Pittsburgh Sleep Quality Index (PSQI).¹⁷ The PSQI is a 19-question self-rated questionnaire which evaluates subjective sleep quality and disturbances over the preceding 1 month. Global PSQI scores range from 0 to 21, with higher numbers indicating worse sleep quality, and a score of greater than 5 indicating poor sleep quality.

Pathways was a prospectively enrolled cohort of 140 individuals who delivered at Magee-Womens Hospital in Pittsburgh, PA, between July 2016 and July 2019. Individuals with an HDP, fetal growth restriction, or preterm birth, along with a control group of individuals were followed through 1 year postpartum with research study visits consisting of BP measurements, activity surveys, and sleep quality assessment including the PSQI.

Individuals with a singleton pregnancy complicated by an HDP (gestational HTN, preeclampsia without severe features, or preeclampsia with severe features) and no prepregnancy history of chronic HTN who had a 1-year postpartum follow-up assessment with the PSQI were included (n = 118 from HH4NM and n = 36 from the Pathways cohort). The diagnosis of an HDP was adjudicated by physicians (A.K.H. and A.J. for both studies) using criteria from the American College of Obstetricians and Gynecologists for the diagnoses of gestational HTN, preeclampsia without severe features, and pre-eclampsia with severe features.¹⁸ Finally, given that there is a poor overall understanding of sleep in the postpartum period regardless of an HDP diagnosis, we performed an additional sensitivity analysis to assess sleep quality at 1 year postpartum in an additional 44 normotensive individuals with uncomplicated pregnancies in the Pathways prospective cohort study.

The resulting combined cohort of 154 individuals with an HDP was included in the analysis. Information on demographics, medical history, and pregnancy outcomes was collected. Race was self-reported by participants with the options of “White,” “Black,” or “None of the above,” provided by the investigators of each respective study. Assessment of race was included given that BP trajectories in the postpartum period following an HDP have been found to differ by self-reported race, with Black individuals having a lower decrease in BP in the 6-week postpartum period compared with White individuals.¹⁹

Participants underwent a research study visit at a mean 48.7 ± 9.5 weeks postpartum with standardized BP measurement and assessment of subjective sleep quality with the PSQI. BP measurements were performed in triplicate after at least 5 minutes at rest using an iHealth, Wireless Blood Pressure Monitor BP5 or an A&D UA-651 (A&D Medical; San Jose, CA) automatic upper arm BP monitor, both validated by Dabl Educational Trust and the British Society for Hypertension, for use in postpartum individuals. The mean of the three measures was used in analysis. Individuals were classified as having persistent postpartum HTN if, at the time of the 1 year postpartum research study visit, they had Stage 1 HTN or greater (mean systolic BP ≥ 130 mm Hg or mean diastolic BP ≥ 80 mm Hg over three measurements at rest) or required antihypertensive medications.

Statistical analysis was performed using Stata (StataCorp LLC, version 17.0, College Station, TX). Univariate analysis was conducted when indicated. Pearson’s chi-square tests were performed to analyze the differences between categorical variables. Two-sample two-sided t-tests and Mann–Whitney U tests were performed to analyze the differences between two groups of continuous, normally distributed, and non-normally distributed variables, respectively. A multivariable logistic regression model to evaluate the relationship between persistent HTN and PSQI score was also performed. We adjusted for prespecified covariates, including age, self-reported race, prepregnancy BMI, parity, and intervention group for individuals enrolled in HH4NM. For all statistical tests, the α level was set at 0.05.

Results

Of the 154 individuals with an HDP included in the analysis, 84 (55%) were normotensive without antihypertensive medication use at 1 year postpartum, and 70 (45%) had persistent postpartum HTN. Of the 70 individuals with persistent postpartum HTN, 10 were on an antihypertensive medication at 1 year postpartum, representing 6% (10/154) of the entire cohort and 14% (10/70) of those with persistent postpartum HTN.

The demographic characteristics of the cohort are listed in ►Table 1. Compared with individuals with HDP who were normotensive at 1 year postpartum, those with persistent postpartum HTN were older (mean 31.8 4.7 vs. 30.0 5.5 years, p = 0.04), more likely to identify as Black race (24 vs. 10%, p = 0.01), have a higher prepregnancy BMI (median 31.7 kg/m² [interquartile range (IQR) 28.2–36.5] vs. median 29.2 kg/m² [IQR 25.7–33.1], p < 0.01), higher BMI at 1 year postpartum (median 33.0 kg/m² [IQR 29.1–40.2] vs. median 29.8 kg/m² [IQR 26.0–35.3], p < 0.01), be multiparous (51 vs. 34%, p = 0.03), and deliver at a slightly earlier gestational age (median 37.3 weeks [IQR 36.6–38.4] vs. median 38.0 weeks [IQR 37.0–39.4], p = 0.04). Pregnancy outcomes including the preterm delivery rate, neonatal birthweight, proportion of neonates who were born small (<10th percentile) for gestational age, infant sex at birth, breastfeeding rates, and smoking rates were similar between the two HDP groups. The type of HDP (gestational HTN, preeclampsia without severe features, and preeclampsia with severe features) was also similar between the groups.

Table 1.

Demographics

Demographics	Normotensive at 1 year postpartuma	Hypertensive at 1 year postpartumb	p-Valuec
	n = 84	n = 70	-
Age at deliveryd	30.0 ± 5.5	31.8 ± 4.7	0.04
Self-reported race
White	76 (90%)	51 (73%)	–
Black	8 (10%)	17 (24%)	0.01
None of the above	0 (0%)	2 (3%)	–
Median BMI prepregnancy (kg/m2)e	29.2 [IQR 25.7–33.1]	31.7 [IQR 28.2–36.5]	<0.01
Median BMI 1 year postpartum (kg/m2)e	29.8 [IQR 26.0–35.3]	33.0 [IQR 29.1–40.2]	<0.01
Multiparous	28 (34%)	35 (51%)	0.03
Median gestational age at delivery (weeks)e	38.0 [IQR 37.0–39.4]	37.3 [IQR 36.6–38.4]	0.04
Preterm	19 (23%)	20 (29%)	0.37
Type of HDP
Gestational hypertension	28 (33%)	21 (30%)	0.68
Preeclampsia without severe features	27 (32%)	20 (29%)
Preeclampsia with severe features	29 (35%)	29 (41%)
Mean systolic BP at 1 year postpartumd	112.8 ± 7.3	127.2 ± 9.9	<0.01
Mean diastolic BP at 1 year postpartumd	72.5 ± 5.6	85.7 ± 6.3	<0.01
Antihypertensive medications at 1 year postpartum	0 (0%)	10 (14%)	<0.01
Median birthweight (g)e	3,080 [IQR 2,625–3,408]	3,145 [IQR 2,590–3,490]	0.59
Small for gestational age	12 (14%)	7 (10%)	0.42
Infant sex
Female	41 (49%)	28 (41%)	0.31
Male	43 (51%)	41 (59%)
Ever breastfed	71 (86%)	55 (82%)	0.57
History of smoking	11 (13%)	11 (16%)	0.61

Abbreviations: BMI, body mass index; BP, blood pressure; HDP, hypertensive disorder of pregnancy.

^aData are reported as n (%), unless otherwise indicated.

^bMean systolic BP ≥ 130 mm Hg or mean diastolic BP ≥ 80 mm Hg over three measurements at rest, or requiring antihypertensive medication.

^cp-Values indicate the statistical testing result from two-sample t-tests for normally distributed continuous variables, Mann–Whitney U tests for non-normally distributed continuous variables, or Pearson’s chi-square for categorical variables.

^dData are reported as mean ± standard deviation.

^eData are reported as median [interquartile range].

Sleep outcomes at 1 year postpartum are listed in ►Table 2. The mean global PSQI score was 8.7 ± 3.7, with 81% (125/154) reporting poor subjective sleep quality (global PSQI > 5). The mean global PSQI score was 7.9 ± 3.6 in individuals with HDP who were normotensive at 1 year postpartum, and 9.6 ± 3.5 in those with persistent postpartum HTN, p < 0.01. The proportion of individuals reporting poor subjective sleep quality (global PSQI > 5) was 73% (61/84) in individuals who were normotensive at 1 year postpartum, and 91% (64/70) of those with persistent postpartum HTN (p < 0.01).

Table 2.

Sleep outcomes at 1 year postpartum

Sleep outcome	Normotensive at 1 year postpartuma	Hypertensive at 1 year postpartumb	p-Valuec
	n = 84	n = 70
Global PSQI scored	7.9 ± 3.6	9.6 ± 3.5	<0.01
Global PSQI score >5 (poor sleep quality)	61 (73%)	64 (91%)	<0.01
Median reported sleep qualitye (scale of 1 = very good; 4 = very bad)	2 [IQR 2–3]	2.5 [IQR 2–3]	0.05
Median reported sleep satisfactione (scale of 1 = very dissatisfied; 5 = very satisfied)	n = 52 3 [IQR 2–4]	n = 66 2 [IQR 2–4]	0.04
Reported hours slept per nightd	6.6 ± 1.3	7.2 ± 2.5	0.12

Abbreviation: PSQI, Pittsburgh Sleep Quality Index.

^aData are reported as n (%), unless otherwise indicated.

^bMean systolic blood pressure (BP) ≥130 mm Hg or mean diastolic BP ≥ 80 mm Hg over three measurements at rest, or requiring antihypertensive medication.

^cp-Values indicate the statistical testing result from two-sample t-tests for normally distributed continuous variables, Mann–Whitney U tests for non-normally distributed continuous variables, or Pearson’s chi-square for categorical variables.

^dData are reported as mean ± standard deviation.

^eData are reported as median [interquartile range].

Select individual questions from the 1-year postpartum assessment were analyzed given that they assessed perceived subjective sleep quality. An individual component of the global PSQI asked participants “During the past month, how would you rate your sleep quality overall?,” with 1 indicating very good sleep quality and 4 indicating very bad sleep quality. Individuals with persistent postpartum HTN reported worse sleep quality overall compared with those who were normotensive at 1 year postpartum (median 2.5 [IQR 2–3]) vs. 2 [IQR 2–3], p = 0.05). A separate question asked participants to rate how satisfied they were with their sleep, with 1 indicating they were very dissatisfied and 5 indicating they were very satisfied. Individuals with persistent postpartum HTN also reported worse sleep satisfaction compared with those who were normotensive at 1 year postpartum (median 2 [IQR 2–4] vs. 3 [IQR 2–4], p = 0.04). Another individual component of the global PSQI asked participants to self-report their sleep duration overnight. Both groups reported a similar number of hours slept per night (6.6 ± 1.3 h for individuals who were normotensive at 1 year postpartum vs. 7.2 ± 2.5 h for those with persistent postpartum HTN; p = 0.12). Given that sleep-disordered breathing (SDB) can be a cause of poor sleep, we also analyzed four individual questions from the PSQI pertaining to symptoms of SDB (see ►Supplementary Table S1, available in the online version).

The findings for our primary outcome of global PSQI scores were unchanged in a multivariable logistic regression model adjusting for age, self-reported race, prepregnancy BMI, and parity. In our adjusted analysis, every 1-point increase in the global PSQI score was associated with a 1.19-fold (95% confidence interval 1.04–1.37) increased odds of persistent postpartum HTN. We repeated our models with inclusion of intervention group as a covariate for individuals enrolled in HH4NM, but ultimately did not include this in our final model given the nonsignificant p-values.

Given the relationship of BMI and BP, we repeated analysis of our primary outcome stratified by prepregnancy BMI (►Table 3). Given the small sample size in each of the strata, a statistical difference (p = 0.02) was only noted in obese individuals.

Table 3.

Global Pittsburgh Sleep Quality Index scores stratified by prepregnancy body mass index

Sleep outcome	Normotensive at 1 year postpartuma	Hypertensive at 1 year postpartumb	p-Valuec
Prepregnancy BMI (kg/m2)
< 25	n = 15 7.7 ± 3.9	n = 2 9.0 ± 1.4	0.66
25–29.9	n = 31 7.2 ± 3.7	n = 26 8.6 ± 3.9	0.16
≥30	n = 38 8.5 ± 3.4	n = 40 10.4 ± 3.3	0.02

Abbreviation: BMI, body mass index.

^aData are reported as mean ± standard deviation.

^bMean systolic blood pressure (BP) ≥130 mm Hg or mean diastolic BP ≥ 80 mm Hg over three measurements at rest, or requiring antihypertensive medication.

^cp-Values indicate the statistical testing result from two-sample t-tests for continuous variables.

Finally, given the limited understanding of sleep quality in the first year postpartum, we performed an additional sensitivity analysis assessing sleep quality in the 44 individuals from the Pathways cohort with uncomplicated pregnancies. All these individuals were normotensive and not on antihypertensive medications at 1 year postpartum. Within this group, the mean global PSQI score was 8.3 ± 3.7, and 73% (32/44) reported poor sleep quality (global PSQI > 5; see ►Supplementary Tables S2 and S3, available in the online version).

Discussion

In this combined cohort of 154 nonpregnant individuals with a history of an HDP, the majority (81%) reported poor subjective sleep quality at 1 year postpartum, and individuals with persistent postpartum HTN reported worse subjective sleep quality compared with normotensive individuals. Individuals with persistent postpartum HTN at 1 year had significantly higher PSQI scores (indicating poorer sleep quality), and worse self-reported sleep quality and sleep satisfaction when compared with those who were normotensive at 1 year postpartum, despite reporting a similar number of hours slept per night.

Outside of pregnancy, women are more commonly affected by poor sleep. There may be sex-specific effects of sleep deprivation, with women who experience shortened sleep at increased risk of HTN compared with men.^20,21 During pregnancy, both mean maternal sleep duration and sleep quality have been shown to decrease with increasing gestational age.^22,23 In addition, disrupted sleep and/or SDB in pregnancy have been associated with increased risk for adverse pregnancy outcomes including preterm birth,²⁴ gestational diabetes mellitus,²⁵ and preeclampsia.²⁶

In the postpartum period, sleep disturbances are attributed to infant sleep and feeding schedules, along with stressors associated with a new caretaker role.^10–14 There is also evidence to suggest that again, women are differentially affected, with women having worse sleep quality in the postpartum period than men.^13,14 Although poor sleep in the postpartum period has been attributed to worsened daytime function,²⁷ little is known about the effects of poor sleep in the postpartum period on future cardiovascular health.

In our cohort, the finding of poor subjective sleep quality at 1 year postpartum in the majority (81%) of individuals is extremely concerning. Although our primary goal was to assess differences in subjective sleep quality at 1 year postpartum in individuals with an HDP, we separately examined sleep quality among individuals with uncomplicated pregnancies and demonstrated that these individuals also report poor subjective sleep quality at 1 year postpartum, providing further evidence that poor sleep is ubiquitous in the postpartum period, but by itself is likely not the sole mechanism contributing to persistent postpartum HTN.

The results from our study support the existing literature regarding the biological plausibility of the effects of poor sleep on cardiovascular risk and BP for individuals outside of the peripartum period and suggests that this pathophysiology may also occur in the peripartum period. Our findings raise the possibility that sleep disruption in the postpartum period following a hypertensive disorder may contribute to persistently elevated BP and potentially increased cardiovascular risk. The cumulative effects of poor sleep quality over an extended time period are unknown, but raise concern that the chronic effects of vascular inflammation and endothelial damage associated with poor sleep may have lasting impact on an individual’s cardiovascular health, particularly in individuals with an HDP who may be especially susceptible to future cardiovascular disease.

Although our study demonstrates an association between poor sleep quality and persistent postpartum HTN, we have not established causality. Additionally, there are likely unmeasured confounders impacting the relationship between self-reported sleep and persistent postpartum HTN. Poor sleep and development of HDP likely share similar risk factors and a common pathway of vascular inflammation and endothelial damage contributing to increased future cardiovascular risk. Of note, we do not have information about sleep quality prepregnancy or during pregnancy, which may differ across the groups.

Strengths of our study include the prospective inclusion of individuals with a diagnosis of an HDP adjudicated by physicians with expertise in the clinical diagnosis and management of these disorders. Follow-up was conducted by trained research personnel and occurred at a mean 48.7 ± 9.5 weeks’ postpartum, thus providing insight into BP trajectories and subjective sleep assessments past the immediate postpartum period. By combining two cohorts, the study included a diverse patient population enrolled from two sites, thus increasing generalizability to a larger population.

There are several limitations to this study. The control group of 44 individuals (see ►Supplementary Tables S2 and S3, available in the online version) from the Pathways cohort who were normotensive and not on antihypertensive medications at 1 year postpartum is a small sample size, thus limiting the ability to draw comparisons or conclusions regarding normative sleep quality in the postpartum period. In addition, we combined two cohorts which, by design, had different inclusion and exclusion criteria for study entry. In particular, overweight or obesity (BMI ≥ 25 kg/m²) was an inclusion criterion for the HH4NM study, and thus our combined cohort has an overrepresentation (89%) of overweight and obese individuals. When stratified by BMI, poorer sleep was seen in those with persistent postpartum HTN across each BMI strata. However, given the small sample size in each of the strata, a statistical difference was only noted in obese individuals. It is, however, possible that the effects of poor sleep may be most pronounced in obese individuals.

Obesity in particular is a known risk factor for SDB. SDB is a cause of poor sleep in many individuals, and has been associated with HTN and metabolic syndrome in pregnant and nonpregnant individuals.^26,28,29 When comparing those who were normotensive at 1 year postpartum and those with persistent postpartum HTN, there were no differences to responses to the four PSQI questions pertaining to symptoms of SDB. This suggests that in our cohort, SDB alone is unlikely to explain differences in BP at 1 year postpartum. However, sleep apnea was not feasibly able to be diagnosed for the participants in the two cohorts included, and thus was not included in the analysis.

In addition, neither extended follow-up of participants’ sleep patterns and assessments, nor longer term follow-up of participants’ cardiovascular health status are available, beyond the first year postpartum. Thus, the sleep characteristics which occur after the first year postpartum and the resultant cardiovascular effects decades later are still unknown. Of note, we relied on the PSQI to assess sleep characteristics including quality and duration. The PSQI has not been validated in the pregnant and/or postpartum population; however, there are actually no validated sleep quality questionnaires for this unique population currently available. Comparing individual components of the PSQI has not been validated particularly in the postpartum population. Other well-known subjective sleep questionnaires including the Epworth Sleepiness Scale³⁰ and the PROMIS Sleep Disturbance questionnaire³¹ have also not been validated in this population. The highly fragmented sleep which is characteristic of the postpartum period is likely not well-captured in these traditional sleep assessments which assumes a nocturnal-dominant sleeping pattern and does not assess for potential sources of sleep disruption such as infant feeding or caretaking. There certainly is a need for improved characterization and understanding of sleep in the postpartum period and its long-term effects on cardiovascular health. Future studies should focus on confirmation of our findings with objective assessment using actigraphy.

Our study suggests that sleep behavior may be a candidate for future intervention to improve maternal cardiovascular health following an HDP. Given the widespread sleep disturbances which occur in the postpartum period in individuals with an HDP, such an intervention has the potential to make a significant improvement and impact on cardiovascular health.

Key Points.

• After an HDP, poor sleep quality was common at 1 year postpartum.

• Those with persistent postpartum HTN reported worse sleep quality at 1 year postpartum.

• Sleep behavior may be a target for intervention to improve maternal cardiovascular health.