The Association between Hypertension and Insomnia: A Bidirectional Meta-Analysis of Prospective Cohort Studies

Background Studies on bidirectional associations between hypertension and insomnia are inconclusive. The purpose of this meta-analysis was to systematically review and summarize the current evidence from epidemiological studies that evaluated this relationship. Materials and Methods PubMed, Embase, China National Knowledge Infrastructure (CNKI), Wan Fang, and VIP databases were searched for studies published up to May 2021. Prospective cohort studies that reported the relationship between hypertension and insomnia in adults were included. Data were extracted or provided by the authors according to the prevalence rate, incidence rate, unadjusted or adjusted odds ratio (OR), and 95% confidence interval (CI). Heterogeneity was assessed by I2 statistics. ORs were pooled by using random-effects models. Results A total of 23 prospective studies were identified. Twenty cohort studies recorded OR-adjusted value with the outcome for hypertension (OR = 1.11, 95% CI: 1.07–1.16; I2 = 83.9%), and three cohort studies reported OR-adjusted value with the outcome for insomnia (OR = 1.20, 95%CI: 1.08–1.32; I2 = 35.1%). Subgroup analysis showed that early morning awakening and composite insomnia were significantly associated with hypertension. Conclusions The result indicates a possible bidirectional association between hypertension and insomnia. Early identification and prevention of insomnia in hypertension patients are needed, and vice versa.


Introduction
Hypertension afects 26.4% of people worldwide and is considered the main risk factor for mortality [1]. Patients with hypertension commonly complain of insomnia. Hypertension adults have reported an increased risk of insomnia, with a risk ratio of 1.5 to 3.18 [2,3]. Several studies have reported that adults with hypertension have an increased risk of insomnia. Still, patients with hypertension also sufer from psychological diseases such as anxiety and depression [4][5][6], which are risk factors for insomnia.
However, there is no systemic evidence available to support this relationship.
Insomnia is associated with a variety of mental and physical health problems. In addition, abnormal sleepers may also sufer from cardiovascular disease [12]. Hernandez-Aceituno et al. found that increased use of antihypertensive medications was signifcantly associated with poor sleep status [13]. Terefore, treating insomnia and ameliorating sleep habits may be crucial to control some chronic diseases [8].
Hypertension and insomnia are major public health issues, and investigations into the association between these diseases have recently attracted broad attention [14]. Li et al. performed a meta-analysis to assess the pooled relative risk (RR) of insomnia on hypertension. Te fndings suggested that the ultimate RR value was 1.21 (1.10 to 1.33) [15]. However, in various epidemiological studies, this association remains inconsistent [16][17][18][19], and comprehensive reviews that focus on the bidirectional association between insomnia and hypertension are lacking [20]. Terefore, we conducted a bidirectional systemic review and meta-analysis to determine the association between insomnia and hypertension.

Materials and Methods
Tis meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Two authors (HK and LDW) independently evaluated eligibility, extracted data, and scored the quality of the study included. Disagreements were settled by a discussion until consensus was reached or determined by a third author (KJ).

Search Strategy.
We searched PubMed, Embase, CNKI, Wan Fang, and VIP (up to May 2021). To minimize bias, two authors (HK and LDW) independently performed an online search using the following combination of search terms: "hypertension," "high blood pressure," "disorders of initiating and maintaining sleep," "sleep disturbance," "sleep disorder," "sleep quality," "insomnia," "agrypnia," and "sleep maintenance," to identify published studies evaluating the association between hypertension and insomnia. Additionally, a search of the reference lists of eligible articles was conducted to determine any missed reports.

Inclusion and Exclusion
Criteria. Studies were selected based on the following inclusion criteria: (1) the study design was prospective; (2) participants aged 18 years or older; (3) insomnia diagnosed through any symptoms (DIS, DFA, SCD, NRS, DMS, and EMA) or diagnostic criteria (e.g., DSM-IV/V, ICSD-1/2/3, and ICD-9/10); (4) hypertension diagnosed was based on a current resting systolic blood pressure (SBP) ≥ 140 mmHg and/or diastolic blood pressure (DBP) ≥ 90 mmHg, by self-reported hypertension, or by antihypertensive treatment; (5) included an OR value and a 95% confdence interval or other sufcient results; (6) published in English and Chinese. Te exclusion criteria included the following: (1) studies with special populations (e.g., child and pregnancy); (2) letters, comments, reviews, or meta-analyses; and (3) the full text was not available. Te methodological quality for  the included studies was assessed based on the Newcastle-Ottawa Scale (NOS), including the quality of study selection (0-4 points), comparability (0-2 points), and exposure and outcome of study participants. A fnal score ≥7 is considered a high-quality article [21].

Data Extraction.
Two researchers (HK and LDW) independently extracted the following information from each study: basic information (author, publication time, nationality, source of literature, number of studies, age, and gender), criteria for evaluation of hypertension and insomnia, number of patients and participants, crude, adjusted OR and confdence intervals, and the variables used in multivariate analyses. A maximum level adjustment was selected if adjusted ORs were shown in diferent adjustment levels.

Statistical Analysis.
Te association between hypertension and insomnia was assessed from the following perspectives: (1) the OR of baseline insomnia and risk of incident hypertension in prospective cohort studies; (2) the OR of baseline hypertension and risk of future insomnia in prospective cohort studies. If the studies reported efect size other than OR, the transformation was performed, and unpublished data were collected by contacting the corresponding author if possible. A random-efects model was used to pool the data, and statistical heterogeneity between summary data was evaluated using the I2 statistic. 25%, 50%, and 75% represent low, moderate, and high heterogeneity [22]. Forest plots were used for the graphical display of the results. Funnel plot, Begg's test, and Egger's test were used to assess publication bias. Visual asymmetry in funnel plot or P ≤ 0.05 in Begg's and Egger's tests was considered statistically signifcant. Subgroup analyses were performed to illustrate the infuence of study results' specifc characteristics, including age, sex, insomnia type, continent, hypertension assessment, insomnia assessment, and follow-up time. Based on hypertension assessment, studies were divided into different types: SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg or use of antihypertensive medication and others (self-report or diferent levels of BP or ICD9/10). For insomnia assessment, studies using sleep questionnaires were defned as "nonclinical insomnia criteria," and others such as DSM-IV and ICSD-I were defned as "clinical diagnostic criteria." Sensitivity analyses were conducted to detect the stability of our results by excluding each included study one at a time. Stata version 16.0 (StataCorp, College Station, TX) was used for all statistical analyses. P values were two-sided, and a signifcance level cutof of 0.05 was used.

Study Selection and Characteristics.
Te literature search yielded a total of 14,738 articles. After removing the duplicate articles, 10,409 articles remained. After reviewing the titles and abstracts, articles were excluded for their irrelevance. Te remaining 157 articles were identifed through full-text screening. One hundred thirty-four articles did not meet the inclusion criteria, and 23 were included. After quality assessment, most studies have shown good quality with scores ranging from 6 to 8.95. 65% (22/23) of the studies' score was ≥6, and 43.48% (10/23) of the studies' score was ≥8. However, three studies did not satisfy the criterion "Demonstration that outcome of interest was not present at start of study." Two studies [23,24] did not meet the "Demonstration that outcome of interest was not present at the start of the study" criterion in insomnia predicting incident hypertension. One study [25] did not meet the criterion in hypertension predicting incident insomnia (Supplementary Table 1). Te fnal meta-analysis ( Figure 1) included 23 cohort studies (Table 1).

Cohort Studies of Baseline Insomnia Predicting the Risk of
Hypertension. Te association between samples with insomnia at baseline and incident hypertension was investigated in twenty studies, with a total of 242,415 participants. Table 1 summarizes the basic characteristics of these studies. Of the 20 studies, hypertension was identifed by measured blood pressure, self-reported hypertension, or antihypertensive treatment. Sleep questionnaire was used in four studies for diagnosing insomnia; one study used the Women's Health Initiative Insomnia Rating Scale (WHIIRS), and four studies used DSM-IV, ICSD-1, and ICD-9/10 instead. Seventeen studies were conducted in North America or Europe and three in Asia. Te follow-up ranges from 1 to 20 years.
Te result was OR = 1.11 (95% CI: 1.07-1.16) with high heterogeneity (I 2 = 83.9%, P < 0.001) detected ( Figure 2). Publication bias was found in the funnel plot (Supplementary Figure 1(a)) and confrmed by Egger's test (P � 0.01) but not in Begg's test (P � 0.347). We further performed subgroup analyses ( Table 2). Te association between insomnia and hypertension were signifcant in the age subgroups ( Figure 2(a)). We also drew a forest plot after excluding two studies [23,24] that did not satisfy the "Demonstration that outcome of interest was not present at the start of the study" criterion (Supplementary Figure 3(a)). Te fgure shows that excluding the two studies does not infuence the main results. We have also plotted the efect size against follow-up time to confrm there is no link between the two (Supplementary Figure 4(a)).  (Figure 3). However, the pooled OR and 95% CI were based on only three studies, which will afect the results. We drew a funnel plot (Supplementary Figure 1(b)

86.9
Others (self-report or diferent levels of BP or ICD9/10) International Journal of Hypertension Figure S2(b)). We also drew a forest plot after excluding one study [25] that did not satisfy the "Demonstration that outcome of interest was not present at the start of the study" criterion (Supplementary Figure S3(b)). Te fgure shows that excluding the study does not infuence the main results.
We have also plotted the efect size against follow-up time to confrm there is no link between the two (Supplementary Figure 4(b)).

Discussion
It is the frst meta-analysis to investigate the bidirectional association between insomnia and hypertension as far as we know. Tis meta-analysis indicated a likely bidirectional association between insomnia and hypertension in the prospective cohort studies. Our results suggested that insomnia and hypertension are signifcantly related. We collected adjusted ORs to test their association, and we found that the OR of insomnia predicting hypertension was 1.11 (95%CI: 1.07-1.16), and the OR of hypertension predicting insomnia risk was 1.20 (95% CI: 1.08-1.32). When stratifed by insomnia assessment, we found an association only between insomnia diagnosed by non-clinical criteria and hypertension, not with insomnia diagnosed by clinical criteria. Studies that used non-clinical criteria, such as sleep questionnaires, may have a higher sensitivity to detect missed sleep issues when collecting only doctor-reported medical diagnoses. In addition, composite insomnia and early morning awakening were found to be signifcantly associated with hypertension, consistent with the previous meta-analysis [42]. It is also worth noting that we found a more substantial efect in using SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg or antihypertensive medication group (OR � 1.21, 95%CI: 1.10-1.33) compared with the other group (self-report or diferent levels of BP or ICD9/10) (OR � 1.04, 95%CI: 1.01-1.08). Te reason may be that some patients tended not to report hypertension without knowing the diagnostic criteria, so patients with hypertension can be classifed as not hypertensive [43]. Using diferent levels of BP may decrease the number of people diagnosed with hypertension. As for gender diference, men with insomnia were more likely to sufer from hypertension than women [38].
Although our study confrmed that insomnia was associated with an increase in hypertension occurrence, the mechanism behind it was not fully elucidated. Generally, insomnia afects blood pressure through 3 pathways. (1) Psychogenic pathways: insomnia leads to mental changes, mainly manifested as anxiety [42], depression [38], and so on. Te sympathetic nervous system becomes overactive leading to peripheral vasoconstriction and blood pressure increase. (2) Neurogenic pathway: it was found that the activity of the sympathetic nervous system (SNS) increased in insomnia patients, which would lead to a series of hypertension events [44,45]. (3) Humoral pathway: insomnia has been proven to increase the release of pulsatile cortisol by afecting its rhythm [46]. In addition, insomnia causes stress dysregulation [47], which is a potential cause of high hypothalamic-pituitary-adrenal (HPA) reactivity [48]. Te renin-angiotensin-aldosterone system (RAAS) was also activated along with the HPA axis [49]. In addition, insomnia is a pathological state accompanied by infammation, oxidative stress, and endothelial dysfunction [47], which may be the potential mechanism of hypertension [50]. At the same time, the melatonin secretion of patients with hypertension could be disturbed [47], circadian rhythm would change, and sleep disorders would occur [8,47]. Indeed, these mechanisms may interact with subsequent pathological conditions. Besides, nocturnal hypertension may also be a likely link between insomnia and increased blood pressure, which may be triggered by specifc triggers (OSA episode, arousal, rapid-eye-movement sleep, and nocturia). However, few studies investigate the biological plausibility between hypertension and insomnia. Te exact mechanism between hypertension and insomnia still needs further elucidation.
Tis meta-analysis has several strengths and limitations to address. Te primary strength is that this is the frst meta-  International Journal of Hypertension analysis that comprehensively examines the bidirectional association of insomnia and hypertension based on a comprehensive literature search of studies. Our meta-analysis provided more reliable results in cohort studies with a larger number of studies than the previous meta-analysis. In addition, our study quantifes the bidirectional association in detail, which was stratifed according to factors such as gender, continents, hypertension assessment, and diferent types of insomnia. Insomnia assessment based on formal criteria was frst considered in the current studies, which helps to shed new light on the exact efect of diferent criteria. However, there are still some limitations in this study. First, the heterogeneity is high in some analyses of insomnia and hypertension. By subgroup analysis, diferent types of hypertension assessment and insomnia subtypes might be the source of the heterogeneity. Secondly, some studies failed to control potential confounders such as psychological symptoms, age, or gender, although most studies adjusted several factors. Tirdly, publication bias of baseline insomnia and risk of hypertension was found in the funnel plot and identifed by Egger's test (P � 0.01). Tree studies have diferent hypertension assessments than most included studies, which may be why they are outside the confdence interval of the funnel plot. Fourth, our analyses did not consider the efect of sleep time on hypertension because insomnia patients often have abnormal sleep time, which could bias our research results. Finally, the results of baseline hypertension predicting the risk of insomnia were based on only three studies, which will afect the stability of the meta-analysis. So, there is an increasing demand for high-quality research in the future.

Conclusions
In summary, our study shows that there may be a statistically signifcant bidirectional association between hypertension and insomnia. Early morning awakening and composite insomnia are potential risk factors for hypertension, while baseline hypertension also serves as a risk factor for insomnia. An assessment of insomnia may be benefcial for patients with hypertension, and treatment for hypertension may include improving sleep quality in those patients who show signifcant clinical symptoms of insomnia.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon reasonable request.

Ethical Approval
Tis article does not contain any studies with human participants performed by any of the authors.

Conflicts of Interest
Te authors declare that they have no conficts of interest.