Restless Legs Syndrome and Hypertension in Mexican Women

Abstract

Background

RLS is a common chronic disorder characterized by an irresistible need to move the lower limbs that affects sleep. Poor sleep has been associated with increased blood pressure (BP). Thus, we evaluated the cross‐sectional relationship between RLS and hypertension (HTN) in a large cohort study in Mexico.

Methods

In 2011, 54,925 female participants from the Mexican Teachers’ Cohort responded to a four‐item questionnaire based on the International Restless Legs Syndrome Study Group's minimal diagnostic criteria. Women also reported diagnosis and treatment of HTN. We used multivariable logistic regression models to estimate prevalence odds ratios (ORs) for HTN, adjusting for lifestyle and dietary factors. We also estimated adjusted prevalence ORs for HTN by frequency of RLS symptoms.

Results

We identified 9,230 cases (17%) of RLS, and the prevalence of HTN was 13.1% among women with RLS and 9.4% among those without RLS. The multivariable‐adjusted prevalence OR for HTN comparing women with to those without RLS was 1.18 (95% confidence interval [CI]: 1.10–1.26). Compared to those without RLS, the prevalence OR of HTN in women reporting a symptom frequency of once a month or less was 1.14 (95% CI: 1.00–1.30); among those with symptoms two to four times a month, the OR was 1.17 (95% CI: 1.05–1.30); and for those with symptoms at least two times a week, the OR was 1.22 (95% CI: 1.10–1.35).

Conclusion

We observed an association between RLS and HTN. Future studies should evaluate the impact of treating RLS on BP.

Restless legs syndrome (RLS) is a common chronic sensorimotor disorder, characterized by unpleasant sensations in the lower limbs and an irresistible urge to move them.1 Symptoms improve with movement and worsen at night, affecting sleep,1 quality of life,2 and productivity.3 The prevalence of this condition varies between 4% and 29% and is more common at older ages and among women.4 In our previous descriptive analysis, we reported a prevalence of this disorder in Mexico similar to that observed in other Western countries.5

More than 70% of people with RLS report sleep disturbances,3 and sleep quality is associated with hypertension (HTN)6 and cardiovascular (CV) disease (CVD).7 RLS has been associated with a high‐risk CV profile,8, 9 and a cross‐sectional study of 65,544 women found a 20% higher risk of HTN comparing individuals with RLS to those without RLS.10 However, the association of RLS with CVD outcomes is less clear. A large prospective study of U.S. women did not find a significant association with incidence of coronary artery disease,11 and two prospective studies in men and women did not observe an association between RLS and major CV events.12

Understanding the relationship of RLS and HTN could help clarify the relationship of this common disorder with CVD and may be useful to guide future research efforts. Therefore, we evaluated the cross‐sectional association of RLS and HTN in a large, prospective study of Mexican women.

Methods

Study Population

The Mexican Teachers’ Cohort or MTC (Estudio de Salud de las Maestras, ESMaestras) is a prospective cohort study that started in 2006 when 27,992 Mexican women within the public education system from the states of Jalisco and Veracruz answered a questionnaire on demographic and reproductive characteristics, diet, lifestyle, therapeutic drug use, and diagnoses of various medical conditions.13 In 2008, the study was expanded to 10 additional states to include a total of 115,343 participants. Questionnaires were delivered and collected using established communication channels through educational authorities. In the first follow‐up questionnaire in December 2011, information on risk factors for chronic disease and medical conditions was updated. This questionnaire included detailed questions regarding symptoms consistent with RLS (described below). The current analysis is based on teachers who had responded to the first follow‐up questionnaire by the time we performed this analysis (N = 69,830). After excluding teachers with invalid answers to the RLS questions, a total of 54,925 were included in this analysis. All participants signed an informed consent and the study was approved by the Bioethics, Biosafety, and Research Commissions at the National Institute of Public Health (Mexico).

Assessment of RLS

The 2011 questionnaire included four items that include the minimal diagnostic criteria for RLS as defined by the International Restless Legs Syndrome Study Group (IRLSSG).14 Participants responded to the following questions: “During the last year, did you have unpleasant leg sensations (tingling, crawling, numbness, or pain) and at the same time an urge or need to move them?”, “Do these symptoms occur only while sitting or resting?”, “Does moving improve the symptoms?”, and “Are these symptoms worse in the evening or at night?.” Individuals were defined as having RLS if they responded “yes” to all four items. Participants were also asked: “On average, how often do you have these symptoms?” and were given the following response options: less than once a month; once a month; two or three times per month; once a week; two to four times per week; and five or more times per week. This set of RLS questions has been used in different settings8, 9, 15, 16, 17 and has been shown to have good validity when compared to clinical evaluation.18

Identification of HTN Cases

At baseline and in 2011, participants responded to a question on physician diagnosis of elevated blood pressure (BP), year of diagnosis, and treatment. We defined HTN as self‐reported physician diagnosis of elevated BP and treatment. We assessed the validity of this definition of HTN in a subsample of 101 study participants using a brief structured phone interview. Standardized interviewers asked participants seven questions: (1) Can you confirm that you have been diagnosed with elevated BP or HTN?; (2) What was your age when you were diagnosed?; (3) Who made the diagnosis: a physician, nurse, or another health professional?; (4) On how many occasions have you been told you have HTN?; (5) What was your BP when you were diagnosed with HTN (systolic/diastolic)?; (6) Has a health care professional provided you with lifestyle recommendations to control your BP?; and (7) Do you use medications to control your BP? Responses were reviewed by a physician that classified as hypertensive 79% of individuals who, before the structured phone interview, reported to be under treatment for elevated BP.

Other Variables

We obtained information on age, family history of myocardial infarction (MI), hormonal contraceptive use, menopause and hormone replacement therapy, body mass index (BMI; weight in kilograms divided by height in meters squared), physical activity, smoking, and the diagnosis of other diseases from the 2011 questionnaire. Usual weekly physical activity was assessed using a seven‐item questionnaire on time spent in common activities in the previous year and transformed into metabolic equivalents (METs; mL/kg per min). For dietary variables and alcohol consumption, we used a previously validated semiquantitative 140‐item food frequency questionnaire19 to which participants responded in 2008. Alcohol intake was evaluated using nine items that included wine, beer, brandy, whisky, tequila, rum, mezcal, and two distilled traditional drinks. For each food and drink item, women were asked to specify how often, on average, they consumed a specified commonly used unit or portion size of the food/beverage over the previous year. We calculated total energy intakes by multiplying the nutrient content of specified portion sizes by the frequency of consumption using the U.S. Department of Agriculture food composition database20 supplemented with a database used in the National Health and Nutrition Survey in Mexico. The validity of our dietary assessment questionnaire has been previously described among women residing in Mexico City.19 We evaluated the validity of self‐reported BMI in a subset of 3,413 study participants and we found a high correlation, as compared to anthropometry performed by standardized technicians (r = 0.89).

Statistical Analyses

We used logistic regression models to estimate the prevalence odds ratio (OR) and 95% confidence intervals (CIs) for treated HTN comparing participants with and without RLS. We calculated prevalence ORs adjusted for age (continuous) and in multivariable models additionally adjusted for family history of MI, hormonal contraceptive use, menopause (premenopause, postmenopause with hormone therapy, and postmenopause without hormone therapy), BMI (<25, 25–29, ≥30 kg/m²), smoking (never, past, or current), physical activity (quartiles in METs/week), migraine (non‐, migraine without aura, and migraine with aura), alcohol intake (no, ≤1 drink per week, 2–4 drinks per week, and ≥1 drinks per day), and consumption of bread, fruits, vegetables (quartiles), and total energy (quartiles). We categorized participants according to symptom frequency based on the distribution of responses: no RLS; RLS symptoms once a month or less; RLS symptoms two to four times per month; and RLS symptoms more than once a week. We included RLS symptom frequency as a categorical variable in the models to test for trend across categories. We evaluated the potential modifying effect of age, menopause, and BMI in analyses stratified by median age (<46 or ≥46 years), menopausal status, and BMI (<25 and ≥25 kg/m²). To test for heterogeneity, we included a cross‐product term of RLS and the two age, menopause, and BMI categories and compared models with and without the cross‐product term with a log‐likelihood ratio test.

We performed the following sensitivity analyses: (1) restriction of the RLS definition to having symptoms at least twice a week; (2) exclusion of cases of HTN diagnosed and treated in or before 2008 (to limit the possibility of long‐standing HTN affecting RLS symptoms); and (3) exclusion of participants with self‐reported medical diagnoses that could be associated to the secondary forms of RLS (diabetes mellitus [DM], rheumatoid arthritis, lupus, lower‐limb venous thrombosis or insufficiency, Parkinson's disease [PD], and multiple sclerosis [MS]) as well as pregnant participants. For all variables used for adjustment, the frequency of missing values was <5%. We handled missing values by including a missing category. All statistical tests were two‐sided (α level = 0.05), and analyses were performed using SAS software (version 9.3; SAS Institute Inc., Cary, NC).

Results

Table 1 shows the characteristics of the 54,925 study participants. Mean age was 44.4 years (standard deviation [SD] ± 7.3) for women without RLS and 46.2 (SD ± 6.7) for women with RLS, and the overall prevalence of RLS was 17%. Women with RLS were older, had a higher BMI (mean BMI: 28.2 vs. 27.4 kg/m²), exercised less, and were more likely to smoke, have a family history of MI, and use hormone therapy, but were less likely to have used hormonal contraceptives.

Table 1.

Characteristics of 54,925 women from the Mexican Teachers’ Cohort according to RLS in 2011

	RLS N = 9,230	No RLS N = 45,695
Characteristicsa
Mean age, years (SD)	46.2 (6.8)	44.4 (7.3)
Family history of MI	21.4	18.1
BMI (kg/m2)
<25	24.5	30.6
25–29	40.4	38.7
≥30	27.6	22.9
Physical activity (METs/week)
<14	27.4	25.6
14–<32	25.8	24.2
32<57	24.0	24.7
≥57	22.8	25.5
Hormonal contraceptive use (ever)	4.9	6.2
Menopausal status
Premenopausal	70.8	77.9
Postmenopausal
With hormone replacement therapy	4.3	3.3
Without	21.7	16.6
Smoking
Never	70.8	74.3
Past	12.4	10.9
Current	9.8	8.2
Migraine, %
Without aura	5.7	4.5
With aura	19.4	12.2
Diet
Total energy, kcal/dayb	1,729 (1,356, 2,218)	1,710 (1,331, 2,177)
Fruits and vegetables, servings/dayb	5.7 (3.6, 8.6)	5.7 (3.6, 8.7)
Bread, servings/dayb	0.73 (0, 2.5)	0.73 (0, 2.5)
Alcohol, drinks
0	33.3	35.6
1 per week	56.8	54.8
2–4 per week	8.5	8.3
1 per day or more	1.5	1.4

^aIn percentage, unless indicated otherwise.

^bMedian (interquartile range).

Prevalence of treated HTN was 13.1% among women with RLS and 9.4% among those without RLS. Prevalence of treated HTN increased with increasing frequency of RLS symptoms (11.7%, 12.5%, and 14.8% for symptoms once a month or less, two to four times per month, and more than once a week, respectively). Age‐adjusted prevalence OR for treated HTN was 1.31 (95% CI: 1.22–1.41) comparing women with RLS relative to those without RLS (Table 2). This estimate was somewhat attenuated after additional adjustment for family history of MI, hormonal contraceptive use, menopause, BMI, smoking, physical activity, migraine, alcohol consumption, total energy, and consumption of bread, fruits, and vegetables, but remained statistically significant. The multivariable‐adjusted OR comparing women with RLS to those without RLS was 1.18 (95% CI: 1.10–1.26). Relative to those without RLS, the multivariable‐adjusted prevalence odds were 14% higher for individuals with RLS symptoms at least once a month (95% CI: 1–30), 17% higher for those with symptoms two to four times per month (95% CI: 0.5–30.0), and 22% higher for participants reporting RLS symptoms twice a week or more (95% CI: 10–35; P value for trend: <0.0001).

Table 2.

Age‐adjusted and multivariable‐adjusted prevalence ORs (95% CIs) of HTN according to RLS and RLS symptom frequency

	Cases (%)	Age‐adjusted	Multivariable‐adjusteda
No RLS	4,288 (9.4)	1.00	1.00
RLS	1,209 (13.1)	1.31 (1.22, 1.40)	1.18 (1.10, 1.26)
RLS symptomsb
Once a month or less	279 (11.6)	1.20 (1.05, 1.36)	1.14 (1.00, 1.30)
Two to four times per month	434 (12.5)	1.27 (1.14, 1.41)	1.17 (1.05, 1.30)
More than once a week	484 (14.8)	1.44 (1.30, 1.60)	1.22 (1.10, 1.35)
P value for trend		<0.0001	<0.0001

^aAdditionally adjusted for family history of MI (yes, no), hormonal contraceptive use (never, ever), menopausal status (premenopausal, postmenopausal with, postmenopausal without hormone therapy), BMI (<25, 25–30, ≥30 kg/m²), physical activity in METs/week (quartiles), smoking (never, past, current), migraine (no migraine, migraine with no aura, migraine with aura), and consumption of alcohol, bread, fruits, vegetables, and total energy (quartiles).

^bInformation missing in 12 cases.

In models stratified by age, menopause, and BMI, we found a similar relationship between RLS and treated HTN within subgroups (Table 3). The multivariable‐adjusted prevalence OR for treated HTN comparing participants with RLS to those without RLS was 1.16 (95% CI: 1.07–1.26) in women younger than 46 years and 1.28 (95% CI: 1.12–1.46) in those 46 and older (P value for heterogeneity: 0.10). Similarly, we found no statistical evidence that this relationship differed by menopausal status and BMI (P value for heterogeneity: 0.52 and 0.12, respectively). However, we observed a suggestion that the relationship of RLS and HTN may be stronger among normal weight, compared to overweight/obese women.

Table 3.

Multivariable prevalence ORs (95% CIs) of HTN according to RLS stratified by age and menopausal status

	No RLS	RLS	P for Heterogeneity
<46 years	1.00	1.16 (1.07, 1.26)
≥46 years	1.00	1.28 (1.12, 1.46)	0.10
Premenopausal	1.00	1.18 (1.08, 1.29)
Postmenopausal	1.00	1.15 (1.03, 1.29)	0.52
BMI <25 m/kg2	1.00	1.30 (1.11, 1.52)
BMI ≥25 m/kg2	1.00	1.17 (1.08,1.27)	0.12

Models adjusted for age, family history of MI (yes, no), hormonal contraceptive use (never, ever), menopausal status (premenopausal, postmenopausal with, postmenopausal without hormone therapy), BMI (<25, 25–30, ≥30 kg/m²), physical activity in METs/week (quartiles), smoking (never, past, current), migraine (no migraine, migraine without aura, migraine with aura), and consumption of alcohol, bread, fruits, vegetables, and total energy (quartiles).

We conducted several sensitivity analyses and our results remained essentially the same. When we restricted the definition of RLS to participants reporting symptoms at least twice weekly, we found similar prevalence odds in women with RLS relative to those without RLS (OR, 1.18; 95% CI: 1.07–1.32). When we excluded cases of HTN diagnosed in or before 2008, ORs increased (OR, 1.2; 95% CI: 1.08–1.33). Similarly, when we excluded pregnant women or those with DM, rheumatoid arthritis, lupus, lower‐extremity thrombosis or varicose veins, PD, and MS (n = 8,228), conditions that could be associated to the secondary forms of RLS, results were similar (OR, 1.12; 95% CI: 1.00–1.24).

Discussion

In this cross‐sectional analysis of 54,925 Mexican women with a mean age of 45 years, we observed a higher prevalence of HTN among women with RLS, relative to those with no RLS, after adjusting for potential confounding factors. Prevalence of HTN appeared to slightly increase with increasing frequency of RLS symptoms. We found no evidence of effect measure modification by age, menopause, or BMI.

The physiological mechanism that could explain the relationship between RLS, HTN, and other CV risk factors remains unclear; however, sleep disruption might play a central role in this relation. Ninety percent of individuals with RLS frequently have periodic limb movements during sleep (PLMS),21 and excessive movements may disrupt sleep through arousals and sleep fragmentation. PLMS are associated with increases in heart rate and BP during sleep and these may be more pronounced in individuals with RLS.22 In a recent study, PLMS was associated with HTN and BP.23 Experimental studies in animal models show that hypothalamic‐pituitary‐adrenal axis activity is increased with sleep deprivation.24 In older, normotensive adults, lack of sleep significantly augments systolic and diastolic BP.25 Furthermore, RLS is hypothesized to be the result of a medullary dopaminergic dysfunction resulting in sympathetic hyperactivity.21 A dopamine‐agonist–decreased heart rate changes during PLMS in RLS patients.26

Several population‐based studies have evaluated the relationship between RLS and HTN using IRLSSG criteria8, 9, 10, 15, 16, 17, 27, 28, 29, 30, 31, 32, 33 and observations are inconsistent. Differences between results may be explained by assessment of RLS and HTN, covariates available for adjustment, and statistical power. In a recent cross‐sectional analysis among 65,544 nurses in the United States that adjusted for several covariates, the magnitude of the association between RLS and HTN was very similar to what we observed (OR, 1.20; 95% CI: 1.10–1.30).10 An important difference between this study and ours is the definition of RLS. In the former, in addition to the IRLSSG criteria,14 the definition was restricted to those who reported symptoms at least five times per month. In contrast, we detected an association with HTN at much lower symptom frequencies, and the last category of RLS frequency in our analysis contains the case definition of that study. In contrast, two large studies in 22,786 men and 30,262 women did not observe a significant association between RLS and HTN. However, these studies used a similar definition of RLS to ours that did not restrict case definition to frequency of RLS symptoms. Two studies with repeated RLS assessment evaluated the relation of HTN and incident RLS,34, 35 assuming that HTN may affect the risk of developing a sensorimotor disorder. The biological mechanism by which this could occur is lacking, and the study that included the largest number of participants did not find an association between HTN and risk of developing RLS, suggesting that RLS affects HTN risk and not vice versa.34

In addition to a large sample size, our study has important strengths. We used a standardized questionnaire to assess the presence of RLS based on the four minimal diagnostic criteria by the IRLSSG14 that has been successfully used in other large‐scale epidemiological studies.8, 9 We had detailed information on the main risk factors of HTN and medical conditions. This allowed us to adjust for all the main risk factors for HTN and conduct sensitivity analyses where individuals with self‐reported conditions predisposing to RLS or that could mimic RLS could be excluded. Nevertheless, there are some limitations to consider. This is a cross‐sectional analysis; thus, we cannot establish the direction of the association between RLS and HTN. Even though we cannot rule out this possibility, our results were not sensitive to the exclusion of HTN diagnosed 3 years before RLS assessment. As with all observational studies, the observed relationship cannot be interpreted as causal without assuming no confounding by unmeasured factors, such as common genetic susceptibility or poorly measured confounders. Measurement error resulting from misclassification of RLS based on self‐reported information is a concern. However, the instrument has been previously used in epidemiological studies and is valid relative to a clinical evaluation.18 One important limitation is our use of self‐reported information to identify cases of HTN and the use of a definition of HTN that required treatment. Thus, we may have misclassified individuals with HTN as nonhypertensive. In our validation study, when we defined HTN as a diagnosis of elevated BP only, independently of the additional report of medical treatment, we only classified 70% as hypertensive using information from the structured interview. In addition, even in a population with health care access underdiagnosis of HTN is frequent; therefore, it is possible that we could have an appreciable number of false negatives. However, the outcome misclassification is probably nondifferential, and, based on our definition that requires treatment, it is very likely that we have few false positives; thus, this misclassification produces limited bias toward the null. We used structured interviews to assess the validity of our definition of HTN; however, we had no access to medical records or BP measurements in this validation study. Finally, the generalizability of the study is limited to middle‐aged women.

Results of this large, cross‐sectional study show that RLS is associated with HTN. Further targeted research is needed to identify whether treatment of RLS reduces the risk of HTN. Our observation indirectly supports poor sleep quality as a potentially important risk factor for HTN.

Author Roles

(1) Research Project: A. Conception, B. Organization, C. Execution; (2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; (3) Manuscript Preparation: A. Writing of the First Draft, B. Review and Critique.

A.C.‐K.: 1B, 1C, 2C, 3B

A.J.‐A.: 2B, 3B

E.O.‐P.: 2A, 2B, 2C

A.M.‐U.: 2B, 3B

K.P.P.: 2A, 2B

C.C.‐B.: 3B

R.L.‐R.: 1A, 1B, 1C, 3B

M.S.R.: 2A, 2B, 3B

T.K.: 1A, 2A, 3B

M.L.: 1A, 1B, 1C, 2A, 3A

Disclosures

Funding Sources and Conflicts of Interest: The authors report no sources of funding and no conflicts of interest.

Financial Disclosures for previous 12 months: The Mexican Teachers’ Cohort was performed with the financial support of the American Institute for Cancer Research (05B047) and Mexico's National Council of Science and Technology (14429). M.S.R. is supported by the National Institutes of Health (NIH; T32 09001). T.K. has received, within the last 2 years, investigator‐initiated research funding from the French National Research Agency and the NIH. Furthermore, he has received honoraria from the BMJ and Cephalalgia for editorial services. M.L. is partly supported by the Bernard Lown Scholars in Cardiovascular Health Program and has a nonrestricted investigator‐initiated grant from AstraZeneca.