Restless legs syndrome (RLS) is a common neurological disorder characterized by dysaesthesias and a strong urge to move the lower limbs. Several studies reported that RLS is associated with cardiovascular risk factors (CVRF) and vascular diseases (VD) (Walters and Rye, 2009). Recent studies observed that subjects with RLS are more likely to have hypertension, higher body mass index (BMI), myocardial infarction and coronary heart disease (Berger et al., 2004; Gao et al., 2009; Ohayon and Roth, 2002; Ulfberg et al., 2001; Winkelman et al., 2008). A high prevalence of RLS was indicated among diabetic patients (Lopes et al., 2005), while most population-based studies failed to find a significant relationship between RLS and diabetes (Ohayon and Roth, 2002; Schlesinger et al., 2009; Winkelman et al., 2008). Finally, there are reports of higher prevalence of stroke in subjects with RLS than in those without RLS (Walters et al., 2010; Winkelman et al., 2008), and a case series suggested that RLS became manifest shortly after the acute stroke event (Lee et al., 2009).
Despite these reports, the association of RLS with CVRF and VD is far from established, as other studies failed to find associations between these conditions (Berger et al., 2004; Rothdach et al., 2000; Winkelman et al., 2008). A major limitation of most previous studies is their cross-sectional design. In a recent prospective analysis of two large cohorts of health professionals and physicians, baseline RLS was not associated with subsequent risk of major cardiovascular events or death (Winter et al., 2012). However, another novel prospective study suggests that longer-duration RLS is an independent predictor of coronary heart disease among women (Li et al., 2012). The relationship in the reverse direction, i.e. between baseline cardiovascular disease and incident RLS, was not reported, even though the existing data raise the possibility that certain cardiovascular risk factors and/or diseases might be predictors of the development of RLS.
The main focus of the present analysis was to evaluate if the presence of four different CVRF and two VD predicts the incidence of RLS in two independently conducted prospective population-based cohort studies in Germany. Because there are controversial findings with regard to RLS as a risk factor for incident cardiovascular events, we also studied the relationship between baseline RLS and the incidence of CVRF as a secondary objective.
The primary aim of the DHS study was to determine the frequency of headache types, cardiovascular and other chronic diseases, and unhealthy habits in the population of the city of Dortmund in the western part of Germany (Happe et al., 2008). From a total population of 587 607 living in Dortmund in 2003, a random sample of 3820 individuals aged 25–75 years was drawn from the municipal registry, stratified by 5-year age groups and gender. Of those sampled, 395 individuals were excluded because they had moved out of the study area, died or did not have sufficient knowledge of the German language. Thus, 3425 individuals were eligible and invited to an interview at the DHS study centre. If personal participation at the study centre was impossible, a questionnaire with a subset of the otherwise identical questions was mailed to the participants. The overall response rate at baseline was 66.9%, yielding 2291 participants. RLS assessment at baseline was restricted to interview participants only (n = 1312), because the respective questions were not included in the questionnaire due to the reduced space availability. Sufficient data to classify RLS at baseline were provided by 1291 interview participants. Median follow-up time (range) was 2.1 (0.07–2.7) years, during which 11 people died. The follow-up was conducted by mailed questionnaire, and 1122 responses were obtained yielding a follow-up response of 86.2%.
The Study of Health In Pomerania (SHIP) is an ongoing population-based study comprising three cities and 29 communities in the rural area close to the Baltic Sea (West Pomerania). It was designed to assess a broad range of health and quality of life indicators in the north-east region of Germany after the German reunification (John et al., 2001). From the total population of 212 157 residents living in the study area in 1995, a sample of 7008 men and women aged 20–79 years stratified by 5-year age groups was drawn using a multi-stage random sampling design. The final number of subjects participating in the study was 4308 (response rate of 68.8%). The baseline examination was conducted from 1997 to 2001, combining an interview, medical and dental examinations performed in one single visit to the study centre. RLS data at baseline were available from 4274 participants. The follow-up of SHIP was performed after a median (range) of 5.0 (4.3–8.5) years on average. From the baseline 4274 participants, 311 people died. The remaining participants were all contacted, and 3300 subjects participated (follow-up response rate of 83.3%). The follow-up examination was also performed during a personal visit to the study centre. Finally, 3264 participants had both baseline and follow-up RLS data.
All participants gave informed written consent and the study protocol was approved by the local ethics committees of the Medical Faculty at the University of Münster (for DHS) and the University of Greifswald (for SHIP).
Baseline RLS assessment was conducted identically in both studies in face-to-face interviews, with a short set of questions that had been validated previously against physician's classification (Berger et al., 2002), and had already been used in previous reports (Berger et al., 2004; Happe et al., 2008; Rothdach et al., 2000; Szentkiralyi et al., 2011). The following questions were used according to the standard criteria of the International Restless Legs Syndrome Study Group (Walters, 1995): (i) ‘do you have sensory discomfort like tingling, crawling with ants or pain in the legs associated with an urge to move?’; (ii); do these symptoms occur at rest, i.e. while sitting down or falling asleep, and do they improve by moving?’; and (iii) ‘are these symptoms worse in the evening or at night, compared with the morning?’. The three answer categories included ‘yes’, ‘no’ or ‘don't know’. Participants were classified as RLS-positive only if they answered all symptom questions with ‘yes’. At the end of the follow-up period in DHS, i.e. after 2.1 years, the same set of questions was used to assess RLS as part of a mailed questionnaire, and participants also had to specify the number of years elapsed since the onset of symptoms. In SHIP, at the end of the 5-year follow-up, the questions were applied in a face-to-face interview, but only symptoms occurring during the follow-up time had to be reported, i.e. from the previous 5 years. Incident RLS was defined as having no RLS at baseline and being RLS-positive at follow-up. Subjects without RLS at both assessments were classified as having no RLS. Participants evaluated as RLS-positive cases at baseline were excluded when calculating incident RLS.
Sociodemographic data, comorbidities and laboratory measurements
Age, gender and level of education were assessed in interview form by the trained and certified interviewers. In both studies, diabetes mellitus, hypertension, myocardial infarction and stroke were assessed as self-reports, with specific questions asking for a physician diagnosis of the respective condition. The phrasing of the questions and the answer categories were similar in the two studies. The current medication, taken within the last 7 days, was recorded and classified subsequently according to the Anatomic Therapeutic Chemical (ATC) Classification System (ATC-Index, 2006). Those who reported to receive antihypertensive (ATC codes ‘C02’, ‘C08’ and ‘C09’) or antidiabetic (either oral or insulin, ATC code ‘A10’) medication were also classified as having hypertension or diabetes, respectively. Body weight, height and blood pressure were measured according to standard protocols. In SHIP and a subset of DHS participants (n = 1152), non-fasting blood samples were drawn under standardized conditions for later measurement of serum cholesterol, creatinine, glucose and haemoglobin. Glomerular filtration rate (GFR) was calculated from serum creatinine according to the Chronic Kidney Disease Epidemiology Collaboration formula (Levey et al., 2009). Hypercholesterolaemia was defined by a 240 mg dL−1 cutoff total serum cholesterol level and/or taking a lipid reducing drug (ATC code ‘C10’). In the following analyses, hypertension, diabetes, obesity (BMI > 30 kg m2) and hypercholesterolaemia were considered as CVRF, while myocardial infarction and stroke were classified as VD. A vascular comorbidity index was calculated by adding up the number of co-occurring CVRF and VD.
Follow-up assessments were conducted through a mailed questionnaire in DHS and in a face-to-face interview in SHIP. The set of questions assessing new-onset comorbidities and medications were similar to those used at baseline, except that body weight and height were self-reported in DHS. Serum blood parameters were not measured at follow-up.
Continuous variables are described with mean and standard deviations or with median and interquartile range, as appropriate. Means and medians were compared with Student's t-test or Mann–Whitney U-test, respectively. Categorical variables across groups were compared using chi-square test or Fisher's exact test (if a cell number was five or less). Multivariate logistic regression models were built separately for both studies to analyse the independent relationship of each condition with RLS. Subjects who did not participate in the follow-up examination or had missing data were excluded from the multivariate analyses. No attempt was made to pool the data, as mean follow-up time varied in both studies. We adjusted for gender, age and individual follow-up time in each model. Variables under examination were added with forced entry, while other potential confounders were subject to backward stepwise selection using a removal condition, P > 0.2. Trends odds ratios (ORs) for the vascular comorbidity index were calculated by adding the index as a continuous variable to the model.
Sensitivity analyses were performed with the exclusion of the following groups of participants: subjects who reported some RLS symptoms at baseline, but did not qualify for definite RLS, as some of them might have been prevalent cases (DHS: n = 174, SHIP: n = 612); individuals taking antidepressants (DHS: n = 36, SHIP: n = 78); subjects with diabetes (DHS: n = 100, SHIP: n = 342); and subjects reporting leg cramps (data available only in SHIP, n = 380). In DHS those subjects with incident RLS, who reported at follow-up that the onset of RLS preceded the baseline measurement, were excluded in another sensitivity analysis (n = 49) as they might have had ongoing RLS at baseline.
Each multivariate model was also rebuilt within gender strata. All tests were two-sided, the level of statistical significance was at P < 0.05. All analyses were performed with spss version 18.0 (SPSS Inc., Chicago, IL, USA).
Demographics and basic characteristics
Table 1 summarizes the baseline characteristics in both studies. The mean age was 52.1 ± 13.8 years and 50.3 ± 16.4 years, and the proportion of women was 52.9 and 50.9% in DHS and SHIP, respectively.
Table 1. Baseline characteristics of participants in the Dortmund Health Study and in the Study of Health in Pomerania
|Age, years (mean ± SD)||52.1 ± 13.8||50.3 ± 16.4|
|None or primary school||49.9||40.0|
|College, technical college, university||29.7||16.3|
|BMI, kg m2 (mean ± SD)||27.5 ± 5.0||27.3 ± 4.8|
|Obesity (BMI > 30 kg m2), %||26.4||25.6|
|History of myocardial infarction, %||3.7||3.4|
|History of stroke, %||2.2||2.3|
|Systolic blood pressure||141.0 ± 21.7||137.1 ± 21.5|
|Diastolic blood pressure||86.8 ± 12.4||83.7 ± 11.3|
|Glomerular filtration rate, mL min−1 1.73 m2 (mean ± SD)||91.1 ± 16.7||82.1 ± 16.9|
|Haemoglobin, g dL−1 (mean ± SD)||14.5 ± 1.3||13.6 ± 1.3|
|Cholesterol, mg dL−1 (mean ± SD)||219.9 ± 40.9||223.2 ± 48.1|
|Glucose, mg dL−1 [median (IQR)]||88 (16)||95.5 (18)|
|RLS prevalence at baseline, %||7.4||10.1|
Risk factors of RLS in multivariable models
In SHIP, 206 of 2929 individuals subjects reported new onset of RLS during 5.2 years of follow-up, yielding a standardized incidence rate of nine per 1000 person-years [95% confidence interval (CI): 8–10]. In DHS, 85 participants from 935 subjects newly developed RLS during 2.2 years, yielding an incidence rate of 22 per 1000 person-years (95% CI: 18–27). The cumulative incidences of RLS were 9.1 and 7.0% in DHS and SHIP, respectively.
Table 2 displays the ORs for incident RLS in models adjusted for age and gender. Age was associated significantly with incident RLS in both studies, while female gender was related significantly to incident RLS only in SHIP. BMI and a physician's diagnosis of diabetes were associated significantly with incident RLS in both studies. Some other factors were related significantly to incident RLS in only one study: obesity and higher serum glucose level in DHS and hypertension, myocardial infarction and hypercholesterolaemia in SHIP (Table 2).
Table 2. Age- and gender-adjusted relationships between cardiovascular risk factors/vascular diseases at baseline and incident restless legs syndrome (RLS) in the Dortmund Health Study and in the Study of Health in Pomeraniaa
|Female gender (adjusted for age only)||1.39||0.88–2.20||0.16||1.80||1.34–2.41||<0.001|
|Age: +10 years (adjusted for gender only)||1.04||1.02–1.05||0.001||1.02||1.01–1.03||<0.001|
|BMI, +1 kg m2||1.09||1.04–1.14||<0.001||1.04||1.01–1.08||<0.01|
|Obesity (BMI > 30 kg m2)||2.45||1.53–3.90||<0.001||1.28||0.94–1.75||0.12|
|History of myocardial infarction||0.80||0.23–2.73||0.72||2.04||1.04–4.00||0.04|
|History of stroke||0.51||0.07–3.98||0.52||0.75||0.23–2.47||0.64|
|Systolic blood pressure, +1 mmHg||1.01||0.99–1.02||0.35||1.00||0.99–1.01||0.78|
|Diastolic blood pressure, +1 mmHg||1.01||0.99–1.03||0.55||0.99||0.98–1.01||0.41|
|Glomerular filtration rate, −1 ml min−1 1.73 m2||0.99||0.97–1.01||0.43||1.01||1.00–1.02||0.24|
|Haemoglobin, −1 g dL−1||1.14||0.92–1.41||0.24||1.13||0.98–1.30||0.09|
|Cholesterol, +10 mg dL−1||0.95||0.89–1.01||0.12||1.03||1.00–1.06||0.11|
|Glucose, +10 mg dL−1||1.11||1.02–1.20||0.01||1.03||0.99–1.07||0.11|
Table 3 shows the multivariable models adjusted simultaneously for each CVRF/VD. This model was also adjusted for education, glomerular filtration rate, haemoglobin level and subsequently for behavioural factors. In DHS, only obesity was associated significantly with incident RLS. In SHIP diabetes, hypertension and hypercholesterolaemia were related significantly to a higher incidence of RLS. The relationship between myocardial infarction and incident RLS was also nearly significant in SHIP (Table 3). When BMI was applied without a cutoff in the same multivariable models, a 1 kg m2 increment of BMI was associated significantly with incident RLS in both studies (DHS: OR: 1.07, 95% CI: 1.01–1.12, P = 0.02; SHIP: OR: 1.04, 95% CI: 1.01–1.07, P = 0.02). There was a significant positive trend between the vascular comorbidity index defined by the number of CVRF/VD and incident RLS in both studies (Table 4).
Table 3. Cardiovascular risk factors/vascular diseases at baseline as risks for incident restless legs syndrome (RLS) in multivariate logistic regression models in the Dortmund Health Study and in the Study of Health in Pomeraniaa
|Model 1: adjusted for age, gender, education, glomerular filtration rate, haemoglobin|
|Cardiovascular risk factors|
|Obesity (BMI > 30 kg m2)||2.27||1.37–3.78||<0.01||1.23||0.88–1.71||0.22|
|History of myocardial infarction||0.69||0.20–2.41||0.56||1.77||0.87–3.61||0.12|
|History of stroke||0.47||0.06–3.74||0.47||0.58||0.17–1.99||0.38|
|Model 2: adjusted as model 1 plus alcohol consumption, smoking, physical activity|
|Cardiovascular risk factors|
|Obesity (BMI > 30 kg m2)||2.06||1.22–3.47||<0.01||1.27||0.91–1.77||0.17|
|History of myocardial infarction||0.64||0.18–2.28||0.49||1.84||0.90–3.76||0.1|
|History of stroke||0.41||0.05–3.35||0.41||0.55||0.16–1.91||0.35|
Table 4. The association between the vascular comorbidity index at baseline and incident restless legs syndrome (RLS) in the Dortmund Health Study and in the Study of Health in Pomeraniaa
|Model 1: adjusted for age, gender, education, glomerular filtration rate, haemoglobin|
|Vascular comorbidity index|
|0||1.0||Reference|| ||1.0||Reference|| |
|3 or more||3.50||1.36–9.03||<0.01||2.61||1.61–4.25||0.001|
|Model 2: adjusted as model 1 plus alcohol consumption, smoking, physical activity|
|Vascular comorbidity index|
|0||1.0||Reference|| ||1.0||Reference|| |
|3 or more||3.03||1.16–7.90||0.02||2.77||1.69–4.52||<0.001|
Next, RLS status at baseline was evaluated as a risk for each CVRF and VD in five separate multivariable models (Table 5). As we had no data about cholesterol levels at follow-up, hypercholesterolaemia was omitted from these analyses. The presence of RLS at baseline was not associated significantly with the incidence of any CVRF or VD.
Table 5. Restless legs syndrome (RLS) at baseline and the risk of incident cardiovascular risk factors/vascular diseases in the Dortmund Health Study and in the Study of Health in Pomerania*
|Dortmund Health Study (2.1-year follow-up)|
|No. of incident cases||17||31||87||11||12|
|Study of Health in Pomerania (5-year follow-up)|
|No. of incident cases||254||133||347||37||35|
In both studies, baseline diabetes showed a stronger relationship with incident RLS in younger participants (≤50 years) than in elderly subjects, and the interaction between age group and diabetes was significant (DHS: P = 0.004; SHIP: P = 0.03). None of the other interactions of each explanatory variable with either gender or age strata reached statistical significance in any of the multivariate models.
Sensitivity analyses were conducted excluding potential prevalent RLS cases at baseline, those taking antidepressants and participants reporting either diabetes or leg cramps. These analyses provided results consistent with those presented above (data not shown). In a further analysis we excluded subjects with incident RLS reporting at follow-up that the onset of symptoms preceded the baseline measurement (in DHS only). In this subanalysis, diabetes (OR: 3.86, 95% CI: 1.41–10.59, P < 0.01), obesity (OR: 3.15, 95% CI: 1.46–6.79, P < 0.01) and hypercholesterolaemia (OR: 2.68, 95% CI: 1.25–5.78, P = 0.01) were also independent significant risk factors for incident RLS.
Because blood pressure and serum glucose values were also available at baseline, we reanalysed the multivariate models with an extended definition of hypertension and diabetes; in these analyses a systolic blood pressure >140 mmHg or a diastolic blood pressure >90 mmHg was also defined as hypertension. The criterion of a non-fasting serum glucose cutoff at 200 mg dL−1 was included for baseline diabetes. These subanalyses also yielded very similar results to those presented previously (data not shown).
We report results from an analysis of the time sequence in the onset of several CVRF/VD and RLS, using two prospective cohort studies in Germany and incident case assessment. We found that the vascular comorbidity index was related strongly to incident RLS, following a dose–response pattern. Obesity was an independent risk factor for incident RLS in DHS, and higher BMI was an independent predictor of incident RLS in both studies. Diabetes, hypertension and hypercholesterolaemia were associated with higher incidence of RLS only in SHIP. In the analysis with reversed sequential order, RLS at baseline was not associated with increased risk of the incidence of CVRF or VD in any of the studies.
To date, mainly cross-sectional and case–control studies have been published concerning the relationship of RLS with CVRF and VD. This difference in study design of the previous literature and the present work might have contributed to discrepancies between our results and previous reports. Therefore, any comparison of our findings with those from the literature is difficult to interpret.
The variability across study populations may also be responsible for the controversial findings. Important parameters including mean blood pressure, haemoglobin and glucose levels or renal function were worse among the participants of SHIP than in DHS. It may be assumed that these different health profiles could have led to differences in the prevalence and incidence of RLS, and risk factors of RLS between the two studies. Different methods (i.e. different follow-up times, follow-up assessment of RLS with mailed questionnaire in DHS as opposed to face-to-face interviews) could also have contributed to the differences between the two studies. In a sensitivity analysis we excluded those participants from DHS who did not report RLS at baseline, but claimed at follow-up to have RLS symptoms preceding the baseline measurement. This subanalysis provided more consistent results with SHIP results; therefore, failing to report RLS symptoms at baseline in DHS could have also contributed to differences between the two studies.
It should also be noted that hypertension, diabetes, hypercholesterolaemia and obesity co-occur frequently and are risk factors for each other. As the presence of these conditions was assessed cross-sectionally at baseline, it was not possible to disentangle and interpret the mutual relationship and complex interaction of these factors and to describe their exact individual role in the pathogenesis of RLS.
The studies regarding the association of RLS with either higher BMI (Gao et al., 2009; Happe et al., 2008; Ohayon and Roth, 2002; Rothdach et al., 2000; Winkelman et al., 2008) or hypertension (Ohayon and Roth, 2002; Schlesinger et al., 2009; Ulfberg et al., 2001; Winkelman et al., 2008) are controversial. One study did not find a significant association between RLS and hypertension, also taking BMI into account (Winkelman et al., 2008), and another study reported a significant relationship independent from obesity (Ohayon and Roth, 2002). Most previous population-based cross-sectional studies did not find a significant relationship between RLS and diabetes (Ohayon and Roth, 2002; Schlesinger et al., 2009; Winkelman et al., 2008). In contrast, case–control studies showed an increased prevalence of RLS among diabetic patients (Lopes et al., 2005).
Some previous cross-sectional studies found an increased frequency of heart diseases in subjects with RLS, although important potential confounders such as diabetes or obesity were often not controlled (Ohayon and Roth, 2002; Ulfberg et al., 2001; Winkelman et al., 2008). A case series and a case–control study reported an association between stroke and RLS (Lee et al., 2009; Walters et al., 2010). Our results did not indicate any relationship between stroke and incident RLS. However, we did not have any information about the type and localization of the stroke, and lesion topography might have a crucial role in poststroke RLS. There is hardly any information about the relationship between cholesterol level and RLS in the literature, although two studies also found a strong relationship between hypercholesterolaemia and RLS (Cosentino et al., 2012; Schlesinger et al., 2009).
According to a recent analysis of two large prospective cohort studies among health-care professionals, RLS was not a significant risk factor of incident cardiovascular events, myocardial infarction and stroke (Winter et al., 2012). Another recent large prospective study indicated that RLS with a duration of at least 3 years predicts the development of coronary heart disease among women (Li et al., 2012). In secondary analyses, we found that baseline RLS was not related significantly to any incident CVRF/VD. We could not investigate further the potential risk associated with either the severity or onset of RLS; thus, more studies are warranted to explain the controversial findings.
Even though we reviewed studies that used the standard minimal criteria for RLS, the wording of questions, the time-frame regarding RLS symptoms as well as the population samples and statistical adjustments used in these studies showed great variability. These differences may explain some of the controversy in the literature and warrant caution in the interpretation of the existing results.
The nature of a potential relationship of RLS with CVRF and VD is currently unclear; however, several plausible biological models have been proposed. Gao et al. (2009) suggested that the dopaminergic dysfunction of the central nervous system may mediate the association between RLS and obesity. The prevalence of iron deficiency, a common risk factor of RLS, is also increased in overweight and obese populations (Nead et al., 2004).
Reduced blood flow due to vascular alterations or sympathetic hyperactivity in the periphery or in the brain are common complications in vascular diseases, and these might be causal factors for RLS and/or periodic leg movements in sleep (PLMS) (Ware et al., 1988). The relief from muscle movement and the original therapeutic use of vasodilators in RLS (Murray, 1967) might indicate the role of tissue hypoxia and consequent peripheral neuronal dysfunction. Genes increasing susceptibility to RLS, such as MEIS1 and LBXCOR1, may also have a role in the pathogenesis of vascular diseases (Walters and Rye, 2009). Alternatively, cardiovascular diseases may trigger RLS mainly in genetically predisposed subjects.
The majority of subjects with RLS also suffer from PLMS (Montplaisir et al., 1997), and PLMS might be related independently to cardiovascular risk and events (Koo et al., 2011; Lindner et al., 2012). The association between PLMS and cardiovascular risk might be mediated by the episodes of elevated sympathetic activity and increased blood pressure accompanying the leg movements (Pennestri et al., 2007), but this hypothesis remains to be tested.
Among the strengths of our investigation is the use of two independent cohort studies, with participants selected randomly from the general population in different regions of Germany. Both studies were prospective, with large sample sizes encompassing a broad age range and high response rates. Interviews were conducted by trained personnel and RLS cases were classified identically according to the RLS standard minimal criteria (Walters, 1995). We accounted for numerous comorbid conditions and behavioural factors, such as physical activity, which allowed us to examine the independent risk posed by each CVRF or VD. This is of special importance, as different cardiovascular problems tend to occur concurrently.
A limitation is that assessment of RLS was not based on a diagnostic interview and misclassification of RLS cases was possible. Nevertheless, exclusion of participants with conditions mimicking RLS symptoms in sensitivity analyses suggested that our results were somewhat robust. Furthermore, the same set of questions had been validated and used successfully in various cohorts (Berger et al., 2004; Happe et al., 2008; Rothdach et al., 2000), and the prevalence values in both cohorts are comparable to those reported in previous studies (Ohayon and Roth, 2002; Rothdach et al., 2000; Schlesinger et al., 2009; Ulfberg et al., 2001; Winkelman et al., 2008). Nevertheless, the studies were not designed specifically to assess RLS, and we relied on questionnaire data regarding RLS symptoms; therefore, the results should be interpreted cautiously. As in all observational studies, it is possible that some confounding factors remained unmeasured. The follow-up time was only 2 years in DHS. Consequently, the number of cases for incident CVRF/VD was very low in DHS, therefore we may have missed a significant relationship with baseline RLS due to lack of statistical power. One might argue that a longer period of follow-up time is required for the manifestation of RLS-related cardiovascular risk. Despite these limitations, we decided to include these results from both cohorts due to their potential relevance. Data from SHIP and a previous prospective study, using the same method to assess RLS (Winter et al., 2012), do not support either that RLS is a significant cardiovascular risk factor.
We found that cardiovascular comorbidity predicted incident RLS in a dose–response pattern. Among these variables higher BMI was an independent risk factor of RLS in both studies. RLS status at baseline was not a predictor of subsequent development of any CVRF or VD.
Our findings suggest that RLS may be a marker of poor cardiovascular health. Further research focusing on the relationship and biological models between RLS and cardiovascular diseases is warranted.
We are indebted to all participants for their outstanding commitment and cooperation, and to the entire staff of each study for their expert and unfailing assistance. Data collection in the Dortmund Health Study was supported by the German Migraine and Headache Society and by unrestricted grants of equal share from Almirall, Astra Zeneca, Berlin Chemie, Boehringer, Boots Health Care, Glaxo-Smith-Kline, Janssen Cilag, McNeil Pharma, MSD Sharp and Dohme and Pfizer to the University of Münster. SHIP is part of the Community Medicine Research Net of the University of Greifswald (available at http://www.medizin.uni-greifswfald.de/cm) and was funded by grant ZZ9603 from the Federal Ministry of Education and Research, Berlin, and the Ministers of Cultural and Social Affairs of the Federal State of Mecklenburg—West Pomerania, Schwerin.
A.Sz., K.F. and W.H.: none to declare. S.H.: lectures including service on speakers' bureaus supported by Boehringer Ingelheim, Cephalon, Janssen Cilag, Glaxo Smith Kline, Hoffman La Roche, Merck Serono, Pfizer, Sanofi Aventis, UCB Pharma (Schwarz Pharma). K.B.: travel to meetings for the study or other purposes supported by NIH and US Restless Legs Foundation.