Restless legs syndrome (RLS), also known as Willis–Ekbom disease, is a common neurological sensorimotor disorder often associated with severe sleep disturbances and an impaired quality of life (1). The reported RLS prevalence varies between 3.9 and 14.3% depending on the population studied and the criteria used (2). Some but not all reports suggest marked geographic differences in the prevalence of RLS with a lower prevalence in Asian countries (3–5). If the former four minimal (6) or later revised four essential diagnostic criteria are used (7), RLS is reported in 7.2–10.6% of the Caucasian population (8–14). When so-called RLS mimics are excluded, significantly lower prevalence rates of 0.4–2.4% are seen in different samples (15, 16). The proportion of patients in need of treatment is estimated to vary between 1.5 and 2.7% (15). In two prospective studies with observation periods of 2 and 5.2 years, cumulative incidences of 7.0 and 9.1%, respectively, were reported (17). Women are affected up to twice as often as men presumably due to parity; the risk of developing RLS increases with increasing number of births (3, 9, 18). Risk factors for RLS include genetic predisposition, age, gender, parity, and lifestyle factors (9, 19). Secondary or comorbid RLS may occur as a result of another condition such as iron deficiency, renal disease, or pregnancy (20). In addition, secondary RLS can be induced by psychotropic drugs and other medications (21–23). Current hypotheses on the etiology of RLS favor a heterogeneously genetically determined, complex disorder with involvement of abnormal iron metabolism, multiple neurotransmitters, and the opioid system. As revealed by neuroimaging and neurophysiological studies, a subtle receptor dysfunction of the central dopaminergic system seems to play a critical role in the pathophysiology of RLS (24). Clinically, levodopa and dopamine receptor agonists improve RLS, whereas dopamine antagonists such as antipsychotics may evoke or worsen RLS suggesting a decreased dopaminergic state (25). Furthermore, tyrosine hydroxylase, the rate-limiting enzyme in dopamine synthesis, requires iron as a cofactor; a low brain iron state has been repeatedly demonstrated in RLS (26). However, current evidence points to a pre-synaptic hyperdopaminergic state in RLS as revealed by a decreased dopamine transporter activity and a decreased dopamine reuptake leading to a downregulation of postsynaptic D2 dopamine receptors (27). A circadian regulation of dopaminergic activity, i.e., a nadir at night and a peak in the morning, may explain why dopaminergic drugs are effective in the treatment of restless legs symptoms in the presence of a hyperdopaminergic state (28). However, by further downregulating the dopamine receptors, medication at night might paradoxically worsen restless legs symptoms, a phenomenon known as augmentation which is the most important long-term adverse effect of dopaminergic therapy (29).

Although the underlying mechanisms are not fully understood, epidemiological studies suggest that psychiatric disorders and RLS may frequently co-occur, leading to complex interactions between both conditions (25, 30, 31). A bidirectional cause-and-effect relationship is postulated, however, the available evidence is scarce (25). Particularly, comorbid anxiety and depressive disorders are major factors influencing the course of RLS and quality of life of affected patients (25). Previous studies found that treatment outcomes of RLS with comorbid depression are worse compared to RLS without depressive symptoms (32). Furthermore, longitudinal studies have demonstrated an increased risk of depression in RLS patients and, vice versa, an increased likelihood of the occurrence of RLS in depression, suggesting a bidirectional relationship between both disorders (33). Moreover, patients with mental disorders often require treatment with psychotropic drugs that are associated with an increased risk to induce or exacerbate RLS (21–23). However, an increased risk of depression in patients with RLS may also be independent of psychopharmacological effects (34).

Despite the close relationship between RLS and psychiatric disorders, studies on the prevalence of RLS in psychiatric patients are scarce and refer to limited patient samples in Asia (35–37) or otherwise strongly selected psychiatric samples (38, 39). The aim of the present study was to estimate the prevalence and assess the clinical correlates of RLS in a larger sample of psychiatric inpatients in Germany and Switzerland.

Overall, 331 patients met the inclusion criteria and were invited to participate in the study. Of these, 14 patients declined the study participation after being fully informed about the purpose of the study. A total of 52 of the 317 patients met the diagnostic criteria of RLS according to the IRLSSG (1), representing a prevalence of 16.4% (95% CI: [12.3; 20.5]). The diagnoses of the patients were classified according to the following ICD-10 codes: F10-19: Mental and behavioral disorders due to psychoactive substance use; F20-29: Schizophrenia, schizotypal, and delusional disorders; F30-39: Mood [affective] disorders; F40-48: Neurotic, stress-related and somatoform disorders; F60-69: Disorders of adult personality and behavior; F90-98: Behavioral and emotional disorders with onset usually occurring in childhood and adolescence. Further sample characteristics of patients with and without RLS for ordinally and nominally scaled variables are given in Table 1. Among patients with RLS, an RLS diagnosis was previously known in 23.1, and 19.2% were treated with an RLS-specific medication. In 54.8% of all patients with RLS, complaints of symptoms had been present for more than 2 years prior to diagnosis and more than 3 years in 28.6% of patients. In women and in first-degree relatives, RLS prevalence was significantly higher in patients with RLS compared to the non-RLS group. 15.7% of the patients with RLS had a positive family history of RLS. In addition, BMI was significantly higher in patients with RLS, while the frequency of alcohol consumption as well as the number of alcoholic drinks per day was significantly lower. Furthermore, PSQI (sleep quality), ISI (insomnia), and ESS (daytime sleepiness) scores were significantly higher in patients with RLS.

No differences between patients with and without RLS were observed for the distribution of primary and secondary psychiatric diagnoses, somatic and sleep medicine diagnoses, or for current psychopharmacological therapy (except for antipsychotics), smoking, educational status, or PHQ-9-scores (Table 1).

Sample characteristics related to the distribution of continuous variables are reported in Table 2. Although patients with RLS tended to be slightly older than patients without RLS, this difference was not significant. Statistically significant differences in the prevalence of patients with increased daytime sleepiness, impaired sleep quality, and insomnia symptoms are reflected in significant differences between patients with and without RLS in the overall distribution of these scores. No significant differences were observed in age at psychiatric diagnosis, duration of psychiatric illness, smoking habits, or the distribution of PHQ-9 scores.

Overall, there was a wide diversity of primary and secondary psychiatric diagnoses. To assess the association between specific psychiatric diagnoses and RLS, only diagnoses with a cumulative frequency of more than 20 were considered; this resulted in 9 concomitant diagnoses in the data set. The frequencies of the corresponding diagnoses according to ICD-10 are shown in Figure 1.

Treatment with antidepressants, hypnotics, mood stabilizers, or anticonvulsants (when considered at the class level and at the level of specific drugs), as well as with opiates or stimulants were not associated with RLS (Supplementary Table 2). A statistically significant association between pharmacotherapy and RLS was observed only for treatment with antipsychotics (p = 0.0491). RLS patients received antipsychotics more often than patients without RLS. When atypical and typical antipsychotics were analyzed separately, the effect was restricted to atypical antipsychotics (p = 0.0317; Supplementary Table 2).

Restless legs syndrome severity according to the IRLS was reported between moderate and severe (mean 20.3; median 22). A comorbidity or secondary RLS was affirmed 21 times, allowing multiple answers. Thyroid disorders were the most frequent comorbidity (15.4%, n = 8) followed by drug-induced RLS (13.5%, n = 7). In addition, drug-induced worsening of RLS was reported in 17.7% (n = 9) of cases. Further details are reported in Tables 3, 4.

To identify significant predictors for the occurrence of RLS, all variables that showed significant differences in distribution between patients with and without RLS at the univariate level were considered for inclusion in a logistic regression model. After exclusion of highly correlated variables, logistic regression analysis was performed and included age, gender, RLS family history, BMI, ESS, and PSQI scores, frequency of alcohol consumption, use of antidepressants, and use of antipsychotics as independent factors (Table 5). Of these, age, gender, positive family history, and ESS score were statistically associated with higher risk of RLS. Psychiatric patients with first-degree relatives who have RLS had the highest RLS risk (OR 3.29, 95% CI [1.1; 9.7]). Women had a 2.7-fold higher risk (95% CI: [1.25; 5.72]) compared to men. Increased daytime sleepiness was also associated with the occurrence of RLS (OR 1.09, 95% CI: [1.01; 1.17]). On the other hand, a higher frequency of alcohol consumption was associated with a lower risk of RLS (OR 0.45, 95% CI: [0.22; 0.94]). Age, BMI, PSQI score, and the use of antidepressant or antipsychotic medications were not significantly associated with an increased RLS risk (Figure 2). A receiver operating characteristic (ROC) curve resulted in an area under the curve of 0.7747, which corresponds to an acceptable discrimination between the RLS yes/no conditions (Figure 3).

In the present multicenter study, an RLS prevalence of 16.4% was observed in a sample of 317 psychiatric inpatients, which is markedly higher than the RLS frequency in the general population, which is estimated to be 5–10% (2, 12, 51). According to the IRLS, RLS in our patients was moderate to severe. A prior study in a smaller sample of 117 hospitalized psychiatric patients reported an RLS prevalence of 19.7% (39). In contrast to the present study, only patients with comorbid severe sleep disturbances such as insomnia, dyssomnia or daytime sleepiness were included in this retrospective analysis, precluding direct comparison with the present results. In a Spanish cross-sectional study of a sample of 100 non-demented psychogeriatric outpatients (mean age 76.9 ± 6 years), a prevalence of 11.1% was shown for definite RLS with an additional prevalence of 10.1% for possible RLS. Because of the specific age range of the included patients (38), there is no direct comparability with the present study sample (mean age 42.2 ± 15.3), as RLS prevalence is age-dependent (2). Studies in Lebanon revealed RLS prevalence rates of 14.3 (37) and 18.0% (35) in a sample of 203 and 126 hospitalized psychiatric patients, respectively. Both studies used the IRLS not only as a tool to determine the severity of RLS, but also as a diagnostic screening instrument. However, the scale is only validated as an instrument for determining severity and not for diagnosing RLS (41), clearly limiting the results of these studies. A more recent study in Singapore yielded a prevalence of RLS/PLMS (periodic leg movements during sleep) of 14.8% in a sample of 400 psychiatric outpatients with schizophrenia, mood disorders, or anxiety disorders, based on ICD-9 diagnostic criteria (36). In this study, two out of four positive responses on RLS-related questions were considered sufficient for an RLS diagnosis. In addition, the criteria according to the IRLSSG were not applied. Comparability of these results with the present study is highly limited due to methodological differences and limitations. In addition, RLS prevalences have been shown to be much lower in Asia (0.9–7.2%; overview in Ref. (3) than that determined by Hombali et al. (36).

Among the psychiatric inpatients with RLS included the present study, a high proportion (76.9%) were previously undiagnosed, although for 54.8% of the patients, symptoms had been present for more than 2 years. It has been repeatedly demonstrated that RLS is often underdiagnosed and untreated (52, 53) because symptoms are primarily subjective and are based almost exclusively on patient statements. Affected persons may have difficulty adequately describing the unpleasant sensations as well as the urge to move. Often, patients with RLS do not complain of restless legs symptoms, but of insomnia or fatigue, leading the symptoms of RLS to be falsely attributed to a mental condition (54, 55). Specifically in a psychiatric context, the differentiation of RLS complaints and neuroleptic-induced akathisia may be challenging (56).

Patients with a first-degree relative with RLS symptoms were significantly more likely to have RLS themselves. Familial clustering of RLS is a frequent finding (9), which can be interpreted as an expression of increased genetic vulnerability. Currently, 22 gene loci are known to be associated with an increased incidence of RLS (57, 58). In line with previous epidemiological studies in the general population, we found an increased RLS prevalence for women (21.7 vs. 9.5% in men) (9, 18).

In this prospective, cross-sectional, multicenter study, RLS was common among psychiatric inpatients. RLS was associated with higher BMI, impaired subjective sleep quality, and lower alcohol consumption. In a logistic regression analysis, after adjusting for confounders, gender, a positive family history of RLS, and increased daytime sleepiness were identified as predictors for the occurrence of RLS, whereas a higher frequency of alcohol consumption was associated with a lower frequency of occurrence. Knowledge of the prevalence of RLS appears important to increase awareness and systematic diagnostic assessment, potentially contributing to improved care for psychiatric patients.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Ethics Committees of the Universities of Regensburg, Berlin, Leipzig, Freiburg, Nuremberg, Heidelberg (Germany), and Bern (Switzerland). The patients provided their written informed consent to participate in this study. The patients/participants provided their written informed consent to participate in this study.

FW, HD-H, LF, NM, KR, CS, CN, and TW contributed to conception and design of the study. FW, HD-H, ED-S, LF, AH, NM, CM, DN, KR, CS, MSe, MSp, CN, and TW acquired data, performed investigations, and wrote and reviewed the manuscript. HD-H and FW performed quality control, data entry, and statistical analysis. FW, HD-H, and TW wrote the original manuscript draft. All authors have read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.