Actigraphic assessment of periodic leg movements in patients with restless legs syndrome


Correspondence Esther Werth, Department of Neurology, University Hospital Zürich, Frauenklinikstrasse 26, 8091 Zürich, Switzerland. Tel.: +41-44-255-5535; fax: +41-44-255-9201; e-mail:


The diagnosis of restless legs syndrome (RLS) relies upon diagnostic criteria which are based on history only, and dopaminergic treatment is not normally the first choice of treatment for all patients. It would be worthwhile to identify patients non-responsive to dopaminergic treatment beforehand, because they may suffer from a restless legs-like syndrome and may require alternative treatment. We included retrospectively 24 adult patients fulfilling the four essential criteria for restless legs and 12 age-matched healthy controls. They were investigated by ambulatory actigraphy from both legs over three nights, and patients started treatment with dopamine agonists after this diagnostic work-up. We examined 12 responders to dopaminergic treatment and 12 non-responders and studied the association between response to dopaminergic treatment and the periodic limb movement index (PLMI) as assessed with actigraphy. Demographic characteristics, excessive daytime sleepiness and fatigue at baseline were similar in all three groups. Baseline RLS severity was similar between responders and non-responders [International Restless Legs Severity Scale (IRLS): 25 ± 9 and 24 ± 8]. Group comparisons of PLMI before treatment initiation showed significant differences between the three groups. Post-hoc pairwise comparisons revealed that healthy controls had significantly lower PLMI (4.9 ± 4.5) than responders (29.3 ± 22.7) and non-responders (13.3 ± 11.2). Similarly, the PLMI in responders was lower than in non-responders. PLMI day-to-day variability did not differ between responders and non-responders and there was no correlation between treatment effect, as assessed by the decrease of the IRLS and baseline PLMI. Our retrospective study indicates that actigraphy to assess periodic limb movements may contribute to a better diagnosis of dopamine-responsive restless legs syndrome.


The diagnosis of restless legs syndrome (RLS) is based on four essential criteria which are assessed by history: (1) an urge to move the legs, accompanied by an uncomfortable sensation that (2) begins or worsens during rest and (3) in the evening or at night, and which (4) is relieved by movement (Allen et al., 2003). Three further criteria support the diagnosis: positive response to dopamine agonists, periodic limb movements (PLM) in sleep and positive family history. Patients fulfilling all essential but no supportive RLS criteria are encountered frequently in clinical practice. As these patients differ in many aspects from RLS patients, we introduced the term ‘RLS-like syndrome’ (Bassetti et al., 2006; Baumann et al., 2007).

Periodic limb movements are periodic episodes of repetitive stereotypical movements of the limbs, predominantly the legs. They occur mainly during sleep, last 0.5–10 s and are separated by intervals of 5–90 s (Iber et al., 2007). PLM can be measured by bilateral anterior electromyography of the legs as a part of nocturnal polysomnography (PSG), or by actigraphy from both legs. The two methods are not necessarily interchangeable, as PSG records electric muscle activity, whereas actigraphy measures mechanically generated movements; despite this, it has been shown that the two methods were highly correlated (Gschliesser et al., 2009; Sforza et al., 2005). Furthermore, actigraphy does not allow distinguishing PLM during sleep from those during wakefulness. Nevertheless, actigraphy is cost effective and can be performed easily at the patient's home during multiple consecutive nights. It is therefore a useful proxy for PLM detection when combined with detailed clinical evaluation (Kohnen et al., 2007).

Thus, our study was designed to examine whether actigraphy distinguishes RLS from RLS-like syndrome, and thus whether or not it might contribute to the accurate diagnosis of dopamine-responsive RLS.


We analysed retrospectively adult patients who: were examined by a neurologist (CRB, MMS, RP); fulfilled the four essential RLS criteria; were investigated by ambulatory actigraphy from both legs over three nights; and started treatment with dopamine agonists after this diagnostic work-up. Because PSG was not always feasible, and despite not being validated, we occasionally performed actigraphy studies for the assessment of periodic limb movements, as in previous publications (Kohnen et al., 2007; Sforza et al., 2005). Patients on dopaminergic treatment before or during actigraphy and subjects with other disorders, including those who might present with RLS complaints, were excluded from this study. Respective examinations were performed in cases with a clinically suspected underlying cause of RLS; e.g. electrodiagnostic testing in cases with suspected polyneuropathy. Patients were divided into two groups according to their response to dopamine agonists: responders revealed an improvement of at least 10 points on the International Restless Legs Severity Scale (IRLS, score 0–40), whereas non-responders did not (Hening and Walters, 2001). We identified retrospectively 12 non-responders for this study and added the same number of age- and gender-matched responders and controls. All non-responders were treated with at least two different dopamine agonists (pramipexole, ropinirole or rotigotine) in up to maximally recommended doses and over at least 2 weeks. Eleven of the 12 responders reacted to the first treatment choice (mainly pramipexole or ropinirole). In addition, we recruited healthy age-matched controls prospectively. All individuals provided signed informed consent, and the protocol was approved by the local ethics committee.

All patients completed the Epworth Sleepiness Scale (ESS, overall score 0–24), the Fatigue Severity Scale (FSS, score 1–7) and the IRLS. PLMs were measured by actigraphy with Actiwatch (Cambridge Neurotechnology Ltd, Cambridge, UK) on both legs for 3 consecutive nights at the subject's home, as described by King et al. (2005). We combined PLM of both legs and calculated an averaged PLM index (PLMI) over three nights.

The three groups were compared with one-way analysis of variance (anova). After confirmation of normal distribution, post-hoc tests between groups were performed with the Student's independent-samples t-test applied using a Bonferroni correction. < 0.017 was considered significant. PLMI day-to-day variability was assessed by analysing the variance over three nights of recording. Correlation analysis was computed to assess the association between treatment effect and the number of PLMs; spss version 20.0 was used for all calculations.


Demographic characteristics [age, gender, body mass index (BMI)], excessive daytime sleepiness and fatigue at baseline were similar in all three groups (Table 1). Baseline RLS severity as assessed with the IRLS was similar between responders and non-responders (25 ± 9 and 24 ± 8, > 0.69). On treatment, IRLS in responders was reduced to 10 ± 7 points, whereas it remained unaltered in non-responders (22 ± 6). Some non-responders were responsive to gabapentin (one of 12), pregabaline (one of 12), clonazepam (two of 12), dihydrocodeine (three of 12) or antidepressants (one of 12).

Table 1. Demographic characteristics of responders (R), non-responders (non-R) and controls (C)
 Rnon-RCP-value all groupsP-value R versus non-R
  1. Age: years, mean ± standard deviation; gender: % women.

  2. BMI, body mass index, kg/m2; IRLS, International Restless Legs Severity Scale; ESS, Epworth Sleepiness Scale; FSS, Fatigue Severity Scale.

Age52.1 ± 11.655.3 ± 6.259.1 ± 17.10.3910.394
BMI23.9 ± 2.224.7 ± 3.623.3 ± 4.30.6330.598
IRLS25.1 ± 8.723.6 ± 8.20 ± 0<0.0010.695
ESS7.0 ± 4.17.2 ± 6.35.5 ± 3.10.4210.948
FSS2.2 ± 1.02.2 ± 1.01.5 ± 0.90.1540.990

Group comparisons of PLMI before treatment initiation as assessed by actigraphy showed significant differences between the three groups (F2,35 = 12.6, < 0.001; Fig. 1). Post-hoc pairwise comparisons revealed that healthy controls had lower PLMI (4.9 ± 4.5) than responders (29.3 ± 22.7, = 0.001) and non-responders (13.3 ± 11.2, = 0.006). Similarly, the PLMI of responders was lower than in non-responders (= 0.016). Day-to-day variability did not differ between responders (74.4 ± 42.2) and non-responders (44.9 ± 68.0, = 0.23). Furthermore, there was no correlation between treatment effect, assessed by the decrease of the IRLS and baseline PLMI (= 0.22, = 0.35).

Figure 1.

Periodic limb movement index (PLMI) in responders, non-responders and healthy controls. Scatterplot; *< 0.017.


Our preliminary study indicates that assessing PLMI with actigraphy may be helpful to distinguish RLS patients with response to dopamine agonists from patients with RLS symptoms who do not benefit from dopaminergic treatment. This distinction is clinically important, because non-responders might suffer from an RLS-like syndrome (Bassetti et al., 2006; Baumann et al., 2007): in our previous studies, we found that RLS-like patients without increased PLMI suffered more often from depression, and RLS symptoms improved with antidepressants rather than with dopaminergic drugs. Finally, some of our non-responders were treated more or less effectively with gabapentin, pregabaline, clonazepam, dihydrocodeine or antidepressants. The present data from this preliminary study suggest that actigraphy may be a useful alternative to expensive PSG for an objective differentiation of RLS from RLS-like syndrome. Furthermore, because of the large day-to-day variability of PLMs, actigraphy over three nights may be more representative than a single-night PSG (Trotti et al., 2009).

In the present small sample, however, we found a broad overlap of PLMI between the three groups, which suggests that this examination may not be specific. In this regard, it is well known that PLM are sometimes found—although less frequently—in normal healthy subjects, particularly in the elderly, and in a variety of other sleep–wake disorders (Bliwise et al., 1985; Montplaisir et al., 2000). In addition, distinguishing PLM during sleep from wakefulness was not possible, which might be important because only PLMs during sleep have been reported to respond to dopamine agonists (Ferri et al., 2012). Apart from the small sample studied here, this study has further significant limitations. It is based on retrospective data analyses, and response to treatment was assessed only by means of interviews and scales, but not with objective tests, including repeated actigraphy. Therefore, larger and PSG-controlled studies in different patient groups will be necessary to gain clearer insight into the measurement of PLMs for the diagnosis of dopamine-responsive, i.e. typical, RLS.


We acknowledge Ulf Kallweit and Pietro Cippà for their support of our study. This research received no specific funding.

Conflict of Interest

No conflicts of interest declared.