Patients were eligible if they were 18 years or older and fulfilled the clinical criteria for FMD as defined by Fahn and Williams (17) (Table 1). Patients were excluded if they had a history of other neurological disorder or psychosis; if another neurological disease was diagnosed during hospitalization for magnetic stimulation; if they had received TMS in the past; or if they had a contraindication to magnetic stimulation. We enrolled consecutive FMD inpatients seen in our Movement Disorders Clinic between April 2013 and July 2015. The study was approved by the local ethics committee (CPP-IdF-Paris 6, Pitié-Salpêtrière University Hospital), and all the patients gave their informed consent.

The patients were informed that their symptoms were linked to a non-lesional brain dysfunction (Figure 1). The patients were prospectively randomized to receive RMS on day 1 and TMS on day 2 (group 1) or the reverse sequence (group 2). A randomization function was used in Excel (Microsoft Excel RAND function) to establish the order of magnetic stimulations before the beginning of the study. A minimal interval of 18 h was respected between the two treatment sessions.

A detailed clinical assessment was done by CH on day 1 (baseline score, before stimulation 1), day 2 (before stimulation 2), and day 3 (after stimulation 2). Each clinical assessment was videorecorded, and one movement-disorder expert (BD) blindly rated the patients before and after each session, using a rating scale that was specifically designed for FMDs (11) (Data Sheet 1 in Supplementary Material), yielding an “FMD score” ranging from 7 (normal) to 41 (most severe). The percentage of improvement was calculated as follows: ((pretreatment score) − (posttreatment score))/((baseline score) − 7). No other treatment was provided during the hospital stay, and the patients’ medication was kept stable during the study. Physiotherapy was not provided during the protocol, but it was systematically prescribed at discharge. A new TMS session was performed when patients relapsed during the follow-up. No psychotherapy was offered during the whole duration of follow-up.

On day 1, all the patients completed self-administered psychiatric questionnaires, including a screening version of the Mini International Neuropsychiatric Interview (MINI) (18), the Hospital Anxiety and Depression (HAD) Scale (19), and the exposure to traumatic life events section of the French version of the Composite International Diagnostic Interview (CIDI 2.1) (20). These evaluations allowed us to explore mood and anxiety disorders, addictive behaviors, post-traumatic disorders, and psychotic disorders. When a psychiatric examination was considered necessary, because of a positive MINI section, a behavioral disorder, or the care team opinion, a psychiatrist (Thomas Mauras) interviewed the patient on day 3 (n = 16, 48.5%) or in the psychiatric clinic after hospital discharge (n = 6, 18.1%).

At inclusion, magnetic stimulation was described to the patients as an effective treatment with excellent results. However, to ensure a reproducible protocol, the neurophysiologist (Iulia Iliescu or Francine Mesrati) said nothing to reinforce perceived treatment efficacy during the stimulation session.

An average of 50 (range: 30–80) consecutive stimuli, at 120–150% of the resting motor threshold (each stimulus lasted 250 µs) was delivered at low frequency (0.25 Hz) over two different sites, namely, the lateral (upper limbs) or medial (lower limbs) motor cortex contralateral to the FMD for repeated TMS, and over the cervical (upper limbs) or lumbar (lower limbs) spinal roots homolateral to the FMD for repeated RMS. The raters (Bertrand Degos and Cécile Hubsch) were blinded to the type of stimulation.

The patients were followed up with telephone interviews at 3 and 12 months (Cécile Hubsch) and with a visit (Bertrand Degos or Cécile Hubsch) 6 months after treatment. The Clinical Global Impression—Improvement (CGI-I) Scale (21) was estimated by the patients at each follow-up. Patients were asked how the severity of movement disorder was as compared to pretreatment. This scale ranges from 1 (very much improved) to 7 (very much worse).

SPSS software (http://www-01.ibm.com/software/analytics/spss/) was used for all analyses, and significance was assumed at p < 0.05. The distribution of the percentage improvements did not follow a normal distribution (Shapiro–Wilk test, p = 0.002 for global improvement). For this reason, and because of the small number of patients in each group, we used non-parametric tests. Spearman’s rank correlation test was used for correlation analyses. The Wilcoxon signed-rank test was used for group comparisons and the Kruskal–Wallis test was used to compare scores across more than two groups. A mixed three-factor ANOVA was used to determine the effect of (i) the type of stimulation (root versus transcranial), (ii) the day of stimulation (day 1 or day 2), and (iii) the intervention group (group 1 or 2). The condition of a normal distribution of residuals was respected; this allowed us to use an ANOVA for this multifactorial analysis.

Thirty-six consecutive patients with FMDs participated in the study (Table 1). Two patients were excluded because of an “organic” condition (Parkinson’s disease and algoneurodystrophy syndrome), and one patient was excluded because he refused magnetic stimulation. The remaining 33 patients (26 F/7 M) were included in the analyses. Their main characteristics are summarized in Table 1, and a detailed description is provided in Table S1 in Supplementary Material.

Median age was 45 years (range: 18–74). The median symptom duration was 2.9 years (range: 0.3–30 years). The predominant movement disorders were tremor (n = 13), dystonia (n = 11,), jerky dystonia (n = 4), myoclonus (n = 2), stereotypies (n = 2), and parkinsonism (n = 1). Associated disorders included motor deficits (n = 7; 21.2%), sensory deficits (n = 7; 21.2%), and pain (n = 8; 24.2%).

Ongoing anxiety disorders were found in 20 patients (60.6%) and included generalized anxiety disorder, panic disorder, social phobia, agoraphobia, and obsessive–compulsive disorder. The anxiety disorder was associated with depression in seven patients (21.2%). Twenty-one patients (63.6%) reported a traumatic life event according to the traumatic life events section of the French version of the CIDI 2.1, including rape in 24.2% of cases and another life-endangering trauma in 39.4%. Only 12 patients (36%) were in employment or study, the remaining 21 patients (64%) being on long-term sick leave (with a disability living allowance), unemployed, retired, or students having interrupted their studies. Fifteen of the latter 21 patients were receiving disability-related benefits.

The median percentage improvement in FMD scores was 29.2% after the first session and 18.2% after the second session (Table 1). The median total percentage improvement was 70% at day 3. Twenty-two patients (66.7%) experienced a significant improvement (>50% improvement) (Figure 2). Motor symptoms resolved completely on day 3 in 10 of these latter patients (30%).

We found no correlation between the final percentage improvement and age (r = 0.018, p = 0.92), gender (z = −1.095, p = 0.27), education (r = −0.004, p = 0.98), HAD scores (r = −0.134, p = 0.45), depression and/or anxiety (z = −0.972, p = 0.33), traumatic life events (z = −1.747, p = 0.08), symptom duration (r = 0.315, p = 0.74), clinical presentation (Kruskal–Wallis: 2.8, p = 0.24, Table S2 in Supplementary Material), presence of any additional symptoms (deficit and/or pain) (z = −0.26, p = 0.811), or initial severity (r = 0.11, p = 0.53).

Patients in group 1 (n = 16) and group 2 (n = 17) did not differ significantly with respect to age (z = −0.20, p = 0.84), gender (Fisher’s test, p = 0.4), symptom duration (z = −0.13, p = 0.9), clinical presentation (χ² = 8.3, p = 0.19), or FMD severity (z = 0.81; p = 0.42) (Table 2).

There was no significant difference between the improvements noted after the first session of repeated TMS (median: 37.5%; IQR: 20–72.7) and after the first session of repeated RMS (median: 23.6%; IQR: 5.5–47.8) (z = −1.05; p = 0.29, Wilcoxon signed-rank test).

There was no significant difference in the final percentage improvement (day 3) between the two groups (z = 0; p = 1).

There was no significant effect of the treatment modality (median improvement 35.8% after TMS versus 24.8% after RMS, p = 0.26), and no significant effect of the treatment order (RMS first or TMS first, p = 0.8) (Table 3). Regardless of the sequence, the percentage improvement was significantly larger after the first magnetic stimulation session (day 1) than after the second session (day 2) (p = 0.03).

All the patients were followed up for 6 months, and 29 patients (87.9%) were followed up for 1 year (Figure 3). On day 3, 24 h after the second session of magnetic stimulation, 60% of the patients were much (CGI score = 2) or very much (CGI score = 1) improved. At 1 year, 56% of the patients were still much or very much improved. Chi-square analysis revealed no statistical difference between the CGI score on day 3 and the CGI score at 3 months, 6 months, or 1 year (p = 0.77). However, 12 patients relapsed (total number of relapses = 19, median time to first relapse = 6 months). All the patients concerned were offered another TMS session at least 3 months after the previous session, and all were significantly improved after each relapse by another session of repeated TMS. These additional TMS sessions obviously influenced the follow-up CGI assessment at 6 months and 1 year.