The tDCS treatment was well tolerated by the patients. There were few, mild adverse events that were equally distributed across the three treatment groups (p= 0.95): mild transient headache (lasting <1 h) (in 14.2%, 11% and 10% of subjects in the DLPFC, occipital and sham tDCS groups, respectively), itching sensation on the site of stimulation (19.1%, 33.3% and 20%, respectively) and mild transient redness of skin at the site of stimulation (5%, 11.1% and 10%, respectively) – that lasted <40 min after the end of stimulation. There were no adverse effects related to application of tDCS during the follow-up period. Demographic characteristics were not significantly different across the treatment groups as detailed in . As previously mentioned, patients were not taking antidepressants.
Demographic and clinical characteristics
For our primary outcome (HDRS), the mixed model revealed a significant main effect of group (F2,111=5.2, p=0.01), time (F3,111=12.0, p<0.0001) and significant interaction term group vs. time (F6,111=2.6, p=0.02). When comparing the effects of tDCS on mood immediately after the end of treatment, there was a significant difference between DLPFC and sham groups (p=0.0018), and between DLPFC and occipital groups (p=0.009), but not between occipital and sham groups (p=0.6). In addition the effects of this treatment lasted for 30 d after the end of treatment as shown by a significant difference between sham and DLPFC at the follow-up assessment (p=0.04) (; see also Supplementary Table 1 – available in the online version of the paper).
Figure 1 Plots showing mood scores changes [as indexed by the Hamilton Depression Rating Scale (HDRS)] over time (t0, baseline; t1, immediately after treatment; t2, 15 d after end of treatment; t3, 30 d after the end of treatment). Each data-point represents HDRS (more ...)
Our model using spline transformation showed a significant interaction term (F2,114=4.87, p=0.009) up to the first evaluation (immediately after treatment) but not for the follow-up – 15 d and 30 d after treatment (F2,114=0.04, p=0.95); suggesting that after the 10 d treatment, the three groups had a similar performance (i.e. no further improvement or worsening over time. Indeed this model explained the data better than the linear model that used time as a continuous variable [Akaike Information Criterion (AIC) for the model with spline function=951.0 and for the linear model=998.3].
The BDI scores revealed similar results: a significant interaction term (F6,111=4.2, p=0.0007) and significant difference between DLPFC and sham groups immediately and 30 d after treatment (p=0.0045 and p=0.03, respectively) () [see Supplementary Table 2 (available online) for further details]. The effect size of Cohen’s d when comparing the DLPFC and sham groups immediately after treatment, as indexed by HDRS, was 1.11.
Depression scores changes immediately after 10 d of stimulation compared with baseline in the three groups of treatment. Each column represents mean Hamilton Depression Rating Scale (HDRS) score reduction; error bars indicate s.e.m.
There was a significantly greater number of responders after DLPFC tDCS (8 responders) compared with sham and occipital tDCS (2 and 0 responders, respectively) (χ21=5.43, p=0.019). In addition, there were five remissions in the DLPFC group and no remissions in the other two controls group (χ21=5.17, p=0.02).
There were no significant correlations between mood improvement as indexed by HDRS and clinical variables (baseline HDRS and duration of disease) or demographic characteristics (age and, gender).