Appendix S3. Supplementary results

Meta-regression analyses performed after the exclusions of cross over studies. Year of publication Similarly to the analysis performed on all studies, responder rate was significantly associated with year of publication (r = 0.545, p < 0.001). Among AEs, somnolence showed a significant reduction over the years (r = -0.341, P = 0.005). Excluding cross-over studies, also ataxia (r = -0.524, P = 0.003) and anxiety (r = -0.623, P = 0.041) resulted negatively related to year of publication. Geographic area Differences between geographic for proportions of patients with at least one AE, dizziness, somnolence, and headache have been confirmed without cross-over studies. In contrast, no significant differences emerged for ataxia and tremor removing cross-over studies. Disease severity All the results emerged in the main analysis have been confirmed excluding cross-over studies. Drug load All the results emerged in the main analysis have been confirmed excluding cross- over studies. Duration of double-blind phase Proportion of patients withdrawing due to AEs was positively correlated with study duration (r = 0.312, p = 0.010). Among specific AEs, headache (r = 0.484, p < 0.001) had a positive correlation with duration of double-blind phase. Unlike the analysis including cross-over studies, the relationship between the proportions of responders and tremor were no longer significantly related to the duration of the double-blind phase. Table. Effect of date of publication on the relationships between placebo and active treatment for all outcome measures. The Pearson product–moment correlation weighted by the population size has been calculated separately for RCTs conducted in the period 1987-2002 and that conducted in the last decade 2003-2013. These grouping has been decided in order to maintain a comparable number of RCTs in the two periods. The Pearson r coefficient, the r-squared, the p value and the number of studies included in the analysis are given.

1987-2002 2003-2013

R Number of R Number of R P R P square studies square studies Responder all 0.588 0.346 < 0.001 31 0.523 0.274 0.001 36 double-blind Withdrawal for 0.490 0.240 0.003 35 0.647 0.419 < 0.001 40 AEs

All AEs 0.917 0.841 < 0.001 17 0.830 0.689 < 0.001 34

Abnormal 0.586 0.343 0.028 14 0.663 0.440 0.037 10 thinking

Anxiety 0.925 0.856 0.001 8 -0.384 0.147 0.616 4

Ataxia 0.607 0.368 0.013 16 0.727 0.529 < 0.001 20

not appli- not appli- not appli- Depression 0.360 0.130 0.308 10 cable cable cable 1

Diplopia 0.772 0.596 < 0.001 21 0.301 0.091 0.173 22

Dizziness 0.648 0.420 < 0.001 30 0.538 0.289 < 0.001 40

Fatigue 0.649 0.421 < 0.001 29 0.453 0.205 0.015 28

Headache 0.595 0.354 < 0.001 32 0.803 0.645 < 0.001 37

Memory im- 0.872 0.760 < 0.001 12 0.110 0.012 0.795 8 pairment

Somnolence 0.492 0.242 0.004 33 0.727 0.529 < 0.001 40

Tremor 0.291 0.085 0.358 12 0.604 0.365 0.065 10 Table. Impact of geographic area on the relationships between placebo and active treatment for all outcome measures. The Pearson product– moment correlation weighted by the population size has been calculated separately for RCTs conducted in Europe, North America, and Asia. The Pearson r coefficient, the r-squared, the p value and the number of studies included in the analysis are given.

Europe North America Asia

R Number of R Number of R Number of R P R P R P square studies square studies square studies Responder all double- 0.479 0.229 0.061 16 0.688 0.473 0.007 14 0.616 0.379 0.104 8 blind Withdrawal for AEs 0.296 0.088 0.218 19 0.506 0.256 0.032 18 0.602 0.362 0.114 8

All AEs 0.878 0.771 < 0.001 14 0.655 0.429 0.158 6 0.528 0.279 0.179 8 not not not Abnormal thinking -0.011 0.000 0.984 6 0.523 0.274 0.184 8 applicable applicable applicable 1 not not not not not not Anxiety 0.934 0.872 0.02 5 applicable applicable applicable 1 applicable applicable applicable 0 not not not Ataxia 0.757 0.573 0.049 7 0.556 0.309 0.060 12 applicable applicable applicable 2 not not not not not not Depression 0.609 0.371 0.147 7 applicable applicable applicable 2 applicable applicable applicable 0 Diplopia 0.365 0.133 0.269 11 0.893 0.797 < 0.001 10 0.587 0.345 0.298 5

Dizziness 0.650 0.423 0.003 19 0.548 0.300 0.028 16 -0.016 0.000 0.973 7

Fatigue 0.756 0.572 0.001 16 0.275 0.076 0.364 13 0.398 0.158 0.739 3

Headache 0.371 0.138 0.118 20 0.813 0.661 0.001 13 0.299 0.089 0.515 7

Memory impairment 0.724 0.524 0.066 7 0.998 0.996 0.002 4 -0.668 0.446 0.332 4

Somnolence 0.698 0.487 0.001 20 0.505 0.255 0.046 16 0.453 0.205 0.307 7 not not not Tremor 0.260 0.068 0.619 6 0.202 0.041 0.664 7 applicable applicable applicable 1