between the treatment groups (P 5 .19); however, decreased appetite was reported more frequently in the topiramatetreated group. Because the UDysRS change slightly favored topiramate, we calculated the effect of not meeting our recruitment goal on statistical power to detect the desired change in the total UDysRS score using a frequentist conditional power model. 6 Given the present results and assuming that additional enrollment to our goal of 50 participants would continue in the current trend, our conditional power was estimated at 5.2%. A final sample of more than 350 participants would have been required to reach the goal of 80% power (see Supplemental Materials). We found a small topiramate-associated improvement in dyskinesia in comparison to placebo, but the impact was far less than predicted. We failed to find a significant change in our primary outcome and, given the failure to retain the blind by patients or raters, we are not pursuing this treatment further at the present time. Our conditional power analysis results suggest that our truncated sample size was not a major determinant in our negative findings. Our findings, along with those of Kobylecki and colleagues, 5 suggest that neither shorter term (3 weeks) nor longer term (8 weeks) topiramate treatment in PD patients either on or off concomitant amantadine is effective in the treatment of dyskinesia associated with levodopa treatment.