There's a power analysis in the paper showing that they've got plenty of statistical power. (Estimated N=34 subjects for 90% power, but they used 77 subjects instead).