Impeaching clinical studies is easy as there are always methodological difficulties. Essentially, Read et al are saying that the statistical advantage for ECT over sham ECT has not been demonstrated, whereas Anderson thinks it may have been. I do think, though, that Anderson ought to spell out which studies he is relying on for that conclusion.
As with antidepressant studies, where I think the statistical difference overall in clinical trials between active and placebo treatment cannot be denied, placebo amplification, for example through unblinding, could still explain any statistical difference (eg. see previous post). Read et al argue that none of the ECT studies convincingly demonstrate they are double-blind.