'Therapeutic potential of valproic acid for retinitis pigmentosa,' by Clemson et al, has significant flaws that militate against their conclusion that short-term valproic acid improved the vision of patients with retinitis pigmentosa.The investigators should have performed a case-control study, comparing patients taking valproic acid with control patients matched for baseline field and acuity. This would have allowed for possible floor effects owing to the low baseline function of some of the treated patients (ie, who might be more likely to improve rather than decline over follow-up by chance variability) and avoided their use of historical rates of change drawn from different populations tested by others with different methods.Statistical analyses should have been performed on patients, instead of eyes, unless controlling for the intraclass correlation between fellow eyes of the same patient. A reanalysis of their results by the signed-rank test for duster-correlated data reveals weaker effects for visual-field improvement (eg, p =0.14 vs no change) and visual acuity improvement (eg, p=0.06 vs no change) in their cohort. In fact, none of the presented comparisons was statistically significant with this test.
展开▼
