Use of covariates in randomized controlled trials

A recent discussion in this journal illustrates some recur-rent misunderstandings about the role of covariates inrandomizedcontrolledtrials(RCTs).Thisletteraimsatclar-ifyingthisroleandatpointingoutapitfallinSPSSrepeatedmeasuresANOVA.We hope that our commentary will con-tribute to a further improvement in the use of advancedstatistics in neuropsychology.Anstey et al. (2006) reported the effects of cataract sur-gery on neuropsychological test performance in an RCT.They used repeated measures ANOVA to test for group bytimeinteraction,adjustingfortwocovariates:ageandbase-line visual acuity. Gilmore (2007) disputed the correctnessof their analyses, pointing out a significant group differ-enceinbaselinevisualacuityandsuggestingthatabetween-subject covariate cannot adjust within-subject effects orinteractions involving within-subject effects. Anstey et al.(2007)repliedthattheirmodelincludedacovariatebytimeinteraction, thereby adjusting the group by time interactionof interest for baseline visual acuity. While the analysis byAnstey et al. (2007) is by and large correct, all three publi-cations mentioned above show some statistical misconcep-tions that we briefly discuss in this commentary.First of all, in an RCT baseline group differences arecaused by chance fluctuation and will not be significant,except because of type I errors or if dropouts are excludedfrom the test on baseline differences. The only significantbaseline difference in Anstey et al. (2006), that in visualacuity, had a