Carolien Zeetsen

54 Table 3.5. Relation between Montreal Cognitive Assessment total score (MoCA–TS) and neurocog- nitive disorders (NCD or no–NCD), at baseline and follow–up. Statistical significance of the Area Under the Curve (AUC) is reported. MoCA at baseline MoCA at follow–up AUC (SE) .676 (.059) .745 (.055) p –value .006** < .001*** Cut–off (≤) Sensitivity Specificity Sensitivity Specificity 20 .289 .919 .200 .973 21 .378 .892 .311 .946 22 .400 .838 .356 .865 23 .444 .811 .444 .865 24 .556# .622# .511 .838 25 .756 .459 .667# .730# 26 .911 .297 .822 .541 27 .933 .162 .889 .378 PPV (%) at # 64.1 75.0 NPV (%) at # 53.5 64.3 Accuracy (%) at # 58.5 69.5 Cohen’s Kappa (%) at # 17.5 39.2 ≤ optimal cut–off ( n , %) 39 (47.6) 40 (48.8) Note: # = MoCA cut–off with most optimal sensitivity and specificity; PPV = positive predictive value; NPV = negative predictive value; * = p < .05; ** = p < .01; *** = p < .001. Systematic differences between MoCA versions 7.1 and 7.2 Paired t –tests between MoCA results for version 7.1 (baseline) and 7.2 (follow–up) showed that only the MoCA–DS abstract reasoning, memory, and orientation did not differ significantly. Scores on all other MoCA–DS and the MoCA–TS differed significantly between both versions. For all MoCA–DS except language, mean scores on MoCA version 7.2 were higher (Table 3.4). Substance type and abstinence duration The logistic regression model was statistically significant ( χ 2 (7) = 16.58, p = .020), correctly classifying 69.0% of cases. However, neither the predictors nor any interaction between the predictors in the model were statistically significant.