Sensorimotor battery
Tracking
A 2 × 2 × 3 mixed ANOVA was conducted to analyse the effect of group (TN/Control), guide (with guide/no guide) and speed (slow/medium/fast) on RMSE. A reciprocal transform was used to achieve homogeneity of variance and a mixed ANOVA was performed on these data. Raw scores pre-transform can be seen in Fig. 2. The Levene’s test remained significant for the ‘Fast With Guide’ condition. Data were available from 36 TN participants and 36 Controls. A main effect of group emerged [F (1,70) = 7.44; p < 0.01, η2 = 0.10] with RMSE being significantly higher in the TN group. Main effects for Guide [F (1,70) = 185.99; p < 0.01, η2 = 0.73] and Speed [F (1.11,78.61) = 1915.63; p < 0.001, η2 = 0.96] emerged, with RMSE significantly increasing (p < 0.001) as the speed of the dot increased. There was also a Speed x Guide interaction [F (1.25, 87.52) =187.44; p < 0.001, η2 = 0.73].
Aiming
An independent t-test was used to test for a difference in movement time (s) between the two groups. Data were included from 40 control and 40 TN participants. Movement Time (MT) was significantly longer for the TN participants (mean = 1.33 s, SD = 0.18) than the controls (mean = 1.24 s, SD = 0.20) [t (78) = 2.10, p < 0.05, Cohen’s d = − 0.47].
Steering
A 2 × 2 mixed ANOVA was conducted to analyse the effects of group (TN/Control) and path (path A and B) on penalised path accuracy. Data were included from 39 control and 40 TN participants (Control path A: mean = 1.33 s, SD = 0.54; Control path B: mean = 1.23 s, SD = 0.45; TN path A: mean = 1.40s, SD = 0.54; TN path B: mean = 1.31 s, SD = 0.49). No main effects or interactions emerged [Group F (1,77) = 0.561, p = 0.456, η2 = 0.007; path F (1,77) = 2.849, p = 0.1, η2 = 0.036; Group x path F (1,77) = 0.001, p = 0.970, η2 = 0.00].
Cognitive battery
Forward digit recall (FDR)
Proportion correct
A χ2 test was used to test for an association between participant group (TN/Control) and obtaining a high proportion correct score. Participants were at ceiling for trials with a sequence length of three so these cannot be analysed. For trials with a sequence length of four there was no significant association between the participant group (TN/Control) and whether they were likely to get a high score [χ2 (1) = 1.89, p = 0.14, odds ratio = 2.25]. For trials with a sequence length of five or six there was a significant association between the group the participants were in (TN/Control) and whether or not they were likely to get a high score [length 5: χ2 (1) = 7.623, p < 0.01, length 6: χ2 (1) = 5.529, p < 0.05]. Based on the odds ratio, the odds of getting a high score were 3.97 (sequence length 5) and 2.93 (sequence length 6) times higher if the person was a Control participant than a TN participant. Raw scores pre categorisation can be seen in Fig. 3.
RT
For the reaction time (RT) data a reciprocal transform was used to achieve homogeneity of variance and a 2 × 4 mixed ANOVA performed on these data to analyse the effect of group (TN/Control), and sequence length (3,4,5,6) on RT. Raw mean RT scores pre-transform can be seen in Fig. 3. A significant main effect of group emerged [F (1,81) = 17.94, p < 0.001, η2 = 0.181] with RTs longer for the TN group than Control group. There was also a main effect of sequence length [F (2.51, 203.64) = 14.96, p < 0.001, η2 = 0.15] caused by significant differences between all sequence lengths (p < 0.05) apart from 3 and 4 (p = 0.09). No group x sequence length interaction emerged [F (2.51, 203.64) = 2.09, p = 0.114, η2 = 0.02].
Backward digit recall (BDR)
Proportion correct
Participants were at ceiling for trials with a sequence length of two, so these cannot be analysed. For trials with a sequence length of four there was no significant association between participant group (TN/Control) and whether they were likely to get a high score [χ2 (1) = 3.046, p = 0.064, odds ratio = 2.25]. For trials with a sequence length of three and five there was a significant association between the group the participants were in (TN/Control) and whether or not they were likely to get a high score [length 3: χ2 (1) = 3.013, p < 0.05, length 5: χ2 (1) = 6.467, p < 0.05]. Based on the odds ratio, the odds of getting a high score were 3.23 (sequence length 3) and 3.17 (sequence length 5) times higher if the person was a control participant than a TN participant. Raw scores pre categorisation can be seen in Fig. 3.
RT
For the reaction time (RT) data a reciprocal transform was used to achieve homogeneity of variance and a 2 × 4 mixed ANOVA performed on these data to analyse the effect of group (TN/Control), and sequence length (2,3,4,5,) on RT. Raw mean RT scores pre-transform can be seen in Fig. 3. A significant main effect of group emerged [F (1,81) = 18.41,p < 0.001, η2 = 0.185] with RTs longer for the TN group than Control group. There was also a main effect of sequence length [F (2.71, 219.78) = 110.98, p < 0.001, η2 = 0.57] caused by significant differences between all sequence lengths (all p < 0.001). A group x sequence length interaction emerged [F (2.71, 219.78) = 2.96, p < 0.05, η2 = 0.015] supporting what can be seen in the figure: that group differences increased as sequence length increased.
Corsi block tapping (Corsi)
Proportion correct
For trials with a sequence length of three there was no significant association between the group a participant was in (TN/Control) and whether they were likely to get a high score [χ2 (1) = 0.863, p = 0.25, odds ratio = 1.59]. For trials with a sequence length of four, five and six there was a significant association between the participant group (TN/Control) and whether or not they were likely to get a high score [length 4: χ2 (1) = 5.23, p < 0.05, length 5: χ2 (1) = 6.37, p < 0.01, length 6: χ2 (1) = 3.49, p < 0.05]. Based on the odds ratio, the odds of getting a high score were 2.81 (sequence length 4), 3.12 (sequence length 5), and 2.30 (sequence length 6) times higher if the person was a control participant than a TN participant. Raw scores pre categorisation can be seen in Fig. 3.
RT
A 2 × 4 mixed ANOVA was conducted to analyse the effect of group (TN/Control), and sequence length (2,3,4,5,) on RT. Mean RT scores can be seen in Fig. 3. Reaction times were longer for the TN group than the control group but the main effect failed to reach conventional levels of significance [F (1,81) = 3.959, p = 0.09, η2 = 0.035]. There was also a main effect of sequence length [F (2.61, 211.41) = 7.69, p < 0.001, η2 = 0.085] caused by differences between sequence lengths of 3 and 4, 4 and 5 and 4 and 6 (all p < 0.01). No group x sequence length interaction emerged [F (2.61, 211.41) = 1.44, p = 0.231, η2 = 0.016].
Processing speed
A Mann-Whitney U test revealed a significant difference in RTs between the groups [U = 465.00, p < 0.001, rank biserial correlation = − 0.460] with RTs being longer for the TN group (mean = 3.69 s, SD = 0.86) compared to Controls (mean = 3.14 s, SD = 0.74).
Inhibition (flankers)
A reciprocal transform was used to achieve homogeneity of variance and a 2 × 2 mixed ANOVA performed on these data to analyse the effect of group (TN/Control), and condition (congruent/incongruent) on RT. A significant main effect of group emerged [F (1,81) = 11.45, p < 0.001, η2 = 0.124] with raw RTs for the TN group (mean = 0.12 s, SD = 0.40) being longer than those for the Control group (mean = 0.86, SD = 0.24). There was also a main effect of condition [F (1,81) = 43.70, p < 0.001, η2 = 0.346] with RTs being longer for incongruent (mean = 1.03 s, SD = 0.37) compared to congruent trials (mean = 0.95, SD = 0.27). There was no group × condition interaction [F (1,81) = 1.51, p = 0.223, η2 = 0.012].
Additional analyses
We first explored monopharmacy versus polypharmacy. We found no main effect of group for any of the sensorimotor measures (Aiming, Tracking, and Steering). There were no differences between groups on reaction time for Processing Speed or Inhibition tasks. For FDR, BDR and Corsi (our working memory tests) we found no associations between group and proportion correct but we did find a significant main effect of group on reaction time for both FDR [F (1,40) = 4.198, p < 0.05, η2 = 0.095] and BDR [F (1,40) = 4.424, p < 0.05, η2 = 0.1]. In both cases RTs were longer for those undergoing polypharmacy (FDR mean = 1.146, BDR mean = 1.74) compared to monopharmacy (FDR mean = 0.896, BDR mean = 1.475). We next explored whether there would be any differences between groups taking different types of AED. We found no main effects of group on any of our measures. Finally, we tested for an association between dose and performance and found a significant association on the Backward Digit Recall task when sequence length was highest (sequence length 5) with higher doses associated with poorer performance [r = 0.321; p = 0.049].