Behavioral Data in Patients with OCD and Healthy Controls
Participants had significantly higher flanker task accuracy on congruent than incongruent trials (paired t (209) = 26.42, P < 0.0001). OCD cases were significantly less accurate than HC in all trial conditions (all P values < 0.02) (Table 1). Overall accuracy was also significantly decreased in both TR OCD and NTR OCD cases compared to HC (both P values < 0.05) (Table 2).
Correct responses were significantly slower than incorrect responses (paired t (209) = 6.33, P < 0.0001). No main effect of group or response type for reaction time and no interaction between group and response type for reaction time reached significance (P = 0.63 and P = 0.40, respectively). Age had significant positive correlations with accuracy (r = 0.17, P = 0.01), post-correct accuracy (r = 0.16, P = 0.02), and post-error accuracy (r = 0.21, P = 0.002). Age had significant negative correlations with reaction time on correct (r = -0.57, P < 0.0001) and incorrect trials (r = -0.46, P < 0.0001) and a significant positive correlation with post-error slowing (r = 0.21, P = 0.002).
There were no significant sex differences in flanker task accuracy, reaction time on correct or incorrect trials, or post-error slowing (all P values > 0.5). There were no significant differences in accuracy, reaction time on correct or error trials, or post-error slowing between patients with a current and past diagnosis of OCD (all P values > 0.09), between medicated and unmedicated patients with OCD (all P values > 0.1), or between patients with TR and NTR OCD (all P values > 0.37).
Event-Related Potential Data in Patients with OCD and Healthy Controls
ERN amplitude was significantly increased (more negative) compared to CRN amplitude (paired t (209) = -9.05, P < 0.0001), and Pe amplitude was significantly increased (more positive) compared to Pc amplitude (paired t (209) = 24.16, P < 0.0001). Age in all participants had significant correlations with the ERN (r = -0.20, P = 0.004), CRN (r = 0.24, P = 0.0005), and ΔERN (r = -0.38, P < 0.0001) but not Pe, Pc, or ΔPe (all P values > 0.14). Accuracy had significant correlations with the ERN (r = -0.17, P = 0.02) but not CRN or ΔERN (both P values > 0.07). Accuracy also had significant correlations with the Pe (r = 0.18, P = 0.009) and ΔPe (r = 0.26, P = 0.0001) but not Pc (P = 0.10). Supplementary Table 1 provides a correlation matrix for age, flanker task accuracy, and ten ERPs.
ERN amplitude was significantly increased in OCD cases compared to HC (F1, 206 = 19.36, P < 0.0001, Cohen’s d = 0.52), with significant effects for accuracy (F1, 206 = 7.55, P = 0.006) and age (F1, 206 = 6.63, P = 0.01) (Table 1; Figure 1). CRN amplitude was significantly enlarged in cases compared to HC (F1, 206 = 5.58, P = 0.02, Cohen’s d = 0.24), with significant effects for accuracy (F1, 206 = 8.73, P = 0.004) and age (F1, 206 = 16.55, P < 0.0001). The ΔERN was significantly enhanced in cases compared to HC (F1, 206 = 6.19, P = 0.014, Cohen’s d = 0.31), with a significant effect for age (F1, 206 = 35.16, P < 0.0001) but not accuracy (P = 0.73).
There was no significant group difference in Pe amplitude (P = .14). Pc amplitude was significantly increased in OCD cases compared to HC (F1, 206 = 4.02, P = 0.046, Cohen’s d = 0.32), without significant effects for accuracy or age (both P values > .08). Moreover, the ΔPe was significantly decreased in cases compared to controls (F1, 206 = 9.88, P = 0.002, Cohen’s d = 0.53), with a significant effect for accuracy (F1, 206 = 12.76, P = .002) but not age (P = .07).
Results for the ERNresid, CRNresid, Peresid, and Pcresid in the two groups are summarized in Table 1. There were no significant sex differences in the ERN, CRN, ΔERN, ERNresid, CRNresid, Pe, Pc, ΔPe, Peresid, or Pcresid (all P values > .05). There were no significant differences in any brain potentials between patients with a current and past OCD diagnosis (all P values > .55), between medicated and unmedicated patients with OCD (all P values > .10), or between patients with TR OCD and NTR OCD (all P values > .35) (Table 2).
Event-Related Potential Data in Patients with TR OCD, Patients with NTR OCD, and Healthy Controls
In a comparison of the ERN in TR OCD cases, NTR OCD cases, and HC, there were significant effects for group (F2, 205 = 9.70, P < .0001), age (F1, 205 = 6.72, P = .01), and accuracy (F1, 205 = 7.52, P = .007) (Table 2). ERN amplitude was significantly increased in TR OCD cases compared to HC (F1, 122 = 6.09, P = .02, Cohen’s d = .47), with no significant effects for age or accuracy (both P values > .07). ERN amplitude was also significantly enlarged in NTR OCD cases compared to HC (F1, 185 = 16.72, P < .0001, Cohen’s d = .53), with significant effects for age (F1, 185 = 8.44, P = .004) and accuracy (F1, 185 = 7.65, P = .006). In a comparison of the ΔERN in the three groups, there were significant effects for group (F2, 205 = 3.69, P < .03) and age (F1, 205 = 36.36, P < .0001) but not accuracy (P = .74). The ΔERN was significantly increased in TR OCD cases compared to HC (F1, 122 = 4.46, P = .04, Cohen’s d = .32), with a significant effect for age (F1, 122 = 25.97, P < .0001) but not accuracy (P = .32). The ΔERN was also significantly augmented in NTR OCD cases compared to HC (F1, 185 = 3.97, P = .048, Cohen’s d = .31), with a significant effect for age (F1, 185 = 37.04, P < .0001) but not accuracy (P = .77).
In a comparison of the ΔPe in TR OCD cases, NTR OCD cases, and HC, there were significant effects for group (F2, 205 = 4.99, P = .008) and accuracy (F2, 205 = 12.72, P = .0005) but not age (P = .06) (Table 2; Figure 3). The ΔPe was significantly decreased in TR OCD cases compared to HC (F1, 122 = 4.43, P = .04, Cohen’s d = .57), with significant effects for accuracy (F1, 122 = 8.02, P = .005) and age (F1, 122 = 4.14, P = .04). The ΔPe was also significantly diminished in NTR OCD cases compared to HC (F1, 185 = 7.73, P = .006, Cohen’s d = .52), with a significant effect for accuracy (F1, 185 = 13.80, P = .0003) but not age (P = .77). Results for the ERNresid, CRNresid, Peresid, and Pcresid in the three groups are summarized in Table 2.
Flanker Task Accuracy and Event-Related Potential Data in Patients with OCD and Healthy Controls
Because of the group differences in the ERN and ΔPe noted above along with the strong correlation of accuracy with age, a multiple linear regression analysis was done using age and both brain potentials as predictors to determine their associations with flanker task accuracy as the dependent variable. Age and both brain potentials were significantly associated with flanker task accuracy in all participants, with a more negative ERN and more positive ΔPe associated with higher accuracy (Table 3).
Clinical and Event-Related Potential Data in Patients with OCD and HC
Separate multiple linear regression analyses were done with all participants with either the ERN or ΔPe as the dependent variable and age, flanker task accuracy, lifetime OCD diagnosis, and CBCL/6-18 DSM-Oriented Scale scores as predictors. The ERN had significant associations with CBCL/6-18 Anxiety Problems Scale scores, age, and accuracy in the full model (Table 4). Backward stepwise regression analysis determined that only Anxiety Problems Scale scores and age were significantly associated with the ERN in the reduced model. The ERN had a significant negative correlation with Anxiety Problems Scale scores in OCD cases (r = -0.30, P = 0.002) but not HC (P = 0.46).
The ΔPe had significant associations with accuracy and age in the full model (Table 5). Backward stepwise regression analysis found that accuracy, CBCL/6-18 Obsessive Compulsive Problems Scale scores, and age were significantly associated with the ΔPe in the reduced model. The ΔPe had no significant correlations with Obsessive Compulsive Problems Scale scores in either OCD cases or HC alone (both P values > 0.18). Supplementary Table 2 provides a correlation matrix for ten brain potentials and seven CBCL/6-18 DSM-Oriented Scales.
Age at Onset of OCD Symptoms and Event-Related Potential Data in Patients with OCD
A multiple linear regression analysis was done with the OCD cases to examine the relation of age, history of tic disorder, ERN, and ΔPe to age at OCD symptom onset. Age at OCD symptom onset had significant associations with age and history of tic disorder in the full model (Table 6). Backward stepwise regression analysis found that tic disorder, age, and ΔPe with significantly associated with age at OCD symptom onset in the reduced model. The ΔPe had a significant positive correlation with age at OCD symptom onset in NTR OCD cases (r = 0.25, P = 0.02) but not TR OCD cases (P = 0.62). ERN amplitude had no significant correlations with age at OCD symptom onset in either TR OCD or NTR OCD cases (both P values > 0.19).