Individuals whose scores were indicative of definite ADHD were excluded from further analyses. This resulted in four out of 18 (22.2%) of the eligible cocaine-dependent individuals and one of 16 (6.25%) of the healthy controls being excluded. The final sample consisted of 29 participants (namely,14 cocaine-dependent patients and 15 healthy controls). provides demographic characteristics of the cocaine-dependent sample (7 males and 7 females; a mean age of 39.00 years) and the healthy control participants (7 males and 8 females; a mean age of 34.27 years).
Demographic characteristics of the subject groups
The groups did not differ significantly in terms of their proportion of males, χ(1) = 0.03, n.s. The two groups did not differ significantly in terms of age at the time of testing, t(27) = 1.98, n.s. The groups did not differ significantly in terms of ethnicity, χ(3)= 6.57, n.s. Subjects were considered smokers if they reported smoking either regularly or socially. The patient group was more likely to include smokers, χ(3)= 12.71, p< .01. The groups were also compared in terms of their scores on the WURS. They did not differ in terms of their mean scores on the scale, t(27) = 1.51, n.s.
3.2 Cocaine use
The mean age when these patients first used cocaine was 20 (range, 13 to 31) years. Their average age when they began regular use of cocaine was 23 (range, 14 to 35) years. Mean duration of use was 19 (range, 1 to 25) years. The CEQ was administered to the cocaine-dependent patients. Although none of the patients reported experiencing psychotic symptoms prior to their illicit drug use, most (71%, or 10 of 14) of the cocaine-dependent individuals reported experiencing paranoid symptoms during their use.
3.3 Electrophysiological measures of attention and information processing
Grand averaged ERPs for the cocaine dependent patients and the healthy controls are shown in . The amplitudes and latencies are summarized in . Analyses revealed a significant group difference between cocaine-dependent subjects and healthy controls in P300 amplitude, with the cocaine-dependent subjects having reduced amplitude compared to the controls, t(27) = 4.04, p < .001. Cocaine dependence resulted in an average decrease of 63% compared to controls. However, the difference between groups in terms of P300 latency failed to reach statistical significance, t(21) = 2.06, n.s. To determine whether there was an association between duration of cocaine use and the P300 indices, we computed Pearson product-moment correlations. There was no significant relationship between duration of cocaine use and either P300 amplitude or latency, r's = −.32 and −.04, n.s., respectively.
Grand Averaged Response to the Infrequent Stimuli
Attention and Information Processing in Cocaine-Dependent Subjects and Controls
3.4 Continuous Performance Test
We examined the association between patients' duration of cocaine use and performance on the CPT. Duration of cocaine was not significantly associated with any of the CPT performance indices, r's ranged from −.28 to .24, n.s. provides the means and standard deviations on the CPT-IP indices for both groups. Analysis of the discrimination sensitivity (d') data revealed that there was a main effect of group, F(1, 27) = 10.33, p < 0.01. However, there was no effect of stimulus type (digits vs. figures), F(1,27) = 0.17, n.s., nor was there a group by stimulus interaction, F(1,27) =0.84, n.s.
In terms of the groups' response criteria (ln b), the between-factor analysis failed to yield a significant main effect of group, F(1,27) =0.91, n.s. Within-subject analysis did not reveal a significant effect of either stimulus type [F(1,27) =0.00, n.s.] or stimulus × group interaction [F(1,27) = 0.26, n.s.]. Taken together, these findings indicate that the two groups differed in terms of their task performance, with the patient group having a significantly lower mean d' than the controls. However, the two groups did not differ in terms of their response criterion.
The latency for correct responses (hits) was also analyzed. We observed a significant effect of stimulus, F (1,27) = 5.60, p < .05, whereby participants displayed a significantly longer reaction time on the digits condition relative to the shapes condition. However, the two groups did not differ in terms of their latency for correct hits, F(1,27) =1.88, n.s., nor was there a significant group × stimulus type effect, F(1,27) = 0.38, n.s. The cocaine-dependent group made more commission errors than the control group, F(1,27) = 8.53, p < .01. However, there was no effect of stimulus type [F(1,27) = 0.26, n.s.], nor was there a significant group by stimulus interaction [F(1,27) = 2.36, n.s.].
Finally, we compared the two groups on an index of impulsivity, operationally defined as the ratio of false alarms to correct detections. There was a significant group effect, F(1,27) = 9.57, p < .01, though there was no effect of stimulus type [F(1,27) = 0.01, n.s.]. The group × stimulus interaction failed to reach significance, F(1,27) = 3.37, n.s.
3.5 Associations between the electrophysiological measures and behavioral and clinical measures
With the exception of RTs for correct responses, there were no significant differences between the CPT-digits condition and the CPT-shapes condition. Thus, for the remainder of the analyses, the data for the two stimulus types were combined. In the entire sample, we observed a significant association between P300 amplitude and discrimination sensitivity (d'), r = 0.43, p = .019. Overall, participants with larger P300 amplitudes displayed better CPT performance, as measured by d'. Behavioral impulsivity, operationalized as the ratio of false alarms to hits, was inversely related to P300 amplitude, r = −0.41, p=.026. Contrary to expectations, P300 latency was not significantly associated with reaction time to correct detections, r= 0.11, n.s. In fact, none of the correlations between P300 latency and the CPT performance indices were significant.
Scores on the Wender-Utah Rating Scale were not significantly correlated with P300 amplitude for the entire sample, healthy controls, or cocaine-dependent patients (r = −.20, .07, and −.19, n.s., respectively). Similarly, the WURS scores were not significantly associated with P300 latency for the groups separately or pooled (r's ranged from −.35 to −.18, n.s.). We observed a trend, whereby WURS scores were negatively associated with the number of correct responses (hits) on the CPT, r = −.36, p = .056. None of the other correlations between WURS scores and CPT performance indices approached statistical significance.
3.6 Cocaine-Induced Psychosis
All of the cocaine-dependent patients who experienced cocaine-induced psychosis had no history of psychosis prior to their drug use. The patients reported a varying amount of time that elapsed before paranoia set in (within an episode of drug use), ranging from less than 5 minutes to more than 120 minutes following use. The CEQ total score was significantly correlated with scores on the distress subscale, r = .79, p < .01. The mean score on the CEQ paranoia distress subscale was 4.2 (SD = 1.55), with half (5 of 10) of the patients with cocaine-induced paranoia reporting frankly psychotic experiences. The patients displayed a considerable range of behaviors associated with their paranoid ideation, ranging from simple fears to needing to attack others in perceived albeit delusional need for self defense (mean behavioral severity subscale rating = 2.60, SD = 1.3). CEQ insight scores ranged from 1 to 3 (mean = 1.70, SD =.68), with 60% of the patients indicating a lack of full insight regarding their paranoia. Scores on the insight subscale were not correlated to CEQ total score, r = .40, n.s.
Among the cocaine-dependent patients who experienced psychosis, P300 amplitude was inversely related to CEQ paranoid distress scores, r = −.66, p < .05. Thus, the greater the self-reported distress during cocaine-induced paranoia, the smaller the patients' P300 amplitudes. Similarly, CEQ total scores were inversely related to P300 amplitude, r = −.62, p=.05. No other significant associations between the CEQ scores and any of the behavioral measures were significant.