Cognitive deficits predict low treatment retention in cocaine dependent patients

Cognitive deficits predict low treatment retention in cocaine dependent patients

Drug and Alcohol Dependence 81 (2006) 313–322 Cognitive deficits predict low treatment retention in cocaine dependent patients Efrat Aharonovich a,b,...

148KB Sizes 2 Downloads 63 Views

Drug and Alcohol Dependence 81 (2006) 313–322

Cognitive deficits predict low treatment retention in cocaine dependent patients Efrat Aharonovich a,b,∗ , Deborah S. Hasin a,b,c , Adam C. Brooks b , Xinhua Liu c , Adam Bisaga a,b , Edward V. Nunes a,b a

College of Physicians and Surgeons, Department of Psychiatry, Columbia University, Division on Substance Abuse, 1051 Riverside Drive, Unit 120, New York, NY 10032, USA b New York State Psychiatric Institute, New York, NY 10032, USA c Mailman School of Public Health, Columbia University, New York, NY 10032, USA Received 25 February 2005; received in revised form 1 August 2005; accepted 9 August 2005

Abstract Previously, we found that impaired cognition predicted treatment dropout from cognitive behavioral therapy (CBT) in a small sample of cocaine dependent patients. To further address the role of impaired cognition in retention and treatment outcome of cocaine-dependent patients in CBT, we expanded a previous investigation to a larger sample, added depressed cocaine patients, and added an additional cognitive assessment. Fifty-six cocaine dependent patients receiving CBT in outpatient clinical trials were assessed for cognitive performance at treatment entry with the computerized MicroCog (MC) and the Wisconsin Card Sort Test (WCST). Treatment completion was defined as 12 or more weeks. Treatment dropouts had significantly lower MC scores (poorer cognitive functioning) than completers on attention, memory, spatial ability, speed, accuracy, global functioning, and cognitive proficiency, with effect sizes in the moderate to large range. These findings were not affected by depression, demographics (age, gender, race, sex, marital status) or drug use (years of cocaine use or average weekly cocaine expenditure in the prior 30 days). In contrast, patients’ performance on the WCST was in the average or near-average range, and WCST scores did not differentiate between completers and dropouts. Consistent with previous research, results suggest that mild cognitive impairments (≤1 S.D. below the mean) negatively affect retention in outpatient CBT treatment for cocaine dependence. Future studies should examine whether there are specific effects of different executive functioning abilities on treatment outcome. Modified behavioral and pharmacologic interventions should be considered to target mild cognitive impairments to improve substance treatment outcome. © 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Cocaine; Treatment; Cognitive deficits; Retention

1. Introduction Dropout from treatment has long been identified as one of the major obstacles in treatment of substance dependence in general (Simpson et al., 1997) and cocaine dependence in particular (Agosti et al., 1991, 1996; Carroll, 1999; Mulvaney et al., 1999; Sayre et al., 2002; Alterman et al., 1996). Early dropout is associated with continued substance use and relapse (Simpson et al., 1999). Attrition also poses problems for clinical research. Attrition rates in clinical ∗

Corresponding author. Tel.: +1 212 740 7257; fax: +1 212 568 3832. E-mail address: [email protected] (E. Aharonovich).

0376-8716/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.drugalcdep.2005.08.003

trials of treatment for cocaine dependence are high, often in the 75% range (Sayre et al., 2002). If there is differential attrition, such high rates may bias results of clinical trials. High attrition can also diminish the ability to detect differences between treatment groups, since dropouts may not receive adequate exposure to study treatment (Carroll, 1997; Nunes, 1997; Claus et al., 2002). Therefore, there are both important clinical and research motivations for identifying factors contributing to low retention of substance abusers in treatment. Over the past decade, research has emerged showing that a substantial number of substance abusers suffer from both detectable damage in cortical and sub-cortical brain regions and impairments across cognitive domains (Vik et al., 2004;

314

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322

Rogers and Robbins, 2001). In chronic cocaine abusers, cognitive domains, such as attention, memory, decision-making, and problem solving are often impaired (Cunha et al., 2004; Tucker et al., 2004; Ardila et al., 1991; Beatty et al., 1995; Bolla et al., 1998, 2003). However, despite abundant empirical evidence on the existence of cognitive deficits in substance abusers, the clinical implications of these impairments on drug treatment outcome have received little attention (Bolla et al., 1998, 2003). To date, most studies of cognitive functioning and treatment outcome have focused mainly on patients with primary or co-existing alcohol use disorders. In such patients, cognitive impairments were found to impede acquisition of new coping behaviors (McCrady and Smith, 1986) or learning and retention of new material (Alterman and Hall, 1989), to disrupt therapeutic interventions (Fals-Stewart, 1993), and to increase likelihood of treatment dropout (FalsStewart and Schafer, 1992; Teichner et al., 2002). However, in one study, impaired executive functioning did not seem to negatively affect treatment outcomes in alcohol patients participating in a 12-step facilitation program (Morgenstern and Bates, 1999). For cocaine dependence, cognitive behavioral relapse prevention therapy (CBT-RP) has been one of the most successful treatments (Carroll et al., 1994). Adequate cognitive functioning is required for many of the activities in psychosocial therapies (Gottschalk et al., 2001), including CBT-RP. It is thus reasonable to assume that cognitively impaired substance abusers participating in a cognitively oriented treatment, such as CBT-RP may encounter difficulties in treatment, since the therapy requires the verbal, memory and learning skills that are often impaired in this population. To date, only two studies with cocaine dependent patients have investigated the clinical association between cognitive impairments and CBT-RP outcome. In a case report on two female cocaine-dependent patients being treated in a residential facility, improvement in cerebral perfusion imaging was associated with improved response to cognitive behavioral treatment (Gottschalk et al., 2001). In an earlier brief report of 18 non-depressed cocaine dependent patients, we showed that a significant majority of cocaine dependent patients (>50%) who dropped out in the early phase of outpatient CBT-RP had performed poorly on neuropsychological measures at treatment entry (Aharonovich et al., 2003). While the earlier report on 18 patients served an important function by providing the first empirical evidence on the research question in cocaine dependent patients, the limited scope of this first study clearly left several important questions unanswered. These included the following: (1) would depression affect the results? Depression exerts adverse effects on memory (Zakzanis et al., 1998), attention, executive functioning (Porter et al., 2003), and other aspects of cognitive functioning (Landro et al., 2001). Many cocaine dependent patients enter treatment with depressive comorbidity, yet the original sample excluded patients with DSM-IV major depression or dysthymia. Therefore, it was important to determine whether an Axis I depressive

disorder intensified the previously found relationship between cognitive impairment and treatment retention. (2) We wished to determine the relative utility of the Wisconsin Card Sort Test (WCST), an indicator of executive functioning related to choices and mental flexibility (Heaton et al., 1999) in predicting treatment outcome. The WCST has been used to examine executive functioning in substance (including cocaine and alcohol) abusers (Grant et al., 2000; Bolla et al., 1999; Goldstein et al., 2004) and opiate-dependent tobacco smokers (Rotheram-Fuller et al., 2004). In one prior study of 12-step facilitation, alcoholics (many of whom were mixed substance abusers), the WCST did not predict treatment outcome (days drinking and/or drug use; Morgenstern and Bates, 1999). However, for the purposes of understanding the treatment of cocaine dependence, the mixed nature of the patients’ substance problems in Morgenstern and Bates (1999) did not provide information on whether the WCST would prove informative about treatment outcome in a “cleaner” sample of cocaine dependent patients who were not concurrently dependent on other substances. (3) Because age, gender, education, race, and level of baseline cocaine use may all potentially impact treatment retention and outcome, controlling for the potential confounding effects of these variables is important when addressing the relationship of cognitive impairment to treatment outcome. However, the sample in Aharonovich et al. (2003) was too small for this to be accomplished. Thus, it was important when examining the research question to use a larger sample to rule out the potential confounding effects of these characteristics. Accordingly, we expanded the earlier sample from 18 to 56, allowing us to examine these questions as well as the stability of our initial results. While we added the WCST, the MicroCog (MC) (Powell et al., 1993) a computerized neuropsychological assessment battery, remained the main measure of attention, memory, and abstract reasoning. These cognitive domains were hypothesized to be relevant to successful treatment outcome. The following principle questions were addressed: (1) is level of cognitive functioning at treatment entry predictive of retention? (2) Which specific cognitive domains are most predictive of retention? and (3) is Axis I depression (DSM-IV major depression or dysthymia) a moderator of the cognitive functioning-treatment retention relationship? As a secondary goal, we also examined the relationship of cognitive functioning at baseline to abstinence during the CBT-RP treatment for cocaine dependence.

2. Method 2.1. Participants A total of 56 treatment seeking cocaine dependent patients were included in the study. These were consecutively enrolled patients in placebo-controlled clinical trials of CBT-RP + medication. Of the patients, 33 were not depressed and participated in a trial of CBT + gabapentin, and 23 were

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322

depressed and participated in a trial of CBT + anti-depressant medication (venlafaxine). All patients approached about participation in this study agreed to participate. After signing a screening consent, participants were administered a comprehensive psychiatric evaluation that included a Structured Clinical Interview for DSM-IV (SCID, DSM-IV Axis I/P version; First et al., 1998). The SCID was administered by trained masters- or doctoral-level clinical psychologists, and yielded diagnoses of major depression and dysthymia, as well as other disorders. All participants also received a physical examination. At the end of the screening process, eligible participants signed a consent form for participation in this sub-study of neurocognitive functioning, following study procedures approved by the New York State Psychiatric Institute Institutional Review Board. Participants were paid $20.00 for completion of the neuropsychological battery. 2.2. Inclusion/exclusion criteria Participants met the following inclusion criteria: (1) DSMIV criteria for cocaine dependence; (2) report of cocaine use at least four times in the month prior to admission and duration of cocaine use for a period of at least 12 months preceding enrolment into the study; (3) DSM-IV criteria for major depression or dysthymia (venlafaxine trial only). Exclusion criteria included: (1) met DSM-IV criteria for bipolar disorder, schizophrenia, ADHD or any psychotic disorder; (2) met DSM-IV criteria for another substance abuse or dependence diagnosis (other than nicotine); (3) history of seizure disorders or head injury with loss of consciousness exceeding 1 h. 2.3. Sample characteristics Age ranged from 23 to 54 with a mean of 37.71 (S.D. = 6.61) years. Years of education ranged from 8 to 20 with a mean of 13.32 (S.D. = 2.25). Among the patients, 76.7% (43/56) were male, 46.4% (26/56) were African American, 53.6% (30/56) were Caucasian, 17.8% (9/56) were married, and 86% (48/56) were right handed. The sample includes the 18 non-depressed patients that were included in the previous report (Aharonovich et al., 2003).

315

2.4.2. Determination of drug use To rule out the neuropsychological testing results being affected by acute intoxication, observed urine samples were taken within 24 h of the administration of the neuropsychological battery to verify participants’ self reported recent drug abstinence. Only one participant reported recent cocaine use <7 h prior to testing. The participant was asked to return for the neuropsychological testing the following day. During the course of the CBT-RP treatment, participants gave observed urine samples. Samples were collected three times a week in the non-depressed group and twice a week in the depressed cocaine group. Urine samples were quantitatively analyzed for cocaine metabolites. Level of baseline cocaine use, a control variable in some analyses, was represented by the average weekly dollar expenditure in the 30 days prior to entering treatment. 2.4.3. Administration of the computerized neuropsychological battery Participants were tested individually in a session lasting approximately 60–70 min. The battery was administered at treatment entry. The neuropsychological battery included the computerized MicroCog and the computerized research edition of the Wisconsin Card Sorting Test (WCST). Computerized administration of the battery eliminated examiner effects and increased reliability. In order to investigate the association between cognitive abilities at treatment entry and retention and outcome and in CBT-RP, all participants received at least ‘one dose’ of CBT-RP, and the computerized battery was administered before the second CBT-RP session. The battery was presented on a 15 color PC monitor. Response buttons were assigned keys on a standard PC keyboard, which was placed in front of the participant. The order of presentation between the MC and WCST was counterbalanced between participants using an ABBA design.

2.4. Procedures

2.4.4. MicroCog: assessment of cognitive functioning This computerized MicroCog battery assesses cognitive domains in the areas of attention, visual–verbal memory, visuospatial facility, reasoning, and mental calculation (Powell et al., 1993). The battery takes 40–55 min to complete. All instructions and testing materials were presented on the computer screen. The following were the subtests administered under each domain.

2.4.1. Cognitive behavioral therapy relapse prevention All participants received manualized CBT sessions focusing on relapse prevention. The manualized CBT-RP was based on Marlatt and Gordon, 1985 relapse prevention and adapted with slight variations from the NIDA CBT-RP manual (Carroll, 1998). Patients attended CBT-RP sessions once a week for 12 weeks. Trained masters- or doctoral-level clinical psychologists conducted the CBT-RP sessions. CBT-RP therapists participated in weekly group supervision to assure compliance and adherence to the manual.

2.4.4.1. Sustained-focused attention. (1) The alphabet task is a continuous performance task where letters of the alphabet are presented in a random order. Participants were asked to select the letters in sequential alphabetical order. (2) A list of 16 words containing groups of category-related items is presented four times. Each time, the participant is instructed to respond to words that are members of one of the four given categories. (3) A list of 36 words including the 16 words from the above list is presented, and the participant must indicate which words were previously shown.

316

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322

2.4.4.2. Visual–verbal memory. (1) Delayed recognition: a name and address are presented for participants to memorize for later recognition (∼20 min later). (2) Immediate and delayed recognition: two brief stories presented, immediately followed by six questions requiring multiple-choice recognition of several items from the story. Multiple-choice questions that tested delayed recognition of each story’s content were presented ∼20–30 min later. (3) Immediate memory was assessed using numbers-forward and numbersreversed tasks. Immediately after the presentation, participants pressed the numbers, in either same or reverse order of their presentation. 2.4.4.3. Visuospatial ability. (1) Tic-tac task: a 3 × 3 block matrix is presented and 3–5 squares are briefly illuminated. Immediately after the presentation, participants use the numbered keypad to indicate which blocks contain the illuminated squares. (2) Clock faces with the hands indicating the time, but without associated hour markings, are presented. On each trial, four digital choices are offered (i.e., “09:00”; “05:30”) from which participants must choose the correct time. 2.4.4.4. Mental reasoning and conceptual flexibility. (1) Incomplete verbal analogies are presented and participants are required to complete them by selecting among three multiple-choice alternatives. (2) An object-match task requires abstraction ability and conceptual flexibility. 2.4.4.5. Tests of mental calculation. Participants are asked to complete a math task including basic simple tasks of arithmetic (addition, subtraction, multiplication, and division) with varying degrees of difficulty. 2.4.5. Wisconsin card sort test Computer Version-3 Research Edition (Heaton, 1999): This ∼20-min non-verbal test is used to assess executive functioning. The test requires cognitive flexibility demonstrating the ability to use environmental feedback to shift responses. The WCST requires participants to sort a total of 128 cards (two sets of 64) on the basis of one of three principles: color, form, or number of symbols displayed on each card. After correctly sorting 10 cards consecutively according to one principle, the sorting principle is changed without informing the participant, who needs to reconfigure the cards according to the new sorting principle. 2.5. Main outcome measures 2.5.1. Treatment retention This variable was measured in two ways. We examined it as a binary variable defined as remaining in treatment for at least 12 weeks of CBT-RP treatment without missing two or more consecutive weeks. All others were classified as dropouts. We used 12 weeks as the cut-off because this is the standard length of CBT treatment (Carroll, 1998) and also because national treatment data from the DATOS study

indicates that outcomes for medium-to-high problem cocaine patients are significantly better when they are treated for at least 3 months (∼12 weeks; Simpson et al., 1999). Secondly, we totaled the number of weeks patients remained in treatment and treated it as a continuous variable. 2.5.2. Drug use/abstinence To define urine tests as positive or negative for cocaine, we used the standard NIDA cutoff for cocaine positive results (≤300 ng/ml). Missed urine samples were assumed to represent use and coded as positive. For analysis, we used the proportion of cocaine-negative urines as an outcome, defined as the number of negative urines divided by total number of collected urine samples. 2.6. Independent variables 2.6.1. MicroCog scores The MC yields three levels of scores. Level 1 consists of an index for the main cognitive domains and the subtests. Level 2 indicates overall accuracy and overall speed (two scores). Level 3 consists of composite scores of global cognitive functioning (GCF), consisting of combined accuracy and speed weighted equally, and global cognitive proficiency (GCP), consisting of combined accuracy and speed, with preferential weight to accuracy. The population mean score for each cognitive domain and composite is 100, with 15 equal to one standard deviation (1 S.D.). For the subtests, the mean score is 10 and 1 S.D. is 1.5. 2.6.2. The WCST The set of four WCST scores examined were number of perseverative errors, perseverative responses, categories completed, and failure to maintain set. We chose to report this set of four scores as they are indicative of abilities to sustain attention, distractibility and cognitive flexibility and are widely used across studies. 2.7. Data analysis Preliminary analyses included t-tests and chi-square tests to check whether there were significant differences between the depressed and non-depressed patients on baseline demographic and drug use variables, baseline cognitive testing data and retention in treatment. Preliminary chi-square tests were also conducted to determine whether there were differences in treatment completion between medication and placebo patients in both the depressed and non-depressed groups. Between-group differences (completers versus dropouts) on cognitive subscales were tested using t-tests. Cohen’s d was computed to provide information on effect size; d’s of 0.3, 0.5, and 0.8 reflect small, medium, and large effect sizes, respectively. Correlations were used to indicate bivariate relationships between cognitive tests and weeks in treatment. Logistic regression was used to examine the relationship between cognitive test scores and dropout controlling for

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322

demographics (age, gender, race, marital status) and drug use (years of cocaine use, level of baseline cocaine use, indicated as average weekly cocaine expenditure (in dollars) in prior 30 days) with treatment retention (completers versus dropouts) as the binary outcome variable. Linear regression was used to examine the relationship between the cognitive scores and the continuous outcome variables, number of weeks in treatment and proportion of cocaine-negative urine samples, transformed as necessary to improve the normality of their distributions. The linear regression analyses used the same control variables as the logistic regressions. The significance level for all analyses was set at 0.05 for two-tailed tests.

3. Results Depressed and non-depressed cocaine patients did not significantly differ in age, years of education, gender, race, marital status, years of drug use, route of cocaine administration and baseline drug use frequency (most p-values > 0.15), with only a trend towards a difference (t = 1.71, p = 0.09) on years of cocaine use (mean in depressed patients 13.3, S.D. = 6.4, mean in non-depressed patients 10.5 years, S.D. = 5.7). There were no differences between depressed and non-depressed patients on baseline cognitive tests (all p-values > 0.50) or on retention in treatment (χ2 (1) = 2.39, p = 0.12). Chi-square tests for medication/placebo differences in treatment completion among patients in both groups showed no significant medication effects on treatment completion for venlafaxine versus placebo in the depressed group (χ2 (1) = 0.98, p = 0.32), or gabapentin versus placebo in the non-depressed group (χ2 (1) = 1.38, p = 0.84). Due to this lack of evidence of differences between the depressed and non-depressed patients, we combined the patients in the analyses examining the relationship of cognitive functioning to treatment retention and outcome.

317

3.1. Treatment retention Among the 56 patients, 71.4% (40) dropped out of treatment as defined above. The mean number of weekly CBT sessions attended by the total sample was 5.96 (S.D. = 3.49). The completers attended a mean of 10.3 (S.D. = 1.13) sessions (completers could miss one but not two or more consecutive sessions) and dropouts attended a mean of 4.22 (S.D. = 2.4) sessions. Completers and dropouts did not differ significantly on age (t(54) = 0.27, p = 0.79), race (χ2 (1) = 0.004, p = 1.00), gender (χ2 (1) = 0.001, p = 1.00), marital status (χ2 (1) = 0.001, p = 1.00), years of cocaine use (t(54) = 0.53, p = 0.59), frequency/quantity of use represented by dollars spent on cocaine per week (t(54) = 1.05, p = 0.30). Dropouts had fewer years of education (M = 12.93, S.D. = 2.00) than the completers (M = 14.31, S.D. = 2.57; t(54) = 2.15, p = 0.04). Note that the cognitive tests adjusted for education, so education was not a confounder of the cognitive results. Mean MicroCog scores for completers and dropouts (Table 1) showed that dropouts had consistently lower scores (indicating poorer cognitive functioning) than completers, with dropout scores in most domains near or exceeding one standard deviation below the mean. T-tests revealed significant differences between dropouts and completers on MicroCog measures of attention, memory, spatial ability, speed, accuracy, global cognitive functioning and global proficiency, while mental reasoning demonstrated a trend towards significance (p = 0.076). Fig. 1 illustrates that as patients’ GCF increased, their odds of dropping out of treatment decreased. Values of Cohen’s d (Table 1) indicated effect sizes in the moderate to large range for each significant group difference in MicroCog scores. In contrast, performance on the WCST was at average or near average range, and completers and dropouts did not differ significantly on any of the WCST measures. To explore the relationship between cognitive functioning scores and a continuous outcome variable for treatment reten-

Table 1 Neuropsychological functioning and treatment completion: cocaine dependent patients in cognitive behavioral therapy-relapse prevention (N = 56) Completers (n = 16) mean score (S.D.)

Dropouts (n = 40) mean score (S.D.)

Cohen’s d

t(d.f.=54) , p-value

MicroCog Attention Mental reasoning Memory Spatial ability Speed Accuracy Global cognitive functioning Global cognitive proficiency

100.7 (9.4) 91.8 (15.3) 95.5 (19.3) 92.6 (12.5) 97.3 (17.6) 89.2 (14.8) 91.5 (15.4) 88.8 (14.0)

90.2 (12.4) 83.5 (15.4) 85.3 (13.5) 82.3 (14.9) 86.2 (13.0) 80.6 (13.8) 79.2 (12.4) 78.7 (11.4)

0.84 0.53 0.64 0.69 0.73 0.59 0.87 0.78

3.05, 0.003 1.81, 0.076 2.24, 0.030 2.45, 0.018 2.58, 0.013 2.05, 0.045 3.11, 0.003 2.81, 0.006

WCST Perseverative errors (SS)a Perseverative responses (SS)a Categories completed Failure to maintain set

104.1 (23.5) 102.5 (22.2) 5.50 (1.20) 0.31 (0.47)

96.8 (24.9) 96.9 (24.9) 4.90 (1.40) 0.30 (0.46)

0.30 0.23 0.42 0.02

1.01, 0.319 0.78, 0.438 1.48, 0.145 0.09, 0.928

Neuropsychological assessment

a WCST: the raw scores of perseverative errors and perseverative responses were transformed to standard scores (SS) with a mean of 100 and a standard deviation of 15. Higher scores reflect less preservative errors and responses.

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322

318

Fig. 1. Odds of attrition from cognitive behavioral therapy relative to global cognitive functioning as measured at baseline (N = 56).

tion, we computed correlations between number of weeks in treatment and MC scores. Increased weeks in treatment were significantly correlated with higher scores on GCF (Spearman r = 0.27, p = 0.048), attention (r = 0.28, p = 0.037), and spatial ability (r = 0.28, p = 0.040), and correlated at a trend level with abstract reasoning (r = 0.23 p = 0.082), and GCP (r = 0.25, p = 0.062). Next, the relationship of cognitive functioning scores (e.g., MC and WCST) to treatment dropout as a binary variable was examined using logistic regression, controlling for demographic characteristics and level of baseline cocaine use (Table 2). An odds ratio of 1.00 indicates no relationship, while odds ratios greater than 1.00 indicate positive relationships and odds ratios smaller than 1.00 indicate negative relationships. Confidence intervals that include 1.00 indicate a non-significant relationship. As shown, the odds ratios for MC scores in Table 2 were all smaller than 1.00, indicating

Table 2 Cognitive functioning at baseline (MicroCog and Wisconsin Card Sort Test scores) as predictor of treatment dropout in logistic regression analysisa Neuropsychological assessment

Odds ratio

95% CI

p-value

MicroCog Attention Mental reasoning Memory Spatial ability Speed Accuracy Global cognitive functioning Global cognitive proficiency

0.90 0.96 0.95 0.93 0.94 0.95 0.92 0.92

0.83–0.97 0.92–1.00 0.91–0.99 0.87–0.98 0.90–0.99 0.90–0.99 0.87–0.97 0.87–0.98

0.007 0.067 0.034 0.012 0.014 0.044 0.005 0.009

WCST Perseverative errors Perseverative responses Categories completed Failure to maintain set

0.98 0.99 0.65 1.14

0.96–1.01 0.96–1.01 0.37–1.15 0.30–4.27

0.213 0.348 0.141 0.850

Note: findings were similar when depression was omitted from the model. a All models controlled for age, gender, education, race, depression, and level of baseline cocaine use.

that lower scores predicted dropout. The significance levels were very similar in terms of significance levels compared to Table 1, indicating that the relationship of cognitive functioning to retention in treatment was robust to the effects of numerous potential confounders. Controlling for these variables also did not change the lack of relationship between treatment dropout and the WCST. Linear regression was applied to the transformed outcome variable of number of weeks in treatment, with main predictors consisting of cognitive functioning test scores, and control variables including the same variables used in the logistic regressions. In general, this analysis yielded similar results to the logistic regressions, although some results appeared attenuated. Two of the four MC cognitive domains predicted greater weeks in treatment: mental reasoning (β = 0.286, p = 0.039) and spatial ability (β = 0.344, p = 0.018), while attention showed a trend towards significance (β = 0.282, p = 0.056). Memory (β = 0.112, p = 0.391) did not retain significance. The MC composite scores also predicted number of weeks in treatment: GCF (β = 0.324, p = 0.024); and GCP (β = 0.306, p = 0.035). Processing accuracy showed a trend (β = 0.275, p = 0.064) towards predicting weeks in treatment, while processing speed (β = 0.218, p = 0.112) and reaction time (β = −0.085, p = 0.537) were not significant predictors. WCST scores were not significant predictors of weeks in treatment: number of perseverative errors (β = 0.112, p = 0.391), perseverative responses (β = 0.004, p = 0.977), categories completed (β = 0.114, p = 0.408), or failure to maintain set (β = 0.060, p = 0.675). 3.2. Cocaine use during treatment Linear regression was employed to determine if MC or WCST scores predicted proportion of cocaine-negative urine samples. The average proportion of cocaine-negative urine samples across all patients was 0.30 (S.D. = 0.29), with a range of 0–1. Linear regressions were computed, separately entering each MC score after controlling for age, education, gender, race, depression and baseline cocaine use. No control variables significantly predicted proportion of cocaine-negative urine samples (not shown). Further, none of the cognitive functioning scores on either the MC or the WCST predicted proportion of cocainenegative urines. These included attention (β = 0.12, t = 0.83, p = 0.411), memory (β = −0.12, t = −0.82, p = 0.416), mental reasoning (β = −0.09, t = −0.65, p = 0.522), spatial ability (β = 0.09, t = 0.59, p = 0.522), GCF (β = −0.16, t = −0.11, p = 0.913), GCP (β = −0.06, t = −0.37, p = 0.712), speed (β = 0.02, t = 0.11, p = 0.91), or accuracy (β = −0.05, t = −0.28, p = 0.781) from the MC. WCST variables were analyzed similarly. None of the WCST scores were related to urine-confirmed abstinence from cocaine, including perseverative responses (β = −0.16, t = −1.13, p = 0.265), perseverative errors (β = −0.14, t = −1.01, p = 0.318), failure-tomaintain set (β = −0.08, t = −0.59, p = 0.559), and completed categories (β = 0.04, t = 0.30, p = 0.762).

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322

4. Discussion The main finding of this study was that cognitive functioning as measured by the MicroCog computerized battery differed significantly between treatment completers and treatment dropouts. The patients who dropped out evidenced significantly poorer performance than completers across most cognitive domains, including attention, memory and spatial ability (all p’s < 0.01). Furthermore, dropouts had significantly poorer global cognitive performance and poorer global proficiency (speed and accuracy) than dropouts as reflected in the proficiency index score. The level of cognitive proficiency among the dropouts (approximately 1.5 standard deviations below the mean) suggests slow and inaccurate cognitive processing. Effect sizes were medium to large for the significant differences, ranging from 0.64 (memory) to 0.87 (global cognitive functioning). Results were largely consistent when weeks in treatment (a continuous variable) was used as the outcome, although findings were more consistently significant with our a priori selected cut-off of 12 weeks of treatment completed. The present study replicated and expanded previous findings on cognitive functioning and retention in non-depressed cocaine dependent outpatients (Aharonovich et al., 2003); this earlier study also found that various MicroCog subtests predicted retention in outpatient CBT treatment. The present study found that WCST scores did not differ significantly between treatment completers and dropouts. Moreover, cognitive flexibility assessed by the WCST was at or near normal average in completers as well as dropouts. This finding is consistent with several other studies showing that although many substance abusers have clear impairment on executive functioning tasks, such as the Iowa Gambling Task, their performance on the WCST is in the normal range (Grant et al., 2000; Bolla et al., 2003; Rotheram-Fuller et al., 2004). In Morgenstern and Bates (1999), few (5.4%) of the patients were impaired on the WCST (perseverative errors score). We speculate as to why the WCST does not appear to be informative in predicting treatment outcome in cocaine dependent patients, while components of the MC do significantly predict retention. The WCST has been shown to engage the dorsolateral region of the prefrontal cortex and to detect impairment in patients with lesions in this region (Milner, 1963; Robinson et al., 1980; Anderson et al., 1991; Berman et al., 1995). The WCST detects gross impairment in other severely disturbed populations, such as patients with schizophrenia (e.g., Goldberg and Weinberger, 1988). However, the specificity and the underlying cognitive processes tapped by the WCST remain debatable (Rogers et al., 1998). On average, cocaine dependent patients do not exhibit cognitive impairments as severe as those of patients with schizophrenia or dorsolateral lesions (Goldstein et al., 2004). Bearing this in mind, we speculate that the tasks on the WCST may not be complex enough, hence, sensitive enough, to tap into deficits in executive functioning that while less severe,

319

nevertheless affect treatment outcome. This suggestion is supported by our findings that in contrast to adequate performance on the WCST, dropouts showed poor performance on the abstract reasoning domain of the MC, more specifically, on the Object Match subtests (A and B). The Object Match subtests also elicit cognitive flexibility by assessing ability to abstract common stimuli features and to shift from one concept to another, but the stimuli sets change constantly in the Object Match subtests, in contrast to the WCST. On Object Match subtest A, the task may be more obvious, e.g., identifying one object whose color differs from the others. On Object Match subtest B, the tasks are more difficult. In our study, the dropouts performed poorly on both levels (Object A and B) as indicated by scores >1 S.D. below the mean. However, while dropouts and completers did not differ significantly on Object Match subtest A (mean among completers 9.18 S.D. = 3.6; among dropouts 8.05 S.D. = 3.03; p = 0.24), they did differ significantly on Object B (mean among completers 9.87; dropout 7.5; p = 0.004). (standardized means for these tests are 10, with S.D.s of 1.5). Additional comparisons of specific executive abilities with varying complexity are needed in prospective treatment trials of cocaine dependent patients to better understand these issues. In this study, attention had the strongest relationship to treatment retention in CBT-RP as indicated by both effect size and significance level. This finding has implications for treatment modification. Note that attention is not a unitary process and at least four attentional components can be distinguished from each other: focused attention; sustained attention; divided attention and alternating/shifting attention (Lezak et al., 2004). The subtests included in the MicroCog are generally considered to measure aspects of focused (i.e., concentration) and sustained attention. The present findings suggest that patients who have difficulty focusing and sustaining attention are less likely to remain in a 12-week course of treatment. As retention in treatment for at least 12 weeks has been linked to significantly decreased relapse rates (Simpson et al., 1999), this relationship between impaired attention and treatment attrition is noteworthy. Ability to attend has consistently been demonstrated to be a strong predictor of treatment outcome in different modalities and substances of abuse. Teichner et al. (2001) found that attention was the best neuropsychological predictor of attainment of treatment objectives in a sample of 85 substance abusers participating in both inpatient and outpatient treatment. They concluded that the ability to maintain attention is a prerequisite to encoding and processing of information and thus, is vital and crucial for successful treatment outcome. The confirmation of attention as an important predictor raises an important clinical question for treatment providers: how much of the intervention (session content) presented in a 45–60 min session is actually processed by patients with impaired focused-sustained attention? We speculate that these patients may encounter difficulties attending to the therapist in such long sessions; therefore, it is not surprising that

320

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322

these patients are more likely to drop out of treatment or have poor outcome. When attention is impaired, the patient’s overall cognitive productivity suffers and consequently other cognitive functions, such as memory and learning can be reduced (Lezak et al., 2004). We can then further speculate how reduction in memory and learning abilities due to impaired attention will hamper capabilities to acquire new coping strategies and to prevent relapse. We believe that future systematic explorations into the different aspects of attention and their relations to treatment outcome are warranted. The findings on attention and memory suggest some possible treatment modifications. To aid focused/sustained attention and memory, sessions may be shorter and more frequent. For example, instead of weekly 60-min sessions, 30-min biweekly sessions may be more suitable. Clinicians may wish to consider use of rehabilitation strategies to accommodate memory difficulties of encoding, retrieving and executing the intervention material. Within sessions, multi-modal presentations, frequent rehearsals and feedback with the patient may be applied. In future treatment development studies, medication known to improve attention and memory in some populations could be investigated (Coull and Sahakian, 2000) to determine if this improves the outcome of psychological interventions with cognitively impaired drug patients. Despite the consistency of findings on the existence of cognitive impairments in substance abuse patients and their potential effect on treatment outcome, this issue has received little attention in the clinical literature or in clinical training (Bolla et al., 1998, 2003). One possible explanation is that these patients retain verbal articulation even in the presence of other cognitive impairments. Good verbal articulation can obscure other impairments (Fals-Stewart, 1997; Weinstein and Shaffer, 1993). The therapist then attributes the resulting behaviors (e.g., irregular attendance, non-compliance with homework assignments, etc.) to other causes, depending on the therapist’s own theoretical framework: resistance to treatment, ambivalence, or pre-contemplation stage. If further research confirms that cognitive impairment affects treatment retention or outcome when controlling for some of these other causes (e.g., motivation), then provision of a cognitive screening measure and professional education about the effects of cognitive impairment on behavioral treatment will become important priorities in the dissemination of relevant research findings to clinical providers. Limitations of the study are noted. Since a post-treatment follow-up was not administered, changes in functioning and their relationship to retention are unknown. Future studies should include, such assessments. Further, CBT effects on drug use have been shown to increase over time after the completion of treatment (Carroll et al., 1994; Farabee et al., 2002; Rawson et al., 2002; Epstein et al., 2003). Thus, a longerterm follow-up might reveal a relationship between cognitive functioning and drug use, which was not found in this study over the course of 12 weeks of treatment. Finally, this study does not address the causes of the cognitive impairments,

including whether they were a predisposing factor or a result of substance toxicity. While such knowledge could further inform the development of tailored treatment modifications, a different study design would be needed to investigate this question. These questions all indicate that the topic of cognitive functioning and its effects on treatment retention and outcome in substance abusing patients has not yet been studied extensively, and the issues warrant further investigation using a variety of approaches. The present study had several important strengths: the neuropsychological battery included the assessment of distinct cognitive domains found to be relevant for successful treatment outcome (attention, memory and executive functioning). The ABBA design precluded order effects between the WCST and the MicroCog. The cocaine dependent patients were free of other drug dependence and none were acutely intoxicated at the time of testing, confirmed with baseline urine toxicology, self-report and clinical observation during the testing, making interpretation of the results clearer. The study examined the relationship between cognitive capabilities at treatment entry with cognitive-behavioral treatment outcome in a sample unselected for cognitive functioning. As these conditions are similar to those in which most patients enter outpatient drug treatment, this expands our understanding of a common phenomenon in substance abuse treatment that deserves wider attention. In summary, general level of cognitive functioning and specific cognitive deficits at treatment entry predicted retention in outpatient CBT treatment for cocaine dependent patients. Knowledge of the presence of cognitive impairments at treatment entry may help tailor psychological and pharmacological interventions for cognitively impaired patients, and further investigation of this area should be an important priority in substance abuse treatment research.

Acknowledgments Support for this study is acknowledged from the National Institutes of Health grants K23 DA016743 (Dr. Aharonovich), PI 50 DA09236 (Dr. Kleber), K05 AA014223 (Dr. Hasin) and K02 DA000288 (Dr. Nunes). The authors thank Drs. Raby and Rothenberg and the staff at STARS. We also thank I. Huang for technical assistance and V. Richmond for manuscript preparation. We also acknowledge the helpful comments of Karen Bolla, Ph.D., on an earlier version of this manuscript.

References Agosti, V., Nunes, E., Stewart, J.W., Quitkin, F.M., 1991. Patient factors leading to early attrition from an outpatient cocaine research clinic: a preliminary report. Int. J. Addict. 26, 327–334. Agosti, V., Nunes, E., Ocepeck-Welikson, K., 1996. Patient factors related to early attrition from an outpatient cocaine research clinic. Am. J. Drug Alcohol Abuse 22, 29–39.

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322 Aharonovich, E., Nunes, E., Hasin, D., 2003. Cognitive impairment, retention and abstinence among cocaine abusers in cognitive-behavioral treatment. Drug Alcohol Depend. 71, 207–211. Alterman, A.I., Hall, J.G., 1989. Effects of social drinking and familial alcoholism risk of cognitive functioning: null findings. Alcohol Clin. Exp. Res. 13, 799–803. Alterman, A.I., McKay, J.R., Mulvaney, F.D., McLellan, A.T., 1996. Prediction of attrition from day hospital treatment in lower socioeconomic cocaine dependent men. Drug Alcohol Depend. 40, 227– 233. Anderson, S.W., Damasio, H., Jones, R.D., Tranel, D., 1991. Wisconsin Card Sorting Test performance as a measure of frontal lobe damage. J. Clin. Exp. Neuropsychol. 13, 909–922. Ardila, A., Rosselli, M., Strumwasser, S., 1991. Neuropsychological deficits in chronic cocaine abusers. Int. J. Neurosci. 57, 73–79. Beatty, W.W., Katzung, V.M., Moreland, V.J., Nixon, S.J., 1995. Neuropsychological performance of recently abstinent alcoholics and cocaine abusers. Drug Alcohol Depend. 37, 247–253. Berman, K.F., Ostrem, J.L., Randolph, C., Gold, J., Goldberg, T.E., Copola, R., Carson, R.E., Herscovitch, P., Weinberger, D.R., 1995. Physiological activation of a cortical network during performance of the Wisconsin Card Sort Test: a positron emission tomography study. Neuropsychologia 33, 1027–1046. Bolla, K.I., Cadet, J.L., London, E.D., 1998. The neuropsychiatry of chronic cocaine abuse. J. Neuropsychiatry Clin. Neurosci. 10, 280–289. Bolla, K.I., Rothman, R., Cadet, J.L., 1999. Dose-related neurobehavioral effects of chronic cocaine use. J. Neuropsychiatry Clin. Neurosci. 11, 361–369. Bolla, K.I., Eldreth, D.A., London, E.D., Kiehl, K.A., Mouratidis, M., Contoreggi, C., Matochik, J.A., Kurian, V., Cadet, J.L., Kimes, A.S., Funderburk, F.R., Ernst, M., 2003. Orbitofrontal cortex dysfunction in abstinent cocaine abusers performing a decision-making task. Neuroimage 19, 1085–1094. Carroll, K.M., Rounsaville, B.J., Gordon, L.T., Nich, C., Jatlow, P., Bisighini, R.M., Gawin, F.H., 1994. Psychotherapy and pharmacotherapy for ambulatory cocaine abusers. Arch. Gen. Psychiat. 53, 177–187. Carroll, K.M., 1997. Integrating Psychotherapy and pharmacotherapy to improve drug abuse outcomes. Addict. Behav. 22, 233–245. Carroll, K.M., 1998. A cognitive-behavioral approach: treating cocaine addiction. In: Therapy Manuals for Drug Addiction. National Institute of Drug Abuse, Rockville, MD. Carroll, K.M., 1999. Old psychotherapies for cocaine dependence revised. Arch. Gen. Psychiat. 56, 505–506. Claus, R.E., Kindleberger, L.R., Dugan, M.C., 2002. Predictors of attrition in a longitudinal study of substance abusers. J. Psychoactive Drugs 34, 69–74. Coull, J.T., Sahakian, B.J., 2000. Psychopharmocology of memory. In: Berrios, G.E., Hodges, J.R. (Eds.), Memory Disorders in Psychiatric Practice. University of Cambridge Press, New York, pp. 75– 98. Cunha, P.J., Nicastri, S., Gomes, L.P., Moino, R.M., Peluso, M.A., 2004. Neuropsychological impairments in crack cocaine-dependent inpatients: preliminary findings. Rev. Bras. Psiquiatr. 26, 103– 106. Epstein, D.H., Hawkins, W.E., Covi, L., Umbricht, A., Preston, K.L., 2003. Cognitive-behavioral therapy plus contingency management for cocaine use: findings during treatment and across 12-month follow-up. Psychol. Addict. Behav. 17, 73–82. Fals-Stewart, W., Schafer, J., 1992. The relationship between neurocognitive functioning and length of stay in drug-free therapeutic communities. J. Clin. Psychol. 48, 539–543. Fals-Stewart, W., 1993. Neurocognitive defects and their impact on substance abuse treatment. J. Addict Offender Counsel 13, 46–57. Fals-Stewart, W., 1997. Detection of neuropsychological impairment among substance-abusing patients: accuracy of the neurobehav-

321

ioral cognitive status examination. Exp. Clin. Psychopharmacol. 5, 269–276. Farabee, D., Rawson, R., McCann, M., 2002. Adoption of drug avoidance activities among patients in contingency management and cognitivebehavioral treatments. J. Subst. Abuse Treat. 23, 343–350. First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W. 1998. The Structured clinical interview for DSM-IV Axis I disorders. Patient Edition (SCID I/P) Version 2.0. New York State Psychiatric Institute. Goldberg, T.F., Weinberger, D.R., 1988. Probing prefrontal function in schizophrenia with neuropsychological paradigms. Schizophr. Res. 14, 179–183. Goldstein, R.Z., Leskovjan, A.C., Hoff, A.L., Hitzemann, R., Bashan, F., Khalsa, S.S., Wang, G.J., Fowler, J.S., Volkow, N.D., 2004. Severity of neuropsychological impairment in cocaine and alcohol addiction: association with metabolism in the prefrontal cortex. Neuropsychologia 42, 1447–1458. Gottschalk, C., Beauvais, J., Hart, R., Kosten, T., 2001. Cognitive function and cerebral perfusion during cocaine abstinence. Am. J. Psychiatry 158, 540–545. Grant, S., Contoreggi, C., London, E.D., 2000. Drug abusers show impaired performance in a laboratory test of decision-making. Neuropsychologia 38, 1180–1187. Heaton, R.K., Thompson, J.A., Gomez, E., 1999. Wisconsin Card Sorting Test: Computer Version 3 for Windows [CD-ROM] (Research Edition). Psychological Assessment Resources, Odessa, FL. Heaton, R.K., 1999. Wisconsin Card Sorting Test: computer version 3 for windows research edition. Psychol. Assess. Resour. Landro, N.I., Stiles, T.C., Sletvold, H., 2001. Neuropsychological function in nonpsychotic unipolar major depression. Neuropsychiatry Neuropsychol. Behav. Neurol. 14, 233–240. Lezak, M., Howieson, D.B., Loring, D.W., 2004. Neurospsychological Assessments, fourth ed. Oxford University Press, New York. Marlatt, G.A., Gordon, J.R. (Eds.), 1985. Relapse Prevention: Maintenance Strategies in the Maintenance of Addictive Behaviors. Guilford, New York. McCrady, B.S., Smith, D.E., 1986. Implications of cognitive impairment for the treatment of alcoholism. Alcohol Clin. Exp. Res. 10, 145– 149. Milner, B., 1963. Effects of different brain lesions on card sorting. Arch. Neurol. 9, 100–110. Morgenstern, J., Bates, M.E., 1999. Effects of executive function impairment on change processes and substance use outcomes in 12-step treatment. J. Stud. Alcohol 60, 846–855. Mulvaney, F.D., Alterman, A.I., Boardman, C.R., Kampman, K., 1999. Cocaine abstinence symptomatology and treatment attrition. J. Subst. Abuse Treat. 16, 129–135. Nunes, E.V. 1997. Methodological recommendations for cocaine abuse clinical trials: a clinician-researcher’s perspective, Rockville MD, National Institute of Drug Abuse, NIDA Research Monograph 175, 73–95. Porter, R.J., Gallagher, P., Thompson, J.M., Young, A.H., 2003. Neurocognitive impairment in drug-free patients with major depressive disorder. Br. J. Psychiatry 182, 214–220. Powell, D.H., Kaplan, E.F., Whitla, D., Catlin, R., Funkenstein, H.H., 1993. MicroCog: Assessment of Cognitive Functioning Version 2.1. The Psychological Corporation, San Antonio. Rawson, R.A., Huber, A., McCann, M., Shoptaw, S., Farabee, D., Reiber, C., Ling, W.A., 2002. A comparison of contingency management and cognitive-behavioral approaches during methadone maintenance treatment for cocaine dependence. Arch. Gen. Psychiat. 59, 817–824. Robinson, A.L., Heaton, R.K., Lehman, R.A., Stilson, D.W., 1980. The utility of the Wisconsin Card Sorting Test in detecting and localizing frontal lobe lesions. J. Consult. Clin. Psychol. 48, 605–614. Rogers, R.D., Sahakian, B.J., Hodges, J.R., Polkey, C.E., Kennard, C., Robbins, T.W., 1998. Dissociating executive mechanisms of tasks control following frontal lobe damage and Parkinson’s disease. Brain 121, 815–842.

322

E. Aharonovich et al. / Drug and Alcohol Dependence 81 (2006) 313–322

Rogers, R.D., Robbins, T.W., 2001. Investigating the neurocognitive deficits associated with chronic drug misuse. Curr. Opin. Neurobiol. 11, 250–257. Rotheram-Fuller, E., Shoptaw, S., Berman, S.M., London, E.D., 2004. Impaired performance in a test of decision-making by opiatedependent tobacco smokers. Drug Alcohol Depend. 73, 79– 86. Sayre, S.L., Schmitz, J.M., Stotts, A.L., Averill, P.M., Rhoades, H.M., Grabowski, J.J., 2002. Determining predictors of attrition in an outpatient substance abuse program. Am. J. Drug Alcohol Abuse 28, 55–72. Simpson, D.D., Joe, G.W., Rowan-Szal, G.A., Greener, J.M., 1997. Drug abuse treatment process components that improve retention. J. Subst. Abuse Treat. 14, 565–573. Simpson, D.D., Joe, G.W., Fletcher, B.W., Hubbard, R.L., Anglin, M.D., 1999. A national evaluation of treatment outcomes for cocaine dependence. Arch. Gen. Psychiatry 56, 516–518.

Teichner, G., Horner, M.D., Harvey, R.T., 2001. Neuropsychological predictors of the attainment of treatment objectives in substance abuse patients. Int. J. Neurosci. 106, 253–263. Teichner, G., Horner, M.D., Roitzsch, J.C., Herron, J., Thevos, A., 2002. Substance abuse treatment outcomes for cognitively impaired and intact outpatients. Addict. Behav. 27, 751–763. Tucker, K.A., Potenza, M.N., Beauvais, J.E., Browndyke, J.N., Gottschalk, P.C., Kosten, T.R., 2004. Perfusion abnormalities and decision making in cocaine dependence. Biol. Psychiatry 56, 527–530. Vik, P.W., Celluci, T., Jarchow, A., Hedt, J., 2004. Cognitive impairment in substance abuse. Psychiatr. Clin. North Am. 27, 97–109. Weinstein, C.S., Shaffer, H.J., 1993. Neurocognitive aspects of substance abuse treatment: A psychotherapist’s primer. Psychotherapy 30, 317–333. Zakzanis, K.K., Leach, L., Kaplan, E., 1998. On the nature and pattern of neurocognitive function in major depressive disorder. Neuropsychiatry Neuropsychol. Behav. Neurol. 11, 111–119.