Appendix B: Clinical Research Supporting CBT
Cognitive-behavioral treatments are among the most frequently evaluated psychosocial approaches for the treatment of substance use disorders and have a comparatively strong level of empirical support (American Psychiatric Association 1995; General Accounting Office 1996; Holder et al. 1991). To date, more than 24 randomized controlled trials have been conducted among adult users of tobacco, alcohol, cocaine, marijuana, opiates, and other types of substances (Carroll 1996).
A review of this group of studies (Carroll 1996) suggests that, across substances of abuse but most strongly for tobacco, there is good evidence for the effectiveness of CBT compared with no-treatment controls. The most rigorous level of testing compared CBT with other active treatments (in effect asking the question, Is CBT more effective than other widely used treatments? rather than, Is CBT better than no treatment or minimal treatment?). These comparisons have led to less consistent results; some studies indicate the superiority of CBT, while others have shown CBT as comparable to but not more effective than other approaches. CBT may hold particular promise in reduction in the severity of relapses when they occur, enhanced durability of effects, and patient-treatment matching, particularly for patients at higher levels of impairment along such dimensions as psychopathology or dependence severity.
As this manual focuses specifically on CBT for cocaine abuse, what follows is a brief review of the series of studies conducted at the Substance Abuse Treatment Unit at Yale University, which has evaluated the CBT approach described in this manual with individuals meeting criteria for cocaine abuse or dependence. Moreover, because this manual is intended to provide practical strategies for therapists working with this population, this review focuses on what these studies may imply about means of more effectively applying these CBT strategies to cocaine-abusing populations.
CBT and Interpersonal Therapy
In our first study (Carroll et al. 1991), we directly compared CBT to another active psychotherapy, Interpersonal Psychotherapy or IPT (Klerman et al. 1984), a treatment that was then in regular use in our clinics. The strategy of comparing two active treatments addressed several methodological and ethical questions associated with no treatment or nonspecific control groups, such as differences in demand characteristics and credibility of the offered treatments; lack of control of common factors in the therapies; and the problem of subjecting severely impaired treatment-seeking individuals to minimal or no-treatment control conditions (Basham 1986; Kazdin 1986; O'Leary and Borkovec 1978).
In this, as in all of our studies on CBT, we used a variety of methodological features that were intended to protect the integrity of the treatments evaluated and control other sources of variability. Subjects were randomly assigned to treatments. All treatments were manual guided and implemented by doctoral-level therapists who received extensive training and ongoing supervision. Patient outcomes were assessed by independent evaluators who were blind to the treatment assignment.
In this 12-week outpatient study, 42 subjects who met DSM-III criteria for cocaine dependence were randomly assigned to either CBT or IPT. Those assigned to CBT were more likely than subjects in IPT to complete treatment (67 versus 38 percent), attain 3 or more continuous weeks of abstinence (57 versus 33 percent), and be continuously abstinent 4 or more weeks when they left treatment (43 versus 19 percent).
Although the sample size was small and these differences did not reach statistical significance, significant differences by treatment group did emerge when subjects were stratified by severity of cocaine abuse. For example, among the subgroup of more severe cocaine users, subjects who received CBT were significantly more likely to achieve abstinence than those assigned to IPT (54 versus 9 percent). Among the subgroups of subjects with lower severity of cocaine abuse, outcomes were comparable for both treatments (Carroll et al. 1991). These findings suggest that more severely dependent cocaine abusers may require the greater structure and direction offered by CBT, which emphasizes learning and rehearsal of specific strategies to interrupt and control cocaine use, whereas the specific type of treatment offered may be less important for less severely dependent cocaine abusers.
CBT and Clinical Management
Our next study was more complex because it involved both psychotherapy and pharmacotherapy (Carroll et al. 1994b). This time we compared CBT to Clinical Management (CM) (Fawcett et al. 1987), a nonspecific psychotherapy that satisfied many of the requirements of a control condition.
- CM provided common elements of a psychotherapeutic relationship, including a supportive doctor-patient relationship, education, empathy, and the instillation of hope, without providing active ingredients specific to relapse prevention.
- CM provided medication management as well as an opportunity to monitor patients' clinical status and treatment response.
- CM provided a convincing therapeutic rationale to foster greater retention in the protocol and compliance with medication.
It is important to note that these features, although desirable in a psychotherapy control condition because they address many ethical and methodological concerns, may be powerfully therapeutic on their own and thus also serve as a more stringent test of active psychotherapies than would alternatives such as no-treatment or waiting-list control conditions. All subjects received a medication, either desipramine (which was the most promising medication for cocaine dependence at the time) or a placebo. In this study, 121 individuals meeting DSM-III-R criteria for cocaine dependence were randomly assigned to one of four treatment conditions:
- CBT in combination with desipramine
- CBT plus placebo
- CM plus desipramine
- CM plus placebo
We hypothesized that both CBT and desipramine would be more effective than CM and placebo, respectively. Moreover, this design permitted detection of combined effects of psychotherapy and pharmacotherapy if these proved to be sufficiently strong.
After 12 weeks of treatment, subjects in all four groups showed significant reductions in cocaine use as well as improvement in several other problem areas. Significant main effects for medication or psychotherapy type were not found; that is, cocaine outcomes were comparable whether the patient received CBT or CM, or desipramine or placebo.
We did find an interaction effect similar to that in our first study. That is, baseline severity of cocaine abuse was found to interact differently with the two forms of psychotherapy. Patients who were more severely dependent on cocaine stayed in treatment longer, attained longer periods of abstinence, and had fewer urine screens positive for cocaine when treated with CBT compared with CM. Again, this suggests that abusers with more intense involvement with cocaine may benefit from the additional structure, intensity, or didactic content of CBT, which focuses specifically on reducing access to cocaine and avoidance of high-risk situations for relapse. These results again suggest that low- intensity approaches may be effective for individuals less severely dependent on cocaine.
Additional effects were found in subsequent analyses of data from the study comparing CBT to CM. However, because these findings were based on exploratory, post hoc analyses, they should be interpreted with caution.
CBT and Depressive Symptoms
Because of the clinical importance of affective disorders among cocaine abusers, we evaluated the role of depressive symptoms in response to study treatments (Carroll et al. 1995). We found that CBT was more effective than CM in retaining depressed subjects in treatment. There was also some evidence that it was more effective in reducing cocaine use. This may have occurred because the depressed subjects experienced more distress, which may have enhanced their motivation for treatment, availability for psychotherapy, and ability to implement and benefit from coping skills.
On the other hand, there was no evidence that CBT was more effective than CM in reducing depressive symptoms. While cognitive-behavioral approaches to treating depression have generally been effective and comparable to antidepressant medication in reducing depressive symptoms (Elkin et al. 1989; Simons et al. 1986), our CBT approach did not specifically address depressive symptoms as a treatment target or convey specific strategies for managing coexistent depression. Rather, we focused almost exclusively on helping patients develop strategies to reduce their cocaine use during the early stages of treatment, although we did address the relationship between negative affect and cocaine use. A possible implication of these findings is the need for CBT therapists to more explicitly address depressive symptoms with patients who experience them (Carroll et al. 1995).
Reductions in cocaine use and depression were closely associated throughout treatment, although the direction of these changes was not clear. One possible explanation for this finding is that reduction in depressive symptoms leads to reduction in cocaine use by reducing distress, thus enabling patients to make better use of their coping resources, become more available for psychotherapy, or reduce their possible self-medication of depressive symptoms with cocaine. Conversely, reduction of cocaine use might lead to improvements in depressive symptoms by decreasing depression associated with cocaine withdrawal, reestablishing normal sleep and eating patterns, and reducing exposure to other negative consequences of cocaine abuse.
CBT and Alexithymia
Alexithymia refers to a cognitive-affective style that results in specific disturbances in the expression and processing of emotions. Literally meaning "no words for feelings," the term was coined by Nemiah and Sifneos (1970) to refer to psychosomatic patients who exhibited four specific affective/cognitive impairments:
- Difficulty in verbalizing affect states
- A tendency to focus primarily on the somatic/physiological components of affective arousal
- An impoverished fantasy life
- A highly concrete cognitive style
We evaluated the rates and significance of alexithymia among cocaine abusers in our CBT and CM comparative study. We found that 39 percent of the cocaine abusers scored in the alexithymic range, based on responses to the Toronto Alexithymia Scale (Taylor et al. 1985). While alexithymic subjects did not differ from nonalexithymic patients with respect to overall treatment retention or outcome, alexithymic subjects did respond differently to psychotherapy. They had better retention and cocaine outcomes when treated with CM, whereas nonalexithymic subjects had better outcomes when treated with CBT.
The finding that cocaine abusers with higher alexithymia scores responded more poorly to CBT has several implications. Patients are asked to identify and articulate internal affect and cognitive states associated with cocaine use - a task particularly difficult for alexithymic patients. CBT encourages patients to identify, monitor, and analyze their cravings, negative affects, and many subtle fleeting cognitions. In essence, it requires patients to have good access to their internal world. These demands may be overwhelming for the alexithymic subjects. For example, one patient, as part of a self-monitoring assignment, was asked to note his feelings and their intensity in response to a variety of situations. Rather than describing feelings such as cheerful, irritable, or bored, he consistently wrote either yes or no, suggesting he had some awareness of strong affects, but little ability to articulate them or relate them to his drug use. Therapists may find it helpful to provide a preparatory phase before starting the monitoring of high-risk situations and skills training to prevent such patients from being overwhelmed and to help them identify their feelings and affect states.
Some of the most intriguing findings from the CBT/CM comparative study emerged from the 1-year followup (Carroll et al. 1994a). As a group, subjects' cocaine abuse decreased overall or remained stable with respect to posttreatment levels, rather than rebounding to pretreatment levels. More importantly, there was consistent evidence of delayed effects for CBT compared with CM for cocaine outcomes, even when we controlled for the proportion of subjects who received some nonstudy treatment during the followup period. After leaving the study treatments, subjects who had received CBT continued to reduce their cocaine abuse, whereas cocaine abuse remained relatively stable in the CM group. These results may be related to delayed emergence of specific effects of CBT. During the acute phase of CBT and CM treatment, subjects in all groups received a variety of nonspecific interventions, including weekly urine monitoring, frequent assessment of cocaine use and other symptoms, support and encouragement from therapists and research staff, and positive expectations for treatment effects. These common factors may have been powerfully therapeutic and overwhelmed treatment-specific effects.
The cessation of these nonspecific interventions may have created the conditions under which the more durable and specific effects of CBT had an opportunity to emerge. CBT is intended to impart generalizable coping skills that can be implemented long after patients leave treatment, while supportive treatments may provide patients with fewer enduring resources (Carroll et al. 1994b).
In other clinical populations, followup studies of cognitive-behavioral treatments have indicated the durability of their effects with some consistency. For example, cognitive-behavioral treatments have been found to be superior or comparable to acute or continued tricyclic pharmacotherapy in preventing relapse of depressive and panic episodes (Miller et al. 1989; Simons et al. 1986). Moreover, some studies (Beutler et al. 1987), including a recent one with alcoholic subjects (O'Malley et al. 1994), have shown continuing improvement or delayed emergence of effects during followup after cognitive-behavioral therapy.
CBT and Alcoholic Cocaine Abusers
Our experience pointed to the significance of alcohol abuse and dependence, which occurs quite frequently among clinical populations of cocaine abusers. In a survey of psychiatric disorders among 298 cocaine abusers, we found that alcohol dependence was the most frequently diagnosed comorbid disorder, with 62 percent of the sample meeting RDC criteria for lifetime alcohol dependence and almost 30 percent meeting criteria for current use (Carroll et al. 1993a). This is consistent with reports from large-scale community samples, such as the Epidemiological Catchment Area study, which found that 85 percent of individuals who met criteria for cocaine dependence also met criteria for alcohol abuse or dependence, a rate far higher than that of alcoholism among those meeting criteria for heroin-opioid (65 percent), cannabis (45 percent), or sedative-hypnotic-anxiolytic (71 percent) dependence (Regier et al. 1990). More importantly, comorbid alcohol-cocaine dependence has been associated with more severe drug dependence, poorer retention in treatment, and poorer outcome with respect to either disorder alone (Brady et al. 1995; Carroll et al. 1993b; Walsh et al. 1991).
We then evaluated CBT and other psychosocial and pharmacologic treatments for this large and challenging population (Carroll et al. in press). We compared CBT to two other treatments, CM and Twelve-Step Facilitation (TSF) (Nowinski et al. 1992), an individual approach consistent with the 12 steps of Alcoholics Anonymous (AA) which has the primary goal of fostering the patient's lasting involvement with the traditional fellowship activities of AA or Cocaine Anonymous. We also evaluated disulfiram (Antabuse) in this study because of pilot data that suggested that reduction in alcohol use through disulfiram may be associated with reductions in cocaine use as well (Carroll et al. 1993c). Preliminary data from this study suggest that the two active psychotherapies - CBT and TSF - were more effective than CM in fostering consecutive periods of abstinence from cocaine and abstinence from both cocaine and alcohol concurrently. The two active psychotherapies also yielded a higher percentage of cocaine-free urine specimens. In addition, CBT and TSF, compared with CM, were associated with significant reductions in cocaine use across time, particularly for subjects who received at least minimal exposure to treatment.
Finding that CBT and TSF were more effective than the psychotherapy control condition underlines the important role that well-defined, competently delivered psychosocial interventions play in the treatment of cocaine dependence. Because CM provided a control for general, nonspecific aspects of psychotherapy (including a supportive doctor-patient relationship), this study provided a rigorous test of the specific, active ingredients of CBT and TSF above and beyond simple support and attention.
The finding that CBT was more effective than CM in reducing cocaine use contrasts with the finding from our previous clinical trial, which did not show overall differences between CBT and CM (Carroll et al. 1994b). However, in that study, CBT was found to be more effective than CM for the subgroup of subjects who were more severely dependent on cocaine. Again, because concurrent cocaine-alcohol dependence has been associated with higher severity of cocaine use and poorer prognosis with respect to cocaine dependence alone, subjects in this study may be similar to the more severely dependent subsample from our earlier study. Thus, findings from these two studies taken together may suggest that more severe groups of cocaine-dependent individuals differentially benefit more from the comparatively intensive active ingredients of CBT or TSF than from the supportive but less structured and less directive CM, which also makes fewer demands on patients to carry out assignments outside of sessions.
It is also important to note that study findings did not show significant differences between the two active psychotherapies, TSF and CBT, in either cocaine or alcohol outcomes. This suggests that these two forms of treatment were equally effective with this population in this study. The comparable outcomes occurred despite clear differences in the theoretical basis of these treatments, the specific interventions used by the therapists (as detected by independent raters blind to subjects' treatment assignments), and the evidence that subjects demonstrated specific behavioral changes consistent with the theoretical mechanisms of action of their study treatments (changes in coping skills in CBT, more AA involvement in TSF). This is consistent with other recent research with cocaine-dependent samples (Wells et al. 1994; Carroll et al. in press).