Cognitive-Behavioral Therapy for Medication-Resistant Schizophrenia: A Review

Antipsychotic-resistant symptoms constitute a significant problem in schizophrenia because they affect a substantial number of patients and their families and have significant cost implications for services. Approximately 20%–45% of patients with schizophrenia continue to experience symptoms that interfere with their functioning despite adequate trials of medication (1, 2). An additional 5%–10% of patients show no benefit at all from antipsychotic medication (3). Persistent symptoms are disabling (4) and distressing (5) and they can be associated with affective symptoms (6) and risk of suicide (7). There has been a growing interest in developing cognitive-behavioral therapy (CBT) for patients with psychosis who, despite optimal treatment with antipsychotic medication, continue to experience distress and functional interference because of persistent symptoms. CBT has been shown to produce improvements in positive and negative symptoms (8, 9), depression (10), adherence to treatment (11), and insight (12), and this effect has also proved to be durable and cost effective (13) over the short term. The principal aim of CBT for persistent psychotic symptoms is to work collaboratively with the patient to facilitate improved understanding and coping in order to reduce suffering and distress and improve functioning.

This article provides an update on the impact of CBT on persistent symptoms of schizophrenia. Although this is not a systematic review, published articles have been selected following a detailed search on Ovid Medline, Psychinfo, and Embase using the key words cognitive, behavior, cognitive behavioral, schizophrenia, psychosis, resistant, treatment, medication. The publications that are discussed here were selected based on methodology (studies of CBT for medication-resistant schizophrenia) and include both reviews and randomized trials. This review does not include studies of cognitive remediation (14), adherence to treatment (11), or insight-oriented therapies (12) based on CBT. The criteria used to define “medication resistance” in different studies vary somewhat; however, in most studies, patients are considered to have treatment-resistant illness if they have experienced symptoms for at least 3 to 6 months or longer despite adequate trials of medication. The type of antipsychotic (typical or atypical) used in the studies reviewed here depends on when the study was carried out.

MODELS USED TO EXPLAIN SCHIZOPHRENIA

A number of models have been used to explain the etiology and psychopathology of schizophrenia.

Biological, psychodynamic, and social models

Biological models, which play a central role in discussions of the etiology of schizophrenia, have emphasized genetic and biological causes for the disorder, including abnormalities in brain structure, development, and/or function (e.g., birth injury, abnormal development, or viral influences [15]).

Psychodynamic models have adopted a range of perspectives, but have focused especially on childhood experiences and complexities in interpersonal relationships.

Social models have emphasized environmental influences (e.g., drug abuse [16]) and other sociocultural (17) pressures.

Cognitive models

The cognitive-behavioral therapy approaches for schizophrenia described in this article are based on a cognitive model of schizophrenia. Cognitive-neuropsychological theories suggest that a number of independent or modular processes, including processing and storing information, planning, and assessing significance, are involved in brain functions (18). The cognitive deficits seen in schizophrenia are similar to those that one expects to find in individuals with non-specific neurological damage (19). Therefore, a cognitive deficit resulting from a breakdown in these brain systems (e.g., dysfunction of the neuronal pathways leading to an inability to distinguish between actions that derive from external or internal stimuli) may be at the root of psychotic experiences (20). As a result, incorrect inferences due to an inability to accurately understand the intentions of other people could be a precursor to the formation of a delusion (21). For example, a patient with psychosis may believe that cars with head lights switched on are monitoring him; or ideas of reference could result from incorrect inferences, such as when patients believe that others are talking about them.

Hemsley suggested that the rapid and automatic assessment of the significance or lack of significance of aspects of sensory input is impaired in schizophrenia (22). These deficits may cause the person to attach great significance to nonrelevant aspects of the social environment, which may thus seem personally significant to the patient (22).

Hemsley also suggested that such deficits in assessment may cause patients to access thoughts and images from their memories that may then be attributed to an external source and provide the basis for hallucinatory experiences (23).

Cognitive models of schizophrenia draw on the stress vulnerability model (24) as well as on cognitive models of depression, intrusive thoughts, panic, and trauma. In cognitive models of schizophrenia, the central focus is on psychotic symptoms and the various affective and behavioral factors that maintain them. The vulnerability stress hypothesis of schizophrenia posits that vulnerabilities and stresses interact and produce the symptoms of psychosis. Once the symptoms develop, patients may catastrophize (i.e., assume the worst) or develop safety behaviors that can lead to symptom exacerbation or maintenance. Symptomatic improvement may, therefore, occur when patients achieve a better understanding of the way in which their vulnerabilities and the stresses they have experienced interact (25).

As a way of protecting themselves against unbearable loss of self-worth and self-esteem (26), some people with schizophrenia may have a propensity to develop delusions or hallucinations rather than, for example, becoming anxious or depressed. Thus delusions may be “self-serving.” For example, paranoid thinking or grandiose delusions may play a defensive or functional role by effectively reducing discrepancies between the actual self and the idealized self. Hallucinations may serve a similar purpose (e.g., a patient who hears the voice of God telling him that he is special or has a special task in this world). These delusions and hallucinations protect self-esteem by allowing the person to attribute negative events to external causes (27) (e.g., blaming others rather than themselves for things that go wrong).

Readers are referred to books by Kingdon and Turkington (25, 28), Fowler et al. (29), and Chadwick et al. (30) for more detailed discussions of CBT models for psychosis. Although these publications differ somewhat in their emphasis, there is general agreement on the main methods of therapy.

DELUSIONS

Delusions have been defined as false beliefs that are held with strong conviction and are not amenable to reasoning (31). This view has been challenged by cognitive therapists who believe that delusions are best explained as occurring on a continuum (i.e., with normal beliefs at one end and full-blown delusions at the other). Sachs et al. (32) described a group of 20 psychotic patients suffering from delusions who passed through three phases: delusional, double-awareness, and non-delusional, with each state representing a different range on the continuum. After an initial deluded phase, the patient moves into a state of double-awareness, in which the person is only partially deluded through the emergence of reality testing. In this state, the patient believes that his delusion may not be quite true and he becomes increasingly able to consider alternative explanations. This state is usually associated with improved social contact as the patient's argumentativeness decreases. Improved social contact can in turn strengthen the patient's still impaired reality testing. Successful treatment will finally result in the patient reaching the non-delusional state, in which the person is fully aware that his or her beliefs were unfounded and that they were part of an illness that affected his thinking. It is important to keep in mind, however, that these three states are by no means always clearly delineated, sequential, or of equal duration, nor are they necessary for patient recovery. Nevertheless, this model represents a pathway that is commonly followed by recovering patients—just as progression in the opposite direction often occurs when the illness starts.

CBT techniques can also be used successfully with patients who are not aware of their illness. In such cases, CBT focuses on developing a problem list. Collaborative empiricism and graded reality testing are used until the patient volunteers to consider the scientific basis for his or her delusions or hallucinations.

Maher (33) also believed that delusions occur on a continuum with normal experiences and suggested that the mechanisms that produce delusions are no different from those that underlie the cognitive processes of an individual with “normal” beliefs. Frith (21) proposed that a deficit in self-monitoring of thoughts and intentions to act was the cause of beliefs reflecting thought insertion and delusions of control, with obsessional thoughts and images perceived as originating outside the mind. Hemsley (23), as discussed above, explained delusions as a result of neuropsychological deficits. Garety and Hemsley (34) noted that delusions are associated with a “jumping to conclusions” style of attribution that occurs in the setting of cognitive deficits (as illustrated by the example of the patient presented above who jumped to the conclusion that a car making a U-turn had a personal significance for him). For a more detailed discussion of cognitive models of delusion formation and maintenance, see Garety et al. (35) and Beck and Rector (36).

To summarize, it appears that multiple etiological factors are involved in the development of delusions and that the same factors appear to operate in the development of any strongly held normal belief. Delusions often appear to have an underlying function, such as explaining situations or relationships that are of personal significance to an individual and giving order and meaning to the person's life. Delusions in any psychotic illness would be expected to be consistent with core beliefs about the self (self schemata). Thus, social, religious, and cultural influences can be important sources of delusional content. It is important to formulate an understanding of the personal circumstances that led up to the emergence of the delusion, focusing especially on the stressors that were present prior to the initial episode. Such an understanding usually provides insights into the origin and significance of the delusional beliefs especially when these are part of a fixed delusional system.

On occasion, individuals, especially if they are floridly psychotic, may present with transient delusions that are held with less conviction. These delusions may be less meaningful but still often reflect the person's current and past experiences. The strength of the belief may depend upon the time period over which the belief has been present and the psychological and social consequences of relinquishing the belief (25, 28). For example, the delusion may prevent the person from facing anxieties associated with the need to work or make friends or may explain anomalous or confusing situations. Thus the person may attribute the fatigue she is experiencing as a result of depression to being slowly poisoned by her neighbors.

HALLUCINATIONS

Hallucinations have been defined as vivid experiences with the quality of external reality in the absence of appropriate stimuli to the sensory apparatus. A number of cognitive theories have attempted to explain the origin of hallucinations. Hoffman (37) postulated that auditory hallucinations occur as a result of random firing of speech-processing mechanisms that result in “parasitic memories” (memories that combine parts of real memories and are fallaciously associated with a number of different inputs) being brought to consciousness. David (38) explained hallucinations using a cognitive model in terms of a deficit in speech-processing mechanisms. Frith (21) suggested that faulty internal monitoring of inner speech was involved. Inner speech is a cognitive process involving subvocal speech articulation. Bentall (39) suggested that specific beliefs and attitudes produce biases that determine how the person interprets inner speech. Using single photon computed tomography (SPECT) studies, McGuire et al. (40) demonstrated that auditory hallucinations are associated with activation of Broca's area, which is normally activated by speech, suggesting that such hallucinations are the result of misattribution of inner speech. It has subsequently been shown that, during auditory hallucinations, the auditory cortex in the temporal lobe and various subcortical areas that are normally involved in processing external speech are also activated. Thus, inner speech is processed by the same system that is normally involved in processing external speech (41). Patients with schizophrenia often describe auditory hallucinations as a running commentary on their thoughts delivered in a voice that is not their own. They may then attribute the voice to an external source such as God, aliens, or neighbors, depending on their experiences and biases.

Therefore, in the cognitive model of schizophrenia, hallucinations are conceptualized as the person's own automatic thoughts, which the person perceives as originating outside his or her own mind. (Automatic thoughts are repetitive, automatic self-statements that occur in certain situations. They can be positive or negative; psychological problems develop when our automatic thoughts are consistently negative.) Auditory, visual, and somatic hallucinations are therefore entirely internal cognitive phenomena that elicit powerful affective and behavioral responses because they have all the implications of externally valid events. As with delusions, it appears that a combination of a vulnerability to externalize and the presence of stressful events and circumstances is likely to precipitate hallucinatory experiences. In these situations, thoughts may be transformed through inner speech to hallucinations (e.g., a voice telling the patient “You are useless.”). The hallucinations, which are associated with high levels of affect (e.g., severe distress), may then be maintained by safety behaviors (42) (e.g., avoiding social contact) as well as by dysfunctional explanations (e.g., “Satan is speaking to me”) (25, 28).

Some patients experience functional hallucinations that occur when the person simultaneously receives a real stimulus in the perceptual field concerned (e.g., hallucinatory voices heard simultaneously with—and specific to—the real sound of running water) (43). The form and content of hallucinations are of interest since they may help determine the origin of the hallucination. For example, auditory hallucinations such as hearing a voice say “You are a bad person” may emerge from a sense of low self-esteem and depressive mood. Hallucinations that are diagnostic of schizophrenia are third person hallucinations that generally involve a running commentary on the person's actions and a thought echo. Such symptoms often take the form of “replays” of situations that have actually occurred or statements that have actually been made, usually in distressing or stressful circumstances. Hallucinations in schizophrenia can also be conceptualized in an obsessional framework as being analogous to obsessional doubt. When the person attributes the hallucinations to an external source, this causes distress. Therefore, one goal in CBT is to help the person explore alternatives to this belief. Other techniques include work on reattribution, debating content, developing coping strategies, and looking at the way the hallucinations reflect beliefs the person has about him or herself (25, 28). These techniques involve rationalizing the hallucinations through alternative explanations, learning to cope with the symptoms to reduce distress, and working to understand the personal significance attached to the experience.

DEPRESSION AND ANXIETY

Symptoms of depression and anxiety are common in patients with schizophrenia. These symptoms may precede, occur concurrently with, or present following an acute psychotic episode. Sometimes mood symptoms develop independently between psychotic episodes (44). Patients with schizophrenia also often develop secondary psychiatric illnesses, such as posttraumatic stress disorder, agoraphobia, or social phobia that further exacerbate core symptoms (45).

Depression, which is often associated with feelings of hopelessness, often occurs in response to the patient's appraisal of his or her psychosis, which embodies feelings of loss, humiliation, and entrapment, and a consequent “down-ranking” of the self (46). Addressing this emotional distress is part of the therapeutic process of CBT. Depression that emerges as a result of improved insight is dealt with by using destigmatizing, normalizing explanations. For example, the clinician could say “Let's consider diabetes. It is usually a chronic illness, but with appropriate treatment, most patients lead a normal life.” Depression that appears when the timeline is unclear can be approached by eliciting the views of other people on what actually happened.

Depression emerging as part of a delusional system is likely to recede when the CBT work focuses on the underlying schema vulnerability. For example, in the case of the patient discussed above, who went bankrupt and then developed grandiose delusions, the work would focus on schema vulnerability related to achievement—compelling reasons within the patient's understanding of the world (schema) that make him feel vulnerable. Thus, the therapeutic work would focus on helping the patient identify the importance of achievement and success in fulfilling his self-schema and hence self-worth.

Anxiety is dealt with by using relaxation training, rational responding, and developing less catastrophic explanations for the anxiety.

This patient managed to keep his diary and found that he was able to relax during these times. Further tasks and relaxation techniques were slowly worked on.

THOUGHT DISORDER

A patient with thought disorder is often striving to communicate but cannot convey thoughts in a logical sequence. The thoughts the person is trying to convey may be quite logical (once they can be understood), but their expression is not. People with active thought disorder may display “knight's move” thinking (jumping from one subject to another with only a tenuous connection, as a knight does in chess), fusion of themes, and neologisms (newly created words). These phenomena are usually aroused by specific concerns (47), which are often delusional and driven by anxiety and anger. Because of their emotional involvement with these issues, individuals with thought disorder may have significant problems discussing them; thus, as they come closer to the topics they feel strongly about, they become more agitated and the thought disorder becomes greater and greater, interfering with communication.

Thought linkage is a useful strategy that can be used to help individuals with thought disorder communicate more effectively. This involves asking the patient to explain the link between seemingly unconnected topics. Proceeding slowly and patiently can help clarify the person's meaning and the connections between his or her ideas. For example, there may be one core theme that lies behind the person's seemingly disorganized thoughts (25, 28). Rational responses aimed at tackling the cognitive distortions that are causing the person's affective disturbance can be very beneficial.

RESEARCH EVIDENCE

Over the past 15 years, a number of studies have reported benefits of CBT in patients with medication-resistant schizophrenia. The definition of medication resistance varies among the trials but generally the studies use criteria involving persistent psychotic symptoms for at least 6 months despite trials with adequate doses of two antipsychotics. However, before discussing those findings, we provide a brief review of early work with CBT for psychosis as background.

Early clinical trials of CBT in psychosis

Cognitive behavioral techniques were first used in the treatment of psychosis in 1952 by Beck (48), who was treating a patient who believed he was being followed and watched by the FBI. The patient was encouraged to trace the antecedents of the delusion and reality testing was used. The patient was eventually able to recognize that all his alleged persecutors were normal people going about their daily business, which resulted in a marked improvement in his level of social functioning. Hole et al. (49) subsequently described eight patients with chronic delusions, half of whom appeared to improve when cognitive behavioral techniques were used.

Very little additional work using CBT for psychosis was done until the late 1980s. Fowler and Morley (50) and Chadwick and Lowe (51) published studies describing small cohorts of patients with delusions. In 1991, Roberts (52) described how delusions could be made more understandable in relation to events in the person's life.

More recent controlled trials in treatment resistant psychosis

In a study published in 1993, Tarrier et al. (53) compared the effectiveness of CBT for medication-resistant psychotic symptoms (residual hallucinations and delusions) with a general cognitive intervention aimed at improving problem-solving skills. Forty-nine patients were recruited, 27 of whom entered the trial and completed a post-treatment assessment; 23 of the patients were reassessed at 6-month follow-up. The participants were randomly assigned to receive five weekly sessions of coping strategy enhancement aimed at reducing psychotic symptoms or a general problem-solving intervention. Problems with this study included a high drop-out rate, lack of diagnostic clarity, and failure to use intention-to-treat methodology, with dropouts excluded from the analysis. The group who received coping strategy enhancement showed benefit on the Delusions Scale of the Psychiatric Assessment Scale (PAS) (54) and in overall symptom severity on the Brief Psychiatric Rating Scale (BPRS) (55) at the end of therapy and at follow-up. Despite the fact that the intervention was briefer, more intensive, and more focused (10 sessions over 5 weeks) than that used in later studies, such as the London-East Anglia Study published in 1997 (56), it had a similar effect size on overall symptoms. Early studies of CBT interventions for resistant hallucinations using both focusing and distraction techniques (e.g., Haddock et al. 1996 [57]) also reported improvements in symptoms.

In a controlled pilot study of CBT for drug-resistant psychosis with non-random allocation of participants, Garety et al. (58) found that an average of 16 sessions of cognitive therapy was significantly more effective than a control group in reducing delusional conviction, overall symptomatology on the BPRS, and level of depression as measured by the Beck Depression Inventory (BDI) (59). In this study, the first subjects referred for therapy (n = 13) were placed in the experimental group, while subjects who were subsequently referred (n = 7) were placed in the waiting list group. This study was designed as a pilot for future larger studies.

Bouchard et al. (60) reviewed 15 studies of cognitive restructuring in the treatment of schizophrenia, 5 of which were found to be methodologically rigorous. Focusing on changes in positive symptoms as the main outcome measure, they concluded that cognitive approaches are effective in reducing or eliminating delusions and hallucinations in patients with schizophrenia and that the effect on delusions may be larger.

As noted above, the London-East Anglia group published its initial findings in 1997 (56) and followed this up with papers on prediction of outcome (61) and cost effectiveness (62). In this study, 60 patients who had at least one distressing positive symptom of psychosis that was medication resistant were randomly assigned to 20 sessions of manualized CBT delivered by therapists who were expert clinical psychologists (n = 28) or to a control condition involving standard care (n = 32). At the end of 9 months of individualized therapy, 50% of the CBT group were treatment responders compared with 31% of the control group. Over 9 months, improvement was significant only in the treatment group, who showed a 25% reduction in scores on the BPRS. Participants had a low drop-out rate from therapy (11%) and expressed high levels of satisfaction with treatment (80%). However, this study did not find significant treatment effects on measures that were more specifically focused on delusional conviction or frequency of hallucinations; and differences in BPRS scores between the groups were not considered clinically meaningful. The raters, though independent, were not blind, and fidelity to the treatment manual as evaluated by an independent rater was not undertaken. Among the 47 participants who were available for follow-up at 18 months, those in the CBT treatment group continued to show significant continued improvement in BPRS scores. Delusional distress and frequency of hallucinations were also significantly reduced in the CBT group.

In a well designed, methodologically robust study in 87 patients with treatment-resistant hallucinations published in 1998, Tarrier et al. (8) compared three treatments: 1) a CBT intervention combining coping strategy enhancement and problem-solving, 2) supportive counseling, and 3) treatment as usual. They used an intensive approach for all treatment groups that involved 2 sessions per week over 10 weeks and a random allocation design. Tarrier et al. found that both CBT and supportive counseling were significantly better than treatment as usual at 3 months. CBT had a significant effect on positive symptoms whereas supportive counseling did not. Significantly more patients who received CBT showed more than a 50% improvement in positive symptoms. The relapse rate and time spent in hospital were significantly worse in the treatment as usual group. However, the improvements favoring the CBT intervention over supportive counselling were not sustained at long-term follow-up. After 1 and 2 years, both therapy groups were comparably less likely to relapse than the routine care group (63, 64). It is possible that the result could have been influenced by medication dose and type, since these variables were not reported. The assessors were blind to group allocation but not to the fact that the assessment was a follow up.

In Italy, Pinto et al. (65) carried out a randomized study of CBT in patients who were beginning treatment with clozapine. Twenty clients were randomly assigned to receive CBT plus social skills training, and 21 were assigned to supportive therapy. Both groups showed statistically significant improvement on the BPRS, the Scale for the Assessment of Positive Symptoms (SAPS) (66), and the Scale for the Assessment of Negative Symptoms (SANS) (67) from baseline to post-intervention. Comparisons between the groups showed that, postintervention, clients who had received CBT plus social skills training had lower BPRS and SAPS scores than the clients who received supportive therapy. No significant differences between treatment groups were found on SANS scores during the evaluation period. Adherence to the therapeutic regimen was high in both groups, ranging from 75% to 85%, but the clients in the experimental group attended more sessions than the clients in the comparison group. Unfortunately the results of the study are difficult to interpret since the onset of clozapine's effects varies and may take up to 6 months, and a number of patients were receiving subtherapeutic doses due to side effects. Despite these confounding factors, the CBT group showed a significant effect on overall symptoms.

Sensky et al. (9) conducted a randomized controlled trial comparing 9 months of CBT with befriending (an intervention designed to be a control for non-specific therapy factors, including time spent with subjects) in a randomized controlled trial of patients with drug-refractory positive symptoms of schizophrenia. Ninety patients received a mean of 19 individual treatment sessions over 9 months, with no significant between-group differences in treatment duration. Both interventions resulted in significant reductions in positive and negative symptoms and depression. At the 9-month follow-up evaluation, patients who had received CBT continued to improve on the same measures, while those in the befriending group did not.

Recently, successful pilot projects of CBT for treatment-resistant psychosis in the United States and Canada have been reported by Rector et al. (68), who undertook a randomized controlled trial comparing CBT plus enriched treatment-as-usual (CBT-ETAU) (n = 24) with enriched treatment-as-usual only (ETAU) (n = 18). Enriched treatment-as-usual involved comprehensive treatment in a specialized schizophrenia treatment service. CBT was conducted on an individual basis for 6 months (20 sessions). Significant clinical effects on positive and negative symptoms and overall symptom severity were observed in patients treated with CBT-ETAU, although there were no statistically significant differences between the treatment groups post treatment. The most pronounced effect of CBT-ETAU compared with ETAU in this study was a reduction in negative symptoms at follow-up. However numbers in this study were small.

Temple and Ho (69) performed another open-label controlled study comparing cognitive therapy with treatment-as-usual in 19 patients, 10 of whom received up to 20 sessions of individual CBT, while the others received treatment as usual which included case management or community-based psychotherapy but not CBT. Improvements were found in the CBT group on the Clinical Global Impressions-Improvement scale (70), psychosocial functioning (measured by occupational impairment, housework impairment, relationship impairment, and level of satisfaction), the Global Assessment Scale (71), and in overall symptoms and delusions. There was a trend toward significance in the reductions in negative symptoms. Limitations of this study included small sample size, lack of randomized assignment, and the fact that raters were not blinded to treatment condition.

In a pilot study by Cather et al. published in 2005 (72), 30 outpatients with schizophrenia or schizoaffective disorder, depressed type, with residual psychotic symptoms, were randomly assigned to either 16 weekly sessions of functional CBT (fCBT) or psychoeducation, with assessments conducted at baseline and post-treatment by blind evaluators. In fCBT, symptom-focused CBT interventions are delivered in the context of working on functional goals: a premise of fCBT is that the therapeutic alliance and patient motivation are enhanced by linking interventions to life goals. The results of this study suggested that both treatments were well tolerated. Effects were not significantly different between the groups, but within group effect sizes indicated that fCBT produced greater treatment benefit for positive symptoms.

Granholm et al. (73) compared cognitive behavioral social skills training with treatment as usual in a randomized, controlled trial. The sample included 76 middle-aged and older outpatients with chronic schizophrenia, who were assigned to either treatment as usual or a combined treatment involving cognitive behavioral social skills training administered in 24 weekly group sessions. The authors concluded that the subjects who received cognitive behavioral social skills training learned coping skills, evaluated anomalous experiences with more objectivity (achieved greater cognitive insight), and had improved social functioning.

In the Netherlands, Valmaggia et al. (74) compared 16 sessions of manualized CBT with supportive counseling in a randomized trial in inpatients with resistant symptoms of schizophrenia. They reported that the participants who received CBT (n = 36) showed improvements in auditory hallucinations and insight at the post-therapy assessment compared with those who received supportive counseling (n = 26) but that these findings were not maintained at follow-up. However, the sample size was small and, due to strict inclusion criteria, the results cannot be widely generalized.

Review articles

Over the past 5–10 years, a number of comprehensive reviews and meta-analyses have discussed findings concerning the impact of CBT on the psychopathology of schizophrenia (75–88). Most of these reviews did not focus solely on treatment resistance; rather, they considered the impact of CBT on the general psychopathology of schizophrenia. One review by Garety et al. (76) did focus on CBT for medication-resistant symptoms. In that study, they looked at four controlled trials and found that CBT reduced symptoms of psychosis, and that there was also some evidence that it may contribute to relapse reduction. In the rest of this section, we discuss several important systematic reviews and meta-analyses that have contributed to the growing evidence for the efficacy of CBT for the psychopathology of schizophrenia.

Rector and Beck (83) examined seven randomized, controlled trials that tested the efficacy of CBT for schizophrenia. A review by Pilling et al. (84) included the results from eight randomized trials. A review by the National Institute of Clinical Excellence (NICE) (85) published in 2002 included 13 randomized clinical trials with data from 1297 patients. The NICE guidelines (85) concluded that “longer treatments with CBT are significantly more effective than shorter ones, which may improve depressive symptoms but are unlikely to improve psychotic symptoms. An adequate course of CBT to generate improvements in psychotic symptoms in these circumstances should be of more than 6 months duration and include more than 10 planned sessions.”

Gould et al. (86) reviewed controlled treatment outcome studies of cognitive therapy (CT) for psychotic symptoms in schizophrenia. Effect sizes were calculated for seven studies involving 340 subjects. The mean effect size (ES) for reduction of psychotic symptoms was 0.65. The findings suggest that CT is an effective treatment for patients with schizophrenia who have persistent psychotic symptoms. Follow-up analyses in four studies indicated that patients receiving CT continued to make gains over time (ES = 0.93).

In 2004, Tarrier and Wykes (87) reviewed 20 controlled trials of CBT for schizophrenia, most of which focused on treatment-resistant schizophrenia, and reported a mean ES of 0.37. They found that the effect size of the trials was significantly and negatively correlated with their methodological quality and concluded that overall there was good evidence for the efficacy and effectiveness of CBT in the treatment of schizophrenia.

Zimmermann et al. (82) reviewed 14 studies published between 1990 and 2004 that included 1484 patients. Based on a meta-analysis of the results of these studies, they concluded that, compared with other adjunctive measures, CBT produced a significant reduction in positive symptoms and had greater benefit for patients suffering from an acute psychotic episode rather than a chronic condition (i.e., presentation characterized by residual symptoms) (effect size of 0.57 vs. 0.27)

Pfammatter et al. (88) reviewed the evidence for the efficacy of psychological therapies for schizophrenia derived from meta-analyses on the topic and supplemented these data with selected findings from their own recent meta-analysis. They concluded that CBT for persistent positive symptoms is an effective adjunct to pharmacotherapy (ES = 0.47)

While the studies described in these reviews differed in various factors such as duration of intervention, number of sessions, comparison treatment, and outcomes post-intervention and at follow-up, and some studies had methodological problems (e.g., small sample sizes (53), different levels of baseline symptoms (53), and absence of experimental blinding to treatment conditions (56, 61), a number of common findings emerged. CBT appears to be particularly effective as an adjunct to medication in helping people with the psychopathology of schizophrenia, especially persistent symptoms of psychosis.

CONCLUSIONS AND FUTURE DIRECTIONS

CBT has emerged as an effective adjuvant to antipsychotic medication in the treatment of schizophrenia. It should be considered as a component of a comprehensive treatment package along with antipsychotic medication. However, unanswered questions remain that need further research. Most of the study publications that reported that CBT is useful as an adjunct for the treatment of medication-resistant schizophrenia did not include an adequate discussion of study limitations. In particular, they did not adequately address the fact that the patients in most of these studies were relatively unrepresentative of the general patient population, which could lead to problems with external validity, nor did they discuss study biases, which could lead to problems with internal validity. As is frequently the case, the patients who were included were often geographical convenience samples. Studies also often excluded patients with comorbid substance addiction. Issues related to investigator allegiance could also have had an impact on the study design (89). Therapists trained in only one modality, such as CBT, should not attempt to deliver a comparator therapy such as supportive counselling. Therefore, future studies should compare CBT with psychosocial interventions such as social skills training or psychoeducation packages that are delivered by experts in these modalities. A number of reviews have acknowledged the superiority of CBT over routine care, but when CBT is compared with other therapies that provide one-on-one attention, relative benefits are less apparent, especially at 1 or 2 year follow-up (64). This is a problem since the active ingredients of CBT have not been identified. No research studies to date have been adequately designed to detect the true differences between therapies (89) or the active ingredients that are effective in CBT.

CBT has limitations and is not effective for all patients. For example, it may not be possible to provide CBT during a very florid first episode of psychosis, during a relapse with severe impairment of insight, or in the setting of persistent refusal of medication. Questions also remain as to whether CBT for schizophrenia as currently developed is effective cross-culturally (12). A study of the effects of CBT on insight in the United Kingdom found some evidence of cultural differences in response to CBT, with higher dropout rates and lower rates of improvement with CBT among African Caribbean participants (12). There may be cultural barriers to forming an alliance or developing formulations, especially when the therapist and patient are from different cultural backgrounds; this is an area that needs further research.

CBT for schizophrenia is more widely practiced in the United Kingdom than in the United States (90, 91). Differences in the approaches to healthcare research and delivery (e.g., the presence of universal health care in the United Kingdom and its absence in the United States) may be a contributing factor. Furthermore, the myth that CBT contradicts the biological model of schizophrenia may form a barrier to implementation in the United States. However, despite the more widespread acceptance of CBT for schizophrenia in the United Kingdom, inadequate numbers of trained providers and other problems with treatment resources also create limitations on service delivery there.

Future directions for CBT research include more robust and well designed studies to determine the active and effective ingredients of CBT, studies focusing on early intervention in psychosis (92, 93), and the development of culturally modified explanations for use in CBT for psychosis.