The Efficacy of Cognitive Behavioral Therapy: A Review of Meta-analyses

Methods

Search Strategy and Study Selection

To obtain the articles for this review, we searched PubMed, PsychInfo, and Cochrane library databases using the following key words: meta-analysis AND cognitive behav*, meta-analysis AND cognitive therapy, quantitative review AND cognitive behav*, quantitative review AND cognitive therapy. This initial search yielded 1,163 hits, of which 355 were duplicates and had to be excluded. The remaining 808 non-duplicate articles were further examined to determine if they met specific inclusionary criteria for the purposes of this review. All included studies had to be quantitative reviews (i.e., meta-analyses) of CBT. In order to limit this review to contemporary studies, only articles published since 2000 were included. The final sample included in this review consisted of 269 meta-analyses (Figure 1). Out of those, we described a representative sample of 106 meta-analytic studies. The complete reference list for the final sample of included meta-analyses can be obtained by accessing the webpage www.bostonanxiety.org/cbtreview.html. As already noted, the majority (84%) of these studies was published after 2004, the most recent year covered by the meta-analysis by Butler and colleagues (2006). The number of meta-analytic reviews per year is depicted in Figure 2.

Figure 1

Flow diagram showing effects of inclusionary and exclusionary criteria on final sample selection.

Figure 2

Number of meta-analyses published by year since 2000. Note that the number of studies corresponding to 2011 only covered studies until September of that year.

Search Strategy and Study Selection

To obtain the articles for this review, we searched PubMed, PsychInfo, and Cochrane library databases using the following key words: meta-analysis AND cognitive behav*, meta-analysis AND cognitive therapy, quantitative review AND cognitive behav*, quantitative review AND cognitive therapy. This initial search yielded 1,163 hits, of which 355 were duplicates and had to be excluded. The remaining 808 non-duplicate articles were further examined to determine if they met specific inclusionary criteria for the purposes of this review. All included studies had to be quantitative reviews (i.e., meta-analyses) of CBT. In order to limit this review to contemporary studies, only articles published since 2000 were included. The final sample included in this review consisted of 269 meta-analyses (Figure 1). Out of those, we described a representative sample of 106 meta-analytic studies. The complete reference list for the final sample of included meta-analyses can be obtained by accessing the webpage www.bostonanxiety.org/cbtreview.html. As already noted, the majority (84%) of these studies was published after 2004, the most recent year covered by the meta-analysis by Butler and colleagues (2006). The number of meta-analytic reviews per year is depicted in Figure 2.

Figure 1

Flow diagram showing effects of inclusionary and exclusionary criteria on final sample selection.

Figure 2

Number of meta-analyses published by year since 2000. Note that the number of studies corresponding to 2011 only covered studies until September of that year.

Categorization of Meta-analyses

The 269 meta-analyses were categorized into groups to provide the most meaningful and extensive examination of the efficacy of CBT across a range of problem areas and study populations. The major groupings were the following: substance use disorder, schizophrenia and other psychotic disorders, depression and dysthymia, bipolar disorder, anxiety disorders, somatoform disorders, eating disorders, insomnia, personality disorders, anger and aggression, criminal behaviors, general stress, distress due to general medical conditions, chronic pain and fatigue, pregnancy complications and female hormonal conditions. In addition, some meta-analyses specifically examined CBT for disorders in children and elderly adults. For each disorder and population grouping, data were described qualitatively, considering the findings of all meta-analyses within that group. The 269 meta-analyses included a wide variety of studies that employed different methodologies and effect size estimates. Therefore, we used the designation small, medium, and large for the magnitude of effect sizes in our review of the 106 representative meta-analyses (Cohen, 1988). In addition, we provide reported response rates, a widely accepted and common metric in psychiatry, from a subsample of 11 studies that examined the efficacy of CBT in randomized controlled trials.

Results

Addiction and Substance Use Disoder

There was evidence for the efficacy of CBT for cannabis dependence, with evidence for higher efficacy of multi-session CBT versus single session or other briefer interventions, and a lower drop out rate compared to control conditions (Dutra et al., 2008). However, the effect size of CBT was small as compared to other psychosocial interventions (e.g. contingency management, relapse prevention, and motivational approaches) for substance dependence, and agonist treatments showed a greater effect size than CBT in certain drug dependencies, such as opioid and alcohol dependence (Powers, Vedel, & Emmelkamp, 2008).

Treatments for smoking cessation found that coping skills, which were partially based on CBT techniques, were highly effective in reducing relapse in a community sample of nicotine quitters (Song, Huttunen-Lenz, & Holland, 2010), and another meta-analysis noted superiority of CBT (either alone or in combination with nicotine replacement therapy) over nicotine replacement therapy alone (Garcia-Vera & Sanz, 2006). Furthermore, there was evidence for superior performance of behavioral approaches in the treatment of problematic gambling as compared to control treatments (Oakley-Browne et al., 2000). One meta-analysis (Leung & Cottler, 2009) reported larger effect sizes of CBT when this treatment was grouped with other non-pharmacological treatments (such as brief interventions) as compared to pharmacological agents (e.g. naltrexone, carbamazepine, and topiramate), but CBT was not more efficacious than these other briefer, less expensive approaches.

Schizophrenia and Other Psychotic Disorders

Meta-analyses examining the efficacy of psychological treatments for schizophrenia revealed a beneficial effect of CBT on positive symptoms (i.e., delusions and/or hallucinations) of schizophrenia (e.g., Gould et al., 2001; Rector & Beck, 2001). There was also evidence (e.g., Zimmerman et al., 2005) that CBT is a particularly promising adjunct to pharmacotherapy for schizophrenia patients who suffer from an acute episode of psychosis rather than a more chronic condition.

CBT appeared to have little effect on relapse or hospital admission compared to other interventions, such as early intervention services or family intervention (e.g., Bird et al., 2010; Álvarez-Jiménez et al., 2011). However, CBT had a beneficial effect on secondary outcomes. For example, a more recent meta-analysis by Wykes and colleagues (2008) examined controlled trials of CBT for schizophrenia and confirmed findings from previous meta-analyses (e.g., Gould et al., 2001; Rector & Beck, 2001), suggesting that CBT had a small to medium effect size as compared to control conditions on both positive and negative symptoms. In addition, this meta-analysis revealed medium effect sizes for improvements in secondary outcomes that were not the direct targets of treatment, including general functioning, mood, and social anxiety.

Depression and Dysthymia

CBT for depression was more effective than control conditions such as waiting list or no treatment, with a medium effect size (van Straten, Geraedts, Verdonck-de Leeuw, Andersson, & Cuijpers, 2010; Beltman, Oude Voshaar, & Speckens, 2010). However, studies that compared CBT to other active treatments, such as psychodynamic treatment, problem-solving therapy, and interpersonal psychotherapy, found mixed results. Specifically, meta-analyses found CBT to be equally effective in comparison to other psychological treatments (e.g., Beltman, Oude Voshaar, & Speckens, 2010; Cuijpers, Smit, Bohlmeijer, Hollon, & Andersson, 2010; Pfeiffer, Heisler, Piette, Rogers, & Valenstein, 2011). Other studies, however, found favorable results for CBT (e.g. Di Giulio, 2010; Jorm, Morgan, & Hetrick, 2008; Tolin, 2010). For example, Jorm and colleagues (2008) found CBT to be superior to relaxation techniques at post-treatment. Additionally, Tolin (2010) showed CBT to be superior to psychodynamic therapy at both post-treatment and at six months follow-up, although this occurred when depression and anxiety symptoms were examined together.

Compared to pharmacological approaches, CBT and medication treatments had similar effects on chronic depressive symptoms, with effect sizes in the medium-large range (Vos, Haby, Barendregt, Kruijshaar, Corry, & Andrews, 2004). Other studies indicated that pharmacotherapy could be a useful addition to CBT; specifically, combination therapy of CBT with pharmacotherapy was more effective in comparison to CBT alone (Chan, 2006).

Bipolar Disorder

Meta-analyses examining the efficacy of CBT for bipolar disorder revealed small to medium overall effect sizes of CBT at post-treatment, with effects typically diminishing slightly at follow-up. These findings emerged from examinations of both manic and depressive symptoms associated with bipolar disorder (e.g., Gregory, 2010a, 2010b). There is little evidence that CBT as a stand-alone treatment (rather than as an adjunct to pharmacotherapy) is effective for the treatment of bipolar disorder.

In addition to examining CBT for attenuating symptoms of bipolar disorder, some meta-analyses focused on the efficacy of CBT for preventing relapse in bipolar patients. One study (Beynon et al., 2008) examined the efficacy of CBT for preventing relapse and found it to be somewhat effective when comparing CBT vs. treatment as usual. Overall, CBT for bipolar disorder was an effective method of preventing or delaying relapses (e.g., Lam, Burbeck, Wright, & Pilling, 2009; Cakir & Ozerdem, 2010). Furthermore, the efficacy of CBT at preventing relapse did not seem to be influenced by the number of previous manic or depressive episodes.

Anxiety Disorders

In general, CBT is a reliable first-line approach for treatment of this class of disorders (Hofmann & Smits, 2008), with support for significant positive effects of CBT on secondary symptoms such as sleep dysfunction and anxiety sensitivity (Ghahramanlou, 2003). Further, internet-delivered or guided self-help CBT showed some promise in immediate symptom relief as compared to no treatment, but the long-term maintenance with this modality of CBT remains unclear (Öst, 2008; Coull & Morris, 2011).

CBT for social anxiety disorder evidenced a medium to large effect size at immediate post-treatment as compared to control or waitlist treatments, with significant maintenance and even improvement of gains at follow-up (Gil, Carrillo, & Meca, 2001). Further, exposure, cognitive restructuring, social skills training and both group/individual formats were equally efficacious (Powers, Sigmarsson, & Emmelkamp, 2008), with superior performance over psychopharmacology in the long term (Fedoroff & Taylor, 2001). Similarly, interoceptive exposure for treatment of panic disorder was moderately effective and superior to control/pill placebo treatments and applied relaxation (Haby, Donnelly, Corry, & Vos, 2006; Furukawa, Watanabe, & Churchill, 2007). For panic disorder without agoraphobia, combination treatment of CBT and applied relaxation was equal in efficacy to use of either therapy approach alone, and use of either or both were superior to use of medications (Mitte, 2005).

Various CBT techniques for specific phobia (systematic desensitization, exposure, cognitive therapy) were as effective as applied relaxation and applied tension, producing effect sizes in the large range, with long-term maintenance of gains (Ruhmland & Margraf, 2001). For generalized anxiety disorder, CBT was superior as compared to control or pill placebo conditions, and equally efficacious as relaxation therapy, supportive therapy, or psychopharmacology, but less efficacious in comparison to attention placebos and in those with more severe generalized anxiety disorder symptoms.

CBT for post-traumatic stress disorder was equal in efficacy to eye movement desensitization and reprocessing (Bisson et al., 2007), with both being superior to treatment as usual, waitlist, or other treatments (such as supportive counseling) for post-traumatic stress disorder (Bisson & Andrew, 2008). However, it is questionable whether the eye-movement technique is an active treatment ingredient.

Clinical trials also revealed a large effect size for CBT and/or exposure response prevention for obsessive compulsive disorder, with evidence suggesting that a combination of in vivo and imaginal exposures outperformed the use of only in vivo exposures (Ruhmland & Margraf, 2001). Furthermore, CBT was found to be similarly efficacious than clomipramine and selective reuptake inhibitors (Eddy, Dutra, Bradley, & Westen, 2004).

Somatoform Disorders

Within the somatoform disorders category of DSM-IV, meta-analyses primarily examined the efficacy of psychological interventions for hypochondriasis and body dysmorphic disorder. One meta-analysis found a large mean effect size for CBT, which outperformed other psychological treatments (i.e., psychoeducation, explanatory therapy, cognitive therapy, exposure and response prevention, and behavioral stress management), with effect sizes in the large range, as well as pharmacotherapy treatments (paroxetine, fluoxetine, fluvoxamine, and nefazodone), which also evidenced large effect sizes (Taylor, Asmundson, & Coons, 2005). The mean effect size for control conditions (e.g., wait-list control) was small. These results were partially supported by other evidence, as a more recent meta-analysis found superior outcomes of CBT for hypochondriasis compared to waiting list control, usual medical care or placebo at twelve-month follow-up (Thomson & Page, 2007). However, this meta-analysis also found no differences between CBT and waiting list/placebo at post-treatment.

Meta-analyses comparing the efficacy of CBT to control treatments found that CBT was superior in significantly reducing body dysmorphic disorder symptoms (Ipser, Sander, & Stein, 2009). In comparing relative efficacy of CBT versus pharmacotherapy, effect sizes were large on body dysmorphic disorder severity measures for CBT, and ranged from medium to large for pharmacotherapy (Williams, Hadjistavropoulos, & Sharpe, 2006). In addition, another meta-analysis found that CBT for body image disturbances was effective, with effect sizes ranging from medium to large (Jarry & Ip, 2005).

Eating Disorders

For bulimia nervosa, meta-analyses compared the efficacy of CBT to control treatments and found effect sizes in the medium range (Thompson-Brenner, 2002). However, the effect of behavior therapy was greater than that of CBT, with the average effect size for behavior therapy in the large range (Thompson-Brenner, 2003). Another meta-analysis comparing CBT with control treatments found remission response rates to be higher for CBT, with a medium relative risk ratio (Hay, Bacaltchuk, Stefano, & Kashyap, 2009). When comparing CBT to other psychotherapies, specifically, interpersonal therapy, dialectical behavioral therapy, hypno-behavioral therapy, supportive psychotherapy, behavioral weight loss treatment, and self-monitoring, CBT fared significantly better in remission response rates for bulimia nervosa, with a large relative risk ratio (Hay et al., 2009).

For binge eating disorder, a recent meta-analysis found that psychotherapy and structured self-help yielded large effect sizes, when compared to pharmacotherapy, which yielded medium effect sizes (Vocks et al., 2010). Although this study did not parse out the efficacy of CBT specifically, a majority of the included trials for psychotherapy involved CBT (19 out of 23 trials). Furthermore, a review and meta-analysis by Reas and Grilo (2008) suggested that combination treatment of psychotherapy and medications did not enhance binge-eating outcomes, but may have enhanced weight loss outcomes.

Insomnia

CBT for insomnia (CBT-I) has long been shown to be more efficacious than control treatments. A recent meta-analysis examined its efficacy on both subjective and objective sleep parameters in comparison to a control group for individuals with primary insomnia (Okajima, Komada, & Inoue, 2011). Effect sizes for the efficacy of CBT-I versus control at the end of treatment on subjective sleep measures, which included sleep onset latency, total sleep time, wake after sleep onset, total wake time, time in bed, early morning awakening, and sleep efficiency, ranged from minimal (total sleep time) to large (early morning awakening) (Okajima et al., 2011). For objective measures using a polysomnogram or actigraphic evaluation, effect sizes ranged from small (total sleep time) to large (total wake time) (Okajima et al., 2011). These findings were consistent with results from another meta-analysis, which examined the relative efficacy of behavioral interventions for insomnia including CBT, relaxation, and only behavioral techniques (Irwin, Cole, & Nicassio, 2006). This study reported effect sizes ranging from −.75 to 1.47 for CBT, −.60 to .53 for relaxation techniques, and −.82 to .91 for only behavioral techniques on subjective sleep outcomes.

Personality Disorders

There was one meta-analysis that examined the relative efficacy of CBT versus psychodynamic therapy for the treatment of personality disorders (Leichsenring & Leibing, 2003). The findings indicated a larger overall effect size for psychodynamic therapy compared to CBT. This was consistent with observer-rated measures, which showed a similar pattern of effect sizes: stronger for psychodynamic therapy than for CBT (although this effect size was also large). Self-report measures, however, indicated larger effect sizes for CBT than for psychodynamic therapy.

Another meta-analysis compared the efficacy of eleven different psychological therapies, including CBT, for antisocial personality disorder (Gibbon et al., 2010). Results suggested that compared to control treatment, CBT plus standard maintenance was more efficacious in terms of leaving the study early and cocaine use for outpatients with antisocial personality disorder and comorbid cocaine dependence. However, CBT plus treatment as usual was not better than a control condition for these antisocial personality disorder patients with regard to levels of recent verbal or physical aggression. The relative efficacy of psychological treatments for borderline personality disorder, in particular, was also examined, which yielded no differences between dialectical behavioral therapy and treatment as usual in individuals meeting criteria for borderline personality disorder at six months, or in hospital admissions in the previous three months (Binks et al., 2009).

Anger and Aggression

Two meta-analytic reviews focused on anger control problems and aggression (Del Vecchio & O’Leary, 2004; Saini, 2009). The findings from these meta-analyses suggested that CBT is moderately effective at reducing anger problems. Findings from these reviews also suggested that CBT may be most effective for patients with issues regarding anger expression.

CBT produced medium effect sizes as compared to other psychosocial treatments and control conditions across the two reviews that conducted quantitative analyses. A meta-analysis on the effectiveness of anger treatments for specific anger problems (Del Vecchio & O’Leary, 2004) included only studies in which subjects met clinically significant levels of anger on standardized anger measurements prior to treatment. This meta-analysis examined the effects of CBT, cognitive therapy, relaxation, and ‘other’ (e.g., social skills training, process group counseling) on various anger problems including driving anger, anger suppression, and anger expression difficulties.

Criminal Behaviors

Four separate meta-analytic studies supported the efficacy of CBT for criminal offenders (Illescas, Sanchez-Meca, & Genovés, 2001; Lösel & Schmucker, 2005; Pearson, Lipton, Cleland, & Yee, 2002; Wilson, Bouffard, Mackenzie, 2005). Out of several theoretical orientations and types of psychological interventions for criminal activity, behavior therapy and CBT appeared to be the superior interventions in reducing recidivism rates, both with medium mean effect sizes (Illescas, Sanchez-Meca, & Genovés, 2001). Effect sizes for other interventions ranged from small to medium (Illescas et al., 2001). Another study demonstrated consistent findings with a small weighted mean effect size of behavior therapy or CBT for reducing recidivism (Pearson, Lipton, Cleland, & Yee, 2002). Similarly, Wilson and colleagues (2005) found an overall small-to-medium mean effect size for CBT programs for convicted offenders.

For sexual offenders in particular, physical treatments, such as surgical castration and hormonal treatment, were demonstrated to have greater efficacy in reducing sexual recidivism in comparison to CBT, with large significant odds ratios for both of these alternative interventions (Lösel & Schmucker, 2005). Of the various psychological interventions for sexual offenders, however, classical behavioral and CBT approaches indicated the strongest efficacy, with odds ratios in the medium to large range (Lösel & Schmucker, 2005) as compared to insight-oriented and therapeutic community interventions.

A study of CBT for domestic violence indicated no differences between CBT and the Duluth model (which is based on a feminist psycho-educational approach) for treating domestically violent males (Babcock, Green, & Robie, 2004). The aggregated data from experimental and quasi-experimental studies showed that CBT had an overall small effect size, and the Duluth model had an overall slightly larger, but still small effect size (Babcock et al., 2004).

General Stress

Four meta-analyses examined occupational stress and the majority of their results were quite similar: CBT interventions were more effective in comparison to other intervention types such as organization focused therapies, especially when CBT focused on psycho-social outcomes in employees (Kim, 2007; Richardson & Rothstein, 2008; van der Klink, Blonk, Schene, & van Dijk, 2001). For example, Richardson and Rothstein (2008) found CBT alone to be more effective in comparison to CBT combined with additional psychological components. These studies found a large effect size for overall CBT interventions, large effect size for single-mode CBT interventions, and small effect size for CBT interventions with four or more components. In contrast, Marine and colleagues (2006) chose not to compare CBT with other interventions, such as relaxation techniques for psychological stress, because most interventions comprised both elements and could not be evaluated separately. With respect to stress in parents of children with developmental disabilities, positive effects were found for CBT, but the effect size was relatively small (Singer, Ethridge, & Aldana, 2007). In contrast to the results of Richardson and Rothstein (2008), this meta-analysis found multiple component interventions which combined CBT, behavioral parent training and in some cases other forms of support services, to have a higher and large effect size in comparison to CBT alone (Singer, Ethridge, & Aldana, 2007).

Distress Due to General Medical Conditions

Limited well-controlled studies existed in the study of non-ulcer dyspepsia, multiple sclerosis, physical disability following traumatic injury, non-epileptic seizures, post-concussion syndrome, chronic obstructive pulmonary disease, hypertension, Type II diabetes, and burning mouth syndrome (e.g. Soo et al., 2004; Thomas, Thomas, Hillier, Galvin, & Baker, 2006; Baker, Brooks, Goodfellow, Bodde, & Aldenkamp, 2007; Ismail, Winkley, & Rabe-Hesketh, 2004). However, cancer was studied more rigorously and with more robust methodological attention, indicating small to medium effect sizes of individual CBT as compared to patient education only in gynecological and head/neck cancers (Zimmerman & Heinrichs, 2006; Luckett, Britton, Clover, & Rankin, 2011), on secondary outcomes such as quality of life, psychological distress (i.e., depression and anxiety), and pain. Further, CBT was shown to be equally effective as exercise interventions in treating cancer-related fatigue (Kangas, Bovbjerg, & Montgomery, 2008).

Small to medium effect sizes were observed in treatment of secondary symptoms (anxiety and stress) experienced by individuals who were HIV positive, with particular efficacy (particularly for stress management) in reducing anger symptoms as compared to supportive therapy (Crepaz et al., 2008), but not for outcomes such as low cell count, medication adherence, or when used with marginalized populations such as ethnic minorities and women (Crepaz et al., 2008; Rueda et al., 2006).

CBT was shown to be superior in the treatment of secondary symptoms of spinal cord injury as compared to controls in assertiveness skills, coping, depression and quality of life (Dorstyn, Mathias, & Denson, 2011), better than placebo or diet/exercise alone (Shaw, O’Rourke, Del Mar, & Kenardy, 2005), but equal to yoga/education in depressive symptoms (Martinez-Devesa, Perera, Theodoulou, & Waddell, 2010). CBT was only slightly more effective than usual care or waitlist condition in the treatment of irritable bowel syndrome, with peppermint oil having greater efficacy in providing relief in this particular disorder (Enck, Junne, Klosterhalfen, Zipfel, & Martens, 2010).

Chronic Pain and Fatigue

Meta-analyses examining the efficacy of psychosocial treatments for chronic pain have investigated chronic low back pain, fibromyalgia, rheumatoid arthritis, chronic fatigue syndrome, chronic musculoskeletal pain, and non-specific chest pain. These reviews have examined the effect of a range on treatments on chronic pain, including relaxation techniques, mindfulness-based techniques, acceptance-based techniques, biofeedback, psycho-education, and behavioral and cognitive-behavioral treatments. Results of these meta-analyses revealed varying effect sizes for these treatments depending on the type of chronic pain targeted; however, CBT treatments for chronic pain were consistently in the small to medium effect size range.

Similar results were found in a meta-analysis examining psychological treatments for fibromyalgia (Glombiewski et al., 2010). This meta-analysis revealed that CBT was superior to other psychological treatments for decreasing pain intensity. Pre-post analyses revealed a medium effect size for CBT as compared to a small effect size for all other psychological treatments combined (excluding CBT). CBT treatments for chronic fatigue syndrome were moderately effective (e.g., Malouff et al., 2008; Price et al., 2008). Malouff and colleagues (2008) conducted a meta-analysis revealing a medium effect size in post-treatment fatigue for participants receiving CBT versus those in control conditions.

Pregnancy Complications and Female Hormonal Conditions

One meta-analysis found CBT to be more effective in comparison to control conditions for perinatal depression (Sockol, Epperson, & Barber, 2011), and another meta-analysis found beneficial effects of CBT for postnatal depression, but these results need to be interpreted with caution because it is difficult to causally link depression with pregnancy and hormonal changes in these studies (Dennis, & Hodnett, 2007). Further, Bledsoe and Grote (2006) found greater decreases in depression for women experiencing non-psychotic major depression in pregnancy and postnatal periods treated with combination treatment in comparison to antidepressant medication alone, which was itself more effective in comparison to CBT alone. The effect size for postnatal treatments was large in comparison to the small to medium effects of prenatal treatments, but when pharmacological treatments were excluded, the effect size for postnatal treatments decreased to the medium range.

For the treatment of premenstrual syndrome, Busse and colleagues (2009) found that CBT significantly reduced depressive and anxiety symptoms associated with this syndrome, as indicated by a medium effect size. Once again, these results need to be interpreted carefully due to the small number of well-controlled studies on which these reviews were based.

CBT for Special Populations

Children

Within internalizing symptoms, there was support for the preferential use of CBT approaches in treatment of anxiety disorders in children and adolescents, with effect sizes in the large range (Santacruz et al., 2002; James, Soler, & Weatherall, 2005). Further, CBT treatment for obsessive compulsive disorder as compared to alternative approaches (no treatment, other psychosocial treatments and medications such as clomipramine and fluvoxamine) resulted in significantly better outcomes (Phillips, 2003; Guggisberg, 2005). The data supporting CBT for depression was less strong, but still in the medium effect size range across meta-analyses, with maintenance in 6-month follow-up periods (Santacruz et al., 2002). In addition, CBT seemed to work equally well as other psychotherapies (i.e. interpersonal therapy and family systems therapy), but was regarded as superior to selective reuptake inhibitors due to reduced chance of side effects and greater cost effectiveness (Haby, Tonge, Littlefield, Carter & Vos, 2004). The studies on efficacy of CBT for addressing suicidal behaviors were scarce (Robinson, Hetrick, & Martin, 2011), and warrant further investigation.

The picture was more mixed for other disorders, with CBT showing equal efficacy in reducing disruptive classroom behaviors and aggressive/antisocial behaviors, as other psychosocial treatments, better efficacy as compared to no treatment or treatment as usual, and less efficacy than pharmacological approaches (Lösel & Beelmann, 2003; Özabaci, 2011). Similarly, CBT for attention deficit hyperactivity disorder showed some efficacy, but was not superior to medications (Van der Oord, Prins, Oosterlaan, & Emmelkamp, 2008). The efficacy of behavioral techniques (e.g. motivational enhancement and behavioral contingencies) was small to medium for the treatment of adolescent smoking and substance use as compared to no treatment, but not more so than other psychotherapies. In addition, there was a medium to large effect size of CBT over waitlist across meta-analyses examining chronic headache pain. Finally, the data on efficacy for CBT in juvenile sex offenders, childhood sexual abuse survivors, childhood obesity, fecal incontinence, and juvenile diabetes was limited, showing preliminary support for CBT as compared to no treatment, but equal efficacy to other psychosocial approaches (Walker, McGovern, Poey, & Otis, 2005; Macdonald, Higgins, & Ramchandani, 2006).

Elderly Adults

With respect to mood disorders, with depression as the most commonly examined disorder, nearly all meta-analyses showed that CBT was more effective than waiting list control conditions, but equally effective in comparison to other active treatment methods, such as reminiscence, (an intervention that uses recall of past events, feelings and thoughts to facilitate pleasure, quality of life or adaptation to the present; Peng, Huang, Chen, & Lu, 2009), psychodynamic therapy, and interpersonal therapy (Krishna et al., 2011; Wilson, Mottram, & Vassilas, 2008). Pinquart and colleagues (2007), however, found a large effect size for CBT, whereas the effect sizes for the other active treatment conditions were in the medium-large range. When long-term outcomes were examined, results of one meta-analysis indicated that treatment gains of CBT for depression were maintained at 11-months follow-up (Krishna et al., 2011), but long-term follow-up data remained scarce in the other meta-analyses. In a meta-analysis assessing the additive effects of CBT and pharmacological approaches, Peng and colleagues (2009) found that CBT was more effective in comparison to placebo, but CBT as an adjunct to antidepressant medication did not increase the effectiveness of antidepressants in this population.

For anxiety disorders in the elderly, CBT (alone or augmented with relaxation training) did not enhance outcomes beyond relaxation training alone (Thorp et al., 2009), although many of these studies were uncontrolled. In contrast to the findings by Thorp and colleagues (2009), Hendriks and colleagues (2008) found that anxiety symptoms were significantly decreased following CBT than after either a waiting-list control condition or other treatment methods. Additionally, CBT significantly alleviated accompanying symptoms of worry and depression when compared to waiting-list control or an active control condition.

Response Rates of Randomized Controlled Studies

The meta-analytic studies that provided response rates are listed in Table 1. The response rates of CBT varied between 38% for treating obsessive compulsive disorder (Eddy et al., 2004) and 82% for treating body dysmorphic disorder (Ipser et al, 2009). In contrast, the response rates of the waitlist groups ranged from 2% for the treatment of bulimia nervosa (Thompson-Brenner, 2003) to 14% for generalized anxiety disorder (Hunot et al., 2007). CBT also demonstrated higher response rates in comparison to treatment as usual in treatment of generalized anxiety disorder and chronic fatigue (Price et al., 2008), and higher or equal response rates as compared to other therapies or psychopharmacological interventions in most studies. CBT only produced a lower response rate than psychodynamic therapy for the personality disorders (47% vs. 59%; Leichsenring & Leibing, 2003).

Discussion

CBT is arguably the most widely studied form of psychotherapy. We identified 269 meta-analytic reviews that examined CBT for a variety of problems, including substance use disorder, schizophrenia and other psychotic disorders, depression and dysthymia, bipolar disorder, anxiety disorders, somatoform disorders, eating disorders, insomnia, personality disorders, anger and aggression, criminal behaviors, general stress, distress due to general medical conditions, chronic pain and fatigue, distress related to pregnancy complications and female hormonal conditions. Additional meta-analytic reviews examined the efficacy of CBT for various problems in children and elderly adults. The vast majority of studies (84%) was published after 2004, which was the last year of coverage of the review by Butler and colleagues (2006), making the present study the most comprehensive and contemporary review of meta-analytic studies of CBT to date.

For the treatment of addiction and substance use disorder, the effect sizes of CBT ranged from small to medium, depending on the type of the substance of abuse. CBT was highly effective for treating cannabis and nicotine dependence, but less effective for treating opioid and alcohol dependence. For treating schizophrenia and other psychotic disorders, the empirical literature suggested appreciable efficacy of CBT particularly for positive symptoms and secondary outcomes in the psychotic disorders, but lesser efficacy than other treatments (e.g. family intervention or psychopharmacology) for chronic symptoms or relapse prevention.

The meta-analytic literature on the efficacy of CBT for depression and dysthymia was mixed with some studies suggesting strong evidence and others reporting weak support. Some authors have suggested that the strong effects in some studies may be an overestimation due to a publication bias (Cuijpers, et al., 2010). Similarly, the efficacy of CBT for bipolar disorder was small to medium in the short-term in comparison to treatment as usual. However, there was limited evidence for the superiority of CBT alone over pharmacological approaches; for the treatment of depressive symptoms in bipolar disorder, the use of CBT was well supported. However, the long-term superiority compared to other treatments is still uncertain.

The efficacy of CBT for anxiety disorders was consistently strong, despite some notable heterogeneity in the specific anxiety pathology, comparison conditions, follow-up data, and severity level. Large effect sizes were reported for the treatment of obsessive compulsive disorder, and at least medium effect sizes for social anxiety disorder, panic disorder, and post-traumatic stress disorder. Medium to large CBT treatment effects were reported for somatoform disorders, such as hypochondriasis and body dysmorphic disorder. However, more studies using larger trials and greater sample sizes are needed to draw more conclusive findings with regard to CBT’s relative efficacy in comparison to other active treatments.

For the treatment of bulimia, CBT was considerably more effective than other forms of psychotherapies, but less is known for other eating disorders. Similarly, CBT demonstrated superior efficacy as compared to other interventions for treating insomnia when examining sleep quality, total sleep time, waking time, and sleep efficiency outcomes. However, although there were small effects of CBT for sleep problems among older adults (aged 60+), these effects may not be long lasting (Montgomery & Dennis, 2009).

For personality disorders, there was some evidence for superior efficacy of CBT as compared to other psychosocial treatments for the personality disorders. However, the studies showed considerable variation in measurement methods, comorbid disorders, and demographic variables. CBT also produced medium to large effect sizes for treating anger and aggression (e.g., Saini, 2009), although a greater number of well-controlled studies are needed to more adequately parse out the specific efficacy of CBT compared to the psychosocial treatments for anger on the whole. Similarly, more studies are needed before any firm conclusions can be drawn about the efficacy of this treatment for criminal behaviors.

As a stress management intervention, CBT was more effective that other treatments, such as organization-focused therapies. However, more research on the long-term effects of CBT for occupational stress is needed. Furthermore, there are open questions about the relative efficacy of CBT versus pharmacological approaches to stress management. Similarly, several common concerns recurred across meta-analytic examinations of CBT for chronic medical conditions, chronic fatigue and chronic pain, namely: (1) a scarcity of studies and small sample sizes; (2) poor methodological design of studies that are included in meta-analyses; and (3) grouping of CBT with a host of other psychotherapies (such as psychodynamic therapy, hypnotherapy, mindfulness, relaxation, and supportive counseling), which made it difficult to parse out whether there are any superior effects of CBT in the majority of medical conditions examined.

There was preliminary evidence for CBT for treating distress related to pregnancy complications and female hormonal conditions. However, more research is needed due to a scarcity of follow-up data and low quality studies. This appeared to be a highly promising area for CBT given that the alternative – pharmacological treatments – can be associated with serious risks of adverse effects for pregnant women and breastfeeding mothers.

In our review of meta-analyses, CBT tailored to children showed robust support for treating internalizing disorders, with benefits outweighing pharmacological approaches in mood and anxiety symptoms. The evidence was more mixed for externalizing disorders, chronic pain, or problems following abuse. Moreover, there remains a need for a greater number of high-quality trials in demographically diverse samples. Similarly, CBT was moderately efficacious for the treatment of emotional symptoms in the elderly, but no conclusions about long-term outcomes of CBT or combination therapies consisting of CBT, and medication could be made.

Finally, our review identified 11 studies that compared response rates between CBT and other treatments or control conditions. In 7 of these reviews, CBT showed higher response rates than the comparison conditions, and in only one review (Leichsenring & Leibig, 2003), which was conducted by authors with a psychodynamic orientation, reported that CBT had lower response rates than comparison treatments.

In sum, our review of meta-analytic studies examining the efficacy of CBT demonstrated that this treatment has been used for a wide range of psychological problems. In general, the evidence-base of CBT is very strong, and especially for treating anxiety disorders. However, despite the enormous literature base, there is still a clear need for high-quality studies examining the efficacy of CBT. Furthermore, the efficacy of CBT is questionable for some problems, which suggests that further improvements in CBT strategies are still needed. In addition, many of the meta-analytic studies included studies with small sample sizes or inadequate control groups. Moreover, except for children and elderly populations, no meta-analytic studies of CBT have been reported on particular subgroups, such as ethnic minorities and low income samples.

Despite these weaknesses in some areas, it is clear that the evidence-base of CBT is enormous. Given the high cost-effectiveness of the intervention, it is surprising that many countries, including many developed nations, have not yet adopted CBT as the first-line intervention for mental disorders. A notable exception is the Improving Access to Psychological Therapies initiative by the National Health Commissioning in the United Kingdom (Rachman & Wilson, 2008). We believe that it is time that others follow suit.

Acknowledgments