Clinical Application of Cognitive Behavioural Treatment of Depression

by William C. Sanderson, PhD

Cognitive behavioral treatment (CBT) of depression involves the application of specific, empirically supported strategies focused on depressogenic information processing (e.g., Beck et al., 1979) and behavior (e.g., Lewinsohn et al., 1986). In order to alleviate depressive affect, treatment is directed at the following three domains: cognition, behavior, physiology (cf: Klosko & Sanderson, 1999 for a session-by-session description). In the cognitive domain, patients learn to apply cognitive restructuring techniques so that negatively distorted thoughts underlying depression can be corrected, leading to more logical and adaptive thinking. Within the behavioral domain, techniques such as activity scheduling, social skills training, and assertiveness training are used to remediate behavioral deficits that contribute to and maintain depression (e.g., social withdrawal, loss of social reinforcement). Finally, within the physiological domain, patients with agitation and anxiety are taught to use imagery, meditation, and relaxation procedures to calm their bodies.

CBT is oriented towards empowering the patient. Within this specific, brief psychotherapeutic treatment modality, the emphasis is on providing patients with skills to offset their depression. One primary goal of CBT is to facilitate the use of treatment techniques outside therapy sessions to create a "positive emotional spiral" wherein patents can implement specific strategies to offset their depressive mood (e.g., cognitive restructuring is used to offset negative thought patterns and the consequent depressive affect, scheduling pleasant activities is used to offset decreased reinforcement secondary to social withdrawal). The major treatment strategies employed within CBT for Depression will be reviewed below in the first section of this paper, with an emphasis on cognitive therapy, the psychological intervention with the most empirical support for the treatment of depression (cf: Depression Guideline Panel, 1993). Research data attesting to the efficacy of CBT is covered in the second part of this chapter.

Cognitive Therapy

Aaron T. Beck and colleagues(1979) initially developed cognitive therapy as a treatment for depression. While cognitive therapy incorporates a range of behavioral techniques within it’s approach (cf. Beck et al., 1979), the current chapter will emphasize specific strategies used to change cognition (i.e., cognitive restructuring).

Cognitive restructuring is a therapeutic strategy that focuses directly on the disturbed information processing maintained by depressed patients. Cognitive theories of depression hypothesize that particular negative ways of thinking increase individuals' likelihood of developing and maintaining depression when they experience stressful life events. According to these theories, individuals who possess specific maladaptive cognitive patterns are vulnerable to depression because they tend to engage in negative information processing about themselves and their experiences. Given their clinical relevance, two widely accepted cognitive theories will be presented.

Cognitive Triad. Beck (1979, 1983) hypothesized that depression-prone individuals possess negative self-schemata (beliefs), which he labeled the "cognitive triad." Specifically, depressed patients have a negative view of themselves (seeing themselves as worthless, inadequate, unlovable, deficient), their environment (seeing it as overwhelming, filled with obstacles and failure), and their future (seeing it as hopeless, no effort will change the course of their life). This negative way of thinking guides one’s perception, interpretation, and memory of personally-relevant experiences, thereby resulting in a negatively biased construal of one's personal world, and ultimately, to the development of depressive symptoms. For example, the depression-prone individual is more likely to notice and remember situations in which he has failed or does not live up to some personal standard and discount or ignore successful situations. As a result, he maintains his negative sense of self, leading to depression.

Hopelessness Theory. Abramson, Metalsky, & Alloy (1989) proposed a hopelessness theory of depression formulated from Seligman’s work on learned helplessness and attribution styles (1975, 1979). Abramson et al. (1989) hypothesize that when confronted with a negative event, people who exhibit a depressogenic inferential (thinking) style, defined as the tendency to attribute negative life events to stable (enduring) and global (widespread) causes, are vulnerable to developing depression because they will infer that: a) negative consequences will follow from the current negative event, and b) that the occurrence of a negative event in their lives means that they are fundamentally flawed or worthless. For example, consider a woman whose fiance breaks off their engagement. If she attributes the cause of her break-up to her personality flaws, a stable-global cause that will lead to many other bad outcomes for her, or if she infers that a consequence of the break-up is that she will never marry or have children, or if she infers that without a lover, she is worthless, she is likely to become hopeless and develop the symptoms of depression. Thus, according to hopelessness theory, a specific cognitive vulnerability operates to increase the risk for depression through its effects on processing or appraisals of personally-relevant life experiences.

The Process of Cognitive Restructuring

Collaborative Empiricism. Beck (1979) coined the term "collaborative empiricism" to characterize the nature of the therapist-patient relationship in cognitive therapy. The therapist is active and directive, and facilitates a rational approach to thinking with regard to the patient's current life circumstances, using the principles of logic and the scientific method. All of the patient's thoughts and assumptions are treated as hypotheses which can be tested to verify their accuracy.

In order to foster this spirit of collaborative empiricism, cognitive behavioral therapists typically begin treatment by educating patients about their disorder. Helping patients understand the cognitive behavioral model of depression is particularly important in strengthening the treatment rationale and subsequent patient compliance. Whenever a new technique is introduced, the therapist always begins by presenting the rationale. Educating patients will build the therapeutic alliance. When appropriate, relevant research findings are also reviewed (e.g., benefits of CBT versus medication, typical rate of response to treatment). Information covered in session can be bolstered by recommending self-help books and websites (e.g., www.cognitivetherapy.com) aimed at educating patients about CBT and depression. The popular self-help book Feeling Good (Burns, 1980) is of enormous value when applying CBT for depression. It is also valuable for patients to audiotape sessions and review them once or more between sessions.

One of the most important functions of the therapist is to provide structure. First, the therapist and patient together discuss the goals of therapy. The goals should be specific and concrete (i.e., decrease depressive symptoms to a score below ten as measured by the Beck Depression Inventory (Beck et al., 1988) rather than "feel better." The therapist continually evaluates with the patient how well therapy is helping the patient progress towards these goals (e.g., is the BDI score decreasing), and modifies treatment strategies and goals when appropriate.

With regard to overall therapeutic style, the CBT therapist aims to communicate a sense of faith in his or her ability to help the patient. Identifying and responding to negative thoughts about therapy will be important in providing a credible treatment plan for the patient. In fact, the patient should be encouraged to express negative feelings about the therapy so that such factors, which may interfere with treatment, can be dealt with up front. Patient's are unlikely to participate in activities that they believe will not ultimately pay off, so the more realistic optimism that the therapist can provide, the better. An optimistic, persistent therapeutic style is especially useful in treating patients with depression.

Homework. In order to maximize rapid treatment response, the therapist places great emphasis upon the use of "homework" outside of therapy sessions. By the end of each session the therapist and patient should agree upon at least one assignment the patient can do to either test beliefs and/or build skills. It is the therapist's responsibility to facilitate compliance with homework assignments. Although it is not always the case, patients are often noncompliant with homework because their therapist does not pay enough attention to it, and as a result, the patient concludes it is not important. Providing a rationale that motivates patients, developing homework assignments in collaboration with the patient, following up with assignments, praising patients when they complete assignments, brainstorming solutions when problems occur rather than labeling the patient as "resistant," and pointing out the positive consequences of carrying out homework assignments are ways the therapist can facilitate compliance.

Guided Discovery. Guided discovery refers to the technique developed by Beck (1979) to help patients identify cognitive errors. Essentially, guided discovery is based upon the Socratic method. Rather than simply pointing out the errors in the patient's thinking (and perhaps seeming to lecture or reprimand patients), the therapist asks specific questions designed to direct (guide) patients to find the errors themselves (discovery). The rationale for this approach is that patients will derive more benefit from discovering their own cognitive distortions than simply being told what their distortions are. In addition, and perhaps most important, by asking directed questions, the therapist models a process that patients can internalize and learn to use on their own. Merely pointing out patients' distortions is not likely to generalize to other situations and will facilitate dependence upon the therapist, whereas teaching patients the process of evaluating their thoughts will provide them with a tool they can use in many other situations.

The use of guided discovery highlights a core principle of CBT: As much as possible, the therapist encourages patients to utilize strategies learned within the context of therapy to generate their own solutions to problems. All strategies employed during therapy sessions are ultimately intended to be used by the patient in his or her "real world." As is evident from this model, patient noncompliance is a major stumbling block to effective treatment, and thus, must be dealt with from the outset. While space does not permit a discussion of dealing with noncompliance, the interested reader should see J. Beck (1995) or Klosko & Sanderson (1999) for a thorough discussion.

Broad Steps of Cognitive Restructuring

As outlined above, in cognitive therapy, the focus is on understanding how patients interpret events in their lives. The therapy is based upon the premise that if distorted thoughts and images can be changed, then the accompanying negative emotional states and behaviors will change as well.In this cognitive mediation model of emotion, for the most part, affect and behavior are seen as "slaves" to cognition. If cognition is distorted, then maladaptive affect and dysfunctional behavior will follow. The affect is always in sync with the appraisal. If the patient feels angry, then cognitions associated with threat should be identified, if a patient feels anxious, then cognitions associated with danger should be identified, and so on.

In order to facilitate cognitive change, patients are encouraged to consider their thoughts and beliefs as hypotheses, to pay attention to all available information, and to revise hypotheses (thoughts) according to incoming information. The following three steps are used to accomplish this goal (see J. Beck (1995) for a detailed explanation of cognitive restructuring).

Step 1. Before patients can change the way they think, they must first recognize what they are thinking. The actual thoughts or images on a person's mind are labeled "automatic thoughts". Thus, the first essential step of cognitive restructuring is to teach people to begin monitoring their automatic thoughts. The best way to begin is to have patients use a change in emotion (e.g., an increase in depression) as a cue to initiate self-monitoring of what was going through his or her mind. According to cognitive theory, the onset or intensification of emotion is an indication that an automatic thought has occurred. In addition to labeling their thoughts, patients will also label the intensity of their negative affect, and note the situation in which the thought occurred. Patients should be encouraged to write whatever went through their mind. Frequently, patients will begin writing their thoughts, but then decide to not put them down because they "do not really believe them." Moreover, sometimes the thought is so extreme that when the patient writes it down in black and white he/she realizes it's inaccuracy and feels embarrassed to acknowledge it. However, even if thoughts are subsequently discounted, the fact that they occurred, even momentarily, suggests that they most likely had an effect on the person's emotional state at the time). Therefore, the therapist must strongly encourage patients to write down any thoughts that passed through their minds and leave the editing for later.

Step 2. Once a patient has elucidated his thoughts and examined how they influence his affect and behavior, the next step is for the therapist and patient to subject each thought to logical analysis by: 1) Examining the evidence for the patient’s thoughts, 2) Determining if any cognitive distortions are present, 3) Attempting to generate alternative hypotheses.

Examining the Evidence. Examining evidence lies at the heart of a rational approach to life. When the patient's thoughts can be framed in an empirical question, then the patient and therapist can test the accuracy of these thoughts by examining the evidence. Facilitating this approach will allow patient’s to break free of the habitual acceptance of implicit negative thoughts generated by depressogenic schema noted above. At this point, cognitions are treated as hypotheses and subjected to logical analysis. Crucial areas to investigate with patients are framed in the following questions:

1. Is there any evidence that supports the thought?
2. Is there any evidence that goes against the thought?
3. If the negative thought is accurate, what can be done to best cope with it?

For example, consider a patient whose depression was exacerbated by the prospect of failing an upcoming exam. The patient’s primary negative thought is, "I’m a loser. I’ll never pass the test." Because of this expectation, the patient may not study very hard, and as a result may fail, creating a self-fulfilling prophecy. However, once the negative thought was identified, the therapist and patient can work together to determine if in fact previous evidence supports the prediction that the patient will fail (e.g., were there previous tests that he failed? passed? If so, what did he do that lead to a positive versus negative outcome?). Once this information is examined in detail, the patient can replace the negative thought with a more accurate thought, such as "I have passed tests in the past, and although this test is difficult, if I put my effort to it, I can pass this one as well."

An important point to be made here is that the therapist is not trying to get the patient to think in a falsely positive manner (ie, "positive thinking). Rather, the goal is to get the patient to see the situation as accurately as possible. If there are problems (e.g., continually failing a test), the therapist can move into a problem-solving mode and help the patient correct behaviors that might be causing the problems (e.g., work on study behavior). This will be much more useful and valid than positive thinking in itself.

Identifying Cognitive Distortions. Beck's (1979) initial work on depression elucidated common errors of logic, or cognitive distortions (cf. Burns, 1985), that existed in the information processing styles of depressed patients. The use of cognitive distortions resulted in the patient's viewing life events in a way that tended to maintain depression. For example, Beck observed that depressed patients frequently thought in extremes (black and white thinking), and as a result, appraised anything less than perfect performance as a failure. Identifying cognitive distortions will facilitate cognitive restructuring, as it will enable patients to clearly see the errors of logic in their thinking, and facilitate the development of more accurate statements (rational responses). Examples of questions that should be applied to the patients' thoughts to uncover cognitive distortions should include:

1. Am I looking at things in extremes?

2. Am I taking one instance and seeing it as a pattern?

3. Am I picking out the negative details of a situation, ignoring the rest of the picture?

4. Am I rejecting positive experiences?

5. Am I jumping to a negative conclusion even though I have no facts to support it?

6. Am I magnifying the importance of things or catastrophizing about the consequences of a situation?

7. Am I assuming that something is going to happen because I feel that way?

8. Am I using emotionally charged words that stimulate my negative feelings?

9. Am I seeing myself responsible for something that I was not in fact responsible for?

For example, a graduate student who had one patient drop out of treatment stated "I’m a lousy therapist and will have difficulty maintaining patients." Upon closer examination, she had a total of eight patients, and one dropped out. This is an example of distortion number 3 (above), where she picked out one negative detail of the situation (ie, the patient dropping out) and ignored the rest of the picture (seven patients currently in treatment). It is also an example of distortion 6, where she is magnifying the significance of one event as indicating her total competence ("I’m a lousy therapist"). Distortions 2 and 8 fit as well, where she is taking one event and predicting that it will become a pattern ("I’ll have difficulty maintaining patients") and since the patient is using an emotionally charged word ("lousy therapist") rather than something more accurate (e.g., a "student therapist" is likely to be inexperienced which may account for the event). As you will notice, other distortions apply, and the goal is to continue identifying them, until all that apply are exhausted.

Generating alternatives. Patients' automatic thoughts represent one interpretation of events. As discussed above, depressed patients interpretation of events is largely influenced by distorted information processing. Thus, before assuming any one interpretation is correct, patients are asked to consider all the possibilities by generating alternative hypotheses to their automatic thoughts. This strategy is intended to move patients away from the exclusive use of negatively biased information processing. Using the example of the student therapist above, one possibility is that her patient dropped out of treatment because the therapist was "lousy." However, there are other alternative hypotheses as well, each of which are equally possible until further information is collected to rule them (or the original hypothesis) out. For example, perhaps the patient dropped out because he felt better, or because the time of the session was inconvenient, or because he decided to pursue an alternative treatment (medication), or he was financially unable to continue, or he had an unexpected major life-event occur that required his full attention, or perhaps the patient dropped out because he did not click with the therapist and that in general that happens, etc.

The goal is to generate as many plausible alternative explanations as possible. Depressive thinking is rigid in its' negativity. When patients step back and generate alternative interpretations of an upsetting event, this process counters and loosens their rigidity. In many instances, none of the alternative hypotheses can be proven, yet each is as plausible as the one that the patient has generated and accepted as true. Thus, increasing patients' awareness of other possibilities gives them a sense of the full picture, demonstrates the frequent subjectivity involved in interpreting events, and highlights their repeated focus on the negative aspects of the situation.

Step 3. Once evidence has been reviewed, distortions examined, and alternatives explored, the next step is to generate a "rational response" (i.e., a more accurate statement about the situation or event). Work done during Step 2 will facilitate their generating a rational (realistic) response. Thus, the goal would be to have the individual modify the original thought ("I’m a lousy therapist and will have difficulty maintaining patients") to something such as "I had one patient drop out and seven patients remain in treatment. While I should examine what went wrong in that one instance, I have to keep in mind that overall, I’m maintaining the majority of my patients. There are many explanations as to why a patient would drop out of treatment that are as plausible as my negative explanation. Losing one patient does not necessarily predict a problem with keeping patients in general. Also, losing patients to some degree is expected since I am not very experienced. That does not make me a lousy therapist, it’s just a function of my current situation that will change over time as I gain more experience and training. "

Hypothesis testing. In many situations, subjectivity does not have to remain. In such instances, hypothesis testing can be conducted. Hypothesis testing involves setting up an "experiment" to test an interpretation or anticipation of an event to provide more definitive information. For example, in the above example, the student therapist could poll other student therapists and learn about drop out rates in their caseloads. Sometimes this figure can come from the director of the clinic or a supervisor (e.g., on average 15% of patients dropout). This would allow the student therapist to determine the accuracy of her thought (if she is having twice as many drop outs as others, perhaps there is an issue, and she should move to problem solving mode; if she is having the same or less, it is just an example of her negative thought pattern). Rather than telling patients how to test the hypothesis, the therapist should ask a series of questions to help the patient uncover the answer herself (guided discovery). Hypothesis testing teaches the patient a process for testing ones thinking pattern, rather than relying on it as fact or as purely subjective.

Problem Solving. More often than not, when depressed patients subject their automatic thoughts to logical principles and empirical testing, they find out their hypotheses are either false or greatly exaggerated. However, sometimes their hypotheses are correct, and the patient is identifying a real problem that requires a solution. Under such circumstances, the therapist should help the patient see their task as problem solving. It is important to note that the pessimistic thinking style that accompanies depression interferes with patients' ability to problem-solve -- depressed patients tend to view situations as overwhelming and hopeless. By elaborating the problem solving process, the therapist provides patients with a strategy to offset this pessimistic thinking style. The following steps are crucial in this process: 1. Brainstorming solutions. This involves generating as many solutions as possible, without stopping to evaluate them. Encourage the patient to be creative and thorough. 2. Looking at the pros and cons of the solutions. Have the patient list the advantages and disadvantages of each proposed solution. 3. Choose the best solution and carry it out. Have the patient consider the importance of the various pros and cons and based upon that analysis, choose the solution that seems best, and take concrete steps to carry it out.

Continuing with the graduate student example from above, if her dropout rate was considerably greater than others in that setting (e.g., her dropout rate was 35% compared to an average of 15%) and there were no alternative explanations (e.g., she tended to take on difficult cases that had a higher likelihood of dropout, such as adolescents), then the next step would be to move into a problem solving mode rather than focus on herself as an eternally lousy therapist and give up. The patient and therapist would brainstorm solutions (e.g., get additional supervision, video tape sessions to get a more precise picture of what is going on, ask for feedback from patients who dropped out, etc.), review the pros and cons of each, then choose the best solution(s) and carry it out in order to resolve the problem (e.g., tape sessions so the supervisor can get a better sense as to what may be causing the problem and ultimately work to correct it).

Behavioral Techniques.

Peter Lewinsohn (1975) advanced the social learning theory of depression which posits that depression is a result of changes in reinforcement from environmental interactions. Routes to depression include a loss of positive outcomes (e.g., being complimented by one's employer for a job well-done) and an increase in negative outcomes (e.g., being criticized by one's spouse). In depression, a vicious cycle ensues, wherein increased feelings of depression lead to decreased activity, resulting in even less reinforcement, which in turn causes increased depression and so on (Lewinsohn, Munoz, Youngren, & Zeiss, 1986). This behavioral model posits multiple pathways to depression (Lewinsohn et al., 1986). First, interactions that have been a source of positive outcomes for one in the past may be no longer available (e.g., death of a close friend or family member preclude positive interactions from that source, an inability to engage in a previously satisfying activity such as stopping work at retirement, or not being able to involve oneself in an enjoyable fitness activity because of an injury). Second, one may lack the skill(s) to achieve positive outcomes in interactions (e.g., a lack of assertiveness results in continuous negative events at work where the person is continually taken advantage of, a lack of social skills leads one to frequently be alone, one sets such high standards for performance at work that it is never met leading to continuous dissatisfaction).

Although it is beyond the scope of the present chapter to present in detail the various behavioral strategies employed to overcome depression (cf. Lewinsohn et al., 1986, Klosko & Sanderson, 1999), a brief overview will be covered.

Scheduling Pleasant Activities. Whether it is a cause or consequence of depression, patients invariably have withdrawn from pleasurable activities. Thus, systematically scheduling and participating in pleasurable events is one way to circumvent the depressive spiral described above. First, the therapist assists the patient in identifying activities that they had found pleasant in the past (e.g., going to lunch with a good friend, bicycle riding in the park). A structured assessment of activities is more likely to get useful information (e.g., administer the Pleasant Events Schedule, Lewinsohn et al., 1986) than merely asking the patient to recall what they found enjoyable in the past, as recollection is likely to be clouded by depression. As noted by Lewinsohn et al. (1986) it is particularly important to include activities from the following three categories which have particularly strong connections to mood: positive social interactions (e.g., spending time with a good friend), activities that make one feel useful (e.g., caring for one's child, doing a job well), and activities that are intrinsically pleasant (e.g., a meal at one's favorite restaurant, listening to music). Once the events are identified, the therapist and patient work out a schedule to reintroduce the activities in one's life by elaborating specific goals (e.g., goal for the week is to call your good friend and have dinner at one of your favorite restaurants). Because the patient's depressed mood will interfere with their ability to anticipate and evaluate positive events, it will be important to focus the patient ahead of time on what was positive about the situation in the past in order to sensitize their information processing. In addition, it is essential to have patients monitor their mood before, during, and after the event to demonstrate the positive effect the event has on one's mood, even if it is temporary.

Improving interpersonal skills. Day to day life involves continuous social interaction. Depending upon the nature of the interaction, it is normal for both positive and negative moods to stem from these interactions. Social learning theory posits that depressed patients may have interpersonal skills deficits that interfere with their ability to gain positive social reinforcement. Two specific areas that have received considerable attention in reducing depression are Assertiveness Training and Social Skills Training (Lewinsohn et al., 1986). Training in these skills will allow patients to be more socially effective, ultimately generating feelings of mastery and positive affect, thereby reducing depression. The techniques are quite elaborate and will not be described here. The interested reader should consult Klosko & Sanderson (1999) or Lewinsohn et al. (1986) for a detailed overview.

Management of Anxiety and Tension. Patients with depression frequently suffer from general tension and anxious overarousal. Symptoms such as anxiety, insomnia, and agitation are characteristic of depression and frequently contribute to it. For example, agitation may result in an individual's inability to enjoy a pleasurable event, thus decreasing reinforcement. Insomnia may lead to daytime fatigue and ineffective performance at work. As a result, it may be essential to provide physical control strategies to decrease unpleasant over arousal associated with anxiety. Progressive muscle relaxation training is an effective strategy to lower overall arousal levels (cf. Bernstein & Borkovec, 1973). Thus, in patients where anxiety and tension exist, utilizing progressive muscle relaxation several times per day to reduce over arousal may be quite effective. However, since as many as fifty percent of patients with depression suffer from a comorbid anxiety disorder (e.g., panic disorder, obsessive compulsive disorder; Sanderson, Beck, & Beck, 1990), in such cases, a specific anxiety disorder focused intervention may be required to directly address the comorbid syndrome (cf. Wetzler & Sanderson, 1997).

Efficacy of Cognitive Behavior Therapy for Depression

Since cognitive therapy was first formulated by Beck (1963), numerous studies have demonstrated the efficacy of cognitive therapy for depression. The first landmark study conducted by Rush and colleagues in the late seventies (Rush, Beck, Kovacs, & Hollon, 1977) demonstrated that cognitive therapy was more effective than tricyclic antidepressant therapy in patients suffering from clinical depression. In contrast with previous outcome research which demonstrated that psychotherapies were no more effective than pill- placebos and less effective than antidepressant medications, the Rush et al. study was the first to show that a psychosocial treatment was superior to pharmacotherapy in the treatment of depression (Hollon, Shelton, & Loosen, 1991). Further, a follow-up study conducted twelve months post-treatment showed that relapse rates were lower among patients who received CT (39%) versus those who received antidepressant medication (65%), although this difference did not reach statistical significance (Kovacs, Rush, Beck, & Hollon, 1981).

In the two decades since the initial trial, many controlled trials have been undertaken to replicate these findings. Although many experts now believe that the Rush study was sufficiently flawed to negate study findings (Hollon et al., 1991), many qualitative and quantitative reviews now conclude that cognitive therapy: 1) effectively treats depression, 2) is at least comparable, if not, superior to medication treatment, and 3) may have lower rates of relapse in comparison to medication treatments (Dobson, 1989; Hollon, 1981; Hollon & Beck, 1986; Hollon & Najavits,1988; Hollon et al., 1991; Miller & Berman, 1983). As a result, cognitive therapy has gained widespread acceptance as a first-line treatment for depression, and cognitive behavioral therapy is one of only two psychotherapies included in the guidelines for the treatment of depression published by the Agency for Health Care and Policy Research (AHCPR).

However, in the midst of what many have termed a golden age of cognitive therapy, a debate has recently arisen about the efficacy of cognitive therapy as a treatment for depression.These discussions stem, in part, from the concerns of some regarding the methodological sufficiency of the studies showing an advantage of cognitive therapy and in large part, from the highly visible results of the National Institute of Mental Health Treatment of Depression Collaborative Research Program study (NIMH TDCRP) which concluded that cognitive behavior therapy was not effective in the treatment of severe depression (Elkin, Shea, Watkins, et al., 1989). Despite the fact that the study is flawed in many respects, this single study is threatening to stem the tide in favor of cognitive therapy as a treatment for depression. We will examine the current state of research on cognitive therapy for depression below, present issues of methodological concern on both sides of the debate, and explore areas of future study. To avoid redundancy with prior reviews conducted on individual studies, the present review is primarily focused on meta-analytic studies.

Is Short-Term Cognitive Therapy Effective? A review of meta-analyses

Several meta-analyses have been conducted to determine the efficacy of cognitive behavior therapies (CBT) in relation to no-treatment, pharmacotherapy, and other psychotherapies, as well as the relative efficacy of pure cognitive (CT) and behavioral components (BT). As will be evident, the meta-analyses themselves are not devoid of flaws and vary widely with regard to the number and methodological stringency of studies reviewed, the size and diagnostic homogeneity of study samples, as well as the outcome measures and methodology used to conduct the meta-analysis.

In an early meta-analysis using data from the Smith, Glass and Miller study (1980), Andrews and Harvey (1981) demonstrated that cognitive and behavioral therapies (CBT) had comparably larger mean effect sizes (ES=0.97) than placebo controls (ES=0.55), other psychotherapies (ES=0.74), and counseling (ES=0.35). However, the efficacy of CBT is difficult to evaluate since many available studies on cognitive behavior therapy were not included in the original Smith et al. (1980) meta-analysis. Further, the specific efficacy of CBT for depression is difficult to assess as the studies used in the meta-analysis were comprised of patients suffering from a gamut of neurotic disorders (although a majority were depressed).

In a meta-analysis of 56 outcome studies, Steinbrueck, Maxwell & Howard (1983) found evidence for a superiority of psychotherapy over no-treatment and pharmacological therapies in the treatment of depression but found no differences in efficacy between the different forms of psychotherapy (cognitive, behavioral, marital, and other). Similarly, Nietzel and colleagues (Nietzel, Russell, Hemmings, & Gretter, 1987) reported only a nonsignificant trend for the superiority of cognitive treatments to other psychotherapies.

By contrast, a meta-analysis combining 48 studies concluded that CBT was significantly more effective than no-treatment, that CBT had somewhat larger treatment effects as compared to "other" psychotherapies, and that CT and BT did not differ with regard to their relative efficacy in treating depression (Miller & Berman, 1983). However, results from this meta-analysis must be interpreted with caution since only 10 out of 48 studies involved patients with "clinical" depression.

A similar flaw is evident in the meta-analysis conducted by Robinson and colleagues (1990) in which only 35% of the 58 studies involved patients who met formal criteria for a depressive disorder (Robinson, Berman, & Neimeyer, 1990). Results of their analysis found that cognitive behavioral therapies (CBT, CT or BT) were more effective than other psychotherapies. Effect sizes for cognitive and behavioral therapies ranged from 0.85-1.02 while those for other psychotherapies was approximately 0.49.

Robinson's meta-analysis also found that CBT was more effective than pharmacotherapy, and that CBT plus medication did no better than medication or CBT alone. Finally, CBT was found to be slightly more effective than either CT or BT alone, and CT alone was slightly more effective than BT alone. However, these findings are tempered by the fact that researcher allegiance was significantly correlated with outcome in the studies used in the meta-analysis. In fact, once researcher allegiance was taken into account, differences in efficacy among the different forms of therapy for depression disappeared.

In one of the better known meta-analyses, Dobson (1989) reviewed twenty-eight studies that included patients with clinical depression only and found that CT was clearly more effective than no-treatment or wait list (ES=-2.15), other psychotherapies (ES=-0.54), medication alone (ES= -0.53), and BT alone (ES=-0.46). Using the Beck Depression Inventory (BDI) as a common measure across studies, his meta-analysis found that the average cognitive therapy (CT) patient did better than 98% of patients in the no-treatment condition, 70% of patients treated with other psychotherapies or medication alone, and 67% of patients treated with BT alone.

The results of this meta-analysis have been criticized on several accounts. Critics have commented on the fact that Dobson's meta-analyses were drawn from a relatively small number of studies with small sample sizes, that some of the studies included were not randomized, while others, such as the Rush et al. (1977) study, utilized less than optimal drug treatments (Glaoguen, Cottraux, Cucherat, & Blackburn, 1998; Hollon et al., 1991; Scott, 1996). Critics also argued that Dobson's findings of such a clear superiority of CT over other treatments may have been influenced by researcher allegiance (which Dobson did not examine) and may have resulted from the fact that he used the BDI as the only outcome measure (Robinson et al, 1990). Robinson and colleagues argue that since the BDI was not as widely used as it is now, many studies were excluded, and hence Dobson's study may have been unrepresentative.

In a re-analysis of the Dobson et al. data, Gaffan and colleagues (Gaffan, Tsaousis, & Kemp-Wheeler, 1995) address some of the criticisms leveled against the Dobson study, particularly the issue of researcher allegiance. Using a larger sample (65 studies including Dobson's original 28), study investigators confirmed the findings of the superiority of CT to other treatments but obtained smaller effect sizes than those found by Dobson. Gaffan and colleagues found that approximately half of the difference observed between CT and other treatments in Dobson's study was attributable to researcher allegiance. However, when they separately examined the additional 37 studies published since the Dobson meta-analysis, Gaffan and colleagues found that despite the fact that outcome in these studies was not correlated with researcher allegiance, the superiority of CT to waiting list controls, other psychotherapy and pharmacotherapy was still present, albeit with smaller effect sizes. Their results also revealed that CT plus medication performed no better than CT alone, although it is important to note that allegiance ratings were much higher for CT alone as compared to CT plus medication.

Two recent meta-analyses conducted are of particular note as both include the multisite NIMH TDCRP study (Elkin et al., 1989) in their analyses and use more stringent selection criteria as compared to prior studies. A meta-analysis conducted by the Agency of Health Care and Policy Research (Depression Guideline Panel, 1993) combined data from twenty-two randomized controlled trials conducted on the treatment of depression. Certain pivotal studies were not included in this meta-analysis because they did not meet the conservative inclusion criteria set by the panel. For example, only studies where treatment outcome was reported categorically and intent to treat samples could be identified were used. Using the confidence profile method, the overall efficacy of CT alone was found to be 46.6. percent, for adult outpatients the efficacy was 46.9 percent, and for geriatric outpatients it was 51.3 percent. CT was 30 % more effective than wait list controls, 15% more effective than pharmacotherapy and 9% more than pill placebo with clinical management. CT alone did not differ in efficacy from other empirically supported psychotherapies as a whole. Individual comparison revealed that CT alone (46.6%) did not differ from BT alone (55.3%) or interpersonal therapy (IPT; 52.3%) but had higher overall efficacy rates than brief dynamic therapy (34.8%).

The latest meta-analysis conducted by Glaoguen and colleagues combined 48 controlled trials conducted on depressed outpatients from 1977 to 1996, excluding thirty trials for methodological reasons (Glaoguen et al., 1998). Confirming Dobson's results, their meta-analysis found that cognitive therapy (CT) was significantly better than waiting list, antidepressant medication, and a group of miscellaneous therapies. However, in contrast to Dobson's finding of a clear superiority of CT over BT, Gloaguen and colleagues found that CT was comparable to behavior therapy alone (BT).

Summary
In sum, all meta-analyses find evidence of a clear superiority of CT over no treatment or wait list controls. Six out of nine studies find significantly greater treatment effects for CT as compared to other psychotherapies. Of the six studies directly comparing cognitive and behavioral therapies, CT outperformed BT in three studies and was comparable to BT in the other three studies. Finally, CT outperformed medication in all five meta-analyses comparing the two treatment modalities.

Taken as a whole, these meta-analyses provide substantial evidence that CBT is an effective treatment for depression, even though some findings are tempered when factors such as researcher allegiance are taken into account. As study designs and methodologies become more sophisticated, it is clear that further meta-analyses will be necessary to confirm these findings.

Is Cognitive Therapy Effective in Preventing Relapse or Recurrence?

The high degree of relapse (i.e., a continuation of the index episode before recovery) or recurrence (i.e., a new episode after recovery) in depression has made the issue of symptom maintenance a critical one for both psychotherapy and pharmacotherapy researchers. Without additional treatment, the range of relapse appears to vary between 50-80% within the first year following recovery from depression (Keller, Shapiro, Lavori, & Wolfe, 1982; Lobel & Hirschfeld, 1985). Medication studies estimate that symptom relapse or recurrence tends to occur within 6-24 months after treatment is discontinued, findings that have led to a recent development of maintenance treatment strategies for depressed patients with recurrent episodes (cf. Jarrett et al., 1998). Preliminary results show that maintenance phase treatment with medications such as fluoxetine and imipramine appear to significantly lower the rate of symptom relapse or recurrence in depression (Fava & Kaji, 1994; Frank et al., 1990; Kupfer, 1992; Montgomery, Dufor, Brion et al., 1988).

CBT has generally been associated with a lower degree of relapse than those for medications (see Hollon et al., 1991 for a review). Although methodologies vary significantly between studies, figures suggest that 0-50% of patients exhibit symptom relapse or recurrence within 1-2 years of discontinuing cognitive behavioral treatment (Jarrett et al., 1998). Six of the nine meta-analyses reported above incorporated follow-up data in their meta-analyses and generally demonstrated that the effects of CBT appear to be maintained over time. Andrews and Harvey (1981) found that the improvement was stable over time for all psychotherapy groups including CBT, with a gradual decline occurring after the first year at an estimated rate of 0.2 ES units per annum. Similarly, Nietzel and colleagues (1987) reported that treatment effects were maintained at follow-up (only 70% of studies performed follow-up evaluations). Results from the meta-analysis conducted by Miller and Berman (1983) also demonstrated that effect sizes were stable at follow-up (mean follow-up = 9.5 weeks), leading to the conclusion that the relative efficacy of CBT at posttreatment appeared to predict relative efficacy at follow-up. For the nine studies that conducted follow-up evaluations in the Robinson meta-analyses, posttreatment and follow-up effect sizes were virtually identical, suggesting that treatment effects were durable over time (Robinson et al., 1990).

A similar trend was observed in the most recent meta-analyses. In the US DHS study (Depression Guideline Panel, 1993), seven of the twenty-two studies incorporated follow-up evaluations and revealed that CT was associated with fewer depressive symptoms and a lower rate of relapse as compared to pharmacotherapy. Three of the seven studies also revealed that CT had fewer depressive symptoms than wait list controls at follow-up. And finally, in the meta-analysis conducted by Gloaguen and colleagues, CT was associated with a lower relapse rate as compared antidepressant treatments (Gloaguen et al., 1998).

Despite these findings, the prophylactic effects of CBT have yet to be fully established, due in large part to the difficulty in interpreting results from follow-up studies. Methodological problems generally observed in such studies include: brief follow-up periods; the inclusion of acute nonresponders in follow-up; naturalistic versus controlled follow-up; lack of controls; and variabilities in the composition of samples followed, timing and nature of assessment, and in defining relapse or recurrence (Depression Guideline Panel, 1993; Hollon et al., 1991). The prophylactic value of CBT (and other treatments for depression) has also come into further question following the results of the NIMH collaborative study. In the 18-month naturalistic follow-up of patients treated in the NIMH TDCRP study, all four treatments (CBT, interpersonal therapy, imipramine and placebo plus clinical management) were associated with similarly high rates of relapse (Shea, Elkin, Imber et al., 1992). Analyses of the intent-to-treat sample revealed that only 15-28% of patients did not suffer from major depression or require further treatment at the 18 month follow-up.

Methodological problems and the results of the NIMH study notwithstanding, the majority of studies have typically found that CBT provides protection against symptom relapse or recurrence, and that it may have a distinct advantage over pharmacotherapy in this respect (Hollon et al., 1991). Nevertheless, relapse or reoccurrence is still a significant issue, and one that has led to a recent focus on developing maintenance strategies for depressed patients with recurrent episodes (Jarrett et al., 1998). Two early trials by Blackburn and colleagues examined the preventive effect of treatment responders who received a continuation or maintenance phase with either CT, pharmacotherapy, or a combined treatment. In the first study, Blackburn and colleagues demonstrated that acute treatment responders who were provided with maintenance treatment every six weeks for an additional six months with CT or CT plus medications had significantly lower rates of relapse (6% and 0%) as compared to patients who were given maintenance treatment with pharmacotherapy alone (30%) (Blackburn, Eunson, Bishop, 1986). At the 24 month follow-up, patients treated with maintenance CT and CT plus medications were significantly less likely (23% and 21%) to have a recurrence as compared to patients treated with medication alone (78%). In the second study, Blackburn and Moore (1997) examined patients with recurrent depression who either received CBT in the short and long-term phase of treatment, medication in the short and long-term phases, or medication in the short-term and CBT in the long-term phase. Follow-up results conducted at 24 months showed comparable relapse and recurrence rates at for all three groups (24-36%).

Two recent pilot studies by Jarrett and colleagues (Jarrett, Basco, Risser, Ramanan, Marwill, Kraft, & Rush, 1998) also shed light on the prophylactic value of CT. These investigators evaluated the rates of relapse or recurrences in patients who received an acute trial of cognitive therapy. Treatment responders in the first study were merely given follow-up assessments on a monthly basis while responders in the second study were treated with CT for an additional 10 sessions, occurring over a period of eight months. Relapse rates in the first study were as follows: 40% at 6 months, 45% at 8 months,, 50% at 12%, 67% at 18 months, and 74% at 24 months. By contrast, symptom relapse occurred at a rate of 20% at the 6 and 8 month follow-up, 27% at 12 months, and 36% at 18 and 24 months. Based on these data, Jarrett and colleagues tentatively conclude that extending acute CT may be successful in achieving symptom maintenance but explore other potential factors such as patient gender, therapist competence, etc. that could have influenced their findings.

Fava and colleagues examined forty patients with recurrent depression who were successfully treated with medications and then randomly assigned to either CBT or clinical management over the next 20 weeks (Fava, Rafanelli, Grandi, Conti, & Belluardo, 1998). Antidepressant treatment was slowly tapered and discontinued over this period. Results demonstrated that after drug therapy was discontinued, patients who received CBT were less likely to have residual symptoms as compared to those receiving clinical management alone. CBT was also associated with significantly lower rates of relapse (25%) as compared to clinical management (80%) at the 2-year follow-up, leading the investigators authors to conclude that not only is CBT useful in preventing symptom relapse but that an alleviation of residual symptoms via CBT may reduce the risk of relapse in depressed patients.

Summary
In sum, a majority of studies have found that CBT provides protection against symptom relapse or recurrence, and may have lower rates of relapse as compared to pharmacotherapy. Six of the nine meta-analyses presented here incorporated follow-up data and demonstrated that the effects of CBT appear to be stable over time. Among the meta-analyses that compared rates of relapse rates between CBT and medication, CBT appears to have a distinct advantage. However, methodological problems inherent in follow-up studies make results difficult to interpret. And, despite the relatively high maintenance rates, relapse or reoccurrence is still a significant problem. Recent studies indicate that adding maintenance phase treatment with CBT may lead to lower rates of symptom relapse or recurrence.

So, Why is the jury still out?

Cognitive therapy is more extensively researched than any other psychotherapy for the treatment of unipolar depression and the breadth of support presented above appears truly impressive. So why is the jury still out (Jacobson & Hollon, 1996a)? Why is there a growing debate about the efficacy of CBT as a first line treatment for severe depression? These concerns arise in large part, from the highly visible results of the National Institute of Mental Health Treatment of Depressions Collaborative Research Program study (NIMH TDCRP) which found that cognitive behavior therapy was no more effective than a pill placebo and somewhat less effective than imipramine in the treatment of patients with severe depression (Elkin et al., 1989). As a result, the treatment guidelines prepared by the American Psychiatric Association as well as the AHCPR both assert that cognitive behavior therapy may be no more effective than a pill placebo in the treatment of severe depression. And despite the fact that CBT did not differ from other active treatments in treating the full sample of depressed patients in the TDCRP study, the less than impressive performance of CBT in the severe subsample, has also sparked questions about the role of CBT as an acute treatment for depression per se.

However, many investigators now suggest that the TDCRP study was flawed in several important respects. For example, Klein (1996) suggested that the study chose too stringent a level of significance which led them to understate the superiority of pharmacotherapy over psychotherapy in the severe subsample. Jacobson and Hollon (1996a and b) suggest that the finding within the severe subsample is suspect since it was not robust across the three different sites (CBT fared comparably to imipramine at one site while IPT fared comparably with imipramine at another site) and because they are not consistent with findings from previous studies. Further, Jacobson and Hollon (1996a and b) also point out that there were no differences in the active treatments at the 18 month follow-up, regardless of level of severity. Finally, questions have also been raised about allegiance effects and possible differences in the quality of CBT offered over the three different sites, which if confirmed, could account for the relatively poor showing of CBT in the study (Jacobson & Hollon, 1996a and b).

What is indeed surprising is that a single study, albeit an important one, is shaping current treatment guidelines when none of the multitudinal studies comparing drugs and CBT have suggested even a remote advantage of drugs over CBT (see Hollon et al., 1991 for a review). Advocates of CBT suggest that the TDCRP study has gained widespread acceptance because it"corresponds so closely to preconceived notions held by many in the psychiatric community" (Jacobson & Hollon, 1996a, pg. 74). On the other side, advocates of pharmacotherapy agree that the results of the TDCRP study need to be replicated but argue that its importance is underscored as it is one the few methodologically adequate studies comparing psychotherapy to drugs (Klein, 1996).

According to Klein, all previous studies comparing CBT and drugs are methodologically flawed and hence invalid since none included a pill placebo comparison (Klein, 1996). According to Klein, unless a pill placebo control is included in studies comparing drugs and psychotherapy, there would be no way of ensuring (a) that pharmacotherapy was adequately implemented and (b) that the study sample was drug responsive. Jacobson and Hollon (1996a and b) agree that a pill placebo control would strengthen the interpretability of drug-psychotherapy studies, both by increasing the likelihood that treatment differences would be discerned, and, in the event that active treatments did not differ, by determining if drugs and psychotherapy were equally effective or ineffective. However, they disagree that the absence of a pill placebo arm nullifies previous findings and assert that studies must independently assess the representativeness of the study sample and the quality of the pharmacotherapy provided. Further, they maintain that it is still important to know the relative efficacy of various treatments for samples regardless of whether or not they are drug responsive. Other advocates of CBT accuse Klein of circular reasoning by asserting among other things, that testing for drug responsiveness is a goal of the study (treatment outcome) and not a pre-study requirement (i.e., whether or not the sample is drug responsive) (McNally, 1993).

Conclusion Regarding Empirical Status of Cognitive Therapy.

Summary of Current Evidence

Based on results from the several meta-analyses reviewed here, we suggest that short-term CBT (and CT alone) is effective in the treatment of clinical depression, and, in the absence of studies replicating the TDCRP findings of it's lesser efficacy among severe depressives, we conclude that CBT is still a viable treatment for severe depression. Based on results from the various meta-analyses reviewed here, we also conclude that CBT is more effective than no-treatment and wait list controls, other psychotherapies, at least comparable to antidepressant medications, and has significant evidence attesting to its prophylactic value.

With regard to the question of whether CT alone is superior to BT or vice versa, findings are muddier. Some meta-analyses find evidence of a clear superiority of CT over BT while others show no difference between the two components. In a recent study by Jacobson and colleagues (Jacobson, Dobson, Truax, Addis, Koerner, Gollan, & Prince, 1996), the comprehensive CBT treatment combining cognitive and behavioral components did no better than the cognitive and behavioral components separately. Further, Jacobson et al. (1996) showed that the two components were just as effective as the comprehensive treatment in altering negative thinking and dysfunctional attributional styles, suggesting that both conditions may arrive at a similar result using different methods.

Future Directions

Although the breadth of support for CBT is impressive, future research is still necessary to address concerns raised above. Additional multicenter studies that compare medications and different forms of psychotherapy are still needed to confirm the efficacy of short-term CBT as a treatment for depression. However, for findings to be valid, it is essential that active treatments are administered by clinicians adequately trained in the various approaches and that issues of researcher allegiance and site differences are appropriately addressed. Further, whether or not both sides agree that the absence of a pill placebo control nullifies prior comparisons between drugs and psychotherapy, it appears that adding pill placebo controls to future drug-psychotherapy studies is imperative, not merely to enhance treatment interpretability but also to address criticisms from pharmacotherapy advocates who will most likely nullify studies which do not include such controls.

Another control that may enhance interpretation of treatment outcome data is that of the combination placebo with CBT control condition. Findings from the recent multisite study on panic disorder demonstrate that although combined treatments were more effective than CBT alone in certain key comparisons, that the combined imipramine plus CBT control was no better than the combined placebo plus CBT condition (Barlow, 1998). Without such a control, investigators would have erroneously concluded that CBT plus medication was more effective than CBT alone for the treatment of panic disorder. Based on these findings, we recommend that future outcome trials on depression include both a pill placebo alone and a pill placebo plus CBT control conditions in order to enhance interpretation of results.

In order to enhance treatment effects, studies must also examine the relative efficacy of CBT in depressive subtypes that may have characteristics that are associated with poorer outcome. For example, it has been demonstrated that patients with atypical depression (AD), a new subtype of the mood disorders in the DSM-IV are less responsive to tricyclic antidepressants (cf: Asnis, McGinn & Sanderson, 1995; McGinn, Asnis, & Rubinson, 1996). Hence, cognitive behavioral treatment for depressed patients with a typical depression may need to be modified to meet their unique symptom needs (McGinn, Ortiz, Sanderson, Kaplan, Asnis & Wetzler, 1998). Along the same lines, there is preliminary evidence that patients with personality disorders may be less responsive to short-term CBT, and that optimal treatment can only be accomplished for these patients if the treatment is modified to address the personality disorder as well (cf, McGinn, Young & Sanderson, 1995 for a review). Identifying specific populations who do not respond as well to short-term CBT will lead to the elucidation of factors that must be modified to provide more appropriate treatment.

Finally, future research studies need to evaluate the effectiveness of CBT for depression outside of clinical research centers. The demonstration of treatment efficacy in controlled research environments is only the first step in treatment research. Once a positive therapeutic effect has been conclusively demonstrated, generalizability becomes of paramount importance. With regard to CBT for depression, it seems fair to conclude from the data presented above that CBT is an effective treatment in clinical research settings. But data are not available on the efficacy of CBT for depression when delivered in non-research clinical settings to a diverse group of patients (This is not unique to CBT, and applied to other empirically supported treatments as well, e.g., pharmacological approaches). Without data, one must be cautious in generalizing the results from research settings to typical clinical settings because there are several factors which might reduce the efficacy of this treatment (cf: Wolfe, 1994) . For example, one area of particular concern is that clinicians in research settings are likely to possess greater expertise in the administration of a particular treatment developed in that setting compared to the average clinician. Thus, since clinician competence may be an important factor for success, one would expect a less favorable outcome in uncontrolled settings where the quality of treatment may not be as good. While caution may be warranted until data are generated specifically on CBT for depression, it is reassuring to note that data are beginning to appear that support the effectiveness of evidence based treatments outside of controlled research environments (e.g., Sanderson, Raue, & Wetzler, 1999), and a recent meta-analysis of psychotherapy studies found that the effect sizes of psychotherapy in "clinically representative settings" is slightly lower (approximately 10%) but comparable to that obtained in clinical research settings (Shadish et al., 1997).

References

Abramson LY, Metalsky GI, Alloy LB: Hopelessnes depression: A theory based subtype of depression. Psychol Rev 96:358-372,1989.

Andrews G, Harvey R: Does psychotherapy benefit neurotic patients? Arch Gen Psychiatry 38:1203-1208, 1981.

Asnis GM, McGinn LK, Sanderson WC: Atypical depression: Clinical aspects and noradrenergic function. Am J Psychiatry 152(1): 31-36, 1995.

Barlow DH (Chair): Results from the multi-center clinical trial on the treatment of panic disorder: Cognitive behavior treatment versus imipramine versus their combination. Symposium presented at the annual convention of the Association for the Advancement of Behavior Therapy (AABT), Washington, DC, November 1998.

Beck AT: Thinking and depression: 1. Idiosyncratic content and cognitive distortions. Arch Gen Psychiatry 9:324-333, 1963.

Beck AT:Cognitive therapy of depression: New perspectives. In Treatment of Depression: Old controversies and new approaches. Edited by PJ Clayton & JE Barrett (Eds.). New York: Raven Press,1983, pp 265-284.

Beck AT, Steer RA, & Garbin MG: Psychometric properties of the Beck Depression Inventory: Twenty-five years later. Clin Psychology Rev 8:77-100, 1988.

Beck AT, Rush A J, Shaw B, Emery G: Cognitive Therapy of Depression. New York: Guilford, 1979.

Beck JS: Cognitive Therapy: Basics and Beyond. New York: Guilford, 1995.

Bernstein D A, Borkovec TD: Progressive Relaxation Training. Champaign, IL: Research Press, 1973.

Blackburn IM, Eunson KM, Bishop S: A two-year naturalistic follow-up of depressed patients treated with cognitive therapy, pharmacotherapy, and a combination of both. J Affect Disord 10: 67-75, 1986.

Blackburn IM, Moore RG: Controlled acute and follow-up trial of cognitive therapy and pharmacotherapy in out-patients with recurrent depression. Br J Psychiatry 171:328-334, 1997.

Burns DD. Feeling Good. New York: New American Library, 1980.

Depression Guideline Panel Depressions in Primary care: Vol. 2. Treatment of Major Depression (AHCPR Publications No 93-0551). Rockville, MD: US Department of Health and Human Services, Agency for Health Care Policy and Research, 1993.

Dobson, K: A meta-analysis of the efficacy of cognitive therapy for depression. J Consult Clin Psychol 57: 414-419, 1989.

Elkin I, Shea T, Watkins J, et al: National Institute of Mental Health of Depression Collaborative Treatment Program. Arch Gen Psychiatry 46: 971-982, 1989.

Fava GA, Rafanelli C, Grandi S, et al: Prevention of recurrent depression with cognitive behavior therapy: Preliminary findings. Arch Gen Psychiatry 55:816-820, 1998.

Fava M, Kaji J: Continuation and maintenance treatments of major depressive disorder. Psychiatry Ann 24: 281-290, 1994.

Frank E, Kupfer DJ, Perel JM, et al:Three year outcomes for maintenance therapies in recurrent depression. Arch Gen Psychiatry 47: 1093-1099, 1990.

Gaffan, E. A., Tsaousis, I., & Kemp-Wheeler, S. M. (1995). Researcher allegiance and meta-analysis: The case of cognitive therapy for depression. J Consult Clin Psychol, 63(6), 966-980.

Glaoguen V, Cottraux J, Cucherat M. et al: A meta-analysis of the effects of cognitive therapy in depressed patients. J Affect Disord 49: 59-72, 1998.

Haaga DAF, Dyck M, Ernst D: Empirical status of cognitive theory of depression. Psychol Bull:110, 215-236, 1991.

Hollon S D: Comparisons and combinations with alternative approaches. Behavior therapy for depression. Edited by LP Rehm, New York, Academic Press, 1981, pp 33-71.

Hollon SD, Beck AT: Cognitive and cognitive-behavioral therapies. The handbook of psychotherapy and behavior change: An empirical analysis. Edited by SL Garlfield & AE Bergin, 2nd edition, New York, Wiley, 1986, pp 443-482.

Hollon SD, Najavits L: Review of empirical studies on cognitive therapy. American Psychiatric Press Review of Psychiatry, Vol 7. Edited by AJ Frances & RE Hales, Washington DC, American Psychiatric Press, 1988, pp 643-666.

Hollon SD, Shelton RC, Loosen PT: Cognitive therapy and pharmacotherapy for depression. J Consult Clin Psychol 59(1): 88-99, 1991.

Jacobson NS, Follette WC, Revenstorf D Psychotherapy outcome research: Methods for reporting variability and evaluating clinical significance. Behav Therapy: 15 336-352, 1984.

Jacobson NS & Hollon SD: Cognitive-behavior therapy versus pharmacotherapy: now that the jury's returned its verdict, it's time to present the rest of the evidence. J Consult Clin Psychol 64(1): 74-80, 1993a.

Jacobson NS, Hollon, SD: Prospects for future comparisons between drugs and psychotherapy: Lessons from the CBT-versus-pharmacotherapy exchange. J Consult Clin Psychol 64(1): 104-108, 1993b.

Jarrett RB, Basco MR, Risser R: Is there a role for continuation phase cognitive therapy for depressed outpatients. J Consult Clin Psychol 66(6): 1036-1040, 1998.

Jacobson, NS, Dobson KS, Truax PA: A component analysis of cognitive behavioral treatment for depression. J Consult Clin Psychol 64(2):295-304, 1996.

Keller MB, Shapiro RW, Lavori P, Relapse in major depressive disorder: Analysis with the life table. Arch Gen Psychiatry 39: 911-915, 1982.

Klein DF: Preventing hung juries about therapy studies. J Consult Clin Psychol 64(1): 81-87, 1996.

Klosko JS, Sanderson WC: Depression: Clinical Application of Empirically Supported Psychotherapy. NJ: Jason Aronson Inc., 1999.

Kovacs M, Rush AJ, Beck AT, et al: Depressed outpatients treated with cognitive therapy or pharmacotherapy. Arch Gen Psychiatry 38: 33-39, 1981.

Kupfer DJ: Maintenance treatment in recurrent depression. Br J Psychiatry 161: 309-316, 1992.

Lewinsohn PM: The behavioral study and treatment of depression. In Progress in Behavior Modification, Edited by M Hersen & RM Eisler,. NY: Academic Press, 1975,pp. 19-64.

Lewinsohn PM, Munoz R, Youngren M, & Zeiss A M: Control Your Depression.NY: Fireside, 1986.

Lobel B, Hirschfeld, RMA: Depression: What we know (ADM publications No.85-1318).Rockville, MD: US Department of Health and Human Services, 1985.

McGinn LK, Asnis GM, Rubinson E: Clinical and Biological Validation of Atypical Depression. Psychiatry Res 60:191-198, 1996.

McGinn LK, Ortiz C, Sanderson WC et al: Prevalence and Implications of Axis II Disorders in Patients With Atypical Depression. Paper presented at the annual convention of the Association for the Advancement of Behavior Therapy (AABT), Washington, DC, November, 1998.

McGinn LK, Young JE, Sanderson WC: When and how to do longer-term therapy.....without feeling guilty. Cognitive Behav Practice: 2(1):187-212, 1995.

McNally RJ: Methodological controversies in the treatment of panic disorder. J Consult ClinPsychol 64(1): 88-91, 1993.

Miller RC, Berman JS: The efficacy of cognitive behavior therapies: A quantitative review of research evidence. Psychol Bull 94(1): 39-53, 1983.

Montgomery SA, Dufor H, Brion S, et al: The prophylactic efficacy of fluoxetine in unipolar depression. Br J Psychiatry 153: 69-76, 1988.

Nietzel, MT, Russell RL, Hemmings KA et al: Clinical significance of psychotherapy for unipolar depression: A meta-analytic approach to social comparison. J Consult Clin Psychol 40: 288-294, 1987.

Robinson L, Berman J, Neimeyer R: Psychotherapy for the treatment of depression: A comprehensive review of controlled outcome research. Psychol Bull 108:30-49, 1990.

Rush A, Beck AT, Kovacs M et al: Comparative efficacy of cognitive therapy and imipramine in the treatment of depressed patients. Cognitive Therapy Res 1:17-37, 1979.

Scott J: Cognitive therapy of affective disorders: A review. J Affect Disord 37:1-11, 1996.

Sanderson WC, Beck AT, Beck J: Syndrome comorbidity in patients with major depression or dysthymia: Prevalence and temporal relationships. Am J Psychiatry 147:1025-1028, 1990.

Sanderson WC, Raue PJ, Wetzler S: The generalizability of cognitive behavior therapy for panic disorder. Journal of Cognitive Psychotherapy 12(4):323-331, 1998

Seligman, MEP: Helplessness: On depression, development, and death. San Franciscso: W.H. Freeman, 1975.

Seligman MEP, Abramson LY, Semmel A, von baeyer, C: Depressive attributional style. J Abnormal Psychology 88:242-247, 1979.

Shadish WR, Navarro, AM, Crits-Christoph P, Jorm AF (1997): Evidence that therapy works in clinically representative conditions. J Consult Clin Psychol 65:355-365.

Shea T, Elkin I, Imber SD et al: Course of depressive symptoms over follow-up findings from the NIMH treatment of depression collaborative research program. Arch Gen Psychiatry 49: 782-787, 1990.

Steinbruck SM, Maxwell SE, Howard GS et al: A meta-analysis of psychotherapy and drug therapy in the treatment of unipolar depression with adults. J Consult Clin Psychol 51(6): 856-863, 1983.

Wolfe BE: Adapting psychotherapy outcome research to clinical reality. Journal of Psychotherapy Integration 4:160-166, 1994.