Early therapy research in the 50s questioned whether the success rate of therapy exceeded the “spontaneous recovery rate”. Eysenck (1952) claimed that there was no evidence for the effectiveness of therapy. Many researchers avoided outcome research and there was a “flight into process”. Subsequently, however, many types of therapies were found to be better than no therapy.
DISCOVERIES AND REALIZATIONS OF THE 60s
Bergin--The variability of outcome in treatment groups is almost always larger than that of the control groups.
1. Perhaps therapeutic benefit depends on certain therapist qualities?
2. Control groups are not “real” control groups. They improve. “Control groups may be better described as unspecified treatment groups.”
3. “WHAT MAY ACTUALLY BE THE EFFECTIVE INGREDIENT IN THE PSYCHOTHERAPEUTIC PROCESS IS NOT SPECIFIC TO PSYCHOTHERAPY.”
Kiesler (1996):
The question “Does psychotherapy work” is a stupidly put question because it assumes a number of MYTHS.
1. The uniformity assumption myths.
A) Patient uniformity--That all patients with the same diagnosis are a homogeneous group.
B) Therapist uniformity--That each and every therapist is an identical social stimulus for all patients.
2. The spontaneous recovery (SR) myth.
More importantly, Kiesler claims that the big 3 theories of psychotherapy (Psychoanalysis, Rogerian, and Behavioral) are inadequate in specificity and/or content for proper assessment in research. It is time to ask the question about psychotherapy differently.
Gordon Paul in 1969 re-poses the question, “What treatment, by whom is most effective for this individual, with that specific problem, and under what specific set of circumstances?”
In the 1st edition of the HANDBOOK (1971), Bergin concludes, “There is modest evidence that therapy works.”
Major outcome studies of the 70s using control groups:
Paul: Behavior therapy>Insight=Placebo>No treatment for speech anxiety.
Sloane: Insight=Behav. Therapy> Wait List control for outpatients with neuroses and personality disorders.
DiLoreto: Ellis=
Rush et al.: Cognitive therapy>Imipramine for depression.
But this may be limited to mild cases of depression. There appears to be no
tri-cyclic effect on those with
1978--2nd edition of the HANDBOOK
Bergin & Lambert cite over 500 studies of psychotherapy outcome show:
60% of studies show therapy helps,
30% of studies show therapy makes no difference,
10% of studies show therapy hurts.
Bergin & Lambert interpret this box score review as GOOD evidence that therapy is effective.
This interpretation is based on the assumption that if therapy were totally un-related to outcome (i.e., that the null hypothesis was true), then at alpha=.05 only 5% of studies would show that therapy works> so 60% cited above is impressive.
But there is an alternative view of hypothesis testing:
In the soft areas of psychology, the null hypothesis is never strictly true. That all molar psychological variables (like psychotherapy and emotional distress) are slightly related, sharing 4-5% of the variance in common, which translates to correlations of .16 to .25. So:
A) If measures are reliable, and
B) The sample size is large enough (There is good statistical power),
One will always reject the null hypothesis even though the theory may be false. If one makes directional predictions, the data will be significant in your predicted direction 50% of the time. The implications of the alternative view are that if the theory is true (i.e., therapy really works) then 100% of well-designed studies should show that therapy>no therapy. From the alternative view, 60% does not look so good.
1977 brings meta-analysis (Smith & Glass) to the
“Outcome question”.
Effect Size (ES) = (M treatment group - Mean control group) / SD control group.
In their book, Smith, Glass, & Miller calculated 1,760 Ess from 475 studies of therapy. The M ES = .85, interpreted as an estimate that after therapy the average patient was better off than 80% of the control group. The following were ESs for different schools of therapy:
2.38--Other cognitive therapy
1.82--Hypnosis
1.13--Cognitive behavior therapy
0.89--Dynamic eclectic
0.89--Eclectic behavioral
0.83--Behavior modification
0.69--Psychodynamic
0.68--RET
0.68--Implosive
0.67--TA
0.65--Vocational personal development
0.64--Gestalt
0.62--Client-centered
0.62--Adlerian
0.56--Placebo
0.28--Undifferentiated counseling
0.14--Reality therapy
The “Super schools”:
Verbal therapies, ES = .77 (78% of normal curve)
Behavioral schools, ES = .96 (83% of normal curve)
The above comparisons are all potentially misleading because they are not based on comparative studies and are likely confounded by variables like reactivity. For example, behavior therapies may be more likely to use more reactive measures than Verbal therapies. Let’s see how reactivity of measure related to effect size in the studies reviewed by Smith et al.
1. Low (ex., GSR & GPA)--ES=.55
2. Low average (blind ratings) ES=.55
3. Average (ex., MMPI) ES=.60
4. HI average (ex., client report to the therapist, experimenter constructed questionnaire) ES=.92
5. Hi (ex., therapist ratings, behavior in the presence of the therapist) ES=1.19
Also, the allegiance of the experimenter to the therapy being studied has a relationship to ES magnitude:
Pos.--ES=.95
Independent allegiance--ES=.78
Negative allegiance--ES=.66
A variety of potential problems with meta-analysis:
1. Garbage in-Garbage out
2. Why the SD of the control group only
3. Only sig. Results get published, some estimate this inflates ES by 60%
4. > 1 ES per study may overweight some studies
5. Other technical statistical problems.
Outcome chapter from the 1986 edition of the Handbook,
written by Lambert, Shapiro, & Bergin:
1. In meta-analyses assignment to treatment vs. No treatment accounts for approx. 10% of the variance in effect sizes.
2. Research consistently shows Therapy>Placebo>No therapy.
3. It is unclear whether the effects of therapy disappear over time, or whether they are constant.
4. Maintaining treatment gains seem to be related to the degree to which patients recognize that the benefits were the result of patient effort-not the therapist or Rx. (Your teacher questions the basis for this conclusion).
5. Comparative studies continue to reflect the Do-Do bird verdict.
Additional conclusions from the 1986 Handbook chapter:
1. No strong evidence for the superiority of 1 therapy over another with non-specific symptoms (e.g., demoralization, anxiety, depression, low self-esteem).
2. With more specific problems (e.g., phobias, compulsive rituals, childhood behavior disorders, sexual dysfunction), some treatments are showing superior outcomes.
3. All studies of therapist experience show paraprofessionals = professionals. They emphasize the role of non-specific factors to account for this.
4. The magnitude of effect sizes is sobering.
5. 9-11% of studies have negative effect sizes, supporting a deterioration effect.
Conclusions from the 1994 (4th ed.) Outcome chapter:
1. For depression, psychotherapy works and is = to Rx, although Rx may work faster (some would dispute this).
2. For anxiety disorders, therapy > Rx, especially for agoraphobia and panic disorder. Rx has high relapse rates here.
3. How important is the 10% of the variance accounted for by therapy, r = .32? Rosenthal provides an example suggesting that 10% may be more powerful than it looks when translated to the following table of Percent Improvement:
Treatment------- 66% improved, 34 % not improved.
No treatment----34% improved, 66% not improved.
4. The effects of therapy seem to be maintained at follow-up
5. Evidence for the superiority of one treatment over another is still small.
6. Common factors loom large as mediators of treatment outcome.
7. There is no strong (or even consistently weak) evidence that professionals are more effective than paraprofessionals.
8. Some therapists are consistently better than others. Other therapists are consistently worse.
9. On the average individual differences across therapists account for 9% of the variance in effect sizes, less when manuals are used.
10. Regarding deterioration: The more severely disturbed borderline and psychotic patients are more likely to deteriorate; especially with therapies that are designed to challenge, break down, or undermine habitual coping strategies. In group settings, authoritarian leaders, using expressive techniques of confrontation or brutal honesty are associated with deterioration. Deterioration in the NIMH study of depression was somewhat higher for CBT (10%) than other treatments (4%). While clinicians may observe neg. effects of individual therapy on a marriage, systematic data supporting this outcome is lacking.
11. A new emphasis on the use of therapy for the prevention of relapses. In 1990, Frank et al. Followed patients successfully treated for their 3rd or greater episode of depression with IPT (Interpersonal Therapy) and Rx in combination. The following are relapse rates at a 3 year follow-up:
Maintained on placebo: 80%
Maintained on Rx: 20%
Maintained on IPT alone or IPT + placebo: 54%
Anti-psychotic (decrease dopamine)
A. Phenothiazines--ex., Thorazine, Milaril, Stelazine
B. Thioxanthenes--ex., Taractan, Navane
C. Butyrophenones--ex., Haldol
D. Atypical--Clozaril
Anti-depressants
A. Tricyclics ( increase norepenephrine and serotonin and increase sensitivity of receptors by inhibiting reuptake)--ex., Imipramine (Tofranil), Amitriptyline (Elavil), Doxepin (Sinequan), Clomipramine (Anafranil).
B. Monoamine Oxidase (MAO) inhibiters--ex., Phenelzine (Nardil), Tranylcypromine (Parnate)
C. Second generation agents--ex., Trazodone (Desyrel), Amozapine (Ascendin), Maprotiline (Ludiomil)
D. Selective Serotonin Reuptake Inhibitors (SSRIs)--ex., Fluoxetine (Prozac), Sertraline (Zoloft), Paroxetine (Paxil), Fluvoxamine (Luvox)
E. Others--Bupropion (Welbutrin), Venlafaxine (Effexor), Nefazodone (Serzone)
F. Stimulants--ex., Benzedrine, Dexadrine, Ritalin
Anti-mania--Lithium
Anti-anxiety
A. Benzodiazepines--ex., Librium, Valium, Tranxene, Serax, Activan
B. Triazolobenzo-diazepine--ex., Zanax
C. Propanediol Carbamates--ex., Miltown
D. Barbituates--ex., Luminol, Seconal
Common beliefs (customs) about the treatment of certain disorders (not necessarily substantiated by research):
Schizophrenia: Psychotherapy (especially verbal) has no advantages by itself. Anti-psychotic drugs are somewhat effective (usually with positive symptoms), and a combination of drugs and psychotherapy is better than drugs alone.
Acute mania--Lithium, Thorazine, or Haldol
Acute depression in bi-polar disorders--Anti-depressants or Lithium
Uni-polar depression: if mild psychotherapy = drugs, if more severe combine therapy and drugs.
Panic disorder--typical drug treatments include imipramine (but “jitter” side effects in 20-40%) and alprazolam.
OCD--typical drug treatments include Prozac or Anafranil
Your instructor’s somewhat sarcastic perspective on fashions in Psychiatry: If one drug fails, change the diagnosis and try another drug. Keep repeating. If all else fails, try ECT.
Effects of psychotherapy on Beck Depression Inventory (BDI) scores, which are normed 10-20 (mild depression), 20-30 (moderate depression), and 30+ (Severe)
In 39 normative studies the general population BDI = 7.0; when samples are restricted to non-distressed people the mean BDI = 4.9.
In 22 studies of psychotherapy of depression the mean BDI at pretreatment was 21.8 and the mean BDI at post-treatment was 11.9. For untreated control patients the pre-treatment BDI was 20.7 and the post was 18.1. So treated patients at the end of therapy are still more depressed than the general population norms.
Westen and Morrison (2001) meta-analyzed high-quality studies of manualized treatments of a) depression, b) panic disorder, and c) generalized anxiety disorder (GAD).
In well-controlled efficacy studies, many patients are excluded from participation, so that relatively homogenous groups of clients can be studied. Reasons for exclusion often include a variety of co-morbid problems that might be considered the primary diagnosis, psychotic, bipolar, or organic disorders. Inclusion rates in the 34 studies were 32% for depression, 36% for Panic, and 35% for GAD. Including screened out subjects in data analyses leads to very conservative estimates of benefit from therapy.
Clients drop out of therapy. Data analyses of all who entered therapy and dropped out generally carry forward the client’s last data point in the analysis of outcome data. This is referred to as an “intent-to-treat” analysis. An intent-to-treat analysis also provides a conservative estimate of the effects of therapy, compared with an analysis of only those that “completed treatment”. 74% completed treatment for depression, 86% completed treatment for panic, and 84% completed treatment for GAD.
Effect sizes (d) are typically labeled “large” when d=.8, “medium” when d=.5, and “small” when d=.2. Effect sizes for post-treatment measures were .3 for depression, .8 for panic, and .9 for GAD. For depression, 51% of completers, 37% of intent-to-treat, and 14% of those screened were “improved”. For panic, 63% of completers, 54% of intent-to-treat, and 20% of those screened were “improved”. For GAD 52% of completers, 44% of intent-to-treat and 10% of those screened were “improved”. Even though many clients improve, at termination a substantial proportion still have mild, but clinically significant, levels of symptoms, especially for depression and GAD. At 12-18 mo. follow-up (based on a small no. of studies) only 37% of those improved from therapy for depression remained improved, the follow-up for people with panic was much better with 86% remaining improved. There is little follow-up information on GAD. Many treatment completers (28% for depression, 35% for panic, and 45% for GAD) sought additional treatment 12-18 months after completing treatment.
Lambert (2001) commentary on Westen and Morrison
should be best understood by considering Lambert’s emphasis on outcomes
management for the improvement of therapy, rather than the reliance on efficacy
studies. First, Lambert criticizes Westen and Morrison’s use of some
statistical analyses, especially those using all screened clients, as providing
underestimates of the effects of therapy. Lambert, however, does recognize that
the data are sobering. The actual proportion of clients that remain in therapy
for the 12-16 sessions usually found in the Westen and Morrison studies is
quite small. If you are in therapy for this long, roughly 50% of clients will
show “improvement”. But the mean no of therapy sessions is about 4 in the
Chapter 5. The Efficacy and Effectiveness of Psychotherapy, 5th
edition of Handbook.
General efficacy: More recent meta-analyses show that most psychotherapies are efficacious, but the magnitude of ES is now estimated at d = .6 rather than the d = .85 reported earlier by Smith et al. This translates to the average psychotherapy patient being better off than 72% of controls. When evaluated in clinical trials (which generally deliver 12-16 sessions of therapy), 40-70% of clients show change that is clinically meaningful (defined in various ways). Most clients in routine clinic practice, however, do not receive sufficient sessions of therapy. The benefits of therapy, as delivered in clinical trials, persist into follow-up; unfortunately, some disorders like the addictions and depression tend to reoccur.
Therapy compared with non-specific (placebo) effects: This is still a contentious issue. The general outcome is that therapy is better than non-specific treatments, which are better than no treatments. By there continues to be debate on the relative magnitude of the specific and non-specific components. The trend over the years is to attribute more of therapy’s effects to the non-specific components. Some (e.g., Wampold, 2001) would argue that no more than 1% of the benefits of therapy can be attributed to specific factors. Lambert and Ogles conclude that, “… common factors across treatments are accounting for a substantial amount of improvement found in psychotherapy patients. These so called common factors may even account for most of the gains that result from psychological interventions.” (p. 172).
How much therapy is necessary for improvement? A realistic summary of dose-response studies indicates that between 13 and 18 sessions of therapy are required for 50% of patients to be improved. It should be noted that the criteria of “improvement” varied across the studies reviewed.
In a review of 28 studies of randomized clinical trials, the average length of treatment was 12.7 session. Using clinical significance criteria, 58% of clients improved after treatment; using reliable change criteria, 67 % of clients improved. These response rates are slightly higher (i.e., longer therapy) than that observed above in the dose-response literature.
A review of OQ data from over 6,000 patients collected from 6 sites on a weekly basis provides a comparison with the clinical trials data. The median no. of sessions at these sites ranged from 2 to 8, with an overall median of 4. The mean no. of sessions was approximately 5. With therapy of this short duration less than 10% of clients show clinically significant change, and less than 20% of clients show reliable improvement (given that therapy ends at the median). Deterioration rates ranged from 3% to 14%, averaging 8.2%. For patients who continue in treatment, there is greater improvement. 21% of clients were reliably improved, with an additional 14% recovered (clinically significant change). Using CSC as a criterion, 21 and 45 sessions of psychotherapy were needed for 50% and 75%, respectively of clients to move from the dysfunctional to the functional distribution. Using RI as a criterion, 7 and 14 sessions were required for 50% and 75% of clients to show a 14 point improvement in OQ scores, on average.
Deterioration: Estimates of deterioration from therapy (5-10%) continue stable from earlier editions of the Handbook.
Comparisons between types of therapy: Most comparative studies of psychotherapy show that there are no differences between the effectiveness of different schools of therapy: “The Dodo Bird Verdict.” One possible explanation for Dodo is that what is effective about therapy lies in what is in common across different schools of therapy. Arguments about Dodo generate a lot of heat and smoke among therapy researchers. Related to the Dodo verdict is the finding that there may be a strong allegiance affect operating in therapy outcome research. That is, researchers with a belief in a theoretical mechanism are more likely to find larger effect sizes when investigating their favorite therapy and small effect sizes when investigating other therapies. Luborski, who (although he is analytic) has an allegiance to Dodo, has reported correlations of therapist allegiance and effect sizes in their studies of r = .85. Luborski contends than any studies showing differences between schools of therapy can be accounted for by the allegiance effect, not any real difference between therapies.
An argument against Dodo is that it does not consider the possibility of treatment by type of client interactions. That is, one treatment may be superior with clients in group A; but a second treatment may be superior with group X. This is generically referred to as Aptitude by Treatment interactions (ATI). Larry Beutler, and others, have an allegiance to this position. The authors of the chapter advocate for more research on ATI. Although some studies have reported ATI effects, they are rare and usually post hoc findings. Wampold (2001), who has a Dodo allegiance, has claimed that in the 30 years since Gordon Paul first proposed the matrix solution, which encompasses an ATI viewpoint, “…not one interaction theoretically derived from hypothesized client deficits has been documented robustly, casting doubt on the specificity of psychological treatments” (p. 147).
Considerations of the failure to fine many ATI interactions and many differences between therapies should also include the concept of statistical power. Many therapy outcome studies do not have sufficient participants to have sufficient power to detect small effect sizes or significant interactions.
Despite the evidence for Dodo, and their misgivings about empirically supported treatments, Lambert and Ogles do claim that there is some evidence for treatment specificity with some disorders. Specifically, they refer to behavioral and cognitive methods having a significant increment of efficacy with phobias, panic, and OCD (as well as childhood aggression, psychotic behaviors, and health-related behaviors).
Differential Effectiveness of Individual therapists. There is a consensus that some therapists have outcomes that are better than other therapists, although this effect becomes less with manualized treatments. Almost all of this research is post hoc. Based on meta-analyses, Lambert and Ogles estimate that 9% of the variance in treatment outcome is associated with individual differences in therapists.
Comparative Effectiveness of Professionals versus Paraprofessionals: Historically, the research literature has shown no consistent, strong support that professional training leads to greater improvement in outcome. If anything, research has shown that paraprofessionals were slightly more effective than professionals. In the last decade, however, some evidence has emerged that professional training may lead to improved outcomes.
Therapeutic
Thase & Jindal (Chapt. 16): Combining Drugs and
Psychotherapy—When to do so.
Historically, most people have believed that a combination of psychotropic medication and psychotherapy would be more effective than either treatment delivered alone. There was, however, little research to support this reasonable hypothesis. Thase and Jindal advocate certain instances where research supports the combination of therapy and drugs. Mono-treatments are more cost effective than combined treatments.
First, there are methodological and statistical reasons why it might be difficult to observe and additive or synergistic effect of adding the two treatments. First consider the drug placebo (PBO) effect, which can be construed as a combination of spontaneous remission plus the effects of repeated measurement plus expectancy effects plus the benefits of a relationship with the person administering medications. “In RCTs (randomized control trials) of antidepressant medications, for example, PBO effects consistently account for 50-75% of the response to pharmacotherapy (p.745). When the active drug is given, treatment effects may reach a ceiling of effectiveness that limits the possible additive effects of psychotherapy. The ceiling effect may also obscure benefits of adding drugs to ongoing psychotherapy. Many subjects are needed in research studies to observe small additive effects.
“No FDA-approved medication approaches universal efficacy, and the effect sizes of antidepressant and antianxiety medications are usually only modestly (i.e., .2 to .4) superior to placebo. In fact, FDA-approved medications fail to show a statistically significant drug-placebo difference in approximately 50% of controlled clinical trials of antidepressants and anxiolytics.” (p.748). “No psychotropic medication is truly curative; similar dilemmas must be faced in treatment of hypertension, diabetes, and arthritis. The aim of pharmacotherapy is to suppress symptoms, achieve return to premorbid (“normal self”) functioning, and prevent relapses or chronicity” (p. 749).
Depression: The results of the large NIMH collaborative treatment study of depression
Elkin (1994) NIMH Depression Study
250 uni-polar
depressed, non psychotic outpatients: 70% female, 89% Caucasian received a drug
wash-out before the study. They were assigned to 16 weeks of therapy in the
following four groups (% refer to drop-out rates) A) Beck’ Cognitive-Behavior
therapy (CBT-32%), (B) Klerman’s Interpersonal
Therapy (IPT-23%), (c) Imiprimine + Clinical
Management (Rx-33%), or D) placebo + Clinical Management (P-40%).
Three samples were analyzed statistically:
The Completer Sample of 155 Ss, who received 12 or more sessions, The end-point 204, who were enrolled at least 3.5 weeks, and
the end-point 239, which included 35 Ss, who attended at least 1 session, but
dropped out before 3.5 weeks.
The four measures: Hamilton Ratings Scale (Ham),
Beck Depression Inventory (BDI), Hopkins Symptoms Checklist (Hop), and the
Global Assessment Scale (GAS) times the 3 samples gives a total of 12
statistical analyses that were conservative in nature. The general direction of
the results is Rx>IPT=CBT>P. There were only 4 statistical significant
analyses.
1, Completer: Rx>P on Hop; 2, End-239:
Rx>P on GAS; 3, End-239: Rx=IPT>P on Ham; and 4, End-204, Rx>P on GAS.
Technically, analyses 3 and 4 should be considered “trends” only.
Recovery was defined as scores < 6
on the Ham and < 9 on the BDI. Recovery rates for the END 239 sample
were IPT (43%) = Rx (42%)>P (21) on the HAM. CBT (36%) was not different
from other groups on the HAM. There were no sig. Differences in recovery on the
BDI.
Secondary analyses split subjects into 2
groups (more vs. Less severe) with cut-off of 20 on Ham & 50 on GAS.
FOR LESS SEVERE DEPRESSION: THERE WERE NO
DIFFERENCES AMONG THE GROUPS.
FOR MORE SEVERE DEPRESSION: 1)
Rx>IPT=CBT>P on GAS, and 2) Rx=IPT>P on the HAM with CBT intermediate
between IPT and P.
CONCLUSIONS:
1. No evidence therapy more or less effective
than drugs.
2. Limited evidence IPT>P; no evidence
CBT>P.
3. Large improvement in P may have masked
treatment effects.
4. Some evidence that drugs worked faster.
The temporal course of change: by 12 weeks
Rx> other 3 treatments, but less of a difference among groups at 16 weeks.
There were also site X treatment interactions.
Generalized prognostic signs:
+ Lower cognitive dysfunction; higher patient
expectance
- Axis II Personality disorders, they still
improved, but were more likely to have some residual depression.
18 month follow-up: Recovery is defined as no
relapse (2+weeks of major depression), and at least 8 consecutive weeks of
little or no depressed symptoms:
CBT(30%); IPT (26%); Rx(19%); P(20%)
Relapse (defined as any treatment for 3
consecutive weeks in 18 mo).
(Figures are % relapsed followed by weeks in
treatment)
CBT(14%, 4,2 wks); IPT(43%, 11.0wks); Rx(44%, 20.3); P
(27%, 7.8wks)
Additional dependent measures were also used
in the NIMH study to investigate the possibility of “mode specific outcomes”.
The notion here is that certain measures may be more sensitive to specific
treatments. For example, The Dysfunctional Attitude Scale (with subscales of
Perfectionism and Need for Social Approval) was hypothersized
to be more sensitive to CBT. The Social Adjustment Scale (with subscales of
work, social, and leisure adjustment) was thought to be more sensitive to IPT.
The Endogenous Scale from the Schedule for Affective Disorders was thought to
be differentially responsive to Rx. However ONLY 1 ANALYSIS OF 11 showed mode
-specific effects: On the completer sample CBT>IPT on N. Social Approval.
Other research studies presented in Chapt.16 show a consistent pattern. When severity of depression is conceptualized as less severe, situational, non-endogenous versus more severe, recurrent, and endogenous; differential treatment recommendations are as follows:
For the less severe depression, treatment form generally does not matter. No psychotherapy has proved better than any other psychotherapy or any less effective than medications (or placebo administered within a case management framework that also provides support and reassurance). In fact, as we shall see, no more complex therapy (i.e., CBT) has proved to be more effective than Behavioral Activation. There may be an advantage to CBT in preventing or postponing relapse. After remission, booster psychotherapy sessions should be strongly considered to minimize relapse.
For more severe, recurrent depression, combined treatments are indicated. Life-long maintenance of drug therapy should be strongly considered.
From your teacher’s perspective: I would first look at the client’s Beck score(s). If above 20 I would strongly consider medication from the beginning. There is some indication from the NIMH study that medication may work more rapidly than psychotherapy. If doing only psychotherapy with someone with depression, drug treatment should be considered if there has not been substantial symptomatic improvement after 2 mo. of treatment.
Panic Disorder: Most forms of CBT, most antidepressants, and potent benzodiazepines (BZ) have been found to be equally effective in treating panic disorder. There should be a preference for CBT, however, given the side effects of the drugs and high relapse rates found after drug withdrawal. CBT seems to have more of a lasting influence. Anti-depressants are generally preferred to BZs. In combined treatments, BZs are thought to undercut the value of exposure treatments by suppressing arousal cues.
OCD and other anxiety disorders: Use combined treatments only with the more severe, chronic, or relapsing forms of the disorder; otherwise use a monotherapy. Discontinuing medication produces greater relapses than discontinuing psychotherapy.
Bulemia and other eating disorders: The authors conclude that CBT, either alone or in combination with pharmacotherapy, is superior to pharmacotherapy alone. There does appear to be a subset of patients that gain extra benefit from the combination, although the authors do not describe characteristics of these patients.
Disorders for which psychotherapy alone should not be used: Schizophrenia and bi-polar disorder. Psychotherapies should be strongly considered as important adjuncts in the treatment of these disorders.
. Stiles et al. (2003). Early Sudden Gains: The phenomenon of “early sudden gains” in therapy is one that has been recently identified. It refers to large early improvements (usually in symptom intensity) that exceed reliable change and often occur around the 5 th session. These gains were first observed in randomized clinical trials (RCT) with depressed clients. This report examines data from routine practice settings in order to study how generalizable the “gains” phenomenon is. The proportion of clients with such gains in this study (17%) was less than previous RCT reports (39%). Early gains in this report accounted for the great majority of total therapeutic change compared with 51% of the total change in the RCT with depressed clients. 43% of clients in this report had a later reversal of gains (loss of over 50% of the gain). (These reversals were often not sustained for more than one session.) Only 17% of the clients in the RCT study showed reversals.
The authors speculate that clients in routine practice settings are likely to show greater volatility in symptom intensity than depressed clients in RCTs. The authors conclude that sudden gains are a real, generalizable phenomenon that predict ultimate success in therapy and are maintained into follow-up. The authors state that on the core outcome measure; which is a composite of subjective well-being, symptom intensity, social/occupational functioning, and risk to self and others; “…clients who did not gain suddenly often did not gain at all in the long run on this measure.” (p. 19)