2018 - Enduring Effects of Psychological
2018 - Enduring Effects of Psychological
Review article
According to treatment guidelines,1–3 psychological therapies and as to whether it is an oversimplification to assume that the differ-
psychopharmacological drugs have shown efficacy for the treatment ences in relapse rates between drug treatment and psychotherapy
of the three major anxiety disorders: panic disorder with or without are substantial. In naturalistic studies following up patients with
agoraphobia (PDA), generalised anxiety disorder (GAD) and social anxiety, considerable relapse rates were found years after CBT treat-
anxiety disorder (SAD). Among psychotherapies, cognitive–behav- ment. For example, in an analysis of 8 controlled studies of CBT
ioural therapy (CBT) is the method studied most, but some trials for anxiety disorders, 48% of patients were still symptomatic after
have also investigated applied relaxation, psychodynamic therapy, 2–14 years of follow-up.7 On the other hand, in relapse prevention
interpersonal therapy, mindfulness meditation and therapies con- studies1,8 in which treatment responders to open drug treat-
ducted via the internet. Medications used for anxiety disorders ment for 8–12 weeks were re-randomised to long-term treatment
include selective serotonin reuptake inhibitors, serotonin–nor- (24–52 months) with the same drug or placebo, the relapse rates
adrenaline reuptake inhibitors, pregabalin, tricyclic antidepressants, of patients randomised to placebo ranged from 8 to 56%.
benzodiazepines and others.1 Available follow-up studies directly comparing the durability of
In a recent meta-analysis of 234 acute treatment studies for CBT with drug therapy did not show clearly longer-lasting effects of
anxiety disorders involving 37 333 patients, we had shown that CBT: in only one9 of one studies of PDA, a longer-lasting effect of
medications were associated with significantly higher average pre- CBT could be demonstrated.10–13 Likewise, in SAD, only two14,15
post effect sizes (Cohen’s d = 2.02) than psychotherapies (Cohen’s of four studies have shown longer-lasting effects for CBT than for
d = 1.22).4 We did not find evidence that this result was influenced medication, whereas two did not.16,17
by heterogeneity, publication bias or allegiance effects.
333
Downloaded from https://www.cambridge.org/core. 10 Nov 2021 at 17:55:01, subject to the Cambridge Core terms of use.
Bandelow et al
To our knowledge, no previous meta-analysis has studied For every follow-up week, we pooled all available studies for the
whether psychological therapies have longer-lasting effects than three anxiety disorders for this time point. We only included studies
control conditions. Therefore, we used studies involving a drug with up to 24 months duration, as there was only one evaluable
treatment arm and studies using a pill placebo or a psychological study with one CBT arm using a longer follow-up period
(attention) placebo as a control to see if there was a significantly (36 months), and no further eligible studies with longer follow-up
larger decline of effect size after termination of drug treatment com- intervals.
pared with psychological therapies. A ‘psychological placebo’ is
defined as conversation of the same length as a psychotherapy
session, in which study staff who do not necessarily have psycho- Meta-analytical procedure
therapeutic training establish a supportive, listening and nondirec- Outcome measures
tive relationship without applying specific techniques. Three reviewers (B.B., Y.G., and A.S.) independently extracted all
The meta-analytic procedure has the advantage that all of the data. Any discrepancy was resolved by consensus. To limit hetero-
many available follow-up studies can be included in the analysis geneity and to achieve maximum comparability, we preferably used
and not only the few head-to-head comparisons of psychotherapy the most commonly applied scales: the Hamilton Rating Scale for
and drug or placebo conditions. Anxiety21 for PDA and GAD and the Liebowitz Social Anxiety
Scale22 for SAD. If these were not available for the follow-up time
points, we chose other scales following an algorithm described in
Method
Bandelow et al.4
Effect sizes (Cohen’s d) were calculated from the differences
Selection of studies between baseline and end-point or follow-up time point by sub-
The present study extends a comprehensive meta-analysis on effi- tracting the post-treatment mean from the pre-treatment mean
cacy of treatments for anxiety disorders in short-term studies to and dividing the difference by the pre-treatment s.d. of the
follow-up studies.4 measure.23 If there was more than two treatment groups in a
Randomised treatment studies from 1980 to 2016 for PDA, study, a pooled baseline s.d. based on all treatment arms in the
GAD and SAD were found by electronic and hand search. Study study was used.
quality was assessed with the Scottish Intercollegiate Guidelines Whenever available, intention-to-treat data were used. Where a
Network statement.19 Reasons for exclusion were missing informa- study only reported data from dichotomous outcomes (the propor-
tion making it impossible to compute effect sizes, a sample size of tion of responders to treatment, e.g. defined by a 50% reduction on
any of the treatment arms at inclusion less than ten, reports that the Hamilton Anxiety Scale), it was assumed that participants who
were restricted to subsamples (e.g. only elderly patients) and ceased to engage in the study from whatever group had an unfavour-
studies that included children and/or adolescents. We did not able outcome. Odds ratios were transformed to Cohen’s d.24
include open studies because these may have been influenced by We calculated the effect sizes for all three anxiety disorders
expectation effects. Drugs were included that had been shown to together, as effect size did not differ significantly between the differ-
be effective in randomised controlled studies and are licensed in ent disorders in our first study.4
at least some countries for the treatment of anxiety disorders.1
Psychological therapies were categorised as follows: ‘CBT’ included
individual or group CBT or exposure techniques or a combination Analysis
of both, as well as CBT treatments conducted via the internet; ‘other Meta-analyses were done by Comprehensive Meta-Analysis
psychotherapies’ comprised psychodynamic therapy (n = 5), inter- Version 3.0 (Biostat, USA). IBM SPSS Statistics 24 was used to
personal therapy (n = 2), relaxation (n = 16), mindfulness therapy conduct further analyses. Because most studies differed consider-
(n = 2) and bibliotherapy (n = 7). Drugs used in the studies were ably in scheduling follow-up assessments, including intervals
alprazolam, citalopram, clomipramine, fluoxetine, fluvoxamine, between 4 and 104 weeks (see online Table DS2 for an overview),
imipramine, lorazepam, moclobemide, paroxetine, phenelzine and we interpolated missing data linearly between the nearest empirical
sertraline. Control conditions included pill placebo (n = 7 studies) follow-up assessments by calculating the mean of the available
and ‘psychological placebo’ (n = 8), and a combination of both follow-up scores before and after the interpolated follow-up time
(n = 2). None of the included control conditions involved treatment point. If no further follow-up scores were available, we used the
as usual. Because of the small sample sizes in the different placebo last-observation-carried-forward (LOCF) method. We preferred
groups, we did not analyse the placebo conditions separately. In this combined model, as solely using the LOCF method can lead
the psychological placebo studies, the number and length of the ses- to an inappropriate bias.25 We chose a general linear model for
sions were the same as in the experimental conditions, except in two repeated measures, as this fits best to analyse the variance
studies, in which patients in the psychological placebo condition for within-subject measurements (follow-up assessments) and for
had fewer sessions than the psychotherapy group. between-subject factors (treatments) within the same model.
From the original database of 234 eligible studies used in the Follow-up effect sizes were added as 13-stage within-subject factor,
meta-analysis, 91 studies with 180 study arms were chosen that also including the interpolated and LOCF scores. Treatment arms
had investigated psychological therapies, medications or a psycho- were included as a four-stage between-subject factor. For multiple
logical placebo and had included at least one follow-up assessment. comparisons, P values were corrected by the Bonferroni method (sig-
In addition, two new follow-up studies that appeared since nificance was set at P < 0.05, two-tailed).
1 October 2013 were added.15,20 Heterogeneity was assessed with the Q statistic and the I2 metric
A total of 93 studies with 185 study arms were included in the with 95% confidence intervals.26 Because moderate (I2 > 50%) to
analysis (CBT, n = 120 study arms; other psychotherapies, n = 32 high (I2 > 75%) heterogeneity27 was found for most comparisons
study arms; medications, n = 16 study arms and placebo conditions, (online Table DS2), the random effects model in which studies
n = 17 study arms). The selection of studies is displayed in a are weighted based on the inverse variances and an additional vari-
Preferred Reporting Items for Systematic Reviews and Meta- ance component reflecting the observed heterogeneity was applied
Analyses statement (Fig. 1). All studies are listed in online in all analyses. In general, including a random effect will lead to
Table DS3. more conservative results than the fixed effects model.
334
Downloaded from https://www.cambridge.org/core. 10 Nov 2021 at 17:55:01, subject to the Cambridge Core terms of use.
Enduring effects of psychological treatments for anxiety disorders
1420 records after duplicates removed 1175 eliminated; reasons: meeting abstract; no full-text available;
Screening
Screened by title and abstract letter; review; meta-analysis; pooled study; double publication;
secondary analyses; open study; naturalistic; case reports; case
series; no randomisation; drug/herbal preparation not licensed;
no DSM/ICD anxiety disorder diagnosis/patients with more than
one anxiety disorder/comorbid patients; treatment of subgroups
(e.g., only students; only elderly patients; children/adolescents
<18 etc.); treatment of non-responders/adjunctive treatment of non-
responders/augmentation/drug combination treatments; long-term;
maintenance; relapse prevention studies; sample size of any of the
arms in a study of <10 at baseline
Eligibility
245 full-text articles assessed for 30 eliminated after reading full-text; reasons: double publication
eligibility (4); scores necessary to calculate effect sizes not presented (8);
data only presented as a graph (6); only subgroups analysed (3);
n<10 at baseline in one of the study arms (2); no adequate control
group (2); methodological flaws (5)
Included
234 studies published in 232 papers 26 additional records indentified through seven eliminated; reasons: scores necessary to
included in qualitative synthesis hand search calculate effect sizes not presented (three); data
only presented as a graph (one); methodological
flaws (three)
91 papers with follow-up assessments two studies added that appeared after
extracted original meta-analysis
93 studies included
Fig. 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.28
The analysis of publication bias has been described in a previous The general linear model for repeated measures showed a sig-
publication.4 Possible allegiance effects for all study arms were ana- nificant difference between the four treatment arms (F(3, 181) =
lysed by two independent raters and were assumed when a medica- 3.268; P = 0.023; see Fig. 2). However, post hoc analyses revealed
tion study was sponsored by the current manufacturer of the that only the difference between CBT and placebo was significant
investigated drug, when authors disclosed financial support from (P = 0.021). The medication arms did not differ significantly from
the manufacturer or when one of the authors was a staff member the CBT and other psychotherapies arms, demonstrating that
of the manufacturer. For psychological treatments, allegiance patients who stopped taking a drug showed the same durable
effects were assumed when authors had developed the treatment, improvement as patients who stopped psychotherapy.
contributed to an aetiological model or published manuals for the Substantial heterogeneity was found in all three conditions, with
treatment. moderate to high I2 values (online Table DS2), indicating that the
distribution did not estimate a common population mean. For 40
(26.3%) of the psychotherapy arms and 5 (31.1%) of the drug
arms, allegiance effects were assumed.
Results
335
Downloaded from https://www.cambridge.org/core. 10 Nov 2021 at 17:55:01, subject to the Cambridge Core terms of use.
Bandelow et al
1.8
1.6
Effect size (Cohen's d)
1.4
1.2
0.8
0.6
Placebo
0.4 CBT
Other psychotherapy
0.2
Drugs
0
0 468 13 16 26 35 39 52 65 78 104
FU week
Fig. 2 Outcome during follow-up period. Error bars indicate 95% CI.
CBT, cognitive–behavioural therapy; FU, follow-up.
The enduring effect of drug treatment was not significantly patients take their medication only for some months and stop the
smaller than the one obtained with psychotherapy. Our study drug soon after remission has occurred, although guidelines recom-
casts doubt on the widespread assumption that only psychological mend a treatment duration of 121 or 62 months after remission.
treatments have enduring effects and gains achieved with medica- Also, the effect sizes in the placebo conditions did not show a
tions are lost soon after they are stopped. However, the good significant decline during the follow-up period. However, in con-
news for patients with anxiety disorders is that the chance of deteri- trast to the other psychotherapies, CBT showed significantly
oration within 2 years after treatment termination is low, and inde- higher effect sizes than the placebo arms, which may have been a
pendent from previous treatment. result of the large number of study arms in this condition. No dif-
We know from relapse prevention studies that drugs may have ferences were found between CBT and the other psychotherapies;
long-lasting effects. These enduring effects may have a neurobio- however, as the latter group comprised various forms of psychother-
logical background. For example, serotonergic drugs may exert sus- apy, no conclusions can be drawn from this finding.
taining effects on serotonin neurotransmitter systems in the brain, Our results will have to be reconciled with relapse prevention
which last longer than the actual treatment period. However, expect- drug studies, which mostly show some deterioration in the drug
ancy effects may also take effect, as patients who have experienced arm and a significantly greater deterioration in the placebo arms.
improvement with a drug know that they can restart the drug at Also, psychotherapy is associated with substantial relapse rates, as
any time in case a relapse occurs. Many patients with anxiety disor- has been shown in naturalistic studies. The most probable explan-
ders have concerns that they ‘will have to take the drug forever’, once ation for this lack of deterioration during the treatment-free
a medication has been started. In clinical practice, however, many period in our meta-analysis is that the long-lasting effects seen in
Table 1 Pre-post effect sizes effect sizes during follow-up period (Cohen’s d).
Treatment Other
All Psychological Psychotherapies Psychological or pill
therapies (n = 152) CBT (n = 120) (n = 32) Medication (n = 16) placebo (n = 17)
d 95% CI d 95% CI d 95% CI d 95% CI d 95% CI
End-point 1.362 1.247–1.477 1.370 1.242–1.499 1.332 1.076–1.587 1.379 0.994–1.764 0.885 0.566–1.204
Follow-up
4 weeks 1.397 1.285–1.509 1.404 1.280–1.528 1.371 1.115–1.627 1.321 0.983–1.659 0.908 0.601–1.215
6 weeks 1.406 1.296–1.516 1.412 1.290–1.534 1.384 1.126–1.642 1.292 0.972–1.612 0.914 0.607–1.220
8 weeks 1.415 1·306–1.524 1.421 1.300–1.541 1.397 1.136–1.657 1.263 0.957–1.569 0.917 0.609–1.224
13 weeks 1.439 1.331–1.547 1.442 1.324–1.559 1.430 1.163–1.697 1.220 0.946–1.494 0.946 0.633–1.258
16 weeks 1.450 1.341–1.559 1.453 1.335–1.570 1.439 1.171–1.707 1.215 0.939–1.491 0.954 0.634–1.274
26 weeks 1.485 1.374–1.596 1.489 1.368–1.610 1.468 1.194–1.743 1.199 0.897–1.501 0.982 0.626–1.338
35 weeks 1.501 1.389–1.613 1.508 1.385–1.630 1.476 1.203–1.749 1.195 0.902–1.489 0.989 0.634–1.344
39 weeks 1.508 1.395–1.621 1.515 1.392–1.638 1.480 1.208–1.753 1.194 0.903–1.485 0.992 0.638–1.347
52 weeks 1.527 1.413–1.641 1.536 1.410–1.662 1.492 1.219–1.764 1.197 0.907–1.487 0.993 0.641–1.346
65 weeks 1.528 1.418–1.648 1.543 1.416–1.669 1.497 1.222–1.771 1.211 0.921–1.501 0.989 0.641–1.337
78 weeks 1.539 1.423–1.655 1.549 1.422–1.676 1.502 1.224–1.779 1.219 0.926–1.512 0.982 0.637–1.327
104 weeks 1.544 1.427–1.661 1.555 1.427–1.683 1.501 1.224–1.779 1.234 0.933–1.535 0.966 0.625–1.308
CBT, cognitive–behavioural therapy; d, Cohen’s d; n = number of study arms.
Missing values were interpolated.
336
Downloaded from https://www.cambridge.org/core. 10 Nov 2021 at 17:55:01, subject to the Cambridge Core terms of use.
Enduring effects of psychological treatments for anxiety disorders
of the acute studies in which for all medications together, an 8 Donovan MR, Glue P, Kolluri S, Emir B. Comparative efficacy of antidepressants
in preventing relapse in anxiety disorders – a meta-analysis. J Affect Disord
effect size of d = 2.02 was calculated. 2010; 123(1–3): 9–16.
Our study has limitations. There were only few follow-up
9 Marks IM, Swinson RP, Basoglu M, Kuch K, Noshirvani H, O’Sullivan G, et al.
studies using medication or placebo conditions. Substantial hetero- Alprazolam and exposure alone and combined in panic disorder with agora-
geneity was found. Follow-up studies with the medications that phobia. A controlled study in London and Toronto. Br J Psychiatry 1993; 162:
showed higher effects sizes in the acute treatment period are 776–87.
lacking. Therefore, our results should be interpreted with caution. 10 Loerch B, Graf-Morgenstern M, Hautzinger M, Schlegel S, Hain C, Sandmann J,
et al. Randomised placebo-controlled trial of moclobemide, cognitive-behav-
Patients on medications should be monitored for a relapse, and ioural therapy and their combination in panic disorder with agoraphobia. Br J
treatment should not be terminated too early. There were substan- Psychiatry 1999; 174: 205–12.
tial differences in the number of available follow-up assessments at 11 Barlow DH, Gorman JM, Shear MK, Woods SW. Cognitive-behavioral therapy,
the various time points and in the intervals between these time imipramine, or their combination for panic disorder: a randomized controlled
points. Therefore, we had to deal with missing values in the statis- trial. JAMA 2000; 283(19): 2529–36.
tical analysis. We pooled studies from all three major anxiety disor- 12 Cohen SD, Monteiro W, Marks IM. Two-year follow-up of agoraphobics after
exposure and imipramine. Br J Psychiatry 1984; 144(3): 276–81.
ders because in the analysis of the acute studies we did not find
13 Mavissakalian M, Michelson L, Dealy RS. Pharmacological treatment of agora-
significant differences between these disorders with respect to phobia: imipramine versus imipramine with programmed practice. Br J
response. However, some anxiety disorders may be more inclined Psychiatry 1983; 143: 348–55.
to respond to a certain treatment than others, and some drugs 14 Clark DM, Ehlers A, McManus F, Hackmann A, Fennell M, Campbell H, et al.
may be more effective than others. Cognitive therapy versus fluoxetine in generalized social phobia: a randomized
In summary, uncontrolled studies that report stable improve- placebo-controlled trial. J Consult Clin Psychol 2003; 71(6): 1058–67.
ments after a treatment-free follow-up period may overestimate 15 Nordahl HM, Vogel PA, Morken G, Stiles TC, Sandvik P, Wells A. Paroxetine,
cognitive therapy or their combination in the treatment of social anxiety dis-
the ‘durability’ of psychotherapies, as these may be caused by unspe- order with and without avoidant personality disorder: a randomized clinical
cific effects. The often-cited advantage of psychotherapy over trial. Psychother Psychosom 2016; 85(6): 346–56.
pharmacotherapy for anxiety disorders – a longer-lasting improve- 16 Liebowitz MR, Heimberg RG, Schneier FR, Hope DA, Davies S, Holt CS, et al.
ment – could not be confirmed in our study. Future follow-up Cognitive-behavioral group therapy versus phenelzine in social phobia: long-
term outcome. Depress Anxiety 1999; 10(3): 89–98.
studies should use a protocol that monitors confounding factors,
e.g. additional, unscheduled medications or psychological therapies 17 Haug TT, Blomhoff S, Hellstrom K, Holme I, Humble M, Madsbu HP, et al.
Exposure therapy and sertraline in social phobia: I-year follow-up of a rando-
during the follow-up period. mised controlled trial. Br J Psychiatry 2003; 182: 312–8.
18 Patterson B, Boyle MH, Kivlenieks M, Van Ameringen M. The use of waitlists
Borwin Bandelow, Anne Sagebiel, Michael Belz, Yvonne Görlich, as control conditions in anxiety disorders research. J Psychiatr Res 2016; 83:
Sophie Michaelis, Dirk Wedekind, Department of Psychiatry and Psychotherapy, 112–20.
University Medical Centre Göttingen, Germany. 19 Scottish Intercollegiate Guidelines Network (SIGN). SIGN 50: A guideline
Correspondence: Dr Borwin Bandelow, Department of Psychiatry and Psychotherapy, developer’s handbook. SIGN, 2015 (www.sign.ac.uk/assets/sign50_2015.pdf).
University of Göttingen, von-Siebold-Straβe 5, D-37075 Göttingen, Germany. 20 Leichsenring F, Salzer S, Beutel ME, Herpertz S, Hiller W, Hoyer J, et al. Long-
Email: [email protected] term outcome of psychodynamic therapy and cognitive-behavioral therapy in
First received 10 Jan 2018, accepted 13 Feb 2018
social anxiety disorder. Am J Psychiatry 2014; 171: 1074–82.
21 Hamilton M. The assessment of anxiety states by rating. Br J Med Psychol 1959;
32(1): 50–5.
22 Liebowitz MR. Social phobia. Mod Probl Pharmacopsychiatry 1987; 22: 141–73.
Supplementary material 23 Dunlap WP, Cortina JM, Vaslow JB, Burke MJ. Meta-analysis of experiments
with matched groups or repeated measures designs. Psychol Methods 1996; 1
Supplementary material is available online at https://doi.org/10.1192/bjp.2018.49. (2): 170–7.
337
Downloaded from https://www.cambridge.org/core. 10 Nov 2021 at 17:55:01, subject to the Cambridge Core terms of use.
Bandelow et al
24 Borenstein M, Hedges L, Higgins J, Rothstein H. Introduction to Meta-Analysis. 27 Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in
Wiley, 2009. meta-analyses. BMJ 2003; 327(7414): 557–60.
25 Little RJ, D’Agostino R, Cohen ML, Dickersin K, Emerson SS, Farrar JT, et al. 28 Moher D, Altman DG, Liberati A, Tetzlaff J. PRISMA statement. Epidemiology
The prevention and treatment of missing data in clinical trials. N Engl J Med 2011; 22(1): 128.
2012; 367(14): 1355–60.
26 Ioannidis JP, Patsopoulos NA, Evangelou E. Uncertainty in heterogeneity esti-
mates in meta-analyses. BMJ 2007; 335(7626): 914–6.
338
Downloaded from https://www.cambridge.org/core. 10 Nov 2021 at 17:55:01, subject to the Cambridge Core terms of use.