Psychodynamic Psychotherapy for Anxiety Disorders

The S3 guideline on the treatment of anxiety disorders ranks cognitive behavioral therapy (CBT) as the first-line treatment for both panic disorder and generalized anxiety disorder. Psychodynamic therapy follows, with a lower grade of recommendation.

Generalized anxiety disorder and panic disorder (with or without agoraphobia) are the two disorders for which the guideline says the most about psychodynamic therapy. Social phobia receives the same conditional recommendation, with a level of evidence of Ib.

It should be offered, but only under three conditions. For panic disorder and agoraphobia, the guideline states verbatim in Recommendation 8: “Patients with panic disorder/agoraphobia should be offered psychodynamic psychotherapy if CBT has not proved effective, is not available, or if the informed patient has a preference for it.”

These three conditions constitute the actual substance of these two recommendations. Psychodynamic therapy comes into play when CBT has not been effective, when it is not available, or when the informed patient prefers it. All three paths lead to the same offer.

For both panic disorder and generalized anxiety disorder, CBT has a level of evidence of Ia and a grade of recommendation of A+, formulated as “shall.” Psychodynamic therapy has a level of evidence of IIa and a grade of recommendation of B+, formulated as “should.” One step softer, one condition longer.

The professional associations for depth psychology and psychoanalysis contributed to this consensus: It is their own assessment.

The lower ranking is based on individual studies that were downgraded for quality deficiencies. According to the guideline, IIa means: “Evidence from at least one methodologically well-controlled study without randomization.” B+ means: “‘Should’ recommendation: Well-conducted clinical studies, but no randomized clinical trials, with direct relevance to the recommendation (levels of evidence II or III) or extrapolation from level of evidence I if the relevance to the specific question is lacking.”

That sounds like a contradiction, because there are randomized studies on psychodynamic therapy for panic disorder. Milrod 2007 is one of them. Nevertheless, the guideline downgrades it and explains why: “Where methodological deficiencies were identified here, the level of evidence was downgraded (e.g., from Ib to IIb for an RCT with moderate deficiencies; to IV for serious deficiencies).”

The guideline illustrates how this applies in specific cases by referencing a study that was conducted simultaneously at two sites. It abbreviates psychodynamic therapy as PDTh and provides a separate statement for each site. Regarding the CORNELL site, it states that the study “shows no superiority of PDTh over the control group; nor can it be used as evidence of the equal efficacy of CBT and PDTh.” Regarding the PENN site, it states that the study “shows inferiority of PDTh compared with CBT and even compared with the control group.” The same study, the same methodology, two different results. It is precisely such findings that lower the level of evidence.

The Cochrane network meta-analysis on psychotherapies for panic disorder provides the clearest explanation. It reviews 54 studies involving 3,021 patients. For the most important outcome, the remission of symptoms after a short period of time, there were 32 studies on CBT compared to only two studies on psychodynamic therapy. The quality of evidence was consistently low across the entire network. Two studies versus thirty-two do not indicate equivalence. However, they also do not indicate inferiority.

The authors also note that psychodynamic therapy showed promising results in the two available studies, that it was the best tolerated of all the psychotherapies examined, as measured by dropout rates, and that, in the long term, CBT and psychodynamic therapy showed the highest remission and response rates. They note that all of these differences are small or imprecise. What is lacking are studies.

There is a counterargument, and it is worth considering. In 2014, Keefe and colleagues reviewed fourteen randomized studies involving 1,073 patients and found psychodynamic therapy to be just as effective as other treatments for anxiety disorders. Nevertheless, the guideline excludes this meta-analysis from its assessment, stating that it did not meet the guideline’s criteria, nor did the majority of the individual studies included in it.

However, since the guideline development group reviewed the literature, two network meta-analyses have been published that examine precisely these two disorders separately. These were not yet available to the guideline development group, and they alter the picture in two ways.

In a 2022 article in the British Journal of Psychiatry, Papola and colleagues analyzed 136 randomized trials on panic disorder with and without agoraphobia. Compared to usual care, two approaches performed best: CBT and short-term psychodynamic therapy. Confidence in the findings was unevenly distributed (moderate for CBT and low for psychodynamic therapy), and after excluding studies with a high risk of bias, only CBT remained superior to usual care. Nevertheless, the authors concluded that both are acceptable first-line treatments.

In a 2024 article in JAMA Psychiatry, Papola and colleagues analyzed 65 randomized trials with 5,048 participants on generalized anxiety disorder. Psychodynamic therapy was one of the eight approaches examined. In that analysis, CBT, third-wave approaches, and relaxation therapy were superior to usual care; psychodynamic therapy was not among them. Three to twelve months after the end of treatment, the advantage held only for CBT. Among the psychotherapies themselves, however, the study found no demonstrable difference in efficacy.

Taken together, this means that for generalized anxiety disorder, CBT is the best-supported choice, even more clearly than the 2021 guideline says. For panic disorder, by contrast, short-term psychodynamic therapy fares better in the more recent paper than in the German guideline text.

Behavior can be measured, but an unconscious conflict cannot.

How often someone avoids a situation, how high the panic level is on a scale of zero to ten, how many fewer points a questionnaire shows after eight weeks: All of this ends up in a table and can be compared between two groups. There is no similarly convenient scale for transference, defense mechanisms, or the maturation of a personality structure. Research follows what can be demonstrated.

In a 2010 article in American Psychologist, Jonathan Shedler quantified the gap: There are possibly ten times as many randomized studies on CBT as on psychodynamic approaches. He attributes part of this to the analysts themselves, because earlier generations considered empirical research unnecessary. The lead held by cognitive approaches dates back decades.

Then there is the issue of treatment duration. Shedler describes comparisons in which psychodynamic treatments consisting of eight sessions were pitted against CBT consisting of sixteen sessions. Even in those cases, the results were comparable. The studies underlying the two recommendations in this guideline are of a similar magnitude, with up to 24 hours for panic disorder and up to 30 hours for generalized anxiety disorder.

And the yardstick is narrow. What these studies measure is the level of symptoms: scores on an anxiety questionnaire, panic attacks per week. Whether someone is subsequently better able to take care of themselves, tolerate closeness, or express anger does not show up in these numbers. Shedler calls this the gap between what psychodynamic treatment aims to achieve and what is measured in the end.

A third point concerns the patients themselves. In 2004, Drew Westen and his colleagues showed that psychotherapy studies often exclude up to two-thirds of applicants, mostly because they have more than one mental health issue. It is precisely these people who are sitting in the therapist’s office.

In my experience, the first thirty sessions for generalized anxiety disorder are spent getting to know one another. More than that happens during this time. However, working through (the same pattern over and over again, in ever-changing contexts) and the subsequent maturation of the personality structure that is tied to it require the time beyond that. That is precisely where the research literature ends.

The decision rule is already stated above, but it is rarely recognized as such. The three conditions from Recommendations 8 and 23 (CBT is not effective, CBT is not available, or patient preference) are the answer to the question of when which approach is an option.

In practice, the two approaches differ in their focus. According to the guideline, modern behavioral therapy includes cognitive, behavior-modifying, emotion-enhancing, and confrontation-based elements. Cognitive behavioral therapy assumes that significant life events leave behind fixed patterns of thought that later help determine how a situation is assessed and what emotion follows; a key component is confronting what triggers fear. Depth psychology focuses more on the unconscious conflict from which the symptom arose and on the relationship patterns that sustain it.

Both have a model of how an anxiety disorder develops and what sustains it, and both take different therapeutic approaches. Those who view anxiety primarily as a signal of an unresolved inner conflict will want to work on that conflict. Those who view it primarily through the lens of the evaluations and avoidance behaviors that keep it alive will focus their efforts there. Both perspectives describe real aspects of the same disorder.

Based on the current data, it is not possible to definitively determine which approach is generally superior.

Depth psychology first asks where the anxiety comes from. It distinguishes between two sources: unconscious conflicts and structure. This distinction determines the course of treatment.

Conflict-Related Anxiety: The Signal for Unconscious Conflicts

One source is conflict. A desire or impulse contradicts an inner prohibition, and anxiety signals this contradiction before it becomes conscious. Because this anxiety signals something, people also speak of its signaling function. It is directed: it can be tied to specific situations, specific relationships, or recurring triggers. Most people can name exactly what they’re afraid of, even if the situation appears harmless from the outside.

Diffuse Anxiety When the Anchoring Fails

The other source is the structure. Diffuse anxiety has no identifiable object. It lacks direction because the ability to anchor anxiety to something is absent. This anchoring of anxiety is considered a sign of ego strength: it serves as a warning when it succeeds. If it fails, anxiety overwhelms the individual. From this perspective, a phobia is an achievement: diffuse anxiety has attached itself to an object and has thus become more bearable.

There is a broad spectrum between the two, and the boundaries are fluid. Ermann classifies fears of feeling helplessly exposed, as in agoraphobia, or fears about one’s own body as falling in the middle between these two extremes.

Depth psychology-based therapy employs two different approaches for these two sources: one insight-oriented and one experience-oriented. In the case of conflict-related anxiety, the focus is on insight: the conflict from which the anxiety arose is understood and addressed. With diffuse anxiety, however, precisely this uncovering approach would be the wrong path. Here, the initial focus is on experience and skills: perceiving, naming, enduring, and being able to calm oneself.

Which of the two sources takes precedence is therefore clarified at the beginning of treatment. Because the boundaries between them are fluid, the question remains open for adjustment throughout the entire course of treatment.

Duration calls for honesty. The studies on which the two recommendations are based lasted up to 24 hours for panic disorder and up to 30 sessions for generalized anxiety disorder. For adults, the Psychotherapy Guidelines initially provide for short-term therapy, in two parts of up to twelve sessions each, 24 in total. The studies on panic disorder fall precisely within this framework. Beyond that, the Psychotherapy Guidelines provide, for this approach in an individual setting, approval steps up to 60 sessions and a maximum limit of 100 sessions. The guideline itself points out this gap, using the example of the even longer analytical approach: “Although up to 300 sessions for analytical therapies can be reimbursed by payers in guideline-based approaches in Germany, there is no evidence for this in clinical studies.” What can be reimbursed thus goes well beyond what was tested in the underlying studies. An unsubstantiated effect is not the same as a refuted one: controlled evidence is lacking for this extended timeframe. I have described elsewhere what the same distinction looks like for depression.

I have never referred anyone with an anxiety disorder to behavioral therapy. The focus has always been on understanding the anxiety.

I have a basic understanding of behavioral therapy, and I myself have sometimes encouraged behavioral experiments to confront anxiety.