Mechanisms of Change in Intensive Short-Term Dynamic Psychotherapy: Systematized Review

Four of the 14 studies refer specifically to the concept of unlocking the unconscious as a key process underpinning change (18–21). Abbass et al. (18) explored instances of unlocking the unconscious during the initial assessment session, referred to as the “trial therapy.” Participants were 500 patients referred to a university- and hospital-based tertiary psychotherapy care unit. The researchers explored whether the therapy was more effective for patients who experienced a major unlocking during their initial session.

Three raters observed either live interviews or video recordings of the initial session. The raters identified sessions with at least one instance of major unlocking, as determined by a clinical assessment of the unlocking process. A major unlocking event was identified if a clear image transfer occurred. In major unlocking, while a patient is looking toward the therapist, an image of a past attachment figure may emerge. For example, a patient may be participating in therapy and have a sudden image of his or her mother transferred onto the therapist. The researchers assessed interrater reliability across 30 video-recorded vignettes and reported adequate interrater agreement (κ≥0.66); however, one of the raters was also the clinical supervisor of the study.

The authors considered resistance as a potential moderating variable. For the purpose of the study, resistance was used as a binary variable, with patients classified as either “psychoneurotic” or “fragile (resistant).” Outcomes were measured by the Brief Symptom Inventory (BSI) (22) and the Inventory of Interpersonal Problems (IIP) (23) at baseline and at 1-month follow-up.

The authors indicated that 25% of participants had experienced a major unlocking event during their first ISTDP session and that these participants had experienced greater improvement on both the BSI and IIP from baseline to the trial’s end. Participants classified as fragile had experienced more symptoms at baseline and were less likely to experience a major unlocking. This finding is consistent with ISTDP theory, which posits that fragile patients have lower emotional capacity and require interventions to build anxiety tolerance prior to experiencing unlocking. Unlocking events had a similar positive impact for patients categorized as fragile or psychoneurotic. The effects seen in the group with major unlocking were reported as moderate to large.

The Abbass et al. findings were correlational. The involvement of the study’s clinical supervisor as a rater to identify instances of unlocking raises questions about the validity of the process followed.

Johansson et al. (19) investigated whether an experience of major unlocking of the unconscious predicted better treatment outcome among 412 patients. Therapists were licensed mental health professionals and ISTDP trainees. The mean±SD for treatment length was 10.20±13.30.

The study provided no operationalized definition of unlocking, instead referring readers to Davanloo’s work for details (4, 24). Patients were categorized into two groups: those who experienced major unlocking and those who did not. Resistance was considered a moderating variable, with patients categorized as either “fragile and/or psychotic” or “psychoneurotic.” An experienced ISTDP supervisor coded unlocking events and resistance by reviewing video recordings of treatment sessions. Outcomes were measured with the BSI and IIP.

This study demonstrated overall effectiveness for ISTDP and that major unlocking of the unconscious during therapy predicted better outcomes. Patients classified as fragile and/or psychotic had more symptom severity pretreatment than those classified as psychoneurotic. Patients classified as fragile and/or psychotic showed a steeper change in BSI scores compared with patients classified as psychoneurotic.

Interestingly, the experienced therapists were no more effective than trainee therapists in the Johansson et al. study (19). This finding contrasts with Abbass and colleagues (15), who found that greater length of therapist training was associated with treatment cost reduction in the long term, suggesting that the utility of treatment may have varied with level of therapist training.

Johansson et al.’s findings were correlational. The rise of complex transference feelings, unlocking of the unconscious, and resistance were coded by using nonvalidated measures. Because the coding was conducted by a single clinical supervisor, it was impossible to assess interrater reliability of the coding measures, raising questions about validity.

The Town et al. study (20) explored differences in self-reported symptom outcomes, cost savings, or functional outcomes between patients who had experienced major unlocking of the unconscious and those who had not. The study used an existing data set (25). Resistance was considered a possible covariate and was dichotomized as lesser or extreme.

The study provided no information on the coding process used to identify instances of unlocking the unconscious. On the basis of Davanloo’s case material and manualized framework (26), the experimental group was divided into patients who experienced one or more major unlocking of the unconscious during therapy and patients who did not.

Within the treated sample (N=89), 57 patients had experienced major unlocking, and 32 had not. Town et al. (20) showed significantly better outcomes for patients with major unlocking on self-reported outcomes (psychiatric symptoms and interpersonal functioning), functional measures (patients’ return to work and discontinuation of medication), and service utilization posttreatment.

The authors concluded that unlocking of the unconscious should be considered a key therapeutic process of ISTDP. A secondary finding showed that patients with high levels of resistance seemed to require major unlocking of the unconscious to benefit. This finding is in alignment with the ISTDP principle that resistance limits patients’ capacities to experience emotions. Major unlocking may be necessary for highly resistant patients to achieve positive change.

A major shortcoming of the Town et al. (20) study was the potential for experimenter bias because the therapist also acted as the rater. The authors argued that bias was less likely because ratings were done without knowledge of the health care cost data outcomes.

Fleury et al. (21) examined a patient’s physiological responses during episodes of unlocking the unconscious. Measures were examined during episodes of unlocking in two ISTDP sessions with a patient suffering from panic disorder. No information was provided on how episodes of unlocking were identified.

Measurements were taken at times when the patient self-identified or discussed with the therapist a strong emotional experience within the session. Measurements included sympathetic and parasympathetic indices, vagal tone, and patient responses. The authors hypothesized that specific self-reported feelings in session (i.e., anxiety, rage, guilt) would be associated with distinct patterns of autonomic nervous system activity (sympathetic and parasympathetic). They also hypothesized that respiratory sinus arrhythmia would increase following an unlocking event.

Consistent with ISTDP theory, the experience of self-reported rage in the transference manifested as a distinct somatic pathway. The authors suggested that the sympathetic system activates during defensive responses associated with anxiety and during the passage of unconscious-aggressive impulses. Following the experience of possible unconscious guilt, the authors noted increased vagal tone corresponding to the phase of reunification with attachment figures.

Although variance in physiological measures throughout a psychological intervention is interesting, the conclusions to be drawn from this study are limited. The study depended on the premise that the authors’ perceptions of the patient’s emotions and unconscious processes were accurate. No operationalized definition or system of identification for unlocking or self-reported measures of distress or emotional activation were provided. Moreover, the single-case design prevents generalization of the findings. Thus, although the Fleury et al. (21) findings are of interest, they are far from conclusive.

Six studies focused on in-session processes and interactions between therapists and patients in ISTDP. Bernardelli et al. (27) coded occurrences of various therapist interventions for a single patient receiving ISTDP. Although no quantitative outcome data were reported, the authors indicated that the intervention was successful, and the patient made good progress. Three sessions (early, middle, and late) of the complete 16-session treatment were analyzed. Therapist statements were categorized by independent raters to explore the pattern of interventions. The Hill Counselor Verbal Response Category System–Revised (28) was used to examine the therapist’s statements.

Results revealed that most of the therapist’s statements could be characterized as information seeking (46%), providing information (21%), interpretation (17%), or confrontation (10%). The results showed an absence of self-disclosure and few restatements, reflections, or direct advice. This study painted a picture of a therapist focused exclusively on the client’s disclosures and drawing upon a psychodynamic frame of reference. The information seeking was consistent with the model’s theoretical framework, with a focus on exploring dynamic patterns that maintained the client’s distress (1).

Although the therapist’s interventions were congruent with ISTDP principles, the lack of reported outcome data makes it impossible to determine whether the interventions were associated with positive change. In addition, the identified interventions were consistent with many forms of therapy, particularly psychodynamic modalities. The Bernardelli et al. findings therefore do not provide specific support for ISTDP, and the single-case design of the study limits the generalizability of the results.

De Stefano et al. (29) investigated the association between therapist interventions and occurrence of “good moments” (occurrences of client movement, progress, or change) (30) of a single patient receiving ISTDP. Independent raters coded therapist responses in three sessions (early, middle, and late) of a 16-session treatment by using the Hill Counselor Verbal Category System–Revised (28) and the Category System of Client Good Moments (30).

Each therapist statement was rated as representing one verbal response mode, and each client statement was rated as to whether it was a good moment. The bulk of therapist response modes were coded as confrontation, interpretation, providing information, and information seeking.

Confrontation and information seeking were associated with the appearance of immediate good moments, whereas interpretation and provision of information were not. The authors concluded that the effectiveness of interpretation varied with factors, such as timing, accuracy, validity, and content. The authors further concluded that while interpretations had established therapeutic value, their use may not be associated with indications of immediate client change. Similarly, provision of information appeared unrelated to occurrence of good moments. This finding may reflect that in ISTDP, provision of information is used primarily as a clarifying statement aimed at helping patients to notice and understand the nature of their psychodynamic conflict (31).

The findings of this single case of ISTDP align with Davanloo’s assertion that the therapist’s confrontation of the patient’s defenses may serve as a mechanism of change by mobilizing intense affective processes. However, this is not a unique feature of ISTDP; such interventions are common across many treatment modalities. Moreover, the single-case design of the study prohibits the generalizability of the findings. The study also did not report pre- to postintervention outcome data. Whereas the authors used observational data to conclude that the treatment was successful, it is difficult to assess how the examined interventions contributed to symptom change.

Stalikas et al. (32) explored clients’ therapy process within and across sessions to identify occurrences of good moments. The main research questions concerned the types of good moments that emerged in ISTDP and the sequential pattern of good moments across sessions. In this single-case study, the authors rated client responses during three sessions (early, middle, and late) of the 16-session treatment.

The Category System of Client Good Moments (30) was used as a measure of therapeutic process. Results suggested that good client change events were related to provision of significant information to the therapist, exploration of feelings, and insight and understanding. Most good moments involved patients sharing significant material about themselves or interpersonal relationships. This pattern suggests a concatenation of exploration, insight, and understanding, leading to behavioral change. This finding is consistent with the ISTDP model; however, the single-case design of the Stalikas et al. study inhibited the generalizability of the findings. The lack of outcome data for the intervention also limited the conclusions regarding efficacy.

Abbass et al. (33) conducted a naturalistic study of the effectiveness of a single session of ISTDP in a tertiary psychotherapy service. Patients were evaluated prior to the session and at 1-month follow-up to assess the need for further treatment. Five randomly selected trial therapy sessions were studied by the therapist and an independent, trained clinician-therapist to gain a deeper understanding of therapy processes. Results supported the effectiveness of a single ISTDP session; BSI scores showed that 43% of those meeting case criteria before intervention no longer did at follow-up.

The session process analysis indicated that the therapists were highly active during the sessions, with an average of 165.50 inventions per hour. Pressure was the most common intervention, occurring an average of 97 times per hour and comprising 59% of all interventions. Pressure refers to any therapist intervention that encourages the patient to feel, acknowledge, examine, or otherwise face avoided feelings, thoughts, relationships, or actions (31). The second most common intervention involved exploring the links between past and present, feelings, anxiety, and defenses. Therapists used such interventions an average of 29 times per hour (19%). Other interventions included clarification (helping patients to notice or understand the nature of their conflict) and challenge of defenses (focusing on patient defenses and encouraging reduced defenses), which together occurred at an average of 23 times per hour (14%); inquiry into problem areas (5%); and dynamic exploration at an average of 5 per hour (3%). These interventions are consistent with ISTDP treatment protocols (2, 4). The study outcomes suggest that these interventions were associated with positive therapeutic change, even after a single session.

Because of the naturalistic nature of the Abbass et al. (33) study, caution should be practiced in generalizing the findings. Given the multiplicity of identified interventions, it is difficult to assess the impact of a singular intervention. However, this is a research issue true of most therapy modalities.

McCullough et al. (34) explored the relationship between interactions in the therapy session and patient affective or defensive responses. Sixteen clients attended weekly 50-minute sessions for an average of 40 sessions. All clients met DSM-III criteria for axis II personality disorders. Three types of therapist intervention were explored: interpretations of the therapeutic relationship, interpretations of relationships with significant others, and clarifications (rephrasing the patient’s words). Affective responses were defined as either verbal or nonverbal (e.g., sighing, crying, lowering head) expression of emotion by the patient. Defensive responses were patient reactions that exemplified mechanisms used to resist facing difficult issues.

Symptom improvement was predicted by interpretations of the therapeutic relationship that were followed by an affective response from the patient. Interpretations of the therapeutic relationship followed by a defensive response were negatively correlated with patient outcome. Therapist interventions independent of patient response were not associated with outcomes, supporting the first hypothesis. Although the McCullough et al. results were correlational, the findings highlighted the importance of the therapeutic relationship and patient responses as potential predictors of therapeutic outcome.

Winston et al. (35) studied patient defensive behaviors and the intervention of “therapist addressing defenses” (TAD) in therapy for 28 outpatients with personality disorders. The authors hypothesized that the frequency of defensive behaviors would decrease over the course of therapy, patient defenses would shift from lower to higher levels of maturity, and a positive correlation between the change in level of defense and patient outcome would emerge.

Additionally, it was expected that the therapist would use more frequent TAD interventions when there was higher frequency of low-level defense reactions. It was expected that more frequent use of TAD to address low-level defenses would predict decreased subsequent use of those defenses.

Two brief psychodynamic therapies were used: short-term dynamic therapy and brief affective psychotherapy (BAP). BAP is a more cognitive form of psychotherapy that focuses more on patients’ maladaptive patterns. Eighteen therapists (12 psychiatrists, four psychologists, and two social workers) delivered the 40-session intervention. Four recorded sessions from each patient were coded for patient and therapist variables.

Winston et al. found a significant decrease in overall defensive behaviors over time, although only intermediate defensive behaviors (e.g., intellectualization, reaction formation) changed. There was no correlation between frequency of defensive behavior and patient outcome. Several changes in patient defensive behavior were found that were significantly related to TAD, highlighting the importance of the therapist-patient relationship.

The Callahan study (36) examined the ISTDP hypothesis that as treatment progresses, resistance and defenses decrease. This study attempted to operationalize defenses according to Davanloo’s (2) assertion that defenses are marked by specific nonverbal cues. These cues include breaks in eye contact, defiant body posture, slowed speech, and involuntary smiling. The authors concluded that breaks in eye contact during anxiety (BECAs) were the easiest variable to objectively assess. Therefore, the study assessed resistance by the frequency of BECAs during a session. Two raters coded instances of BECA, defined as eye contact that was broken for more than 1 second and accompanied by nonverbal anxiety cues. It was hypothesized that BECAs would decrease from early to late therapy and that their frequency would correlate with measures of therapeutic improvement, so that as patients improved, BECA frequency would decrease.

Six participants received weekly ISTDP. Total number of sessions ranged broadly, from 17 to 155 sessions (mean=60±53.1). Outcome was assessed via change in patient scores, as rated by the treating therapist, on the Global Assessment of Functioning (GAF) from the first to last sessions. Results showed that BECAs decreased from early to late therapy. There was a marginally significant correlation between improvements in GAF scores and decreases in BECAs, from early to late therapy. The author argued that the findings provided limited support for the idea that degree of therapeutic improvement is related to degree of reduction in resistance in the transference.

A marginally significant correlation, however, should be considered very limited evidence of a relationship between BECAs and therapeutic outcome. In addition, reliance on a single, therapist-rated measure of improvement is problematic in itself. This measure also provided no information regarding self-reported distress or psychological symptoms. Similarly, the reliance on a single, observer-rated, nonverbal cue as an indication of resistance makes it difficult to conclude that resistance was validly measured. Any conclusions that can be drawn from the Callahan results are therefore extremely limited.

The Town et al. (37) study examined process-outcome associations across 20 sessions of ISTDP. Participants were four clients who met diagnostic criteria for major depressive disorder. The study explored the relationship between affect experiencing (AE) and subsequent self-reported distress. AE relates to the degree of adaptive emotional arousal in session and is part of a broader measure—the Achievement of Therapeutic Objectives Scale. AE was assessed by independent raters who watched recordings of therapy sessions.

Town et al. (37) showed that for participants who had significant improvement in distress and depressive symptoms, higher peak AE in a session was associated with decreased distress in the next session. These findings offer some evidence that directly experiencing the somatic component of feelings contributed to significant therapeutic change (4). However, there was no association between in-session AE and reduced stress during sessions 1–10. For patients classified as fragile, ISTDP initially focuses on capacity building rather than affect experiencing. This result may therefore be understood through the moderating influence of the patient’s capacity to process feelings and the use of the graded format of ISTDP (2).

Town et al. (37) concluded that their results were consistent with ISTDP theory, suggesting that increase in a patient’s emotional experience was associated with positive symptom change. The study also found that increased affective experiencing was associated with an increase in therapeutic alliance. Furthermore, the authors concluded that their findings provided preliminary support for another ISTDP assumption: that a patient’s capacity to process feelings (associated with resistance) might be a moderating factor in the relationship between AE and symptom improvement. However, it should be noted that an association between increased capacity to experience emotions and decreased distress is not a concept unique to ISTDP.

In an unpublished dissertation, Salvadori (2010, unpublished) used a case series design to explore the relationship between AE, inhibition, and distress in the context of ISTDP. Salvadori predicted that an increase in AE and a decrease in inhibition would occur with ISTDP, correlated with an improvement in working alliance and patient distress. Affective capacity referred to the ratio of AE to the degree of inhibition. Participants were six outpatients with common mental health issues (e.g., depression, social phobia). The study examined video recordings of sessions 1–20 of treatment. Sessions were coded for peak affect and average inhibition, and patients gave weekly self-reported measures of distress. Results of the study were mixed. For two participants who were considered recovered at the conclusion of therapy, distress decreased over the course of therapy, and this improvement was positively correlated with increases in affective capacity. In these two cases, affective capacity was also found to correlate with working alliance. Two participants showed no relationship among affective capacity, time in therapy, and distress. Other participants showed mixed results. The study was limited in that the participants had a wide range of disorders and the results were correlational. However, the case series design of the study provided for an in-depth analysis of each participant. Although Salvadori’s results were mixed, they suggest that AE in psychotherapy may be associated with positive treatment outcomes.

In another unpublished dissertation, Paserpskytė (2012, unpublished) used the data from four of the six participants of the Salvadori study (2010, unpublished). Similarly, it used a case series design. The study aimed to explore insight as a mediating variable between AE and patient outcomes. It was hypothesized that participants would show greater insight following therapy segments in which they showed greater AE and that insight would act as a mediator between peak affect in session and session outcome. Paserpskytė specifically chose to study two participants who were considered recovered and two participants who did not change during therapy. Neither hypothesis was supported. Interestingly, insight and AE were found to be possible independent predictors of patient outcomes.