Process and Technique Factors Associated with Patient Ratings of Session Safety during Psychodynamic Psychotherapy

The participants in this study (N = 94) were all were patients admitted to a Psychodynamic Psychotherapy Treatment Team (PPTT) at a university-based community out-patient clinic (Hilsenroth, 2007). Cases were assigned to treatment practica and clinicians in an ecologically valid manner that was based on real world issues regarding aspects of clinician availability, such as caseload, etc. Moreover, patients were accepted into treatment regardless of disorder or comorbidity. In this sample of 94 individuals, 66 (71%) patients were female and 27 (29%) were male. The mean age for this sample was 30.00 (SD = 11.6). Table 1 displays demographic information and the distribution of patients’ primary Axes I and II diagnoses for the entire sample in accordance with the DSM-IV (American Psychiatric Association, 1994; based on the psychological assessment process described below). All patients in this study received a DSM-IV Axis I diagnosis. Fifty-two (51%) of the patients received a diagnosis of an Axis II disorder, while 24 (24%) were assessed to have subclinical, but prominent, Axis II traits or features [Cluster A = 9 (12%), Cluster B = 39 (52%), Cluster C = 27 (36%)]. Thus, this sample consisted of primarily patients with mood disorders with relational problems manifested either in Axis II personality disorders or in subclinical traits/features of Axis II personality disorders, which were generally in the mild-to-moderate range of psychopathology.

Clinicians in the study were 26 advanced doctoral students (13 male and 13 female) enrolled in an American Psychological Association approved Clinical Ph.D. program. Each clinician received a minimum of 3.5 hours of supervision per week (1.5 hours of individual and 2 hours of group) on the Therapeutic Model of Assessment ([TMA] Finn & Tonsanger 1997; Hilsenroth 2007), clinical interventions, the organization of collaborative feedback, psychodynamic theory and review of videotaped case material. Individual and group supervision focused heavily on the review of videotaped case material and technical interventions. All clinicians were trained in psychodynamic psychotherapy using guidelines delineated by Book (1998), Luborsky (1984), McCullough et al., (2003), Wachtel (1993) and selected readings on psychological assessment, psychodynamic theory, and psychodynamic psychotherapy (for a more detailed description of this training process, see Hilsenroth, DeFife, Blagys, & Ackerman, 2006).

Pairwise two-tailed Pearson correlations (N = 94) revealed patient-rated SEQ Session Safety to have significant positive relations to SEQ indices Depth (r = .28, p = .006), Smoothness (r = .36, p = .003), and Positivity (r = .37, p = .0002; see Table 2). Session Safety did not positively correlate with the SEQ index, arousal (r = .04, p = .67) or the Overall Session, Good/Bad item (r = .13, p = .22). The relationships between SEQ Session Safety and Therapeutic Alliance were explored with regard to patient ratings of alliance on the CASF-P. Pairwise two-tailed Pearson correlations revealed signification correlations between Session Safety and CASF-P indices confident collaboration (r = .22, p = .03), Bond (r = .30, p = .004), and Total (r = .20, p = .05). Session Safety was not significantly related to CASF-P indices Goal and Task Agreement (r = .07, p = .49) or Idealized Therapist (r = –.02, p = .87).

Next, pairwise two-tailed Pearson correlations revealed patient rated SEQ session Safety to have significant positive correlations with CPPS Interaction PI – CB (r = .22, p = .04; see Table 3). We also found patient-rated SEQ session Safety to show a trend toward significance in relation to the CPPS-CB scale (r = .20, p = .06) and to have a nonsignificant positive relation to the CPPS-PI scale (r = .12, p = .26). Finally, we observed no significant association between patient-rated SEQ session Safety and CPPS Percent (%) of CB Techniques (r = .12, p = .26).

Though there is a deficit in the empirical literature addressing patient perceptions of session safety, it is widely acknowledged as a key element in effective treatment. Theoretical work on the topic presents a sense of safety as a necessary ingredient in the process of patient emotional growth. According to Control Mastery Theory (CMT), achieving safety with the therapist is a pivotal step early in treatment, and the constant support provided by the therapist is essential for the patient to challenge maladaptive patterns (Silberschatz, 2008). The accepting novel relationship allows patients to take risks and provides corrective experiences (Rappaport, 1996). Safety makes the session a place more conducive to facing uncomfortable affect (Ogrodniczuk, Joyce, & Piper, 2005), as well as safety disarms defenses and enables “working through” the subsequent experiences (Rappaport, 1996). A strong emotional connection with the therapist, as perceived by the patient, facilitates such a setting (Friedlander, et al., 2005, p. 9; Escudero et al., 2011). Other building blocks of a safe space are said to include varied elements, ranging from structured interventions (Rappaport, 1997) to high receptivity and “being with the patient” (Havens, 1989). These contributions make a great deal of theoretical and conceptual sense and are based on valuable clinical observation, but to date had little research support. Friedlander, Escudero, and colleagues (2005; 2006; 2011; 2012) succinctly describe safety as “a sense of comfort and an expectation that new experiences and learning will take place,” by which a patient can be vulnerable to take risks (Friedlander et al., 2006, p. 216). Considering safety a factor of therapeutic alliance, they created the System for Observing Family Therapy Alliances ([SOFTA-o] Friedlander et al., 2005; Escudero et al., 2011) to understand its relevance in the therapeutic process. Through systematic research, their group laid an empirical foundation for the preceding theoretical work. Because their research focuses on couple and family therapy, a gap remains in the study of safety in individual treatment. The present study begins to address this need by including safety as a factor in the systematic study of psychotherapy process. Thus, we examined the associations between Session Safety and other process factors, therapeutic techniques, and dimensions of therapeutic alliance.

Consistent with our hypothesis, Session Safety was significantly related with patient ratings of in-session Depth and Smoothness. Regarding Depth—patient experiences of Session Safety during STDP are consistent with the exploration and Depth of session content. Our results suggest that safety may allow for exploration of affective content, resulting in a more deep, powerful, valuable, special, or full experience of the session (all components of the Depth scale). Aligning with the hypotheses of Ogrodniczuk et al. (2005), safety may enable thorough investigation of uncomfortable affect and in turn be more deep, valuable, and even productive. Our finding of a significant positive relationship with Smoothness supports our hypothesis, and infers that patients who perceived the session as safer also evaluated it as more easy, relaxed, pleasant, smooth, and comfortable (all components of the Smoothness scale). This suggests that even though session content or focus may also be consistent with Depth, a session could be perceived as comfortable and smooth, while concurrently being powerful and deep. The deep exploration of material may not decrease a patient’s comfort, but have the opposite effect when safety is present. The direction of this relationship is intuitive, and further emphasizes the connection between a safe environment and the experience of a session. Although not included in our hypotheses, Safety also significantly correlated with Positivity. Patients who felt more happy, pleased, definite, confident, and friendly also experienced a greater sense of safety with the therapist (all components of the Positivity scale). Feeling safe in the therapeutic environment may be associated with a positive attitude towards treatment and the therapeutic relationship. In addition, other aforementioned aspects potentially associated with safety such as deep exploration of content and comfort, may lead patients to assess sessions as positive. The lack of association between Bad/Good and Safety may imply that a patient’s overall value judgment of the session does not influence their perception of safety. That is, a patient may still experience safety despite viewing a session as “less good” or “bad.” In addition, our findings infer that feeling safe in session is not related to a patient’s sensations of excitation or calmness, as indicated by the absence of a significant relation with the Arousal index.

When we examined the relationship of SEQ Safety with the therapeutic alliance on the CASF-P, there was a significant relation both with alliance overall as well as different facets of alliance. Safety was significantly positively associated with Confident Collaboration and Bond. The relationship of Session Safety with Confident Collaboration may be explained as a connection between the dyadic dynamic and the environmental tone that is set as a result. High Confident Collaboration implies that the patient feels a strong sense of self-efficacy in the success of his or her own treatment, as well as a confidence in the therapist to help collaboratively achieve goals. This may relate to the collaborative nature of feedback sessions in TMA, which are highly individualized and address current problems and goal-setting. A patient may feel safe in therapy if he or she believes that together with the therapist, they are engaging in meaningful work for which there is an attainable, helpful goal. Bond signifies a patient’s perceived personal acknowledgement and acceptance by the therapist, and it may be deeply intertwined with feeling safe in session. Consistent with our hypothesis and with the prior work by Friedlander, Escudero, and colleagues (2005; 2006; 2011), our analyses rendered a strong significant correlation between Bond and SEQ Safety. The Bond factor may reflect the interpersonal environment that the therapist fosters, and the patient’s reaction to this by how he or she processes therapeutic interactions. The cumulative interchange of relational information within the dyad may influence the strength of the bond, and the patient’s perceived positivity of the therapist’s views of him or her. If the bond flourishes and the patient believes that the therapist holds positive and respectful views of him or her, then it is possible that the feeling of safety will also increase. We found no significant relationships between SEQ Safety and Goals and Task Agreement or Idealized Therapist dimensions. Although a more successful outcome may be achieved when Goals and Task Agreement are high, and when patient and therapist agree concerning treatment course, our results imply that this type of synchronicity may not relate to patient comfort. Hatcher and Barends (1996) found that treatment was less effective when patients were hesitant to disagree with their therapists. Our finding of a lack of a significant association with Idealized Therapist may be informed by the result regarding SEQ Depth. Patients who felt safe may be comfortable in disagreeing with their therapists, and the introduction of intense affective material may have enabled deeper therapeutic work.

Contrary to our hypothesis stating that Safety would significantly relate to Psychodynamic-Interpersonal (PI) interventions on the CPPS, we found no significant relation. The correlation with Cognitive-Behavioral interventions (CB) approached significance. Although when we examined the overall percentage of CB techniques used in the session we did not find a strong relationship with session safety. This suggests that the introduction of a limited number of CB techniques early in treatment may increase the sense of safety. It should be noted that within this primarily psychodynamic treatment, neither modality alone was significantly related to safety in session. Our competing hypothesis, however, was supported by the data. The interaction between PI and CB techniques was significantly correlated with Safety. The data from individual sessions in the sample point to an interaction of a moderate use of PI interventions with a limited implementation of CB interventions [M CPPS-PI = 3.29 (SD = .73); M CPPS-CB = 1.27 (SD = .57]. Taken together, it may be inferred from our analyses that the use of a moderate level of PI with integration of limited CB techniques within a predominantly psychodynamic model may facilitate a safer therapeutic space. Our results provide support for integration of therapeutic techniques, and specifically for the concept of therapist responsiveness. Stiles and colleagues (1988, 1998) describe an effective dynamic of responsiveness, in which moment-to-moment therapeutic interventions and adjustments in treatment plans are shaped by ever-changing patients needs. Norcross and Wampold (2010, 2011) stress the importance of adapting treatment to each distinctive person, not by diagnosis alone. They delineate six domains of patient characteristics: reactance level, stage of change, preferences, culture, coping style, and religion/spirituality. Each domain must be considered in treatment formulation for maximal effectiveness and sensitivity to patient needs.

It is possible that the collaborative and individualized treatment plans implemented in this sample by the Therapeutic Model of Assessment (TMA) may be more responsive to patient needs, increasing safety or comfort. TMA is an interpersonally-oriented approach that emphasizes the following principles: “(a) developing and maintaining empathic connections with clients, (b) working collaboratively with clients to define individualized assessment goals, and (c) sharing and exploring assessment results with clients’’ (Finn & Tonsager, 1997, p. 378). Therapist receptivity and open interchanges concerning treatment and progress may help to cultivate safety. Overall, it may be optimal for the therapist to be responsive by utilizing various styles commensurate with the necessities of each patient to insure the critical element of the patient’s sense of safety.

A notable limitation of this area is the absence of prior research literature. As a result, this study was somewhat exploratory, but established a key point of reference for future empirical investigation. There are few existing validated measures of safety in a therapeutic context, which also limited the potency of this study. The only measure of safety presently in use is within the SEQ, and consists of one item. Results may carry more weight and better capture complexity if safety is assessed using a measure with more items specifically pertaining to this construct. Currently, with the exception for the SOFTA-o, such a measure does not exist. The SOFTA-o is designed primarily for couples and family treatment, and thus has limited utility in the assessment of safety in individual therapy. Future work in safety may progress in three ways:

However, one might argue that the most clinically applicable and parsimonious measurement of session safety would be to ask the patient to rate on a Likert scale how safe the session was (which was the basis of the safety rating used in this study).

Future research in this area would benefit from the use of assessments of patient characteristics in the analyses. It is possible that patients with specific diagnoses (e.g. major depressive disorder) may experience more feelings of safety in session than others (e.g. borderline personality disorder). Safety ratings of these specific diagnostic groups may inform relationships between safety and other constructs such as alliance. Interpersonal or attachment style and other personality factors may also be pertinent to safety ratings. Likewise, studies that assess issues pertaining to therapist characteristics and perspectives would be useful. Future research could include therapist ratings of SEQ-measured session process including safety, and therapist views of alliance. Shedding light on this fundamental element of the therapeutic experience may provide data about whether some therapists are more successful at creating a safe space than others and what the key ingredients in doing so are.

Another key suggestion for future studies is the inclusion of outcome analyses to assess the relationship between session safety and psychotherapy outcome. This is essential for future research, as the impetus for this work is to identify which factors are effective. Put more simply: If a patient feels safe in the therapeutic environment, is he more likely to make significant progress than if he did not feel safe? The results could be defined more specifically by reduction of psychiatric symptoms and interpersonal problems and an increase in adaptive behaviors. Findings of this study may stimulate discussion about how to help patients feel safer in therapy. We are uncertain about the underlying mechanisms of patient experiences of safety at this time. More empirical study of this topic may answer questions pertaining to increased patient feelings of safety and to facilitating more positive therapeutic outcomes.