| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ALL ISSUES | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Regular Article |
Received February 4, 1998; revised April 22, 1998; accepted April 23, 1998. From the University of Pennsylvania Medical Center, Center for Psychotherapy Research, Department of Psychiatry, 3600 Market Street, 7th Floor, Philadelphia, Pennsylvania 19104-2648. Send correspondence to Dr. Connolly at the above address.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Supportive-Expressive therapy (SE)5 is a short-term psychodynamic treatment that has been manualized for a variety of disorders, including depression,6 generalized anxiety disorder,7 opiate drug dependence,8 and cocaine abuse.9 The goal of SE psychotherapy is to help patients gain understanding of conflictual relationship patterns in the context of a supportive relationship. The main techniques of SE psychotherapy include supportive techniques to bolster the therapeutic alliance and interpretations to help patients gain self-understanding. An adherence/competence measure rated on whole sessions has been developed to assess the frequency and quality of therapist techniques as specified in the SE manuals.1,10 To date, four outcome trials have been reported evaluating the efficacy of SE psychotherapy for specific disorders,11–14 and an additional efficacy study of SE for cocaine abuse is being conducted.15 However, a molecular-level description of individual therapist interventions in an average SE session has not yet been reported.
Such molecular-level assessments can provide important descriptive information for manualized treatments. Although manuals provide overviews of treatment techniques, a systematic description of what the treatment actually looks like would have great value. SE psychotherapy falls under the broad rubric of dynamic psychotherapy, yet the various models of psychodynamic psychotherapy are quite different.16,17 Although most psychodynamic models include interpretation as a primary therapeutic technique, psychodynamic treatments vary regarding the frequency and content of interpretations prescribed by the manuals. For example, Luborsky's SE model5 has little emphasis on the interpretation of the transference with the therapist and greater emphasis on the patient's maladaptive interpersonal patterns as experienced in current relationships outside of therapy. In contrast, the short-term, dynamically oriented treatment implemented by Piper et al.,18 based on recommendations by Malan,19 contained an average of five transference interpretations per session. A molecular-level assessment of SE interventions would be useful in more fully describing SE psychotherapy in the context of the broad psychodynamic camp.
A second rationale for developing a molecular method for assessing therapist interventions within SE is that it may be useful to explore the relationship between molar methods (global session measures) and molecular methods for describing therapist techniques. Heaton et al.20 evaluated 23 sessions of psychotherapy provided to student volunteers describing problems with assertiveness, using the molecular-level Hill Counselor Verbal Response Category System (HCVRCS)21 to describe each therapist utterance and two molar measures of therapist technique that provided global session ratings. The results revealed that although the two molar assessments were highly correlated, these measures were not associated with the HCVRCS. The authors conclude that different methods for assessing technique provide very different results and that further research is needed to understand the benefits of each assessment method. Although measures of adherence and competence have been developed and evaluated on SE sessions with various diagnostic groups,10,1,22 these molar measures have not yet been compared with a measure of individual therapist response modes.
It is possible that molecular descriptions of therapist interventions would be useful in conjunction with the more molar adherence/competence measures to provide a comprehensive description of a treatment. Barber et al.10 found that therapist competence on expressive techniques predicted subsequent symptom improvement for depressed patients treated with SE psychotherapy. A molecular-level evaluation of therapist response modes in these sessions might be useful for unpacking the relationship between therapist competence and treatment outcome. An examination of the relationship between specific therapist response modes and treatment competence could help indicate ways to enhance competence and thus improve treatment outcome.
Various molecular methods have been developed for classifying individual therapist interventions. In a review of the literature, Elliott et al.23 report as many as 20 to 30 systems for rating therapist response modes. Elliott et al.23 used six common systems to evaluate therapy sessions, for the purpose of assessing the reliability and convergence of the systems. The results indicated that reliability values were comparable across methods and that a set of fundamental response-mode categories existed, including question, advisement, information, reflection, interpretation, and self-disclosure.
Multiple studies have implemented rating systems of therapist response modes to describe psychotherapy and to compare various treatments. Although studies have rated response modes in nonmanualized psychotherapies,21,24,25 few have evaluated therapist verbal response modes in more clearly defined treatments. Hill et al.26 compared therapist utterances—defined as independent therapist clauses—in three sessions conducted by expert therapists with the same patient (the Gloria case). Therapist response mode frequencies were consistent with the theoretical models of the therapists, with Carl Rogers using predominantly encouragements, restatements, and reflections; Frederick Perls using confrontations and interpretations; and Albert Ellis focusing on information giving, direct guidance, and interpretation. Interestingly, all sessions revealed a considerable level of interpretation, defined as a statement that goes beyond what the patient currently recognizes, including 7% of Rogers's utterances and 12% of the utterances made by Perls and Ellis.
Stiles et al.27 evaluated therapist response mode categories in prescriptive and exploratory psychotherapy sessions. Manuals were used to describe the primary techniques of each treatment. The exploratory therapy, described as interpersonal/psychodynamic, revealed more interpretations and reflections than the prescriptive therapy, which represented a cognitive-behavioral orientation. In fact, more than 20% of therapist utterances across the exploratory sessions were rated as interpretations in terms of therapist intent. Within this system, an interpretation was defined as a therapist statement that describes the patient's experience from the therapist's frame of reference, including statements that make connections, explain, or draw conclusions. The investigators later reported significant therapist effects in the use of different response mode categories,28 suggesting stable differences among therapists in verbal styles despite the use of the brief treatment manuals.
Piper et al.29 used the Therapist Intervention Rating Scale (TIRS) to evaluate therapist statements in short-term individual dynamic psychotherapy. Although formal manuals were not implemented, treatment guidelines derived from Malan19 were used to standardize the treatment. The results revealed that 14% of therapist statements were classified as interpretations, with an average frequency of 10 interpretations per session. In this investigation, an interpretation was defined as a statement that referred to an internal conflict of the patient. Interpretations were statements that included a dynamic component, such as a reference to the patient's impulses, anxiety, or defenses.
The purpose of the present investigation was twofold. First, we attempted to characterize therapists' interventions in brief SE psychotherapy for depression. We describe the kinds and frequencies of therapist interventions that typically occur in early SE sessions for depression and provide an analysis of the pattern of techniques from session to session. The second purpose was to further evaluate the relationship between molar and molecular methods for assessing therapist techniques in brief SE treatment. We compared the proportion of therapist interpretations as derived from ratings of individual therapist interventions to a molar measure of adherence to SE therapy. Finally, the relationship between therapist response modes and treatment competence was examined in order to uncover further directions for developing SE therapy.
|
METHODS |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In general, patient outcomes were favorable across the sample. For example, the 17-item Hamilton Rating Scale for Depression32 scores decreased significantly, from 17.9 ± 3.7 to 7.9 ± 6.4 (mean ± SD), over the course of the 16-week treatment (t = 9.40, df = 32, P = 0.000). Complete descriptions of the study procedures are provided by Diguer et al.30 and Luborsky et al.33
SE therapy for depression6 is a short-term, focused, dynamic psychotherapy. The first goal of SE therapy is to establish a trusting, supportive therapeutic relationship that allows the patient to explore thoughts and experiences. Within this framework, the therapist uses the narratives shared during therapy, as well as the transference as experienced in the therapeutic relationship, to interpret the patient's core conflictual relationship themes (CCRT).34 The CCRT consists of the main maladaptive wishes, responses of self, and responses of others experienced by the patient. Because of the time limits of the treatment, specific treatment goals are established at the beginning of treatment in order to work within the time limit and also to strengthen the therapeutic alliance, the sense that the patient and therapist are working together toward a common goal. SE psychotherapy for depression is based on the premise that symptoms will subside as patients begin to understand their maladaptive relationship patterns and begin to implement more adaptive interpersonal responses.
Four Ph.D.-level therapists participated in this investigation. All therapists were psychodynamically oriented and had at least 5 years of postdoctoral clinical experience. Although the therapists were not precertified as competent in supportive-expressive psychotherapy for depression before this trial, 3 of the 4 had previous experience conducting SE treatment in a research protocol for another disorder. The fourth therapist was trained for 2 years in SE psychotherapy by Lester Luborsky, Ph.D. Thus, all therapists were highly familiar with SE therapy in general before beginning the study.
Coding of Therapist Statements
Therapist response modes were rated by our own system, which is designed to classify therapist statements into general categories consistent with the techniques of SE therapy. Three independent judges rated therapist statements from three early psychotherapy session transcripts for each of the 33 patients. Sessions 2, 3, and 4 were transcribed when possible, with later sessions used as needed. All transcribed sessions occurred between sessions 2 and 9 of the 16-week treatment, and 93% of sessions used occurred between sessions 2 and 5. One patient had only two sessions available for scoring, so a total of 98 sessions were scored across the sample. Therapist statements were identified in the session transcripts as therapist "speaking turns." Speaking turns were combined as a single therapist statement only if one statement was interrupted midsentence by a brief patient remark and continued in the next therapist statement.
Each statement was then independently coded by each of three trained advanced graduate student judges into one of four categories: interpretation, clarification, question, or other. Judges rated each statement in the context of the therapy session transcript. Judges were instructed to choose the category that best represented the statement.
Interpretations were defined as statements that linked the patient's current and past life experiences or statements that explained reasons for a patient's thoughts, feelings, or behaviors. Statements were scored as interpretations only if they went beyond the patient's awareness. An example of a therapist interpretation from the transcripts is: "It looks like your relationship with your father is indeed very complex. . . . It's made you extremely angry and extremely afraid of people."
Clarifications were defined as statements that commented on some aspect of a patient's thoughts, feelings, or behavior without offering reasons for their occurrence or without relating them to circumstances other than the present ones. An example of a clarification from the current study is: "What I'm hearing you say is you weren't feeling good."
Questions were defined as statements made for the purpose of gathering information; for example: "Do you think you still might be affected by it?"
Other statements were defined as any statements that did not fit into one of the above categories. They most often consisted of therapist statements that acknowledged that the therapist was listening to the patient, such as "Uh-huh."
The frequencies and percentages of interpretations, clarifications, questions, and other statements were computed across sessions for each judge. The final frequencies and percentages for each response mode were computed by averaging across the three judges for each session.
Because interpretations are considered a primary therapeutic technique in SE treatment, additional codes were applied to statements scored as interpretations in order to better describe the characteristics of interpretations in SE therapy. Any statement rated by at least two of the three judges as an interpretation was included. Three additional trained judges (graduate students or experienced research assistants) independently scored the time frames and persons included in each therapist interpretation. Time frame categories included childhood past, adult past, present, and future. Person categories included therapist, parents, significant others, self, siblings, or none of the above. Judges were instructed to rate all time frames and persons that were present in each interpretation. The percentages of interpretations containing each time frame and person code were computed for each patient for each of the 3 judges. The final percentages of person codes and time frames for each patient were computed by averaging across the three judges.
Ratings of Adherence and Competence
In a separate investigation using 29 of the 33 patients from the current sample, Barber and Crits-Christoph1 evaluated the adherence and competence of therapists in the same sessions by using the Penn Adherence/Competence Scale for SE Dynamic Psychotherapy (PACS-SE). The PACS-SE consists of 45 items designed to assess the basic techniques of SE psychotherapy. Each item is rated on two 7-point Likert scales designed to assess the frequency of the technique and the competence with which the technique is implemented. The items form three theoretically derived subscales, including General Therapeutic Skills, Supportive Skills, and Expressive Skills. An example of an item on the expressive skills subscale is: "The therapist focuses attention on similarities among the patient's past and present relationships." An example of an item on the supportive skills subscale is: "Therapist and patient work as a team to help the patient with better self-understanding." A total adherence and total competence score can be computed across items.
In the investigation by Barber and Crits-Christoph,1 two doctoral-level clinical psychologists (not involved in rating therapist statements) with considerable experience rated adherence/competence from session audiotapes. The pooled judge intraclass reliability coefficients ranged from 0.35 to 0.79 across the subscales and total scores. The interjudge agreement was in the acceptable range for all subscales except the Adherence to Supportive Skills subscale. The subscales also demonstrated acceptable internal consistency, with alpha coefficients ranging from 0.62 to 0.95.
|
RESULTS |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Pooled judge intraclass correlations were computed to evaluate the interjudge reliability of scoring interpretation time frames and persons. The percentages of interpretations containing each time frame category and person category were computed for each patient for each of the three judges. Because of their low frequencies of occurrence, the sibling category and future time frame were dropped from all further analyses. The ICC(2,3) coefficients for percentages of time frame categories were 0.79 for childhood past, 0.50 for adult past, and 0.62 for present. The reliability coefficients for person categories were 0.56 for therapist, 0.94 for parents, 0.88 for significant other, 0.66 for self, and 0.87 for no other person.
The frequencies and percentages per session for each therapist statement category are presented in Table 1. Ninety-eight sessions were scored across the 33 patients (3 sessions were scored per patient except for 1 patient with only 2 session transcripts available). There were between 24 and 319 statements in each of these sessions, with a mean of 125 therapist statements per session. The average session contained 5 interpretations and 43 clarifications. Questions accounted for 27% of therapist statements in the average session, and 33% of statements were coded as "other," meaning the statements did not fit coherently into the interpretation, clarification, or question categories.
|
|
To evaluate the convergence between the molar and molecular measures of therapist techniques, we compared the frequency of and percentage of interpretations per session with scores on the Adherence to Expressive Skills subscale of the PACS-SE. An examination of the variance components for the frequencies and percentages of response modes revealed some degree of non-independence, with between 0% and 13% of the variance attributable to differences between therapists. For this reason, we present correlations descriptively, attending to Cohen's37 ranges for small, medium, and large associations. The Adherence to Expressive Skills subscale is designed to assess the frequency with which the therapist addresses the patient's interpersonal and intra-psychic conflicts across the session. As expected, both the frequency of interpretations and the percentage of interpretations per session revealed large associations with the Adherence to Expressive Skills score (r = 0.58 and r = 0.50, respectively).
On average, therapists scored 4.15 (SD = 0.90) on the 1-to-7 scale assessing competence to expressive skills, indicating that therapists in this sample were of adequate competence in delivering supportive-expressive therapy. (Of the three theoretically derived subscales of the PACS-SE, the current investigation focuses only on the Competence on Expressive Skills subscale, since this scale demonstrated a significant relationship to treatment outcome.10) The comparison of therapist competence and response modes revealed that therapist competence on expressive skills had a moderate to large association with the frequency of interpretations (r = 0.41); small associations with the raw frequencies of clarifications and questions and the total number of therapist statements (r = –0.09, 0.10, and 0.14, respectively); and a moderate association with the frequencies of other statements (r = 0.22). Competence on expressive skills had a large correlation with the percentage of clarifications per session (r = –0.56) and small to moderate relationships with the percentages of interpretations, questions, and other statements (r = 0.27, –0.01, and 0.32, respectively).
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Because SE is a psychodynamic treatment, the manuals for SE psychotherapy advocate a focus on interpretation as a primary therapeutic technique. However, in the current study, only an average of 5 therapist statements per session were coded as interpretations, compared with 10 therapist statements in the psychoanalytically oriented treatment sessions investigated by Piper et al.29 This difference may be partly due to the focus on early psychotherapy sessions in the current study; the Piper et al.29 study evaluated therapist interventions across treatment. However, in the three early sessions examined in the current investigation, there was not a significant increase in the use of interpretations across sessions, indicating that therapists in training with SE treatment did not systematically increase their use of interpretation across the early phase of treatment. It is possible, however, that therapists increase their use of interpretation in later phases of treatment. Other studies reported seemingly higher proportions of interpretations,24,26,17 yet these other studies are not directly comparable given their focus on therapist utterances, defined as independent speaking clauses, as compared with the system implemented in the current investigation, which focuses on therapist speaking turns.
The differences in the proportion of interpretations per session may also be attributable to the varying definitions of interpretation implemented in these investigations. In the investigations by Piper and colleagues, an interpretation is defined as a statement that refers to an internal conflict of the patient, whereas in the current investigation, an interpretation is a statement that explains possible reasons for a patient's thoughts, feelings, or behaviors or that links current experiences to past experiences. If anything, our own definition appears more broad than the one implemented in the TIRS and is therefore unlikely to be the reason for the low rate of interpretation in the present investigation.
It is also possible that brief SE psychotherapy sessions may contain few interpretations per session because of the focus on the patient's core maladaptive relationship patterns. Although traditional analytic therapists might interpret other contents of the patient's experience, such as defenses, the SE therapist is trained to focus interpretations on the patient's maladaptive relationship patterns. During early sessions, the task of the therapist is to elicit material on the patient's interactions with others. Interpretations are then used by the therapist once enough clinical material has been developed to formulate the relationship pattern. In contrast to models that encourage extensive interpretation, in SE the majority of interventions in early sessions are clarifications and questions designed to elicit clinical material regarding the patient's maladaptive relationship themes.
The analyses of interpretation time frame and person codes in the current investigation also shed light on important differences between interpretive techniques as implemented in SE psychotherapy and in other psychoanalytic models. The results revealed that the majority of interpretations focused on patients' parents, significant others, and the self. An example of a typical interpretation focusing on a patient's relationship with a significant other in the current time frame is: "With your wife, and in other relationships, you come into it with the feeling of peacemaker, and if things are relatively good, you don't make it worse by talking about how you feel." The results revealed that only 5% of interpretations early in treatment focused on the patient's childhood past, suggesting minimal exploration of the patient's early life experiences in the SE model. Although exploration of childhood relationship experiences is considered one aspect of the SE model, the predominant focus remains on the exploration of relationship themes in the patient's current world.
Only about 8.6% of interpretations included the therapist in the statement. Only an average of 5 statements per session were rated as interpretations, so an 8.6% rate means that less than 1 interpretation per session focused on the therapeutic relationship. Although no studies are directly comparable to the current investigation because of methodological differences, Piper et al.18 report an average of 5 interpretations per session that focus on the therapeutic relationship. In the SE model, interpretations designed to foster understanding of the therapeutic relationship are included as an important technique, yet they are used relatively infrequently. This finding highlights a major difference between short-term SE treatment and more traditional psychoanalytic treatment. Although psychoanalysis involves a systematic interpretation of the transference of maladaptive themes in the therapeutic relationship, SE treatment focuses on the exploration of transference in patients' relationships outside of therapy.
The comparison of molar and molecular methods for assessing therapist techniques revealed convergence, particularly for the assessment of therapist interpretations. Unlike the results presented by Heaton et al.,20 the current investigation suggests that molar and molecular methods for assessing specific therapist response modes can lead to similar results. Heaton et al.20 suggest that it is possible that their methods did not converge because judges at the molar level were biased by the quantity of information that must be considered at the global level. Perhaps the current investigation found greater convergence because judges were encouraged to take notes while listening to the session audiotapes. These notes may have decreased the typical perceptual biases described by Heaton et al.20 The results indicate that, in general, molar and molecular methods can result in similar indices of therapist techniques; however, it may be important to include procedures in molar methods that decrease the probability of biased ratings.
Although the current investigation provides evidence of convergence between molar and molecular methods for assessing therapist techniques, the results also indicate that the two assessment methods implemented are not redundant. For the assessment of therapist interpretations, the primary technique in dynamic psychotherapy, the correlation between methods was large, indicating that the methods shared approximately 25% to 34% of the variance. These results indicate that each method is capturing unique variance in the use of response modes. The two measures appear to complement each other in providing a comprehensive description of the techniques within SE psychotherapy.
These results also suggest directions for further developing brief SE psychotherapy for depression. The previous investigation by Barber et al.10 indicated that therapist competence significantly predicts treatment outcome, and the results of the current investigation shed light on how changes in specific therapist response modes might enhance therapist competence and thus improve treatment response. Higher proportions of interpretation and lower rates of clarification were significantly associated with greater overall competence. In light of the results presented by Barber et al.,10 training therapists to use a relatively greater number of interpretations and fewer clarifications in early sessions should lead to higher competence ratings, and consequently should predict better treatment outcome. Competent therapists should move from a focus on exploration and clarification of patients' maladaptive themes to a focus on interpretation, even during early sessions of treatment. Thus, training should focus on helping therapists to formulate patients' CCRTs earlier in treatment, especially in time-limited treatments.
The current investigation has a number of limitations that should be considered in interpreting the results. First, these results pertain only to SE psychotherapy as practiced under strict time limits with depressed patients. It is likely that therapists' activity levels as well as the frequency of therapist response modes may vary in SE sessions conducted in different treatment lengths and with different patient populations. The sessions evaluated in the current investigation were early sessions, occurring mostly prior to session 6. Although the manuals for SE advocate the ongoing use of the prescribed techniques throughout the treatment, it is possible that later sessions may reveal different relative proportions of therapist response modes and different interpretive time frames and person codes.
In addition, the interjudge reliability for identifying the therapist as a person code in an interpretation was modest. This was most likely a result of our instructions to the judges to rate the therapist as a person code in a statement only if the statement refers to an interaction or process between the patient and therapist. There are many therapist statements that use "we" to indicate the collaborative nature between the patient and therapist, but many of these statements do not directly address the relationship between the patient and therapist. Thus, the judgment as to whether the therapist is a person in a therapist interpretation may be more difficult that the judgment of other person codes.
In summary, the current investigation attempted to provide a molecular description of the kinds and frequencies of therapist interventions implemented in the average SE psychotherapy session by trained therapists. The average SE session included relatively few interpretations, and the interpretations made were focused primarily on patients' relationship patterns in their current life experiences. The molecular assessment of therapist response modes provided similar results to the more global adherence measure in assessing the use of therapist interpretations. Although the results provide convergent validity for the two methods of assessment, the two methods may account for unique portions of the construct, suggesting that molar and molecular measures might be used in conjunction to provide comprehensive descriptions of the treatment. Finally, the results suggested that brief SE therapy might be further developed by focusing training on helping therapists to formulate and interpret patients' CCRTs in early therapy sessions.
|
Acknowledgments |
|---|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Luborsky, J. Stuart, S. Friedman, L. Diguer, D. A. Seligman, W. Bucci, S. Pulver, E. D. Krause, J. Ermold, W. T. Davison, et al. The Penn Psychoanalytic Treatment Collection: a Set of Complete and Recorded Psychoanalyses as a Research Resource J Am Psychoanal Assoc, March 1, 2001; 49(1): 217 - 234. [Abstract] [PDF]
|
|
|
|
 |
|
 K.-P. Bøgwald, P. Høglend, and O. Sørbye Measurement of Transference Interpretations J Psychother Pract Res., October 1, 1999; 8(4): 264 - 273. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ALL ISSUES | SEARCH | TABLE OF CONTENTS |
