A therapist’s emotional response to a patient can inform both diagnostic and therapeutic interventions (in this context, we use the term “emotional response” interchangeably with “emotional reaction” and “countertransference”) (
1–
5). Concepts such as complementary and concordant countertransference, role responsiveness, projective identification, cognitive interpersonal cycle, and interpersonal complementarity suggest that a patient may engage with a clinician in a manner that leads the therapist to experience emotions and thoughts that may in turn provide greater awareness of the patients’ feelings and perspectives (
6–
10). Personality disorders are by definition dysfunctional schemas of the self, others, and relational interactions. These patterns of relating often appear in the therapeutic relationship, drawing the clinician into interactions that reflect the patient’s enduring and maladaptive relationships (
3,
5,
8,
11–
18). As a consequence, therapists’ recognition of their emotional responses and experience is an important vehicle for assessing and understanding patients’ relationship patterns.
Although clinical descriptions of therapists’ emotional responses to patients are often rich, systematic empirical research to date has been limited. Only a few studies have examined the associations between specific personality disorders and therapists’ emotional responses (
19,
20). Betan et al. (
19) asked 181 clinicians of various theoretical orientations to evaluate their emotional responses to a nonpsychotic patient with the Therapist Response Questionnaire (
21). The therapists also rated their patient’s personality on the presence or absence of each DSM-IV axis II criterion. Factor analysis yielded eight countertransference dimensions (for a detailed description, see the
data supplement that accompanies the online edition of this article): 1) overwhelmed/disorganized indicates a desire to avoid or flee the patient and strong negative feelings, including dread, repulsion, and resentment; 2) helpless/inadequate describes feelings of inadequacy, incompetence, hopelessness, and anxiety; 3) positive indicates the experience of a positive working alliance and close connection with the patient; 4) special/overinvolved describes a sense of the patient as special relative to other patients and includes “soft signs” of problems in maintaining boundaries, including self-disclosure, ending sessions on time, and feeling guilty, responsible, or overly concerned about the patient; 5) sexualized describes sexual feelings toward the patient or experiences of sexual tension; 6) disengaged is marked by feeling distracted, withdrawn, annoyed, or bored in sessions; 7) parental/protective is marked by a wish to protect and nurture the patient in a parental way, above and beyond normal positive feelings toward the patient 8); criticized/mistreated describes feelings of being unappreciated, dismissed, or devalued by the patient. These patterns were associated with the three DSM-IV axis II clusters: cluster A correlated with the criticized/mistreated pattern; cluster B was associated with overwhelmed feelings, helplessness, hostility, disengagement, and sexual attraction; and cluster C correlated with therapists’ protective and warm feelings. In general, cluster B was associated with a broader range of therapist emotional responses than the other two clusters. Clinicians working with patients with narcissistic personality disorder reported feelings of inadequacy, devaluation, and ambivalence.
These results were consistent with the findings of other studies that have found that patients with cluster A and B disorders evoke more negative therapist reactions than cluster C patients, and that cluster B patients evoke more mixed feelings in therapists (
20). Some research has also demonstrated that cluster B patients, especially those with borderline pathology, elicited higher levels of anger and irritation and lower levels of liking, empathy, and nurturance (
22,
23) and tend to be perceived as more dominant, hostile, and punitive than patients with depressive disorders (
23).
These studies have generally focused on comparing therapist reactions in relation to DSM diagnosis at cluster level (
19,
20) or on single disorders, mostly borderline personality disorder (
22,
23). They have not examined the differential responses of clinicians to the broadest possible scope of personality disorders. In addition, some of this work has been constrained by the use of artificial stimuli, such as responses to case vignettes or recordings rather than ongoing interaction with actual patients (
22,
23). There have been other limitations, such as the use of the same therapist to evaluate several patients and thus allowing for potential interdependencies among the ratings (i.e., therapist effects) (
20).
Method
Sampling
From the rosters of the two largest Italian associations of psychodynamic and cognitive-behavioral psychotherapy and from centers specializing exclusively in the treatment of personality disorders, we recruited by e-mail a random sample of clinicians with at least 3 years’ postpsychotherapy licensure experience who performed at least 10 hours per week of direct patient care. We requested that they select a patient who was at least 18 years old; who had no psychotic disorder or syndrome with psychotic symptoms or any pathology that could complicate differentiation between psychological states and personality traits (for example, severe depressive or bipolar disorders); who was not on drug therapy for psychotic symptoms; and whom the therapist had seen for a minimum of eight sessions and a maximum of 6 months (one session per week). To minimize selection biases, we directed clinicians to consult their calendar to select the last patient they saw during the previous week who met the study criteria. To minimize rater-dependent biases, each clinician was allowed to describe only one patient. Clinicians did not receive any remuneration, and we had a response rate of approximately 81% (203 therapists). All participants provided written informed consent.
Therapists.
The therapist sample consisted of 203 Caucasians, 111 of whom were women; 65% were psychologists and 35% were psychiatrists. Their mean age was 43 years (SD=9, range=34–52). Two main clinical-theoretical approaches were represented: psychodynamic (N=103) and cognitive-behavioral (N=100). A portion of cognitive-behavioral clinicians (N=30) had a metacognitive interpersonal orientation (
13,
14). The average length of clinical experience as a psychotherapist was 10 years (SD=3, range=3–17), and the average time spent per week practicing psychotherapy was 16 hours (SD=3.9, range=13–25). Seventy percent of the patients described were from private practice and the remaining 30% from public mental health institutions.
Patients.
The patient sample consisted of 203 Caucasians, 118 of whom were women; their mean age was 34 years (SD=4.5, range=29.5–38.5). Fifty-nine patients had only a DSM-IV axis I diagnosis, 71 had only an axis II diagnosis, 46 had comorbid axis I and axis II diagnoses, and 27 had a double axis II diagnosis.
Among patients with axis I diagnoses (alone and comorbid with axis II disorders), 28 had a generalized anxiety disorder, 25 had a panic disorder, 23 had an eating disorder, 15 had a substance (cannabis) use disorder, and 14 had a dysthymic disorder. The mean Global Assessment of Functioning (GAF) score was 56 (SD=11.9). The length of treatment (one session per week) averaged 5 months (SD=0.9; range=2–6).
Measures
Shedler-Westen Assessment Procedure–200 (SWAP-200).
The SWAP-200 is a psychometric system designed to provide a comprehensive assessment of personality and personality pathology (
24–
28). It consists of 200 items that the assessor sorts into eight categories, from not descriptive to most descriptive of the person.
The SWAP-200 assessment furnishes 1) a personality diagnosis expressed as the matching of the patient assessment with 10 personality disorder scales, which are prototypical descriptions of DSM-IV axis II disorders, and 2) a personality diagnosis based on the correlation/matching of the patient’s SWAP description with 11 Q-factors/styles of personality derived empirically. It also includes a dimensional measure of psychological strengths and adaptive functioning and makes it possible to obtain both categorical and dimensional diagnoses.
Therapist Response Questionnaire.
The Therapist Response Questionnaire (
21), which is filled out by a clinician, is designed to assess countertransference patterns in psychotherapy. It consists of 79 items measuring a wide range of thoughts, feelings, and behaviors expressed by therapists toward their patients (see the online
data supplement). The statements are written in everyday language so that clinicians of any theoretical approach can use the tool without bias. The questionnaire comprises eight countertransference dimensions derived by factor analysis: overwhelmed/disorganized, helpless/inadequate, positive, special/overinvolved, sexualized, disengaged, parental/protective, and criticized/mistreated (
19).
In the present study, the eight factor-derived scales demonstrated excellent internal consistency (
29). The following Cronbach’s alpha values were obtained: overwhelmed/disorganized, 0.83; helpless/inadequate, 0.81; positive, 0.78; special/overinvolved, 0.76; sexualized, 0.71; disengaged, 0.79; parental/protective, 0.73; and criticized/mistreated, 0.81.
Clinical questionnaire.
We constructed an ad hoc questionnaire for clinicians to provide general information about themselves, their patients, and the therapies they used. Clinicians provided basic demographic and professional data, including discipline (psychiatry or psychology), theoretical approach, employment address, hours of work, number of patients in treatment, and gender, as well as patients’ age, gender, race, education level, socioeconomic status, and DSM-IV axis I diagnoses. Clinicians also provided data on the therapies, such as length of treatment and number of sessions.
Procedure
After we received the clinicians’ agreement to participate, we provided them with the material to conduct the study. They were asked first to evaluate their emotional response concerning the selected patient using the Therapist Response Questionnaire, and then, between 1 and 3 weeks later, to evaluate the same patient’s personality using the SWAP-200. We used this interval because of the different time commitment required by the measures. That is, we wanted the clinicians to complete the faster and user-friendly Therapist Response Questionnaire immediately after a session with the designated patient, and to complete the more structured and time-consuming SWAP-200 later, allowing them to plan for this more involved personality assessment. Separating the two evaluations was also aimed at reducing any possible effect that evaluating their emotional response might have on a concurrent rating of that same patient’s personality. Thus, we sought to limit the impact of the clinicians’ emotional response when they completed personality ratings of that same patient.
Statistical Analysis
All analyses were conducted with SPSS 20 for Windows (IBM, Armonk, N.Y.) To study the relationship between countertransference patterns and specific personality disorders, we calculated the partial correlations (partial r, two-tailed) between Therapist Response Questionnaire factors and each personality disorder scale in SWAP-200, removing the effect of the other nine personality disorders in each analysis. We used these partial correlations to obtain results specific and unique to each disorder/countertransference pattern, controlling for the overlap between different personality disorder diagnoses.
To explore whether specific associations were dependent on clinicians’ approaches, we calculated the partial correlations between each personality disorder and therapist response pattern, eliminating from the sample all the psychodynamic clinicians (remaining N=100).
Finally, to investigate the relationship between patients’ psychological functioning and clinicians’ emotional responses, we calculated the bivariate correlations (Pearson’s r, two-tailed) between the SWAP-200 high-functioning scale and Therapist Response Questionnaire factors.
Discussion
The primary goal of this study was to examine the relationship between therapist response and patient personality pathology. The findings support our hypothesis that there would be significant and consistent relationships between therapist reactions and specific personality disorders (
Table 1). Moreover, confirming the results of Betan et al. (
19), we found that clinicians of different therapeutic approaches produced similar data, suggesting that these results are not artifacts of their theoretical preferences and that patient interpersonal patterns are quite robust in evoking emotional responses from therapists employing different technical styles. This demonstrates that if clinicians have and recognize countertransference feelings, they can use them to inform themselves about their patients’ interpersonal patterns (
1,
17).
Another aim of this study was to investigate the relationship between therapist response and patient level of personality functioning. Our results are consistent with those of Dahl et al. (
30), suggesting that therapists feel more helpless, inadequate, and disorganized with low-functioning patients.
Turning to the more specific and nuanced findings of our study, patients with cluster B personality disorders seem to elicit more mixed and negative responses in their therapists than do patients with cluster A and C disorders. Our results further support previous findings that cluster B patients evoke more negative and difficult-to-manage emotions in their therapists (
11,
12,
19,
20,
22,
23,
31,
32). Among cluster B disorders, borderline patients seem to arouse stronger and more heterogeneous reactions in clinicians, who tend to feel overwhelmed with high levels of anxiety, tension, and concern.
Clinicians treating borderline patients report feeling incompetent or inadequate and experiencing a sense of confusion and frustration in sessions. They report apprehension about failing to help these patients, and they experience guilt when they see these patients distressed or deteriorating (Table 2). This heterogeneity among therapists’ emotional responses could reflect the contradictory self and other representations that characterize borderline patients (2, 11).Such intense feelings in work with borderline patients could lead therapists to perform in an erratic manner. For example, therapists could have difficulties in setting and maintaining boundaries or, conversely, could set extensive and rigid limits on their patients’ requests. Clinicians could also avoid the expression of their thoughts and feelings during a session because they fear an angry reaction, or instead offer sudden and aggressive interpretations or confrontations of patient behavior. Likewise, with narcissistic patients, therapists may come to feel bored, distracted, disengaged, and frustrated (
Table 2). These kinds of emotional responses could provoke an emotional disattunement, with lack of interest and empathy ultimately leading to impasse and treatment termination.
Regarding cluster C patients, we found several significant patterns of therapist response. Of particular note was the protective and positive feelings of therapists toward avoidant patients, perhaps experiencing a wish to repair some deficiencies or failures in their patients’ relationships with parents or significant others. Overprotective feelings could induce the therapist to avoid the exploration of the patients’ painful feelings or aggressive affects, considering these individuals to be too fragile and vulnerable.
Our results partially diverge from previous studies that found that cluster C patients do not seem to evoke negative feelings in their therapists (
19,
20).
In our sample, clinician responses to patients with dependent personality were characterized by positive and protective feelings but also by feelings of helplessness and inadequacy; therapists can feel like their hands have been tied or that they have been put into an impossible bind. These results may indicate that if the therapist “buys into” the patient’s view of him- or herself as helpless or incapable, such a perception is capable of arousing first parental and warm feelings, and secondly negative feelings (3, 33, 34).This study has some limitations. First, the same clinician provided data about both a patient’s personality pathology and his or her own countertransference, which may be a source of measurement bias. A more rigorous research design would include an independent assessment of patients’ personality disorders or the use of an observer-rated analysis of therapists’ reactions, or both. Second, our sample may not be representative of all patients with psychiatric disorders, as it contained a substantial proportion of patients with axis II disorders and a limited proportion with axis I disorders. Also, our population had a narrow age range, with patients between their mid-30s and early 50s. Finally, it is possible that social desirability biases influenced the therapist ratings. For example, contrary to clinical expectations (
3,
5,
11,
34), no significant correlations with the sexualized countertransference factor emerged. The countertransference measure we used was limited to therapist self-report, and defensive biases and failure to recognize unconscious feelings are inherent to this method of evaluation. The analysis of countertransference in session video or transcripts using an observer perspective may help overcome this limitation in future work.
Some important factors, however, partially mitigate concerns that the results simply reflect clinician biases, in particular in relation to patients’ personality diagnoses. Previous research has suggested that clinicians tend to make highly reliable and valid judgments if their observations and inferences are quantified using psychometrically sophisticated instruments such as those used in our study (
19,
24,
28,
35,
36). Also, the validity of SWAP diagnoses relies on therapist experience rather than specific instrumental training (
28,
37). In regard to the high prevalence of axis II compared with axis I diagnoses in our patient sample, it is important to note that this is likely due to our recruitment strategy, as several of the psychological associations, institutes, and clinicians assisting us in our research specialized in the treatment of personality disorders, and personality disorders are probably more frequent among their clinical populations.
Previous studies examining therapist emotional responses to axis II patients (
19,
20) have aggregated patients at the cluster level rather than at the level of the individual disorder as we did. Using the latter approach in this study may have obscured important information regarding specific therapist responses in relation to specific personality pathologies. We found similarities in therapist responses to patients suffering from personality disorders of different clusters. That is, some therapist reactions to borderline and dependent personality disorders were quite similar, such as the capacity of both types of patients to evoke helpless and overinvolved reactions in therapists. Although these two disorders belong to different clusters, they share important core pathological dynamics, such as interpersonal neediness and anxious attachment (
3,
5,
11–
13,
33). Also of note, the effect sizes obtained in our study are generally larger than those of Betan et al. (
19). One possible explanation is that Betan et al. assessed personality disorders only through the DSM-IV axis II personality disorder criteria set. This procedure may offer less breadth and depth with regard to the construct of personality than the use of the SWAP-200. It will be important in future studies to evaluate the impact of therapist emotional response and patient personality style with other process variables directly related to treatment process, such as interaction structures (
38), use of technique, or therapeutic alliance ruptures and resolutions (
16,
17).