Countertransference Phenomena and Personality Pathology in Clinical Practice: An Empirical Investigation

Participants were 181 clinicians who constituted a random national sample of experienced psychiatrists and psychologists from the membership registers of the American Psychiatric Association and American Psychological Association. We requested mailing lists of clinicians with at least 3 years’ postlicensure or postresidency experience who indicated that they performed at least 10 hours per week of direct patient care. As in prior research with this method, psychologists responded at a substantially higher rate than did psychiatrists to the solicitation, allowing us to assess for biases imposed by differential training or response rates. We found no differences between patients described by psychologists and those described by psychiatrists on any variable of interest despite a roughly 3:1 response rate ratio. (Variables of interest included age, sex, race, socioeconomic status, education level, treatment length, and countertransference factor scores [14 t tests, not significant at p<0.01].) We also compared the patients of this sample of clinicians to those of the first 181 clinicians in another sample with whom we used a similar method but paid a substantially higher honorarium and obtained a correspondingly higher response rate (>30%) for the psychiatrists; we found no differences between the samples of patients. Together, these data suggest that any potential biases in the tendency to respond had minimal impact on the representativeness of the sample (see also the discussion of limitations in the Discussion section).

To obtain a cross-section of psychotherapy patients seen in clinical practice, we asked clinicians to describe a nonpsychotic patient at least 18 years old whom they had treated for a minimum of eight sessions (to maximize the likelihood that they would know the patient well enough to provide a reasonably accurate description of the patient). To minimize selection biases, we directed clinicians to consult their calendar to select the last patient they saw during the prior week who met study criteria. Each clinician described only one patient in order to minimize rater-dependent biases. Clinicians received a modest honorarium ($85) for a procedure that took 3–4 hours to complete, with a response rate of approximately 10%.

Clinicians could participate either by pen-and-paper forms or on an interactive web site (http://www.psychsystems.net). Web versus paper participants did not differ on any variable studied here (e.g., countertransference factor scores; eight t tests). Clinicians provided no identifying information about the patient (such as name, initials, or social security number) and were instructed to use only information already available to them from their contacts with the patient so that data collection would not compromise patient confidentiality or interfere in any way with ongoing clinical work.

We employed a number of measures in standardized sequence. We describe those of relevance to the present study here.

The clinician sample consisted of 141 (77.9%) psychologists and 40 (22.1%) psychiatrists; 58.6% (N=106) of the clinicians were male. The majority saw patients in private practice (N=145, 80.1%), but they also worked in other settings, including hospital (N=57, 31.5%), forensic (N=15, 8.3%), clinic (N=14, 7.7%), or school (N=9, 5.0%) settings. (As might be expected, psychiatrists were more likely to have primary or secondary employment in hospital settings.) The most common self-reported theoretical orientations included psychodynamic (N=73, 40.3%), eclectic (N=55, 30.4%), and cognitive behavioral (N=37, 20.4%).

Reflecting our efforts to obtain a patient sample stratified by sex, about one-half of the patients were male and one-half were female, with an average age of 40.5 years (SD=13.4). The sample was predominantly Caucasian (N=168, 92.8%). Most were middle class (N=102, 56.4%), with 2.8% (N=5) rated as poor, 24.3% (N=44) as working class, and 16.6% (N=30) as upper class. The mean Global Assessment of Functioning Scale score was 58.0 (SD=12.9). Length of treatment averaged 19 months (SD=30.0), with a median of 13 months, indicating that the clinicians knew the patients very well. The most common diagnoses reported by the clinicians were major depressive disorder (N=89, 49.2%), dysthymic disorder (N=68, 37.6%), generalized anxiety disorder (N=46, 25.4%), and adjustment disorder (N=45, 24.9%).

To identify the factor structure of the Countertransference Questionnaire, we first subjected the items to a principal-component analysis using Kaiser’s criteria (eigenvalues >1). We used the scree plot, percentage of variance accounted for, and parallel analysis (33–35) to select the number of factors to rotate. The scree plot indicated a break between eight and nine factors, and parallel analysis indicated eight factors with eigenvalues larger than would be expected by chance (p<0.05 with 100 random data sets). We therefore conducted factor analyses with seven, eight, and nine factors to maximize interpretability.

Several factors emerged across algorithms, rotations, and estimation procedures. We report here the most coherent solution, the eight-factor promax (oblique) solution using maximum likelihood estimation, which we favored a priori because of the characteristics of promax rotations (notably the absence of the assumption of orthogonal factors and the tendency to maximize factor loadings within factors) and maximum likelihood estimation (notably the advantages for use in subsequent confirmatory factor analyses). This solution accounted for 69% of the variance and included factors well marked by at least five items each, suggesting a stable factor structure unlikely to be substantially affected by sample size (36).

Table 1 presents the factor structure. To create factor-based scores for use in this and subsequent studies, we included items loading ≥0.50 for factors 1 and 2, ≥0.40 for factor 3, and ≥0.375 for factors 4–8 to maximize reliability (coefficient alpha). Intercorrelations among the eight factors ranged from –0.16 to 0.58, with a median of 0.30.

This factor structure is conceptually coherent and clinically recognizable. However, an important question is the extent to which its coherence simply reflects the theoretical beliefs of participating clinicians, particularly given that 40% of clinicians in the sample reported a psychodynamic orientation. To evaluate this possibility, we conducted a second factor analysis, this time eliminating all clinicians who reported a psychoanalytic or psychodynamic orientation (remaining N=108). Using the same rotation and estimation procedures, the factor analysis reproduced the same factor structure as in the complete sample, except that the factors appeared in a slightly different order. Thus, the factor structure does not appear to be an artifact of clinicians’ theoretical preconceptions.

As a first test of the validity and clinical applicability of the Countertransference Questionnaire, we examined the relationship between each of the eight factors and dimensional measures of the DSM-IV personality disorders. Because of the extensive comorbidity of the axis II disorders, we analyzed the data at the personality disorder cluster level (clusters A, B, and C) by summing the number of symptoms endorsed for each of the personality disorders in each cluster. To control for comorbidity across clusters (and for general severity of personality disturbance), we used partial correlations, controlling for pathology associated with the other two clusters in all analyses.

Based on the item content of the factors, we made the following a priori predictions: 1) that the cluster A (odd/eccentric) disorders would be associated with the disengaged factor and secondarily with the criticized/mistreated factor; 2) that the cluster B (dramatic/erratic) disorders would be associated with the overwhelmed/disorganized, helpless/inadequate, special/overinvolved, and sexualized factors; and 3) that the cluster C (anxious) disorders would be associated with the parental/protective factor.

The primary findings are reported in Table 2. As predicted, cluster A showed a significant association with the criticized/mistreated factor, although it was not correlated with the disengaged factor. The data strongly supported the associations for cluster B, except for the special/overinvolved factor. The cluster B (dramatic/erratic) disorders showed an additional (unpredicted) association with the disengaged factor and a negative correlation with positive countertransference. The data supported the hypothesis for cluster C.

In secondary analyses, we followed up on some of these patterns (particularly those that were contrary to our expectations) with hypotheses specific to particular disorders (rather than clusters) using partial correlation analysis, holding constant the other nine disorders in each analysis. We hypothesized that borderline personality disorder would show the expected association with the special/overinvolved factor. This hypothesis was supported (partial r=0.23, df=170, p=0.002). We also hypothesized that narcissistic personality disorder would account for the correlation between cluster B disorders and the disengaged factor. In fact, whereas the other cluster B disorders showed no significant associations with therapist disengagement, narcissistic personality disorder was associated with this factor (partial r=0.30, df=170, p<0.001).

To illustrate the uses of the Countertransference Questionnaire in clinical practice and to examine the extent to which it can be used to create empirical prototypes of common countertransference patterns in specific types of pathology, we created a composite description of countertransference patterns in the treatments of patients who met the DSM-IV criteria for narcissistic personality disorder. We standardized (z-scored) the items across patients and then averaged the item scores from patients meeting the DSM-IV criteria for narcissistic personality disorder assessed from the axis II checklist. By standardizing items (setting means to 0) before aggregating, we reduced the salience of items that were descriptive of all patients in the sample (e.g., positive feelings) but not specific to patients with narcissistic personality disorder.

Table 3 presents the items most and least descriptive of therapists’ descriptions of countertransference responses to patients with narcissistic personality disorder (N=13). The composite description is remarkably similar to theoretical and clinical accounts (e.g., references 39–42). Clinicians reported feeling anger, resentment, and dread in working with narcissistic personality disorder patients; feeling devalued and criticized by the patient; and finding themselves distracted, avoidant, and wishing to terminate the treatment.

As with the factor analysis, to see whether this portrait of countertransference responses to narcissistic personality disorder patients could be accounted for by clinicians’ theoretical preconceptions, we created a second composite description excluding the data for the three clinicians reporting a psychoanalytic or psychodynamic orientation (N=10). The items and means were virtually identical, suggesting once again that the data were not theory dependent.

This study has three primary limitations. First, although clinicians are the most obvious informants to report on their own countertransference responses, the countertransference measure we used shares the inherent limits of self-report measures, such as defensive biases and failure to recognize processes that an outside observer might identify. Thus, it would have been useful to have ratings of therapy process by an independent observer (perhaps based on audiotaped sessions) to identify patterns of clinicians’ behavior that would likely converge with clinicians’ self-reports in some ways and diverge from them in others.

A related concern is that clinicians provided all the data and hence that their responses on one questionnaire may not be independent of their responses on others (e.g., that their diagnosis of narcissistic personality disorder may not have been independent of their observations of the patient’s behavior in the room with them). This limitation is common to virtually all studies in psychiatry, in which a single observer (usually the patient) provides data on one measure that are then correlated with data from another measure completed by the same informant (either by self-report questionnaire or structured interview). It would have been preferable to collect diagnostic data independently of clinicians’ reports of their countertransference responses, and future research should clearly do so.

Several factors, however, mitigate the concern that the results primarily reflect clinicians’ biases or preconceptions. First, as noted earlier, clinicians of widely different theoretical perspectives and with widely different training (M.D. training versus Ph.D. training) produced highly similar data. If, for example, cognitive behavior clinicians share a theory of countertransference with psychoanalytic therapists, we are unaware of such a theory. Indeed, the similarity across theoretical orientations is one of the most interesting findings of this study, suggesting that patients’ interpersonal patterns are quite robust in the face of different technical styles. Second, because we used dimensional rather than categorical measures of axis II pathology, clinicians’ responses on the countertransference measure were not likely to be influenced by their beliefs about whether the patient had one personality disorder or another. Third, previous research suggests that clinicians tend to make highly reliable and valid judgments if their observations and inferences are quantified using psychometric instruments such as the ones used in this study. For example, correlations between treating clinicians’ and independent interviewers’ assessments of a range of variables, such as measures of personality pathology and adaptive functioning, tend to be large, typically >0.50 (44–46). Empirically, clinicians’ theoretical orientation predicts little variance in descriptions of clinical phenomena when clinicians were asked to describe a specific patient rather than their beliefs or theories of psychopathology (47, 48). Nevertheless, future research using this measure should assess the association between countertransference phenomena and patients’ personality pathology by using data about the patient provided by other observers.

The second limitation was clinicians’ response rate to our request for participation (approximately 10%, for a study described as requiring 3–4 hours of work for a token honorarium). Three factors, however, limit the likelihood that the results reflect response rate biases. First and foremost, it is hard to imagine a response rate hypothesis that could explain the pattern of results. By virtue of their willingness to donate 3–4 hours of their time, the clinicians who participated in the study may have been characterized by greater interest in research, altruism, financial distress, or a host of other factors, compared with colleagues who did not participate, but it is difficult to see how any of these variables could have produced the obtained findings. Second, because we solicited data from clinicians in multiple practice settings across North America and provided them with a method of randomly selecting a patient within their practice, we were able to obtain a broad cross-section of patients. (Clinicians who agreed to participate were unaware that countertransference was one of the constructs we intended to study, so we were not selecting clinicians with a particular interest in or knowledge of this domain.) Third, as noted earlier, psychologists’ response rate was almost three times the rate of psychiatrists, yet the two sets of informants provided similar data, suggesting that neither training nor response rate was responsible for the findings. Finally, all studies have selection biases, some avoidable and others less so. An alternative method would have been to sample clinicians at a single hospital or institution, but this strategy would likely have resulted in less generalizable results. It seems unlikely that the sample in our study was less representative of the population of clinicians or patients in psychotherapy than samples in typical studies of psychotherapy or psychopathology, which tend to rely on patient populations from a single site. Nevertheless, the data clearly require replication and extension with new groups of patients, and we hope this article will generate interest in the measure for future research.

A third potential objection is sample size, given the possibility of some instability of factor structure with a ratio of cases to items of <3:1. However, recent thinking about factor analysis, based on data from Monte Carlo simulations and other studies, suggests that factor solutions stabilize with far fewer cases than previously believed (typically by 100 cases), as long as the factors are well marked by a sufficient number of items with loadings above 0.40 or 0.50 (as they were here), and that conventional case-to-item ratios do not take into consideration a range of variables that qualifies them in one direction or the other (see, e.g., references 36, 49). Clearly the next step in this research, however, is a larger-N replication study using confirmatory factor analysis and external ratings of variables such as personality disorder diagnosis and treatment outcome independent of the clinician’s reports.

The Countertransference Questionnaire represents an effort to develop a readily administered measure that reflects shared clinical wisdom in its item content and statistical “wisdom” in its factor structure. This measure is germane to future research on countertransference phenomena, as well as to practice, allowing clinicians to clarify the diagnostic relevance and utility of their reactions by comparing their own responses to normed psychometric data. A broadband measure of countertransference processes with known correlates such as this can turn clinicians’ experiences into quantifiable dimensions that capture interpersonal patterns that emerge in sessions, allowing clinicians who normally attend to countertransference phenomena to hone and systematize their self-reflections and providing clinicians whose theoretical orientations do not emphasize such processes a language and method with which to capture information about the patient and the treatment process that may be diagnostically and therapeutically significant.