Predicting Expressed Emotion: A Study With Families of Obsessive–Compulsive and Agoraphobic Outpatients

Predicting Expressed Emotion: A Study With Families of Obsessive–Compulsive and Agoraphobic Outpatients
[Families and Emotions]
Chambless, Dianne L.1,5; Bryan, Angela D.2; Aiken, Leona S.2; Steketee, Gail3; Hooley, Jill M.4
1Department of Psychology, University of North Carolina at Chapel Hill
2Department of Psychology, Arizona State University
3School of Social Work, Boston University
4Department of Psychology, Harvard University.
5Correspondence concerning this article should be addressed to Dianne L. Chambless, Department of Psychology, University of North Carolina, Chapel Hill, North Carolina 27599-3270. Electronic mail may be sent to chambles@email.unc.edu.
Angela D. Bryan is now at the Department of Psychology, University of Colorado at Boulder.
Chambless, Bryan, Aiken, Steketee, and Hooley (1999) and Chambless, Gillis, Tran, and Steketee (1996) have reported other data on these relatives and patients. This work was supported by the National Institute of Mental Health Grant R01-MH44190. We thank Camberwell Family Interview (CFI) and the Structured Clinical Interview for the DSM–III–R interviewers Marty Gillis, Cheryl Sheffler Rubenstein, Giao Tran, Hope Worden, Sheila Woody, and Whitney Wycoff; CFI coders Henry Chung, Delphine Mattison, Scott Maynard, and Brian McCorkle; Interaction Coding System coders Deborah Dowdall and Peggy Gaver; and Tom Rodebaugh for his assistance in the preparation of this article.
Received Date: March 27, 2000; Revised Date: October 27, 2000; Accepted Date: January 18, 2001
Abstract
The authors used structural equation modeling to examine expressed emotion (EE) in relatives of outpatients with panic disorder with agoraphobia (n = 42) or obsessive–compulsive disorder (n = 60). EE was examined as a function of patients' illness and personality and as a function of characteristics of relatives themselves. EE was operationalized in terms of hostility on the Camberwell Family Interview (C. E. Vaughn & J. P. Leff, 1976) and patients' ratings of their relatives' criticism (perceived criticism). Key findings include the identification of a characteristic of the relative (self-reported angry thoughts, feelings, and behaviors) that is directly linked to both hostility toward the patient and to perceived criticism, as well as a direct path between relatives' low rates of observed problem solving and their hostility toward the patient. Patient Pathology predicted perceived criticism but not observer-rated hostility.
________________________________________


When relatives who live with schizophrenic and depressed patients are high in expressed emotion (EE), the patients are at higher risk for relapse after hospitalization (see meta-analysis by Butzlaff & Hooley, 1998). EE has shown promise as a predictor of poor treatment outcome in anxiety disorders as well (Chambless & Steketee, 1999; Leonard et al., 1993; Tarrier, Sommerfield, & Pilgrim, 1999). Further, controlled trials have demonstrated that family interventions designed to reduce EE result in significantly lower relapse rates (see meta-analysis by de Jesus Mari & Streiner, 1994), although whether reduction in EE is the mechanism for these benefits remains untested. These findings argue for the importance of a better understanding of the antecedent and maintaining factors of EE itself. Toward this end, we examine here the utility of a transactional stress and coping model, including both relatives' and patients' characteristics.
EE refers to emotions a relative expresses about a psychiatric patient, most commonly assessed during an interview given to the relative in the patient's absence, the Camberwell Family Interview (CFI; Vaughn & Leff, 1976). Of the CFI measures, the number of critical comments, often combined with hostility,1 has proved the most consistent predictor of relapse (see Hooley, Rosen, & Richters, 1995). An alternative approach to measuring EE involves obtaining patients' perspectives on their relatives' behavior, rather than relying on codes assigned by objective raters. Indeed, Hooley and Teasdale (1989) have argued that patients' perceptions may be the proximal factor in EE's relationship to treatment response and relapse, in that EE must be apprehended by the patient in some way to have an effect. Hooley and Teasdale devised the Perceived Criticism Measure, a simple rating scale the patient completes to describe a relative's degree of criticism of him or her. In two investigations, this measure proved to be a strong predictor of posthospitalization relapse of patients with major depression (Hooley & Teasdale, 1989) or of initial poor outcome for outpatients with obsessive–compulsive disorder (OCD) or panic disorder with agoraphobia (PDA; Chambless & Steketee, 1999). Similarly, Tompson et al. (1995) found that a measure of perceived criticism derived from an interview with schizophrenic patients predicted subsequent psychotic exacerbation. Perceived criticism is, at best, moderately related to CFI critical comments and hostility (rs = .16–.33, Chambless, Bryan, Aiken, Steketee, & Hooley, 1999; Hooley & Teasdale, 1989). Thus, it requires separate examination, as is accomplished in the present study.
It is hardly surprising that relatives living with patients with serious behavioral disorders often have negative feelings about the patients, given the strain such disorders place on family life and the burden on the relatives themselves (e.g., Chakrabarti, Kulhara, & Verma, 1993). Indeed, after a number of years of living with a patient with a psychiatric illness, the majority of relatives are classified as high EE (see Hooley & Richters, 1995), especially if relatives are assessed during an acute episode of the patient's illness. It is, perhaps, more remarkable that some relatives remain low in EE; yet until recently, little attention was paid to factors that might elucidate why some relatives become critical or hostile and others do not. Although interventions designed to reduce EE have nonetheless proceeded, these interventions lack the guidance of empirical evidence about the antecedent or maintaining factors of EE. An emerging literature has pointed to individual variables that might be associated with EE, and these findings will be reviewed below. However, a more comprehensive, multivariate model of contributors to EE has not been examined. Guided by findings from the empirical literature on correlates of EE, as well as by theoretical models of family stress and coping, we propose and test such a model here. In particular, we adapted the transactional stress and coping model described by Thompson and Gustafson (1996) for families coping with a child's chronic medical illness. Thompson and Gustafson posited that negative outcomes on the relatives' part (in our case, EE) are the outcome of patient illness factors, as well as three relatives' factors: resources, cognitive appraisals, and coping skills. We now review the empirical findings pertinent to this model.
Patient Factors
The model adapted here predicts that, in part, relatives become negative because of the stress of the patients' illness. Supporting this idea are findings that CFI critical or hostile behavior scores are higher during acute episodes of the patients' disorder than after discharge (reviewed by Hooley et al., 1995). However, some relatives do not become negative even during acute episodes, and others remain negative once the episode has passed. Furthermore, severity of patients' symptoms and indices of functioning within a given sample typically fail to predict relatives' CFI critical comments or hostility scores (reviewed by Hooley et al., 1995; see also Chambless & Steketee, 1999; Peter & Hand, 1988; Tarrier, 1996), and in one study, parents were found to be as critical of their well offspring as of their schizophrenic child (Schreiber, Breier, & Pickar, 1995). Although these univariate tests of the relationship between patients' functioning and hostility have generally produced negative results, patients' functioning may still play a significant role in a multivariate model of hostility. Therefore, we include measures of functioning and symptom severity as part of a Patient Illness factor in our model.
Scattered findings suggest that relatives' CFI critical comments or hostility scores may be more closely related to patients' personality traits or to subclinical symptoms rather than to frank expressions of their major psychopathology. For example, relatives and caretakers of patients were more likely to be critical or hostile when patients were disruptive and impulsive (Asarnow, Tompson, Hamilton, Goldstein, & Guthrie, 1994; Heinssen et al., 1994, as cited by Hooley et al., 1995; Rosenfarb, Goldstein, Mintz, & Nuechterlein, 1995), or were hostile or critical themselves (Heinssen et al. as cited by Hooley et al., 1995; Mueser et al., 1993; Strachan, Feingold, Goldstein, Miklowitz, & Nuechterlein, 1989). In observational studies of patient–family interactions, schizophrenic patients in either highly critical or high EE families were more likely to be demanding (Mueser et al., 1993) or to display odd (but not flagrantly psychotic) behavior (Rosenfarb et al., 1995). Thus, in this investigation, we included a second Patient Personality factor representing dramatic (borderline, histrionic, narcissistic, and antisocial) and odd (schizoid, schizotypal, and paranoid) personality traits. Such traits include hostile, impulsive, demanding, self-centered, and odd behavior patterns. In addition, we included patients' self-reported hostile feelings and behaviors in the Patient Personality factor.
Relatives' Factors
Resources
The EE literature does not provide tests of typical resources such as relatives' financial status and sources of social support. However, relatives' own psychological health and personality may be considered resources on which the relative of the patient can draw to manage stress without negative outcomes. Two groups of investigators found that relatives' history of psychiatric disorders predicted high EE (Hibbs et al., 1991) or critical and hostile statements about the patient during a diagnostic interview (Goodman, Adamson, Riniti, & Cole, 1994).
Relatives' personality may also be viewed as a psychological resource or liability. EE researchers have been slow to investigate the possibility that relatives may have personality traits that predispose them to be hostile to or critical of patients. In two articles from her study of relatives of schizophrenic patients, Hooley (1998; Hooley & Hiller, 2000) found that high EE relatives described themselves as more concerned with social convention than low EE relatives on a personality inventory, as well as less self-realized and less flexible. Moreover, on a locus of control measure, high EE relatives endorsed a more internal locus of control than their low EE counterparts. Thus, high EE relatives may in general (rather than just in relation to the patient) believe that people can and should behave properly, which may foster their intolerance of the patient's deviance. Although our data set does not include measures of relatives' personality traits per se, we do have a measure of relatives' self-reported hostile feelings and behaviors, for which no particular target is specified. This measure was used as a proxy for a trait measure of relatives' hostility. We hypothesized that relatives who tend to be more hostile in general are more likely to be hostile toward the patient.
Cognitive Appraisals
Hooley (1987) suggested that relatives are likely to become critical, and eventually hostile, if they believe patients are able but unwilling to control their deviant behavior (cf. Vaughn & Leff, 1981). A number of studies have yielded broadly supportive results. For families of schizophrenic or depressed patients, criticism and/or hostility have correlated with relatives' attributions that the patients' behavior is controllable and personal to the patient rather than universal and stable, and is due to causes internal to the patient (Barrowclough, Johnston, & Tarrier, 1994; Brewin, MacCarthy, Duda, & Vaughn, 1991; Harrison & Dadds, 1992; Hooley & Licht, 1997; Weisman, Lopez, Karno, & Jenkins, 1993). These findings are consistent with Hooley's (1998) report that high EE relatives have a more internal locus of control in general. In our model we include a test of the contribution of relatives' self-reported attributions about the voluntary nature of patients' symptoms.
Coping Skills
Given the stress they face in living with a patient with a serious psychiatric disorder, relatives' coping repertoire may be important in understanding EE. One important coping skill is problem solving. Training in generation of solutions to problems plays a large role in programs designed to reduce EE, implying that investigators believe that deficits in problem solving underlie these behaviors (e.g., Falloon, 1984). Consistent with these ideas, Calvocoressi et al. (1995) found greater self-reported rejection of OCD patients in relatives with poorer self-reported problem-solving abilities. Two studies have provided a test of this hypothesis with CFI-defined EE and observed problem-solving ability during an interaction with the patient. Mueser et al. (1993) reported that the quality of relatives' problem solving was not related to their frequency of critical comments. In contrast, Cole, Kane, Zastowny, Grolnick, and Lehman (1993) found fathers who demonstrated higher rates of problem–solution focused behaviors were significantly less critical; mothers showed a similar pattern, but the findings were not statistically significant. In light of these conflicting data, we offer another test of this hypothesis within the context of the overall stress and adaptation model.
Hypothesized Model
Predictors
The bits and pieces of empirical data reviewed appear to fit well within the transactional stress and coping model. In our model (see Figure 1), patient illness factors are represented by indices of patients' global functioning and their overall psychological distress, whereas patients' personality is represented by dramatic and odd personality traits and general hostility. For relatives, (a) psychological resources are represented by psychological health and lack of general hostility; (b) cognitive appraisals, by relatives' attributions concerning the volitional nature of patients' symptoms; and (c) coping skills, by relatives' constructive problem-solving behavior. We predicted that relatives with poorer coping and higher maladjustment and general hostility, and those living with patients who were more ill, who functioned more poorly, and who had more negative personality characteristics, would be more likely to evince EE.

________________________________________

[Help with image viewing]
[Email Jumpstart To Image]
Figure 1. Proposed model for application of the transactional stress and coping model to expressed emotion. The person completing the measure is designated by P = patient, R = relative, and I = independent coder or interviewer. Pt. = Patient; Rel. = Relative; GAF = Global Assessment of Functioning; SCL–90–R = Symptom Checklist 90—Revised Global Severity Index (excepting hostility items); SAS = Social Adjustment Scale; CFI = Camberwell Family Interview; KPI = Interaction Coding System; Dramatic/Odd traits = sum of dramatic and odd criteria met on the Structured Clinical Interview for the DSM–III–R, Axis II.
________________________________________
Criterion Variables
Criterion variables for relatives' EE in this model were hostility, as coded from the CFI, and perceived criticism, as rated by the patient. We now briefly review our rationale for selecting these variables. CFI hostility has been examined independently of CFI critical comments rather infrequently, no doubt due to early research in which the addition of hostility ratings added little to the frequency of CFI critical comments in predicting relapse after hospitalization (see Hooley, 1986b; Leff & Vaughn, 1985). However, hostility's predictive utility vis-à-vis CFI critical comments may be sample specific, depending on the number of hostile relatives as well as the correlation of hostility to CFI critical comments in a particular sample. In fact, in a number of recent studies hostility scores emerged as a stronger predictor than CFI critical comments scores of patients' pathology, treatment outcome, or relatives' behavior (Brewin et al., 1991; Hooley & Licht, 1997; Vostanis, Nicholls, & Harrington, 1994; see also review by Gottschalk & Keatinge, 1993). This result also proved true in the present sample of outpatients with OCD or PDA (Chambless & Steketee, 1999). Relatives' hostility toward the patient was the most consistent risk factor for poor response to behavior therapy, predicting a sixfold increase in the odds of treatment dropout as well as poorer treatment outcome for completers on measures of anxiety symptoms and general functioning. Thus, the present study is focused on CFI hostility rather than CFI critical comments. Similarly, in the present sample, perceived criticism predicted poor treatment outcome significantly on the major outcome variable whereas CFI critical comments did not (Chambless & Steketee, 1999). Thus, prediction of perceived criticism is a second focus of this study. To date, the empirical literature is barren of findings identifying the antecedents or correlates of perceived criticism, other than alternative measures of EE. To our knowledge, this study represents the first such investigation.
Method
Participants
Participants were 60 outpatients with OCD and 42 with PDA, treated at McLean Hospital in Belmont, MA, or at American University in Washington, DC, and all English-speaking adult relatives living in the household with the patients. All patients met criteria for a primary diagnosis of OCD or PDA according to the Structured Clinical Interview for the DSM–III–R—Patient Version (SCID–P; Spitzer, Williams, Gibbon, & First, 1989). In addition, PDA patients were required to be at least moderately severe in avoidance according to the SCID, and OCD participants were accepted only if they were performing at least 1 hr of rituals per day. Patients were excluded if they met current criteria for or had histories of psychosis, reported evidence of organic etiology of the disorder, met current criteria for alcohol or substance dependence, or had family members unwilling to participate in the research program. Additional criteria for inclusion were age range from 18 to 65, duration of symptoms for at least 1 year, having lived with a relative or partner for at least 3 months prior to beginning the study and intending to do so until at least until the 6-month follow-up.
Seventy percent of patients were married or living with partners, and the remainder were living with parents or, more rarely, other family members. Patients averaged (M) 35 years in age (range = 18–62) and reported a mean symptom duration of 15 years for OCD (range = 1–37 years) and 10.4 years for PDA (range = 1–41 years). The ethnic/racial distribution was 87% White, 10% African American, 3% other; 72% of the patients were women. A total of 143 of their relatives participated (see Results section). The household socioeconomic status indicators spanned almost the entire range of the Hollingshead (1975) Four-Factor Index of Social Position; median status was 48 (e.g., minor professional or owner of a medium-sized business).
Measures
CFI.
This is a 1–2 hr semistructured interview of a psychiatric patient's relative, which is audiotaped and later scored to yield indices of the relative's attitudes toward the patient, as described in our introduction. The measure of interest for this investigation was the global rating of hostility, which ranges from 0 (no hostility) to 3 (generalization of criticism and rejection). Ratings of hostility (labeled Rel. CFI Hostility in Figure 1) indicate the relative has expressed dislike of the patient as a person, rather than disapproval of specific behaviors, through comments indicating generalized criticism, rejection, or both (e.g., “She's just a slob. … There's no doubt about it, she's a whacko.” and “I'm sick of it. I don't want it. Go sponge off somebody else.”).
The CFI was administered by trained interviewers and coded by raters trained to reliability (defined as minimum interrater reliability of .80) with their instructor before coding for the present study. During the study, a second rater independently coded 12% of CFI tapes. Reliability for the hostility rating was acceptable (intraclass correlation coefficient [ICC] = .74, rs = .65).
Perceived Criticism Measure (PCM; Hooley & Teasdale, 1989).
Patients rated each adult relative with whom they lived on a 10-point scale ranging from 1 (not at all critical) to 10 (very critical). Patients in the present investigation were asked to complete the PCM (labeled Rel. Criticism in Figure 1) twice, about 2 weeks apart, before starting behavioral treatment for their disorder. Test–retest reliability was acceptable (r = .75, n = 80). Similar findings were reported by Hooley and Teasdale (1989) over a test–retest interval of approximately 20 weeks (during hospitalization for major depression to 3 months postdischarge), r = .75. Only the data from the first administration were used in the structural equation model.
Discriminant validity of the PCM has been supported by low and nonsignificant correlations with measures of patients' depression, clinical anxiety, and personality disorder traits (Chambless & Steketee, 1999; Hooley & Teasdale, 1989; Riso, Klein, Anderson, Ouimette, & Lizardi, 1996), and predictive validity by its prediction of poor treatment outcome (Chambless & Steketee, 1999; Hooley & Teasdale, 1989). Convergent validity evidence is somewhat mixed, with Hooley and Teasdale (1989) and Chambless et al. (1999) reporting low to moderate correlations with the CFI measure of critical comments (rs = .27 and .16, respectively). In contrast, when studying a sample of PDA women and women without psychiatric disorder, Chambless et al. (1997) found wives' perceived criticism ratings predicted their husbands' actual rates of criticism during a face-to-face interaction (r = .52), and in a confirmatory factor analysis of the present sample, Chambless et al. (1999) found the perceived criticism rating loaded on the same factor as observed criticism and negative nonverbal behavior during a relative–patient interaction, the CFI frequency of critical comments, and the relatives' own report of their criticism of the patient.
Interaction coding system.
The Kategoriensystem für Partnerschaftliche Interaktion ([Interaction Coding System], KPI; Hahlweg & Conrad, 1983) was used to code the videotaped patient–relative interactions. This system requires that each meaningful unit of speech be assigned both a verbal code (1 of 12 possible codes) and a nonverbal code (positive, negative, or neutral); the codes were designed on the basis of behavioral marital theory of communications. The coding system has shown good interrater reliability and convergent and criterion-related validity in previous research (Chambless et al., 1997; Hahlweg et al., 1987; Halford, Hahlweg, & Dunne, 1990; Hooley, 1986a). Of interest in this study was the KPI code indicating the frequency with which the relative offered a positive solution to the problem under discussion. These solutions must incorporate something constructive the patient and relative could do to ameliorate the problem, including compromise. The solution must not include demands or suggestions that the patient cease some behavior to solve the problem. Suggestions are rated according to quantity, not quality. This code was used to represent relatives' coping in the structural equation model (labeled Rel. KPI Problem Solving in Figure 1).
In approximately 25% of cases (n = 38), videotapes were independently coded by two raters. Interrater reliability for the frequency of problem-solving codes, as assessed by intraclass correlation, was acceptable (.79). KPI coders were a separate set from CFI coders and were uninformed as to patients' or relatives' scores on other measures.
The Control of Symptoms Scale.
This scale is based on an unpublished measure developed by Hooley (1985) but altered for the present investigation. Relatives were asked to rate the degree to which they believed the patient, without professional assistance, could control each of 14 symptoms on a scale ranging from 0 (no control at all) to 4 (marked control). Relatives were to rate a given symptom only if the patient had that particular symptom. Responses were averaged to yield a single score for each relative. The schedule includes symptoms that are, by definition, part of PDA or OCD (e.g., obsessions, rituals, panic, and avoidance), as well as common associated problems such as social isolation and depression.
The measure was developed as follows: To form the initial pool of items, we developed 15 questions to reflect the basic psychopathology of these disorders and commonly co-occurring symptoms. These were embedded in a list containing 30 other symptoms, and two samples of relatives were asked to provide ratings. Items were retained if both samples indicated they noticed these symptoms in the patients at least some of the time (median score of 2 on a scale ranging from 1 to 5). Moreover, we determined that all symptoms from the 45-item list rated as present at least some of the time (according to median scores) were represented on the Control of Symptoms Schedule. The final 14-item scale is internally consistent ([alpha] = .80) and normally distributed, with higher scores indicating stronger belief in patients' ability to control their symptoms. Test–retest reliability, available for only a small sample (n = 17), was rather low (r = .55). The Control of Symptoms Scale yielded the attribution measure in the model tested (labeled Pt. Control of Symptoms in Figure 1).
SCID.
Interviewers were graduate students in social work, clinical psychology, or counseling psychology trained in the administration and scoring of the SCID. Interviews were audiotaped, and 25% of the tapes were randomly selected for a cross-site check of reliability. Tapes from studies of other diagnostic groups added to the reliability samples were to help keep reliability interviewers uninformed of the sample composition; the total reliability sample size was 33. Indicating excellent reliability, kappa was .94 for the diagnosis of panic disorder and 1.00 for the diagnosis of OCD. Because of empty cells, kappa could not be computed for presence or absence of agoraphobia (rated only if the diagnosis of panic disorder has been assigned); the two raters both rated patients as agoraphobic in 17 cases and disagreed on the presence of agoraphobia in 2 cases.
In addition to diagnoses for admission, interviewers provided data for two additional measures used in the structural equation model. The Global Assessment of Functioning (GAF; labeled Pt. GAF in Figure 1) rating takes into account all information available to the interviewer from the SCID for Axis I and Axis II disorders and yields one score that includes an assessment of patients' severity of symptoms and an assessment of social, leisure, family, and vocational functioning. Scores may range from 1 (persistent, very severe symptoms and persistent, very poor functioning) to 90 (absent or minimal symptoms, functioning well). Interrater reliability for GAF was adequate (r = .76). GAF was included as an indicator of the Patient Illness factor (see Figure 1). The second measure, dramatic and odd personality traits (labeled Pt. Dramatic/Odd in Figure 1), was defined as the sum of the number of criteria meeting threshold (a score of 3) over all the personality disorders included in the odd and dramatic clusters: paranoid, schizotypal, schizoid, narcissistic, histrionic, borderline, and antisocial. The possible range for this measure was 0–60. Interrater reliability was acceptable (r = .75). These two SCID measures yielded indicators of the Patient Personality factor for the model to be tested.
Symptom Checklist 90—Revised (SCL–90–R; Derogatis, 1994).
Relatives' and patients' scores on an abbreviated form of the SCL–90–R Global Severity Index (GSI), a total score for general psychopathology, were used as an indicator of overall distress and psychopathology. This index includes both the number and intensity of symptoms rated as present during the last week. The possible range for raw scores is 0–4, and scores represent an average of individual item scores. Derogatis (1994) reported evidence of excellent convergent validity of the GSI with another broad measure of psychopathology, the Middlesex Hospital Questionnaire (Crown & Crisp, 1966). He noted excellent test–retest reliability and internal consistency for each of the nine subscales but did not report these properties for the GSI itself. For the purposes of this investigation, we removed the Hostility subscale items from the calculation of the GSI. This abbreviated GSI score, used as a measure of relatives' psychological resources and as an indicator of the Patient Illness factor, is represented as Pt. SCL–90–R or Rel. SCL–90–R in Figure 1. The Hostility subscale includes items designed to tap actions, feelings, and thoughts that typify anger. These are rated on 0–4 scales without specification of a target (if any) for the respondent's anger. This measure is internally consistent (e.g., [alpha] = .84–.85), reliable over a 10-week test–retest interval (r = .73), and strongly correlated with scales derived from the Minnesota Multiphasic Personality Inventory (Hathaway & McKinley, 1940) that measure hostility and aggression (Derogatis, 1994). Hostility scores were calculated for both relatives and patients (labeled SCL Hostility in Figure 1). In the case of patients, Hostility was included as part of the latent Patient Personality factor. For relatives it formed a second, negative psychological resource variable.
The Social Adjustment Scale—Self-Report (Weissman & Bothwell, 1976).
This measure, labeled Pt. SAS in Figure 1, is a widely used measure of social adjustment in multiple areas of functioning, including work, school, family, and social and leisure spheres. Adequate test–retest reliability (mean r = .78; Edwards, Yarvis, Mueller, Zingale, & Wagman, 1978) and agreement with the interview version of the same measure (ICC = .72; Weissman & Bothwell, 1976) have been reported, as well as good sensitivity to treatment effects in a depressed sample (Weissman & Bothwell, 1976). The SAS was included in the Patient Illness factor in Figure 1.
Procedure
Subsequent to telephone screening, patients were first interviewed by one of the principal investigators (Dianne L. Chambless or Gail Steketee) who made an initial determination of eligibility. Those who passed this screening and who agreed to participate were asked to provide written informed consent and to attend a second, independent diagnostic interview (the SCID) before a final decision concerning admission to the study was made.
Applicants accepted for treatment after the SCID were asked to complete the self-report measures. Relatives were also asked to complete a packet of self-report questionnaires once they had provided written informed consent. Patients and relatives were asked to complete their forms separately and privately and to return these to the research assistant on their next office visit. Depending on scheduling requirements and the number of relatives living in the home, families attended one or more pretreatment assessment sessions to complete the videotaped interaction and the CFI, given in randomized order.
For the videotaped interactions, ultimately coded by the KPI system, each patient was taped in separate interactions with each adult relative with whom he or she lived. The dyads were asked to pick the top two problems in their relationship to discuss during a 10-min problem-solving interaction. They were instructed to begin with the most important problem, and, if they solved this before the 10 mins had ended, to go on to the next problem. When dyads had difficulty identifying a problem to discuss, the research assistant provided suggestions from items both patient and relative had indicated to be a problem on the Areas of Change Questionnaire (Weiss, Hops, & Patterson, 1973). Once a clear agreement had been reached about the problems to be discussed, the research assistant turned on the video camera and left the pair alone for 10 mins.
Results
Sample Description
To avoid introducing dependence into the data, when patients lived with more than one adult relative (n = 30), we selected one relative per family for inclusion in this study. Following customary practice in EE research, we included the most critical relative on the CFI. Of these relatives, 70% were spouses; 17%, mothers; 5%, fathers; and 8%, other relatives (e.g., siblings, adult children). On average (median), the patient had been living with this relative for 9.5 consecutive years before assessment (range = 4 months to 40 years). Descriptive data are presented in Table 1 and zero-order correlations among measures are presented in Table 2. Of the relatives included, 35% evinced some degree of hostility on the CFI. Ratings on the GAF indicate a wide range in the degree to which patients were impaired by their problems; on average, patients were rated as having serious symptoms or being seriously impaired in social, school, or vocational functioning.

________________________________________

[Help with image viewing]
[Email Jumpstart To Image]
Table 1 Means, Standard Deviations, and Ranges of Variables Included in the Structural Model
________________________________________

________________________________________

[Help with image viewing]
[Email Jumpstart To Image]
Table 2 Correlations Between Predictor and Criterion Variables Included in the Structural Model
________________________________________
Structural Equation Modeling
All structural equation modeling analysis was conducted with the Software Package EQS (Bentler, 1995). Model fit was evaluated with the likelihood ratio chi-square test, the comparative fit index, and the root-mean-square error of approximation (RMSEA). The likelihood ratio chi-square is a test of the null hypothesis of the equivalence of the observed covariance matrix and the covariance matrix implied by the model (Bollen, 1989). We seek to retain the null hypothesis in order to conclude that the implied and observed covariance matrices are identical (within sampling error). The comparative fit index is a measure that compares a baseline model in which no relationships are estimated between the variables to a theoretical model in which hypothesized paths are estimated. This index ranges from 0 to 1, where 1 indicates perfect fit, .9 indicates adequate fit, and .8 is considered marginal fit (Bentler, 1990). The closer the comparative fit index is to 1, the more of an improvement the hypothesized model is over the baseline model of no relationships. The RMSEA ranges from 0 to [infinity], with fit values of less than .05 indicating close fit, and values of less than .10 indicating reasonable fit (Browne & Cudeck, 1993). The RMSEA is sensitive to overfit; that is, it begins to increase when too many paths have been included. Thus, the RMSEA is especially useful when exploratory model testing indicates the inclusion of additional paths that were not hypothesized a priori (Rigdon, 1996). Our sample size was small, and so another reason for the inclusion of both the comparative fit index and the RMSEA is that, unlike the likelihood ratio chi-square, they are both sensitive to model misspecification and minimally affected by sample size (Hu & Bentler, 1998). All modeling was conducted on the pairwise deleted correlation matrix, using the median pairwise n of 101 for all significance tests.
Preliminary Analyses
Subsample comparisons.
Prior to model estimation, tests of the equality of the covariance matrices for relatives of OCD and PDA patients, for spouses versus other relatives, and for relatives from homes where the relative included was the only person residing with the patient versus one of two or more adult relatives were conducted with Box's M tests. This conservative test (Stevens, 1996) indicated that we were justified in combining the data for OCD and PDA patients: Box's M = 61.52, approximate F(66, 10734) = 0.76, p > .92. This was also the case for data for relatives and patients from homes with single versus multiple relatives: Box's M = 55.26, approximate F(45, 3011) = 0.95, p > .56. However, results indicated that there was some difference in the structure of the covariance matrices for spouses versus other relatives, Box's M = 114.55, approximate F(66, 4479) = 1.33, p < .05. Nonetheless, only in the case of one correlation did the difference in the coefficients for spouses versus other relatives approach significance. The correlation between the Control of Symptoms Scale and the PCM score was not significant for either subsample, but it was small and positive for the spouses (r = .13), versus small and negative for the other relatives (r = -.21, z = 1.74, p < .10).
Patient Illness and Personality factors.
As an initial step, we conducted a confirmatory factor analysis (CFA) to determine whether the latent variables of Patient Illness and Personality indeed represented two distinct constructs. Patient functioning (GAF), patient general psychopathology (SCL–90–R with hostility items removed), and patient social adjustment (SAS) served as indicators of the Patient Illness factor, whereas patient hostility (SCL Hostility) and dramatic and odd traits (Dramatic/Odd) from the SCID served as indicators of the Personality factor. Indicators were permitted to load only on their hypothesized latent factor, and the correlation between the factors was left free to vary. Cross-correlations between the errors of indicators of the two constructs were constrained to zero. The fit of this two-factor model was borderline, [chi]2(4, N = 101) = 10.209, p < .05; comparative fit index = .95; RMSEA = .13, and the correlation between the latent factors was very high (r = .78, p < .001), indicating a possible failure of construct distinctiveness.
We examined Lagrange multiplier statistics (MacCallum, 1995) to determine whether there were nonzero cross-construct correlations between the errors of the indicators, a further diagnostic indicator of the lack of distinctiveness of the constructs. For each parameter constrained to a particular value in the estimated model (e.g., a cross-construct correlation between errors in indicators of two constructs that is constrained to zero), a Lagrange multiplier statistic is computed (Bentler, 1990; MacCallum, 1995). It assesses the improvement in model fit that would result if the parameter were to be freely estimated. A chi-square test of improvement is provided (Hayduk, 1987).
Examination of Lagrange multiplier statistics in this case indicated the presence of two significant cross-construct correlated errors. Given the high correlation between the factors and the presence of cross-construct correlated errors, we estimated a model in which the correlation between the latent factors was constrained to be 1.00—essentially specifying a model with one latent factor and five indicators. The chi-square test for change in fit was not significant, [DELTA][chi]2(1, N = 101) = 2.74, ns, indicating that the two patient factors were best represented statistically as one latent construct. With a single correlated error between the indicators of GAF and SCL Hostility, the one-construct patient model provided good overall fit to the data, [chi]2(4, N = 101) = 5.70, ns; comparative fit index = .99; RMSEA = .07. Thus, this single Patient Pathology factor with five indicators was included in the structural model, as illustrated in Figure 2.

________________________________________

[Help with image viewing]
[Email Jumpstart To Image]
Figure 2. Estimation of the structural model with one correlated error on the patient factor. Coefficients are standardized path coefficients. All exogenous correlations are estimated; only significant correlations are shown. Straight lines represent structural paths. Curved lines with double-headed arrows represent correlations between exogenous constructs or between errors of the indicators of the latent construct. Overall model fit: [chi]2(29, N = 101) = 44.57, p < .05; comparative fit index = .91; root-mean-square error of approximation = .07. The person completing the measure is designated by P = patient, R = relative, and I = independent coder or interviewer. Pt. = Patient; Rel. = Relative; GAF = Global Assessment of Functioning; SCL–90–R = Symptom Checklist 90—Revised Global Severity Index (excepting hostility items); SAS = Social Adjustment Scale; CFI = Camberwell Family Interview; KPI = Interaction Coding System; Dramatic/Odd traits = sum of dramatic and odd criteria met on the Structured Clinical Interview for the DSM–III–R, Axis II. +p < .10 (marginally significant). *p < .05. **p < .01. ***p < .001.
________________________________________
Test of Proposed Model
The theoretical paths in the model of hostility and criticism, including the correlations between all pairs of exogenous variables and the correlated error on the Patient Pathology factor, were estimated using EQS. The hypothesized model provided an adequate fit to the data, [chi]2(29, N = 101) = 44.57, p < .05; comparative fit index = .91; RMSEA = .07. No model modifications were necessary. The final model with all standardized coefficients, significance tests of the coefficients, and significant exogenous correlations appears in Figure 2. As is standard practice in the estimation of structural equation models, all exogenous correlations are estimated, but for ease of interpretation, only significant correlations are depicted in the figure. The model accounts for 16% of the variance in perceived criticism and 18% of the variance in hostility, both medium effect sizes for multivariate models in social science (Cohen, 1988).
A striking feature of the model is the absence of a significant correlation between perceived criticism and hostility as rated by the CFI coder (r = .09, ns, see Table 2). Under those circumstances, it is not surprising that different indicators were significant predictors of these two criteria. The one exception was relatives' self-rated hostility on the SCL–90–R: As expected, higher self-rated hostility was significantly predictive of both patients' perceptions of criticism and hostility toward the patient as rated by the CFI coder.
In accordance with hypotheses, relatives who were low on problem solving were also more likely to express hostile feelings about the patient than were those with higher rates of positive problem solutions. Contrary to our prediction, relatives' ratings that patients had control over their symptoms, Patient Pathology, and relatives' psychological distress did not predict greater hostility. In fact, there was a trend for relatives who were more distressed to express less hostile feelings about the patient. However, except for the aforementioned path from relatives' self-reported hostility to perceived criticism, patients' perceptions of their relatives' criticism were predicted only by Patient Pathology, such that patients who were more dysfunctional and had more negative personality traits were more likely to report that their relatives were critical of them.
The correlation between control of symptoms and perceived criticism tended to differ in the spouse versus nonspouse subsamples. In a final model test, we assessed whether the model, with path coefficients constrained to be equal to the estimates obtained with the full sample, fit well in the spouse subsample. Model fit with these parameter constraints imposed was good, [chi]2(45, N = 71) = 37.21, ns, comparative fit index = 1.00, RMSEA = .00. Furthermore, freeing the path from control to perceived criticism did not improve model fit, [DELTA][chi](1, N = 71) = 1.08, ns. We conclude that the full model represented the spouse data well. We could not repeat the test for the nonspouse sample, given sample size limitation.
Discussion
The hypothesized model for the development or maintenance of relatives' hostility and perceived criticism was partly confirmed. With minimal revision, the overall model fit the data well, and several predictors had significant path coefficients. The findings were applicable to OCD and PDA samples alike.
Perhaps the most important finding in this investigation is the identification of a direct link between relatives' low rates of observed constructive problem solving and their hostility toward the patient. In more rigorous fashion, these data replicate the previous report of Calvocoressi et al. (1995) with OCD patients, findings that were based on self-report data alone. They are also consistent with Cole et al.'s findings (1993) with schizophrenic patients and their families, but our findings extend this research through a multivariate approach accounting for other variables such as patients' functioning in the examination of the contribution of problem solving.
Contrary to hypothesis from the transactional stress and coping model, Patient Pathology was not related to relatives' hostility toward patients. However, it was significantly related to perceived criticism, such that more disturbed patients perceived their relatives to be more critical of them. Thus, findings are mixed on this aspect of the theoretical model. Recall that CFI hostility and perceived criticism bear little relationship to one another. Thus, it is not surprising that factors predicting one may not be associated with the other. Given perceived criticism's track record as a predictor of treatment outcome and relapse, these data emphasize the importance of achieving a better understanding of the discrepancy between objectively rated measures of relatives' criticism and hostility and patients' reports.
In keeping with expectations, relatives of patients who functioned more poorly and had more difficult personality traits were themselves more psychologically distressed. In addition, these relatives were more likely to view patients' problems as volitional. These findings seem reasonable, in that acceptance or resignation seems less likely to occur under such circumstances than if the relative viewed the patient as ill. However, contrary to our prediction, more distressed relatives were not more hostile or perceived to be more critical of the patient. Indeed, there was a trend for the more distressed relatives to be less hostile. Also contrary to our prediction, relatives' reported beliefs that patients could, with effort, control their symptoms without professional assistance were not associated with higher hostility or criticism. This finding is at odds with the extant literature (Barrowclough et al., 1994; Brewin et al., 1991; Harrison & Dadds, 1992; Hooley & Licht, 1997; Weisman et al., 1993). The present investigation differs from the bulk of prior attribution–EE research in at least two ways. First, unlike relatives in most studies, our participants were relatives of patients with anxiety disorders rather than schizophrenia. OCD and PDA patients may have relatively little immediate control over their fear, but their observable symptoms (avoidance behavior and rituals) are surely more voluntary in nature than many symptoms of schizophrenia. This may make it more difficult for relatives to provide valid ratings of controllability. Second, we used a new self-report measure to assess relatives' beliefs about controllability of symptoms, whereas most other authors coded attributions from relatives' spontaneous expressions during the CFI. Additional data are required on the reliability and validity of this new measure. The wording on the Control of Symptoms Scale asked relatives to rate the degree to which they perceive the patient to have control over each symptom listed. However, relatives are not asked to indicate which behaviors they find particularly distressing. A measure that takes into account which symptoms are particularly bothersome to relatives might be more sensitive in assessing attributions.
Finally, this investigation adds to the emerging literature on relatives' personality variables as predictors of EE (cf. Hooley, 1998; Hooley & Hiller, 2000). Relatives' self-reported hostility on the SCL–90–R predicted both patients' perceptions of relatives' criticism and relatives' objectively coded hostility. Thus, relatives who behave negatively toward the patient appear to be angrier, more aggressive people in general. This finding must be interpreted cautiously in light of the nature of the SCL–90–R Hostility subscale. Because the target of the relative's anger was not specified when the relative completed the rating scales, it is possible that the relative was thinking about his or her behavior toward the patient, rather than anger in general. Also, the SCL–90–R scale is not designed to measure anger as a trait; rather, respondents rate their anger for the last week. Thus, additional research including more extensive assessment of relatives' general hostility and aggression would be desirable.
The relative sample consisted of both spouses and nonspouses. There was some evidence from the first-order correlations within each relative subsample that the relationship of control of symptoms to CFI hostility might differ across spouses versus nonspouses. We found the overall model given in Figure 2 to fit the correlational structure in the spouse sample well. Exploration of the relationship of symptom control to hostility in a nonspouse sample requires further research.
This study represents an important step in testing an overall model for the development and maintenance of EE, with this initial model accounting for a moderate proportion of the variance in hostility. In future research it should be possible to increase the effect size by further attention to relatives' personality traits, such as flexibility, and by increasing the representation of the constructs of the transactional stress and coping model in the data set. In the present study, for example, constructs of coping and relatives' resources were each represented by only one variable. There are multiple methods of coping, including palliative as well as active methods (see, e.g., Thompson & Gustafson, 1996), and a variety of resources, including the social support available to the relative.
Implications for Application and Public Policy
Although the concurrent collection of data in the present study precludes conclusions about the causal relationship of predictors and hostility, the findings provide correlational support for inclusion of at least two elements in intervention programs for OCD and PDA patients and their families. Family interventions designed to reduce EE and the rate of relapse in families of schizophrenic and bipolar patients include a large dose of training in problem solving (e.g., Falloon, 1984; Miklowitz & Goldstein, 1997; Mueser & Glynn, 1995). Our findings suggest that programs for relatives of anxious patients should include similar training. Such an approach with relatives of anxious patients, coupled with education, might be helpful in reducing relatives' anger. Next, the present data suggest that clinicians focus on increasing patients' overall functioning rather than on symptom reduction alone. If functioning does not improve spontaneously with symptom reduction, patients may be at risk for relapse in a climate of continued criticism from their relatives (Butzlaff & Hooley, 1998). Furthermore, relatives' own psychological distress is linked to patients' functioning. Short-term, research-based treatments for anxiety disorders tend to be concentrated on symptom reduction, whereas broad changes in functioning tend to occur in a later stage of treatment, subsequent to symptom reduction (Howard, Lueger, Maling, & Martinovich, 1993). Given the important role EE appears to play in treatment outcome for a number of psychological disorders, continued development of reliable methods for reducing EE is required. Such development would be furthered by additional basic research on the contributing factors in relatives' emotional response to the strain of living with patients with severe psychopathology.
References
Asarnow, J. R., Tompson, M., Hamilton, E. B., Goldstein, M. J., & Guthrie, D. (1994). Family-expressed emotion, childhood-onset depression, and childhood-onset schizophrenia spectrum disorders: Is expressed emotion a nonspecific correlate of child psychopathology or a specific risk factor for depression? Journal of Abnormal Child Psychology, 22, 129–146. [Context Link]
Barrowclough, C., Johnston, M., & Tarrier, N. (1994). Attributions, expressed emotion, and patient relapse: An attributional model of relatives' response to schizophrenic illness. Behavior Therapy, 25, 67–88. Bibliographic Links [Context Link]
Bentler, P. M. (1990). Comparative fit indices in structural models. Psychological Bulletin, 107, 588–606. [Context Link]
Bentler, P. M. (1995). EQS: Structural equations program manual. Encino, CA: Multivariate Statistics. [Context Link]
Bollen, K. A. (1989). Structural equations with latent variables. New York: Wiley. [Context Link]
Brewin, C. R., MacCarthy, B., Duda, K., & Vaughn, C. E. (1991). Attribution and expressed emotion in the relatives of patients with schizophrenia. Journal of Abnormal Psychology, 100, 546–554. Ovid Full Text Bibliographic Links [Context Link]
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. Bollen & K. Long (Eds.), Testing structural equation models (pp. 136–162). Newbury Park, CA: Sage. [Context Link]
Butzlaff, R. L., & Hooley, J. M. (1998). Expressed emotion and psychiatric relapse: A meta-analysis. Archives of General Psychiatry, 55, 547–552. Bibliographic Links [Context Link]
Calvocoressi, L., Lewis, B., Harris, M., Trufan, S. J., Goodman, W. K., McDougle, C. J., & Price, L. H. (1995). Family accommodation in obsessive–compulsive disorder. American Journal of Psychiatry, 152, 441–443. Bibliographic Links [Context Link]
Chakrabarti, S., Kulhara, P., & Verma, S. K. (1993). The pattern of burden in families of neurotic patients. Social Psychiatry and Psychiatric Epidemiology, 28, 172–177. [Context Link]
Chambless, D. L., Bryan, A., Aiken, L. S., Steketee, G. S., & Hooley, J. M. (1999). The structure of expressed emotion: A three-factor representation. Psychological Assessment, 11, 67–76. Ovid Full Text [Context Link]
Chambless, D. L., Fauerbach, J., Floyd, F. J., Wilson, K. A., Remen, A., & Renneberg, B. (1997, November). Behavioral observation of the marital interaction of agoraphobic women: A comparison with community couples. Paper presented at the meeting of the Association for Advancement of Behavior Therapy, Miami Beach, FL. [Context Link]
Chambless, D. L., Gillis, M. M., Tran, G. Q., & Steketee, G. S. (1996). Parental bonding reports of clients with obsessive–compulsive disorder and agoraphobia. Clinical Psychology and Psychotherapy, 3, 77–85. [Context Link]
Chambless, D. L., & Steketee, G. (1999). Expressed emotion and behavior therapy outcome: A prospective study with obsessive–compulsive and agoraphobic outpatients. Journal of Consulting and Clinical Psychology, 67, 658–665. [Context Link]
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. [Context Link]
Cole, R. E., Kane, C. F., Zastowny, T., Grolnick, W., & Lehman, A. (1993). Expressed emotion, communication, and problem solving in the families of chronic schizophrenic young adults. In R. E. Cole & D. Reiss (Eds.), How do families cope with chronic illness? (pp. 141–172). Hillsdale, NJ: Erlbaum. [Context Link]
Crown, S., & Crisp, A. H. (1966). A short clinical diagnostic self-rating scale for psychoneurotic patients: The Middlesex Hospital Questionnaire (M.H.Q.) British Journal of Psychiatry, 112, 917–923. [Context Link]
de Jesus Mari, J., & Streiner, D. L. (1994). An overview of family interventions and relapse on schizophrenia: Meta-analysis of research findings. Psychological Medicine, 24, 565–578. [Context Link]
Derogatis, L. R. (1994). SCL–90–R: Administration, scoring and procedures manual. Minneapolis, MN: National Computer Systems. [Context Link]
Edwards, D. W., Yarvis, R. M., Mueller, D. P., Zingale, H. C., & Wagman, W. J. (1978). Test-taking and stability of adjustment scales. Evaluation Quarterly, 2, 275–291. [Context Link]
Falloon, I. R. H. (1984). Family care of schizophrenia: A problem-solving approach to the treatment of mental illness. New York: Guilford Press. [Context Link]
Goodman, S. H., Adamson, L. B., Riniti, J., & Cole, S. (1994). Mothers' expressed attitudes: Associations with maternal depression and children's self-esteem and psychopathology. Journal of the American Academy of Child and Adolescent Psychiatry, 33, 1265–1274. [Context Link]
Gottschalk, L. A., & Keatinge, C. (1993). Influence of patient caregivers on course of patient illness: “Expressed emotion” and alternative measures. Journal of Clinical Psychology, 49, 898–912. [Context Link]
Hahlweg, K., & Conrad, M. (1983). Kategoriensystem zur Beobachtung Partnerschaftlicher Interaktion [Interactional Coding System]. (Available from Prof. Dr. Kurt Hahlweg, Institut für Psychologie, TU Braunschweig, Speilmannstrasse 12a, 38106, Braunschweig, Germany.) [Context Link]
Hahlweg, K., Nuechterlein, K. H., Goldstein, M. J., Magaña, A., Doane, J. A., & Snyder, K. S. (1987). Parental expressed emotion attitudes and intrafamilial communication behavior. In K. Hahlweg & M. J. Goldstein (Eds.), Understanding major mental disorder: The contribution of family interaction research (pp. 156–174). New York: Family Process Press. [Context Link]
Halford, W. K., Hahlweg, K., & Dunne, M. (1990). The cross-cultural consistency of marital communication associated with marital distress. Journal of Marriage and the Family, 52, 523–532. [Context Link]
Harrison, C. A., & Dadds, M. R. (1992). Attributions of symptomatology: An exploration of family factors associated with expressed emotion. Australian and New Zealand Journal of Psychiatry, 26, 408–416. [Context Link]
Hathaway, S. R., & McKinley, J. C. (1940). A multiphasic personality schedule (Minnesota): I. Construction of the schedule. Journal of Psychology, 10, 249–254. [Context Link]
Hayduk, L. A. (1987). Structural equation modeling with LISREL: Essentials and advances. Baltimore: Johns Hopkins University Press. [Context Link]
Hibbs, E. D., Hamburger, S. D., Lenane, M., Rapoport, J. L., Kruesi, M. J. P., Keyson, C. S., & Goldstein, M. J. (1991). Determinants of expressed emotion in families of disturbed and normal children. Journal of Child Psychology and Psychiatry, 32, 757–770. [Context Link]
Hollingshead, A. B. (1975). Four-Factor Index of Social Status. (Available from A. B. Hollingshead, Dept. of Sociology, Yale University, P.O. Box 1965, New Haven, CT 06520) [Context Link]
Hooley, J. M. (1985). Control of Symptoms Scale. Unpublished measure. Department of Psychology, Harvard University. [Context Link]
Hooley, J. M. (1986a). Expressed emotion and depression: Interactions between patients and high- versus low-expressed-emotion spouses. Journal of Abnormal Psychology, 95, 237–246. Ovid Full Text [Context Link]
Hooley, J. M. (1986b). An introduction to EE measurement and research. In M. J. Goldstein, I. Hand, & K. Hahlweg (Eds.), Treatment of schizophrenia: Family assessment and intervention (pp. 25–34). New York: Springer Verlag. [Context Link]
Hooley, J. M. (1987). The nature and origins of expressed emotion. In K. Hahlweg & M. J. Goldstein (Eds.), Understanding major mental disorder: The contribution of family interaction research (pp. 176–194). New York: Family Process Press. [Context Link]
Hooley, J. M. (1998). Expressed emotion and locus of control. Journal of Nervous and Mental Disease, 186, 374–378. [Context Link]
Hooley, J. M., & Hiller, J. B. (2000). Expressed emotion and personality. Journal of Abnormal Psychology, 109, 40–44. Ovid Full Text Bibliographic Links [Context Link]
Hooley, J. M., & Licht, D. M. (1997). Expressed emotion and causal attributions in the spouses of depressed patients. Journal of Abnormal Psychology, 106, 298–306. [Context Link]
Hooley, J. M., & Richters, J. E. (1995). Expressed emotion: A developmental perspective. In D. Cicchetti & S. L. Toth (Eds.), Emotion, cognition, and representation (Vol. 6, pp. 133–166). Rochester, NY: University of Rochester Press. [Context Link]
Hooley, J. M., Rosen, L. R., & Richters, J. E. (1995). Expressed emotion: Toward clarification of a critical construct. In G. A. Miller (Ed.), The behavioral high-risk paradigm in psychopathology (pp. 88–120). New York: Springer Verlag. [Context Link]
Hooley, J. M., & Teasdale, J. D. (1989). Predictors of relapse in unipolar depressives: Expressed emotion, marital distress, and perceived criticism. Journal of Abnormal Psychology, 98, 229–235. Ovid Full Text [Context Link]
Howard, K. I., Lueger, R. J., Maling, M. S., & Martinovich, Z. (1993). A phase model of psychotherapy outcome: Causal mediation of change. Journal of Consulting and Clinical Psychology, 61, 678–685. [Context Link]
Hu, L., & Bentler, P. M. (1998). Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychological Methods, 3, 424–453. Ovid Full Text [Context Link]
Leff, J., & Vaughn, C. (1985). Expressed emotion in families: Its significance for mental illness. New York: Guilford Press. [Context Link]
Leonard, H. L., Swedo, S. E., Lenane, M. C., Rettew, D. C., Hamburger, S. D., Bartko, J. J., & Rapoport, J. L. (1993). A 2- to 7-year follow-up study of 54 obsessive–compulsive children and adolescents. Archives of General Psychiatry, 50, 429–439. Bibliographic Links [Context Link]
MacCallum, R. C. (1995). Model specification: Procedures, strategies and related issues. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 16–36). Thousand Oaks, CA: Sage. [Context Link]
Miklowitz, D. J., & Goldstein, M. J. (1997). Bipolar disorder: A family-focused approach. New York: Guilford Press. [Context Link]
Mueser, K. T., Bellack, A. S., Wade, J. H., Sayers, S. L., Tierney, A., & Morrison, R. L. (1993). Expressed emotion, social skill, and response to negative affect in schizophrenia. Journal of Abnormal Psychology, 102, 339–35. Ovid Full Text Bibliographic Links [Context Link]
Mueser, K. T., & Glynn, S. M. (1995). Behavioral family therapy for psychiatric disorders. Boston: Allyn & Bacon. [Context Link]
Peter, H., & Hand, I. (1988). Patterns of patient–spouse interaction in agoraphobics: Assessment by Camberwell Family Interview (CFI) and impact on outcome of self-exposure treatment. In I. Hand & H.-U. Wittchen (Eds.), Panic and phobias, 2: Treatments and variables affecting course and outcome (pp. 240–251). Berlin, Germany: Springer Verlag. [Context Link]
Rigdon, E. E. (1996). CFI versus RMSEA: A comparison of two fit indexes for structural equation modeling. Structural Equation Modeling, 3, 369–379. [Context Link]
Riso, L. P., Klein, D. N., Anderson, R. L., Ouimette, P. C., & Lizardi, H. (1996). Convergent and discriminant validity of perceived criticism from spouses and family members. Behavior Therapy, 27, 129–139. Bibliographic Links [Context Link]
Rosenfarb, I. S., Goldstein, M. J., Mintz, J., & Nuechterlein, K. H. (1995). Expressed emotion and subclinical psychopathology observable within the transactions between schizophrenic patients and their family members. Journal of Abnormal Psychology, 104, 259–267. Ovid Full Text Bibliographic Links [Context Link]
Schreiber, J. L., Breier, A., & Pickar, D. (1995). Expressed emotion: Trait or state? British Journal of Psychiatry, 166, 647–649. Bibliographic Links [Context Link]
Spitzer, R. L., Williams, J. B. W., Gibbon, M., & First, M. B. (1989). Structured Clinical Interview for DSM–III–R—Patient Version. New York: Biometrics Research Department, New York State Psychiatric Institute. [Context Link]
Stevens, J. (1996). Applied multivariate statistics for the social sciences (3rd ed.). Mahwah, NJ: Erlbaum. [Context Link]
Strachan, A. M., Feingold, D., Goldstein, M. J., Miklowitz, D. J., & Nuechterlein, K. H. (1989). Is expressed emotion an index of a transactional process? II. Patient's coping style. Family Process, 28, 169–181. [Context Link]
Tarrier, N. (1996). An application of expressed emotion to the study of PTSD: Preliminary findings. Clinical Psychology and Psychotherapy, 3, 220–229. [Context Link]
Tarrier, N., Sommerfield, C., & Pilgrim, H. (1999). Relatives' expressed emotion and PTSD treatment outcome. Psychological Medicine, 29, 801–811. Buy Now [Context Link]
Thompson, R. J., Jr., & Gustafson, K. (1996). Adaptation to chronic childhood illness. Washington, DC: American Psychological Association. [Context Link]
Tompson, M. C., Goldstein, M. J., Lebell, M. B., Mintz, L. I., Marder, S. R., & Mintz, J. (1995). Schizophrenic patients' perceptions of their relatives' attitudes. Psychiatry Research, 57, 155–167. Bibliographic Links [Context Link]
Vaughn, C., & Leff, J. (1976). The measurement of expressed emotion in the families of psychiatric patients. British Journal of Social and Clinical Psychology, 15, 157–165. [Context Link]
Vaughn, C. E., & Leff, J. P. (1981). Patterns of emotional response in relatives of schizophrenic patients. Schizophrenia Bulletin, 7, 43–44. Bibliographic Links [Context Link]
Vostanis, P., Nicholls, J., & Harrington, R. (1994). Maternal expressed emotion in conduct and emotional disorders of childhood. Journal of Child Psychology and Psychiatry, 35, 365–376. [Context Link]
Weisman, A., Lopez, S. R., Karno, M., & Jenkins, J. (1993). An attributional analysis of expressed emotion in Mexican-American families with schizophrenia. Journal of Abnormal Psychology, 102, 601–606. Ovid Full Text Bibliographic Links [Context Link]
Weiss, R. L., Hops, H., & Patterson, G. R. (1973). A framework for conceptualizing marital conflict: A technology for altering it, some data for evaluating it. In L. D. Handy & E. L. Mash (Eds.), Behavior change: Methodology, concepts, and practice (pp. 309–342). Champaign, IL: Research Press. [Context Link]
Weissman, M. M., & Bothwell, S. (1976). Assessment of social adjustment by patient self-report. Archives of General Psychiatry, 33, 1111–1115. [Context Link]
1In the CFI scoring system, criticism is noted whenever relatives indicate, through voice tone or a clear statement, that they dislike something about the patient. Such criticism is not necessarily hostile. Hostility is scored when the relative's criticism becomes global or when the relative indicates dislike of the patient as a person, rather than disapproval of some aspect of the patient's behavior. Thus, all hostile relatives are critical, but not all critical relatives are hostile. [Context Link]
________________________________________

Accession Number: 00012003-200106000-00004
Copyright (c) 2000-2007 Ovid Technologies, Inc.
Version: rel10.5.8, SourceID 1.13281.2.32.1.0.2.197.1.4.1.5