Social Psychiatry and Psychiatric Epidemiology

, Volume 45, Issue 2, pp 165–174

The relationship of case managers’ expressed emotion to clients’ outcomes


    • School of Social Policy and PracticeUniversity of Pennsylvania
  • Leslie Alexander
    • Bryn Mawr College
  • Stacey Uhl
    • ECRI Institute
Original Paper

DOI: 10.1007/s00127-009-0051-3

Cite this article as:
Solomon, P., Alexander, L. & Uhl, S. Soc Psychiat Epidemiol (2010) 45: 165. doi:10.1007/s00127-009-0051-3



Expressed emotion (EE) has been studied in families of a relative with schizophrenia as well as other psychiatric disorders; and high EE (hostile, critical, and overinvolved) families have been found to be strongly related to relapse among their relatives. EE has been assessed on a limited basis among non-familial care providers and determined that providers can also have high EE which results in poor quality of life and negative consequences for their clients.


The present study assessed 42 case managers serving clients with schizophrenia spectrum disorder regarding their EE for specific clients enrolled in a larger study examining the reliability and validity of two alliance measures. Case managers and clients were personally interviewed at baseline, 3, 6, 6 plus 2 weeks, and 9 months post-client entry into case management. The EE measure was inserted into the 6 months plus 2 week case manager interview. Generalized Estimating Equation analysis was employed to examine predicted outcomes of EE.


High EE was found to be related to client attitudes toward medication compliance and social contact.


Family psychoeducation interventions, an evidence-based practice, have been demonstrated to be effective in reducing relapse of relatives with serious mental illness. Given the clinical evidence that EE is modifiable, it is expected that such educational training for non-familial caregivers will have the same potential as for family caregivers. Providers dealing with challenging clients may also need support and skills to better handle difficult situations, especially direct support providers like case managers who are not clinically trained.


Case managementCase managerExpressed emotionMedication adherence


In the United States, family psychoeducation interventions are currently considered one of five evidence-based psychosocial interventions for persons with severe mental illness, particularly those with schizophrenia. The development of family psychoeducation interventions was based on earlier research that found that patients released from hospitals to families who expressed a high degree of hostility, criticism, and emotional overinvolvement were more likely to have an exacerbation of their symptoms and to relapse and be rehospitalized [35]. These interventions were designed to reduce family environmental stress by educating families about the illness, ways to communicate more effectively with the patient, to improve problem solving skills, to cope better with the illness and hence to reduce rehospitalizations. These factors of criticism, hostility, and emotional overinvolvement were termed “expressed emotion” or “EE” and those families with a high level of these characteristics were identified as high expressed emotion (EE) families.

Extensive research on EE has well established this phenomenon as a reliable and robust predictor of relapse in a variety of disorders besides schizophrenia, including mood, eating, alcoholism, and depression as well as physical illnesses [6, 52]. However, there continues to be a lack of understanding of the mechanisms and processes that result in this consistent association between EE and relapse. Research has provided some evidence of EE being a reflection of transactional behavioral patterns between the patient and family coping style, suggesting a bidirectional relationship [42]. Stanhope and Solomon [41] noted that EE research among providers gives more support for EE being a transactional phenomenon.

Since most persons with severe mental illness now spend the majority of their time in the community often living in settings other than with families, research on EE has moved into formal care provision, most often in non-familial residential facilities. Although the experiences and roles of non-familial and family caregivers differ, staff who work closely with persons with severe mental illness experience some of the same feelings as families and engage in behaviors characteristics of EE when frustrated with their lack of success with clients [12, 35]. Furthermore, staff have to manage problematic behaviors of their clients, including symptomatic behaviors, verbal abuse, as well as aggressive and acting out behaviors, as do family members [46]. Given that case managers are currently primary providers who work with persons with severe mental illness in the public mental health system in the United States, we considered it important to examine EE among case managers in relation to client functioning in the community and psychiatric hospitalizations. Specifically, we hypothesized EE as a predictor of hospitalizations, client treatment participation, satisfaction with case management, attitudes towards medication, social contact, and quality of life, when controlling for client symptomatology and alliance, and case manager’s job tenure. This analysis was exploratory work that was included as part of a larger study that assessed the psychometric properties of two different measures of therapeutic alliance, which were adapted for use with both case managers and clients.

The few EE studies that have been conducted on non-family caregivers have been limited in scope, used very small samples, and often nonstandardized measures to assess EE. However, these studies generally suggest that non-familial caregiver EE is a promising direction for research, particularly in light of the increased reliance on community caregivers for persons with severe mental illness [17, 24]. Given the infancy of this domain of research and the negative consequences on client functioning and quality of life of high EE in client–professional dyads, there is a need for further research [46], particularly, in non-residential lower intensity case management service as generally provided in the public mental health system in the US.

There are many ways that the experience and roles of non-familial and family caregivers differ, yet staff who work closely with clients with severe mental illness over long periods of time also experience some of the same negative attitudes as family members about these clients. If non-familial caregivers have minimal training, they may bring “a similar level of expertise to their work as relatives who learn as a result of (sometimes bitter) experience what is helpful to their relative and what is not” [24]. Drake and Osher [12] (p. 276) point out that non-familial caregivers “often feel responsible, just like family members, when exacerbations of illness occur. They frequently deny illness, misconstrue symptoms as willful misbehavior, and stress patients unnecessarily with emotionally intense interactions or unrealistically high expectations. Like family members, they can easily become hostile, critical, or overinvolved when their desires to help are frustrated”.

Pilot studies, all of them carried out in Europe, have reported high EE levels among a range of different staff who work with persons with severe mental illness, including nurses [31]; day hospital staff [30]; and non-professional staff in sheltered living facilities [47] inpatient settings, particularly high security units [2], and hostels [1]. Critical environments in these settings were not only associated with a poorer quality of life for residents [38], but also with more clients leaving hostel settings for negative reasons [1] and resulting in more behavioral disturbances [2]. Snyder et al. [38] found that the emotional climate of a client’s immediate interpersonal environment—in this case, residential care operators—was related to the residents’ quality of life and symptoms at 1-year follow-up. EE was also related to client satisfaction with their living situation, satisfaction with personal finances, and feelings of personal safety [38]. Contrary to findings of family EE research, patient rejection by providers was not related to relapse of board and care clients, but was related to negative behaviors such as hostility, deteriorated appearance, and disorganized thinking [7].

Siol and Stark [37] found that experienced therapists of patients with severe mental illness, queried after an acute psychotic episode had subsided, were as likely to report high EE attitudes as relatives, but unlike relatives, were less inclined to be emotionally overinvolved. Other EE provider research has found less overinvolvement than with families [32, 38]. Furthermore, patients responded differently to parents than therapists with high EE. Parents with high EE were rated as inscrutable and demanding by patients, while therapists with high EE were rated as no more or less demanding than therapists with low EE.

Moore et al. [32] noted that provider EE was not independent of patient symptoms and that workers tended to report higher EE with those patients who were violent, argumentative, and irritable. A recent review of provider EE concluded that high EE seems to be consistently associated with poorer client functioning rather than increased symptoms and to have a negative impact on client functioning and quality of life [45]. With regard to characteristics of professional caregivers, the reviewers concluded that providers who were flexible and interested regarding new situations tended to have lower EE. Also, similar to family EE research, providers who perceived symptomatic behaviors as being controllable and internal to the client precipitated more criticism and hostility, while those who attributed these problematic behaviors to being uncontrollable or external to the client were associated with lower EE [46, 53]. Moore et al. [31] found that those high in EE not only made more negative comments, but also fewer supportive comments. This is consistent with other research where high EE providers had lower educational levels and were less open to new situations [48]. However, provider characteristics, including stress, burnout, and a variety of coping mechanisms were not related to the level of EE [4850]. Another study found that high EE was related to the depersonalization dimension of burnout [10].

To date, there have been two studies of EE and case managers. Oliver and Kuipers [33] assessed case managers’ EE as part of an RCT of intensive clinical case management. Thirty-nine percent of the dyads or 70% of the case managers were rated as having high EE. A strong positive association was found between symptomatology and the number of critical comments made by case managers. The researchers were unable to determine whether difficult clients led to high EE or whether those providers who were more critical impacted the mental status of their clients. More recently, Tattan and Tarrier [44] (p. 202) found that among 18 case managers serving 120 clients involved in a large randomized trial of varying intensities of case management, 27% were rated as high in EE. Their analysis suggested that high EE was characteristic of case managers’ response style rather than a response to a specific client. Furthermore, EE was not found to be related to clinical outcomes, but rather the quality of the relationship was associated with positive clinical outcomes. The critical factor seemed to be the “absence of a positive attitude” of case managers, since a positive attitude may be indicative of not believing that aversive client behaviors are volitional and therefore not require coercive responses. The researchers also noted that negative attitudes may have little impact due to the limited contact between case managers and clients which ranged from 2–4 per month totaling 42 min.


The present exploratory analysis was part of a larger study that examined the reliability and validity of two alliance measures—the WAI (Horvath, 1983, 1989) and the HA-q-II (Luborsky et al., 1993)—during the clients first 9 months in intensive case management. Between December, 1994 and July 1997, consecutive dyads of case managers and their new clients were recruited from 20 intensive case management sites, in four suburban counties surrounding Philadelphia and five additional counties within 100 miles of Philadelphia. Inclusion criteria for study clients were: (1) a clinical DSM-IV diagnosis of schizophrenia or schizoaffective disorder; (2) at least 18 years of age; and (3) the client was just referred to ICM services. Exclusion criteria included a history of violent behavior and recent homelessness. The ICM supervisor made the assessment regarding these criteria; exclusion criteria were for client and interviewer protection. When a new client was referred to the agency, if the study criteria were met, the ICM supervisor first asked the case manager if he/she was willing to participate. If the answer was affirmative, the client was then asked if he/she wanted to participate. If both members of the dyad agreed, a study interviewer was assigned and an initial interview was scheduled with the client. Informed consent was obtained from the client prior to the baseline interview being conducted in a mutually agreed upon location. Upon completion of the client baseline interview, an interview was scheduled with the case manager and consent was then obtained from the case manager prior to conducting the baseline interview. The study was approved by the institutional review board of the second author’s college. All nine study interviewers had Master’s degrees (7 in clinical social work, 1 each in nursing and psychology) and extensive clinical experience with persons with severe mental illnesses.

At the time in Pennsylvania, adult ICM was a county-based, intensive broker model, providing 24-h, 7 days per week coverage, with caseloads no larger than 30 (generally much lower), with services delivered in vivo, and a state mandate of a minimum of one contact every 2 weeks. Intensive case management functions included assessment, planning, linking, monitoring, accessing services, building support networks, using community resources, problem solving, and advocacy.

Each client and case manager had five in-person interviews: at baseline (within 6 weeks of the client’s enrollment in ICM); and then after 3, 6 months, 6 months plus 2 weeks, and 9 months post-client entry into ICM. Each client interview took between 1 and 1.5 h, at a location mutually convenient for client and interviewer. Each case manager interview took between 30–45 min, and usually was held in the case manager’s office. The EE measure was intruded into the case manager interview at 6 months plus 2 weeks which involved a less extensive battery of instruments than the other four interviews in the series, therefore enabling us to include additional measures without increasing worker burden.


For the present analysis, we used the following variables in addition to the EE measure: from the client interviews: Working Alliance Inventory (WAI, [21, 22] (this scale was modified in consultation with Adam Horvath for use in this study; basic modification of terms was used: “case manager” instead of “therapist, “consumer” rather than “client”, and “case management” instead of “therapy”). Client WAI scores were used as a control variable in the GEE regression. The WAI is a 36-item self-report alliance measure which uses a 7-point Likert rating scale to assess three primary components of the client–worker relationship: (a) goals and (b) tasks of treatment, comprising the collaborative components of the relationship and (c) bonds, which reflect a complex network of personal attachments between client and worker, including mutual acceptance, liking, trust, and confidence. There are parallel forms for both workers and clients, though we only employed the client’s alliance in the present analysis. This measure was selected as it has been used in other case management studies [40].

In addition, treatment participation and job tenure were assessed from the worker’s interview and from the client interviews: clients’ working alliance, attitudes toward medication compliance, satisfaction with case management, social contact, and quality of life measures were assessed. Each of these measures, except for worker job tenure, was assessed via interviews at 3, 6, and 9 months. Data on the number of hospitalizations during the 9-month period was drawn from agency case records.

Case manager interview variables

In this study, EE was not assessed by the most frequently used, existing measures of EE—the Camberwell Family Interview (CFI, [51] or the 5-min speech method sample (FMSS, [28]. The CFI takes between 1.5 and 2 h to complete and requires extensive training to code. The FMSS measure has a high false negative rate of detecting EE (e.g., overidentifies those scoring low) when compared to the CFI. The FMSS requires tape-recording and extensive training for scoring, and like the CFI, is limited in practical application in general community settings, such as case management. There is also a 60-item true/false, self-report inventory, the Level of Expressed Emotion Scale (LEE, [8], which has undergone some psychometric assessment [15]. Hooley and Parker [20] indicate the need for further assessment of the psychometric properties of this measure, particularly, its validity.

We used Hooley and Hiller [18] scale, the Expressed Emotion Screening Scale (EESS) consisting of 54 items from the California Psychological Inventory (CPI, [16], a personality inventory, which includes items such as “There is something wrong with a person who can’t take orders without getting angry or resentful”. The EESS has been found to discriminate between high and low levels of EE at the P < 0.05 level or higher, as measured by the CFI. The normative sample for the EESS consisted of 45 relatives of 32 psychiatric inpatients, diagnosed with schizophrenia or schizoaffective, and psychotic disorders not otherwise specified; all were participants in the Harvard Patients and Families study, a longitudinal investigation of EE and relapse [19].

In our study, we assessed case manager’s level of EE rather than parental level of EE, for which the scale was originally developed. The alpha reliability for this scale was 0.88 in the Harvard study, with a cutting score ranging between 19 and 22, which correctly classified 98% of the relatives as either high or low EE. In other words, a cutoff score of 19, 20, 21, or 22 produced equally valid discrimination. The average phi co-efficient associated with this range of cutting scores was 0.93. Because this finding has not yet been cross-validated on other samples, Hooley and Hiller (personal conversation, April 13, 1998) did not endorse a universal, optimal cut score for all populations. Hence, we treated it as a continuous variable in the present analyses. Alpha reliability for the scale in this study was 0.70. The EESS scale views EE as more of a disposition or trait, than a situational response, which is the way we used it in this investigation. We were therefore able to examine 3-, 6-, and 9-month client outcomes.

Treatment participation was a single Likert item from full participation to no participation, developed by McGurrin and Worley [29]. Job tenure at the particular community mental health center where the case manager was currently employed was also a single-item scale, where the case manager indicated on a Likert format from 1 = less than 6 months, 2 = less than 1 year; 3 = 1–3 years; 4 = 4–5 years to 5 = more than 5 years. As mentioned earlier, although the worker WAI, whose items parallel the client WAI, was assessed at 3, 6, and 9 months, these scores were not used in the GEE regression.

Client interview variables

Attitudes Towards Medication is a scale which assesses the extent of clients’ agreement with 10 statements regarding psychiatric medication [11, 43]. Sample items are “taking psychiatric medication is a necessary part of your rehabilitation,” “you think your psychiatric medication helps control your symptoms,” and “you fear you will become too dependent on your psychiatric medication.” The measure has documented construct validity in explaining both self-report of compliance and third party report of non-compliance [11]. In this study, Chronbach’s alpha internal consistencies for the 3-, 6-, and 9-month interview points were 0.73, 0.82 and 0.83, respectively.

The measure of symptomatology was the Brief Psychiatric Rating Scale [34] which is a widely used psychometrically sound measure. The objective and subjective quality of life variables were drawn from Lehman [27], using the summary scores from the Delighted–Terrible scale for each of the subjective and objective measures of quality of life. His measure was specifically developed for the population of adults with severe mental illness, has good psychometric properties, and is extensively used in mental health services research. Satisfaction with Intensive Case Management was adapted from Hoult et al. [23] by Solomon and Draine [39]. Objective social contacts used the following scale from Lehman’s Quality of Life interview [27]: In the last 3 months, about how often did you (client) do the following activities (e.g., visit, telephone, write a letter) with a friend, rating on a Likert format from 1 = about daily; 2 = about weekly; 3 = about monthly, 4 = Less than monthly to 5 = never.


We used the generalized estimating equation of Zeger and Liang [55] to examine the relationship between case managers’ level of EE and client outcomes. The GEE approach is a general method for fitting regression models to data involving repeated measurements on the same subject. This method accounts for within-subject correlations, which if left unaccounted for, can produce standard errors that are underestimates of the true standard errors. By using SAS PROC GENMOD, the GEE regression analyses in this study produced robust standard errors.


Characteristics of case managers

EE data were collected on 42 of the 49 case managers who participated in the study. The instrument was administered after the 6 month plus 2-week assessment point, with 7 case managers no longer in the study. If either the case manager or client dropped out, the dyad was eliminated from further data gathering. Only one case manager completed more than one EE measure, i.e., two were completed. The average level of EE for the case managers was 20.6 (SD = 5.9) with a range of 9–33. Using the cutoff scores of Hooley, just under half (48%) of the case managers had low EE, with scores ranging from 9 to 20, and over half (52%) had scores between 21 and 33. The client WAI scores, used as a control in the GEE analysis were: at 3 months, M = 195.9 (SD = 37.88); at 6 months, M = 195.29 (SD = 36.31); and at 9 months, M = 194.33 (SD = 40.68). Fifty-two percent (22) were male, with a mean age of 32.6 (SD = 10.2) and a range of 22–63; all but one were Caucasian, and most (83%) had a bachelor’s degree. About one-third (34.7%) had spent less than a year, about one quarter (26.5%) 1–3 years, and over a third (38.8%) spent 3 or more years working in their current job position. About two-thirds had caseloads which consisted almost entirely of clients with schizophrenia. Average caseload size across all agencies was 18. The majority of study case managers carried only one study case.

Characteristics of case management clients

The 60 clients served by these 42 case managers were 56% male (n = 33). Eighty-one percent were white, 12% were African American, 2 Hispanic, 1 Asian, and 1 unspecified. The median age was 37, with a range from 19 to 60. Two-thirds (66%) were high school graduates (n = 39), with 24% having some higher education. Thirty-one percent spent time in jail in their lifetime. Fifty-two percent had a DSM-IV diagnosis of schizophrenia, with the remainder having a diagnosis of schizoaffective disorder. To provide some understanding of the generally low-functioning of this client group, the mean total baseline score for the BPRS was 45.03(SD = 11.3) and 43.4 (SD = 11.7) at 9 months. Overall, clients had been hospitalized an average of 9.3 (SD = 8.3) times in their lives with a range of 1–37, and their mean age at the time of first hospitalization was 23.6 (SD = 8.9).

Descriptive findings

Means and standard deviations for each study outcome variable are shown in Table 1. At 6 months correlations of EE with client WAI (0.04, P = 0.798), BPRS (−0.286, P = 0.08) and baseline job tenure (−0.02, P = 0.894) were not associated with EE.
Table 1

Mean scores and standard deviations for consumer outcome measures



3 months

6 months

9 months

Mean (SD)


Mean (SD)


Mean (SD)


Mean (SD)


Outcome measure

Attitude toward medication compliance

31.1 (4.90)


31.0 (4.20)


31.1 (5.61)


30.6 (5.21)


Objective social contact

3.47 (0.89)


3.49 (0.98)


3.62 (0.91)


3.78 (0.83)


QOL: overall life

4.60 (1.73)


4.35 (1.58)


4.20 (1.72)


4.54 (1.72)


QOL: living arrangements

4.56 (1.37)


4.44 (1.48)


4.71 (1.19)


4.50 (1.64)


QOL: family relations

4.01 (1.65)


4.30 (1.62)


4.41 (1.63)


4.26 (1.71)


QOL: social relations

4.57 (1.25)


4.48 (1.54)


4.55 (1.16)


4.63 (1.31)


QOL: finances

3.52 (1.56)


3.95 (1.31)


3.77 (1.44)


3.96 (1.48)



1.31 (0.67)


1.31 (0.67)


3.01 (0.59)


Treatment participationa

1.56 (0.84)


1.93 (1.11)


2.14 (1.59)


2.02 (1.27)


Total Sample









aCompleted by case manager


The EESS correlations with days hospitalized between baseline and 3 months, 3–6 months, and 6–9 months were 0.03, 0.04, and −0.01, respectively, and were not significant; and number of hospitalizations between baseline to 3 months, 3–6 months, and 6–9 months were 0.10, −0.01, and 0.03, respectively; none were significant. Given that the correlations were extremely low and non significant, hospitalizations were not used in any further analyses.

Parameter estimates, standard errors, and confidence intervals from the GEE models that examined the relationship between case managers’ level of EE and clients’ treatment participation, satisfaction with case management, attitudes toward medication, subjective quality of life, and number of social contacts are presented in Table 2. After controlling for clients’ symptom level, clients’ alliance and worker’s job tenure, we found that case managers’ level of EE had a negative relationship with clients’ attitudes toward medication compliance, a positive relationship with clients’ number of social contacts, and no relationship with treatment participation, subjective and objective quality of life measures, and satisfaction with case management. Thus, higher levels of EE were associated with less positive attitudes toward medication treatment (co-efficient of −0.2397, P < 0.05). Conversely, higher levels of EE was associated with greater social contact (co-efficient of 0.0432, P < 0.05).
Table 2

The effect of case manager expressed emotion rating on consumer outcomes results of GEE regression analysis for consumers in ICM

Outcome variable

Explanatory variable



Lower 95% CI

Upper 95% CI

Treatment participationc

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





Satisfaction with case managementd

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





Medication attitudese

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





Social contact

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





QOL overallf

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





QOL family relations

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





QOL social relations

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





QOL finances

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





QOL living arrangements

Expressed emotion score





Control variables

Client BPRS





Tenure on the job





WAISUM at 6 months





WAISUM at 9 months





Overall WAISUM





CI confidence interval, BPRS Brief Symptom Rating Scale, WAISUM Working Alliance Inventory summary score

* P < 0.05; **P < 0.01; ***P < 0.001

aUnstandardized beta co-efficients

bRobust standard errors

c1 = full participation; 2 = participation; 3 = moderate participation; 4 = marginal participation; 5 = low participation; 6 = very low participation; 7 = non-participation. Therefore, a negative co-efficient indicated greater treatment participation

dClients were asked about 12 services that they may receive in case management. Scale: 0 = not provided; 1 = not helpful; 2 = a little helpful; 3 = helpful; 4 = very helpful

eScale: 1 = strongly agree; 2 = agree; 3 = disagree; 4 = strongly disagree

fResponses for the Quality of Life Scale: 1 = terrible; 2 = unhappy; 3 = mostly dissatisfied; 4 = mixed; 5 = mostly satisfied; 6 = pleased; 7 = delighted


This exploratory study provides evidence that case managers, like other non-familial caretakers previously researched, can have high EE attitudes towards clients with severe mental illness. In the present study, the average score fell in the high EE family cutoffs in Hooley and Hiller [18] study of EE in families. While there are certainly some similarities in roles and experiences for both of these groups of caregivers of clients with severe mental illness, non-familial caregivers have not experienced the upheaval in finances and family life over extended periods of time, nor have they experienced the myriad of emotional reactions, such as feelings of loss, guilt, embarrassment, anger, and depression that living with a relative with severe mental illness may engender [24]. This may be a reason why our model did not predict quite as well as Hooley and Hiller [18] model, where only family caretakers were assessed. However, like family caretakers, providers may too feel unappreciated by clients [46].

This exploratory analysis did find that EE predicted clients’ attitudes about medication compliance. This is a proximal measure of relapse, given that non-compliance frequently results in symptom exacerbation. Prior research has found that quality of relationship between clients and prescribers affects attitudes of clients toward antipsychotic medication [9], a phenomenon that may well spill over to case managers who frequently enforce medication adherence. High EE predicted social contact, but this finding was anomalous to what was hypothesized. However, the degree of association was extremely small and not of clinical significance. Basically, the study findings are somewhat consistent with prior research on professional EE in that high EE results in negative influences on clients, such as is indicative of poor medication compliance. Most provider studies, as in our study, did not find support for EE being related to hospitalizations. However, quality of life outcomes were not supported in this study, although found to be the case in other studies [46]. Some of the explanation of the findings may result from the EE measure employed in the present study, which has not previously been used in professional EE research and only on a very limited basis in family EE research. Further investigations that examine EE among case managers as a state as opposed to a trait, as the specific measure used in the present study, is needed before concluding that EE among field-based providers may function differently than within a familial context. Further exploration of whether low worker EE and client perception of low worker EE predict the same or different client outcomes is also important. Positive supportive relational environments are clearly needed for this population to maximize positive adjustments and improve their quality of life. Research has found that low EE providers are more able to “control their own feelings and are warm and able to motivate clients…maintain a good balance between over- and under-stimulating clients…restore hope in clients [46] (p. 233).

As Hansen et al. [17] indicated, at the same time that non-familial caregivers have become more prominent in the provision of care in the community for clients with severe mental illness, there has been amazingly little research on the characteristics of community caregivers and what effect caregiver attitudes and behaviors may have on client–worker interactions and on client outcomes. Since the impact of these kinds of stressors may well have negative consequences for clients, interventions for workers may be essential to avoid perpetuating detrimental environmental contexts. Family psychoeducation has been demonstrated to be effective for families of severely mentally ill individuals in improving family life, decreasing burden, improving patient medication compliance, and reducing relapse of their ill relative [36]. Given this clinical evidence that EE is modifiable, we can expect that similar educational training for non-familial caregivers will have the same potential as for family caregivers. Like families, providers dealing with challenging clients may well need support and skills to better handle difficult situations; especially direct support providers who are not clinically trained, as was the case in present study, with most only having a bachelor’s degree.

Based on pilot work of training nurses on family work that indicated some improvement in the knowledge and attitudes toward schizophrenia and families [14, 25], an intervention specifically designed to reduce EE in staff–patient relationships was tested [54]. While the results of the intervention did not change levels of EE, the investigators did find that there was an increase in the use of strategies that were likely to affect client change. For example, there was a shift from staff only expressing their own perspective to staff eliciting and considering the client’s perspective, thus decreasing the likelihood of leading to criticisms of the client. This approach of paying attention to the client perspective is consistent with a recovery orientation that is being promoted by the US federal government. Van Humbeck and Van Audenhove [46] also noted that these interventions were not sufficiently designed to meet the needs of providers, but rather require more tailoring for providers. Further development of such EE interventions along with teaching recovery principles and skills is quite timely. Implementing a recovery orientation requires a collaborative partnership between providers and clients with clients taking the lead in setting their own goals. Since directionality of the EE relationship is unknown, psychoeducational training for both providers and clients may be a more productive approach.

Given that the study was limited in its sample size, further research is needed with larger samples. Some of the outcome measures, such as treatment participation, have limited psychometric properties. Also, the EE measure employed in the present may not have been the most ideal as it assumes that EE is a trait as opposed to a state which may not be the situation for providers, therefore, assuming that high EE of providers affects client outcomes. However, the evidence of EE research is that the process is transactional and consequently the directionality of the causal relationship is questionable. Future research on EE among providers may employ the 5-min speech sample (FMSS) as a measure, given its wide use and its brevity. But, if time permits the CFI should be employed. As Van Humbeeck et al. [49, 50] in their review of three EE instruments concluded: CFI “remains the best instrument for assessing EE in a therapeutic relationship (between a professional and a client)” (p. 237). They warn that other measures, if needed for efficiency of time and resources, should be used with caution.


This research was funded by NIMH Grant R03MH52734-02 to Leslie B. Alexander.

Copyright information

© Springer-Verlag 2009