Compassion Fatigue and Psychological Distress Among Social Workers: A Validation Study

Abstract

Few studies have focused on caring professionals and their emotional exhaustion from working with traumatized clients, referred to as compassion fatigue (CF). The present study had 2 goals: (a) to assess the psychometric properties of a CF scale, and (b) to examine the scale's predictive validity in a multivariate model. The data came from a survey of social workers living in New York City following the September 11, 2001, terrorist attacks on the World Trade Center. Factor analyses indicated that the CF scale measured multiple dimensions. After overlapping items were eliminated, the scale measured 2 key underlying dimensions—secondary trauma and job burnout. In a multivariate model, these dimensions were related to psychological distress, even after other risk factors were controlled. The authors discuss the results in light of increasing the ability of professional caregivers to meet the emotional needs of their clients within a stressful environment without experiencing CF.

Although the psychological consequences of providing social support and care to traumatized individuals have been noted for over 2 decades, relatively few studies have focused on formal caregivers (i.e., therapists, child protection workers, nurses, etc.) and their emotional response to dealing with traumatized clients (Figley, 1995). Studies have shown that providing such care can be both highly rewarding and highly stressful (Ohaeri, 2003). Individuals working in the caring professions, though, may have occupational environments and caregiving demands that increase the likelihood of adverse psychological outcomes (Figley, 2002a; Sabin-Farrell & Turpin, 2003).

Theoretically, individuals working in the caring professions often attempt to alter the behaviors and emotions of their clients by providing emotional support (e.g., empathy), strategies for coping with emotions, or better cognitive management skills (Boscarino, 1997; Francis, 1997; Thoits, 1986). Within the context of formal caregiving, providing therapy to clients who have survived a traumatic event can be particularly stressful (Figley, 1995). Many researchers have indicated that therapists who work with traumatized clients often show signs of psychological distress as a result of these interactions (Figley, 1995; Nelson-Gardell & Harris, 2003; Schauben & Frazier, 1995). The adverse impact of working with clients who have a history of psychological trauma (e.g., sexual and physical abuse, military combat, or community disaster) has been described under a variety of terms: vicarious traumatization, secondary traumatic stress, and compassion fatigue (CF; Jenkins & Baird, 2002).

Compassion Fatigue (CF)

For the present study, we use the term CF. Consistent with most current usage, we defined this as the formal caregiver's reduced capacity or interest in being empathic or “bearing the suffering of clients” and is “the natural consequent behaviors and emotions resulting from knowing about a traumatizing event experienced or suffered by a person” (Figley, 1995, p. 7; see also Figley 2002a, 2002b). Thus, CF is a hazard associated primarily with the clinical setting and with first responders to traumatic events.

With the inclusion of posttraumatic stress disorder in the third edition of the Diagnostic and Statistical Manual (American Psychiatric Association, 1980), common symptoms related to severe psychological trauma were defined as a psychiatric disorder. Figley (1995) developed the concept of CF when he began to focus on the unique work environment of trauma workers and mental health professionals and how they appeared to vicariously experience the effects of trauma. In particular, CF appeared to be the consequence of working with traumatized individuals, if the professional was exposed to significant numbers of them and had a strong empathic orientation (Figley, 1995). Empathic engagement with traumatized clients often requires the professional to discuss details of the traumatic experience, including role playing and dramatic reenactment of the events, which are thought to be vital to the therapeutic process but can have an adverse emotional impact on the caregiver (Figley 2002a, 2002b). This type of vicarious trauma often is referred to as secondary traumatic stress in the literature (Boscarino, Figley, & Adams, 2004). Thus, when therapists, doctors, nurses, or child protection workers report symptoms related to reexperiencing the client's traumatic event, wishing to avoid both the client and reminders of the client's trauma, and feeling persistent arousal due to intimate knowledge about the client's traumatic experiences, they are likely suffering from secondary trauma (Figley, 1995, 2002b; Jenkins & Baird, 2002; Schauben & Frazier, 1995).

More recently, Figley (1995, 2002a, 2002b) and others (e.g., Gentry, Baranowsky, & Dunning, 2002; Jenkins & Baird, 2002; Nelson-Gardell & Harris, 2003; Salston & Figley, 2003; Stamm, 2002) have observed that secondary trauma and what has been termed job burnout overlap, in that both are characterized by the emotionally exhausting nature of working with survivors of trauma. Figley (2002b) has suggested, however, that secondary trauma is not the same as burnout syndrome and that each should be treated as having a unique effect on a professional's well-being (see also Jenkins & Baird, 2002; Sabin-Farrell & Turpin, 2003; Salston & Figley, 2003). Burnout syndrome is often defined as a response to prolonged exposure to demanding interpersonal situations and is characterized by emotional exhaustion, depersonalization, and reduced personal accomplishment (Maslach, Schaufeli, & Leiter, 2001). High emotional involvement without adequate social support or feelings of personal work accomplishments (i.e., job satisfaction) may leave the caring professional vulnerable to burnout. Thus, we suggest that both secondary trauma and job burnout are likely central and critical clinical features of CF. In this study, we assess the extent to which secondary trauma and job burnout are related to and independent of one another.

In their review of the literature on CF, Sabin-Farrell and Turpin (2003) suggested several possible psychological and psychoanalytic mechanisms (e.g., countertransference, emotional contagion) by which working with clients may result in CF (see also Salston & Figley, 2003). In this study, we conceptualize CF within a stress process framework (Pearlin, 1989; Thoits, 1995). This framework contends that challenging environments (stressors) typically require individuals to respond both physiologically, through alterations in the neuroendocrine and hormonal systems (Boscarino, 1997), and psychologically, usually through alterations in cognitive functioning (Francis, 1997; Thoits, 1995).

Both stress (Thoits, 1995) and CF researchers (Figley, 1995, 2002b; Kassam-Adams, 1999) have also suggested that other aspects of the formal caregiver's life can influence his or her likelihood of developing CF. For example, a history of trauma, lower social support, and an inability to cope with the demands of caregiving are thought to increase the likelihood of developing CF. Previous research has tended to support the hypothesized relationships between these risk factors, CF, and psychological distress (Figley, 1995, 2002a; Nelson-Gardell & Harris, 2003; Sabin-Farrell & Turpin, 2003; Salston & Figley, 2003; Schauben & Frazier, 1995). Schauben and Frazier (1995), for example, found that female psychologists and violence counselors with a higher percentage of sexual violence victims as clients reported more vicarious trauma and that this was related to greater psychological problems. A history of personal trauma is also related to poor psychological health among child welfare workers (Nelson-Gardell & Harris, 2003) and psychotherapists (Kassam-Adams, 1999).

Despite these findings, research on CF has had several problems. First, there has been a lack of conceptual clarity about what constitutes CF and how it differs from other adverse work outcomes, such as job burnout (Jenkins & Baird, 2002). There are also a number of CF scales, with many dissimilar items (e.g., Figley, 1995; Gentry et al., 2002; Stamm, 2002). Finally, no study fully incorporates all aspects of Figley's (1995, 2002b) description of CF or key variables in the stress process model.

To address these gaps in previous research, we focused on social workers in clinical practice in a region recently affected by a major traumatic event—the September 11, 2001, terrorist attacks in New York City. We were interested in this caregiving profession because the work environment of social workers is often characterized by high case loads and inadequate resources (Duffy et al., 2003). In addition, the delivery of mental health care services is increasingly being performed by social workers (Mechanic, 1999).

Method

The data for this study are from a survey of social workers living in New York City. The sampling frame included all social workers with a master's degree in social work or higher who were current members of the National Association of Social Workers (NASW). The NASW is the main national social work organization, with approximately 50% of all practicing social workers as members. From the membership list, we randomly selected 600 individuals to be sent a mailed questionnaire between May 12 and May 15, 2003. We mailed a second questionnaire 2 weeks later and a follow-up letter 2 weeks after the second mailing, reminding the person to return the survey. We accepted returned surveys until August 31, 2003. Because we were mainly concerned with social workers who potentially suffered from CF, we asked those who were not engaged in direct practice to return the survey indicating that they were not involved in clinical practice. We eliminated these surveys from our study. Overall, 236 social workers returned completed surveys, and 38 returned surveys indicating that they were not providing services. All of the addresses appeared to be correct, as none of the questionnaires/letters from the three mailings was returned by the post office. Thus, the overall survey completion rate (returned surveys [274]/all surveys sent [600]) was 46%. The Institutional Review Board for the New York Academy of Medicine reviewed and approved the study's protocols.

Dependent Variables

Our study focused on two conceptual domains: CF and psychological distress. We measured CF using the 30-item Compassion Fatigue (CF) Scale—Revised (Gentry et al., 2002). This scale was developed by Figley (1995) on the basis of clinical experience, and versions of the scale have been used in several previous studies (Jenkins & Baird, 2002; Stamm, 2002). The scale was designed to assess both vicarious trauma and job burnout. The present study, however, was the first to assess the basic psychometric properties of this scale. The survey asked respondents to consider each scale item and indicate how closely it currently reflects their experience, using a 10-point, visual analog–type Likert scale (rarely/never = 1 to very often = 10).

We assessed general psychological distress using the 12-item version of the General Health Questionnaire (GHQ-12; Goldberg & Huxley, 1992; McDowell & Newell, 1996). This scale, based on a 4-point Likert scale, was designed to be a broad screening instrument for psychological problems in a general population and has excellent validity and reliability (McDowell & Newell, 1996). In our study, the GHQ-12 (Cronbach's α = .80) was scored so that higher scores reflected poorer psychological status (M = 24.2, SD = 3.5, range = 15−38).

Independent Variables

On the basis of Figley's (1995, 2002b) conceptional of CF and psychosocial stress theory (Pearlin, 1989; Thoits, 1995), we included demographics, stressor exposures, and psychological resources as independent variables. The five demographic variables were gender, race/ethnicity, age, marital status, and years working in professional counseling. Age and years in professional counseling were coded to the nearest year (age, M = 55.9, SD = 8.0; years in counseling, M = 23.4, SD = 7.9). Gender, marital status, and race/ethnicity were coded as binary variables, with male, not married/not living together, and person of color coded as 0.

We included four variables measuring exposure to stressful events. First, the survey inquired about eight negative life events (e.g., getting divorced, having problems at work) that could have happened to the respondent in the past 24 months (Freedy, Kilpatrick, & Resnick, 1993). We summed these events to produce a negative life events scale (M = 1.1, SD = 1.2). Second, we asked about eight lifetime traumatic events (e.g., attacked with a gun, a situation where being killed was possible) that could have happened during the respondent's lifetime (Freedy et al., 1993). As with negative life events, we summed these events to produce a lifetime traumatic events scale (M = 1.9, SD = 1.4). Third, there were seven questions about whether or not the respondent counseled people exposed to the September 11 World Trade Center disaster (WTCD), such as counseling those who directly witnessed the events or who lost a spouse. We summed these questions into a September 11 counseling experience scale (M = 3.2, SD = 1.6). As a measure of exposure to other traumatized clients, we asked participants what percentage of their clients were survivors of physical or sexual violence. We coded responses into a binary variable, indicating low exposure if fewer than 20% of the clients were survivors of violence and high exposure if 20% or more were survivors. The measures of negative life events, traumatic events, and counseling individuals exposed to the WTCD discussed were used and validated in other WTCD studies in New York City (Boscarino, Galea, et al., 2004; Boscarino et al., 2002; Galea et al., 2003).

Finally, we assessed three measures of psychological resources: having information to work effectively with clients, social support, and sense of mastery. Work information was the sum of two items (“Information needed to enhance the delivery of my services to my clients or patients has been readily available to me,” and “I have adequate information about how to control my emotional fatigue through such strategies as deep breathing, positive self-talk, and the appropriate use of humor”) and related to how organizations helped formal caregivers cope with the stressful events associated with their work (Figley, 1995, 2002b). The response options for these two questions were based on 5-point Likert scales, ranging from strongly disagree (1) to strongly agree (5). Higher scores indicated greater agreement that information to be an effective social worker was available (M = 8.3, SD = 1.2). The Social Support Scale (Sherbourne & Stewart, 1991) consisted of four items (e.g., someone was available to confide in), summed so that higher scores reflected higher social support, based on a 4-point Likert scale ranging from none of the time to all of the time (M = 12.8, SD = 3.2; Cronbach's α = .89). Finally, we measured sense of mastery (Cronbach's α = .76) using six items from Pearlin, Lieberman, Menaghan, and Mullan's (1981) Sense of Mastery Scale. The response categories were based on a 5-point Likert scale (strongly disagree = 1 to strongly agree = 5). These items were summed, with higher scores on this scale indicating a greater sense of mastery (M = 23.4, SD = 3.8). Both the Social Support Scale and the Sense of Mastery Scale have been validated in previous studies and showed good reliability (Boscarino, Galea, et al., 2004; Boscarino et al., 2002; Pearlin et al., 1981).

Statistical Analysis

Our analytical goals were to develop a measure of CF that was parsimonious, measured core dimensions, was valid and reliable, and was a good predictor of psychological distress. To achieve these aims, we used principal-components analysis with a varimax rotation to identify the number of underlying factors in the CF scale (Pett, Lackey, & Sullivan, 2003). Once the factor analyses specified items to be included, we used reliability analyses, using Cronbach's alpha, to assess item consistency within each scale. After establishing that the scales measured only one factor and had acceptable internal reliability, we assessed the validity of the scales by correlating them with stress exposure, psychological resource, and psychological distress variables and to each other. Finally, to test the predictive utility of the scales, we estimated a series of ordinary least-squares (OLS) regressions, with the GHQ-12 as the dependent variable and the demographics, stress exposure, psychological resource, and CF scales as independent variables, respectively. We used SPSS Version 11.5 for all data analysis, and all significance values shown were based two-tailed tests.

Results

As the demographic profile of our sample has been discussed elsewhere (Boscarino, Figley, & Adams, 2004), we present only a brief description here. The sample was predominately female (80%), White (89%), in a long-term relationship (married or living as if married; 63%), and older, with more than 80% of the respondents 50 years or older. The majority of the sample also had more than 20 years experience in professional counseling and had had at least one negative life event in the past 2 years and at least one traumatic event in their lifetime. Finally, our sample was very involved in working with clients who had some exposure to the events of September 11, 2001, with 94% reporting at least some involvement. Conversely, fewer than 20% had a practice with a high percentage of clients who were survivors of violence.

Because a central focus of the study was to assess the psychometric properties of the CF Scale, we first conducted a factor analysis on the 30 items in the scale (available on request). We then eliminated items from future factor analyses on the basis of three criteria: loading on a factor with the lowest eigenvalue when that factor was dropped in subsequent factor analyses, having a factor score of .400 or greater on two or more components, or having a communality of less than .500. We also assessed the items in terms of whether or not they pertained to stress symptoms related to working with clients or job burnout.

We dropped items until only two factors remained (see Table 1). Component 1 consisted of items related to work burnout (feeling trapped by work). Those in Component 2 focused on secondary trauma (flashbacks, troubling dreams related to client interactions). We summed the items to form a Work Burnout scale (8 items) and a Secondary Trauma scale (5 items). Each of the scales showed good internal reliability, with the Work Burnout scale having a Cronbach's alpha of .90 and the Secondary Trauma scale an alpha of .80. We also combined all 13 items into a CF-Short Scale (α = .90).

Table 1.

Final Rotated Component Matrix for Factor Analysis of 13 Items in Figley's (1995) Compassion Fatigue Scale (n = 233)

	Component
Compassion Fatigue item	1	2
Q9: Flashbacks connected to clients		0.781
Q14: Troubling dreams similar to client's		0.615
Q15: Intrusive thoughts after working with difficult clients		0.718
Q16: Suddenly recalled frightening experience while working with client		0.846
Q17: Losing sleep over client's traumatic experience		0.630
Q20: I have felt trapped by my work	0.766
Q21: Sense of hopelessness working with clients	0.701
Q25: Felt tired due to work as caregiver	0.616
Q26: Felt depressed as a result of work	0.788
Q27: Unsuccessful at separating work from personal life	0.731
Q28: Sense of worthlessness associated with work	0.872
Q29: Feel like a “failure” in work	0.765
Q30: Thoughts about not achieving goals	0.747
Eigenvalues for rotated components	4.741	3.269

Note. Q = Question.

Table 2 presents Pearson correlation coefficients for the original 30-item CF Scale (CF-Long), the Work Burnout scale, the Secondary Trauma scale, and the 13-item CF-Short Scale with the GHQ-12 stress and psychological resource variables. The primary purpose of these analyses was to assess the scales’ concurrent validity. First, it is clear that the Work Burnout, Secondary Trauma, and CF-Short Scales were highly correlated with the original CF-Long Scale and with each other. Second, all four scales were significantly related to psychological distress as measured by the GHQ12. Third, work burnout was negatively associated with social support, having information to work effectively, and sense of mastery, which suggests that an unsupportive work environment can increase this work-related hazard. Secondary trauma was related to none of the stress/resource variables, whereas the CF-Short Scale was related to the same four variables that burnout was. Finally, none of four scales was significantly correlated to either percentage of clients who were survivors of violence or September 11 counseling activities. Thus, these analyses seem to indicate that exposure to traumatized clients does not, in itself, lead to CF.

Table 2.

Correlations Between Compassion Fatigue-Long Scale, Work Burnout, Secondary Trauma, and Compassion Fatigue-Short Scale and the General Health Questionnaire, Stress, and Psychological Resource Scales (n = 206)

Measure	CF (Long)	Work burnout	Secondary trauma	CF (Short)
General Health Questionnaire	.462**	.481**	.421**	.494**
Negative life events	.293**	.242**	.131	.233**
Trauma Scale	.202*	.092	.132	.098
9/11 counseling involvement	.152	.106	.145	.123
% clients survivors of violence	.065	.034	.020	.039
Social Support Scale	−.235**	−.204*	−.078	−.193*
Have info to work effectively	−.211*	−.191*	−.144	−.186*
Sense of Mastery Scale	−.322**	−.329**	−.147	−.290**
CF (Long)	–	.801**	.645**	.832**
Work Burnout Scale		–	.545**	.939**
Secondary Trauma Scale			–	.753**

Note. CF = Compassion Fatigue; info = information.

^*p < .01.

^**p < .001.

Overall, our multivariate OLS regressions (see Table 3) explained about 40% of the variance in our dependent variable, the GHQ-12 Scale, regardless of which CF scales were in the models. All four scales were significantly related to psychological distress, even after we controlled for other variables. That is, when we took into account recent negative life events, lifetime trauma, social support, the work environment, and sense of mastery, the associations between CF, secondary trauma or burnout, and psychological distress were not eliminated. The regression models also support the notion that job burnout and secondary trauma were separate contributors to psychological distress. As shown in Model 4, both burnout and secondary trauma were statistically significant when we controlled for other variables in the model. Finally, the CF-Short Scale (Model 5) did just as well at predicting psychological distress as the 30-item CF-Long Scale.

Table 3.

Linear Regressions for Compassion Fatigue-Long, Work Burnout, Secondary Trauma, and Compassion Fatigue-Short Predicting Scores on the General Health Questionnaire (n = 206)

	Model 1: Compassion Fatigue-Long		Model 2: Work Burnout		Model 3: Secondary Trauma		Model 4: Burnout and Secondary Trauma		Model 5: Compassion Fatigue-Short
Independent variable	B	beta	B	beta	B	beta	B	beta	B	beta
Age	−0.05	−.12	−0.05	−.11	−0.07	−.16*	−0.06	−.14*	−0.06	−.13
Gender of respondent (female)	−0.65	−.08	−0.50	−.06	−0.82	−.09	−0.71	−.08	−0.65	−.07
Marital status (married)	0.99	.14*	0.71	.10	0.70	.10	0.76	.11	0.72	.10
Race/ethnicity (White)	0.26	.02	0.26	.02	0.14	.01	0.07	.01	0.11	.10
Years as prof. counselor	0.02	.05	0.02	.05	0.03	.06	0.03	.06	0.04	.08
Negative life events	0.48	.17**	0.48	.17**	0.58	.20***	0.51	.18**	0.48	.17**
Trauma History Scale	0.01	.01	0.08	.04	0.02	.01	0.03	.01	0.07	.03
9/11 counseling involvement	−0.13	−.06	−0.11	−.05	−0.18	−.08	−0.18	−.08	−0.13	−.06
% clients survivors of violence	−0.40	−.05	−0.33	−.04	−0.31	−.03	−0.32	−.04	−0.35	−.04
Social Support Scale	−0.11	−.10	−0.11	−.10	−0.13	−.12	−0.12	−.11	−0.11	−.10
Have info to work effectively	−0.38	−.13*	−0.39	−.13*	−0.36	−.12*	−0.33	−.11	−0.37	−.13*
Sense of Mastery Scale	−0.23	−.25***	−0.22	−.24***	−0.27	−.29***	−0.23	−.25***	−0.23	−.24***
Compassion Fatigue-Long Scale	1.33	.33***
Work Burnout Scale			1.06	.33***			0.58	.18*
Secondary Trauma Scale					1.34	.36***	1.01	.27***
Compassion Fatigue-Short Scale									1.22	.37***
Constant	31.564		33.526		36.614		34.215		33.301
R2	.37		.38		.40		.42		.40

Note. prof. = professional.

^*p < .05.

^**p < .01.

^***p < .001.

The regression equations also support several other expected associations. Respondents reporting more negative life events also had higher levels of distress, whereas those with a higher sense of mastery were less distressed. It is interesting that having adequate information to work effectively with clients was significantly associated with lower distress in all but one of the five models. Finally, none of the demographic characteristics consistently predicted scores on the GHQ-12. This last finding is surprising, given that most research shows some statistically significant associations between demographic characteristics and psychological distress (Pearlin, 1981; Thoits, 1995).

Discussion

Using a random sample of social work practitioners in New York City, we assessed the psychometric properties of Figley's (1995) CF Scale—Revised (Gentry et al., 2002). Our analyses showed that the original 30-item scale measured multiple underlying factors. Eliminating items with the goal of producing measures consistent with secondary trauma and burnout, we were successful in developing two reliable and parsimonious scales. The strength of the separate scales (or the combined CF-Short Scale) is that they contain fewer items overall (13 vs. 30) while remaining highly correlated with the original scale. In addition, with separate scales, we could explicitly test the claim that secondary trauma is different from burnout, which is supported by our analyses. Finally, the original scale contained items that could be interpreted as measuring psychological distress (e.g., “I have difficulty falling or staying asleep,” “I startle easily”) or direct, personal trauma (e.g., “I have had first-hand experience with traumatic events in my childhood”) rather than vicarious trauma. Thus, the associations between psychological distress and secondary trauma, burnout, and the CF-Short Scale were not artificially high as a result of the scales containing similar items.

A second goal of our study was to gauge the validity of the CF-Long, Burnout, Secondary Trauma, and CF-Short Scales by examining their associations with other reliable and valid measures. All four scales were correlated with psychological distress (as measured by the GHQ-12), but none of them had correlations with stress exposure or resource variables greater than .33. These results support the hypothesis that secondary trauma, burnout, and, more generally, CF are unique features of the workplace environment and not merely different designations for negative life events, personal trauma, lack of social support, or low mastery.

Finally, we tested the predictive power of the CF scales, burnout, and secondary trauma in a multivariate model. The regression findings clearly indicate that the original CF Scale and the reduced Burnout, Secondary Trauma, and CF-Short Scales predicted psychological distress very well, even after we controlled for demographic, stress exposure, and psychological resource factors. The Burnout and Secondary Trauma scales, therefore, seem to be quite appropriate assessment tools, either separately or combined into the CF-Short Scale, for identifying caregiving professionals at risk for CF and psychological problems.

There are limitations that make our conclusions tentative. First, the sample was smaller than desired for factor analyzing a 30-item scale. Nevertheless, the Secondary Trauma, Burnout, and CF-Short Scales match the clinically based concepts as developed by Figley (1995, 2002b). Given the overall consistency of our results, there is little reason to believe that they were seriously biased by sample size. Second, we cannot fully disentangle the causal ordering of psychological distress and CF because of the cross-sectional nature of our data. Theoretically, Figley (1995) and others (Jenkins & Baird, 2002; Schauben & Frazier, 1995) have assumed a causal direction from CF to psychological problems. It is possible, however, that psychologically distressed individuals may interpret their interactions with clients such that they increase their vulnerability to CF. Only longitudinal data and further theoretical work can adequately address these issues.

A third limitation reflects the evolving conceptualization of CF. In her history of the concept, Stamm (2002) argued that the scale should have both positive and negative items to measure both CF and compassion satisfaction. The present study uses a version of the Figley (1995) scale that does not contain satisfaction items. This may be part of the reason that exposure to survivors of violence was not significantly related to CF. The ongoing discussion of how to measure the concept makes for analytic difficulties and obviously necessitates further research on both the concept and its operationalization.

Another limitation is the fact that only certain types of social workers might be members of NASW, the professional database from which we drew our sample. Thus, our sampling frame might have been biased in some way. Furthermore, our response rate was somewhat below 50%, which, although not significantly lower than for other mail surveys in urban areas (Kessler, Little & Groves, 1995), might have introduced some selection bias. Finally, it may not be possible to generalize our results beyond social workers practicing in New York City. Although social workers around the country face similar work environment problems, there may be other aspects of work unique to less urban settings. In addition, psychiatrists, psychologists, and other trauma workers may have different responses to working with survivors of traumatic experiences as a consequence of working in different occupational settings.

Conclusion

These limitations should not overshadow the strengths of this study. It is one of the few to closely examine the psychometric properties of Figley's (1995) CF Scale and use it to predict psychological distress. The CF-Short, Secondary Trauma, and Burnout scales employed in the present study have good reliability and good concurrent and predictive validity and include only 13 items. Even after we accounted for other factors, both scales were clearly related to psychological distress. In addition, other scales used in the study are all well established, with good reliability and validity.

An additional strength of this study was our attempt to clarify the conceptual differences between secondary trauma, burnout, and CF. Such clarity is necessary for scale construction and has been generally lacking for CF (Jenkins & Baird, 2002). Thus, we argue that therapists can suffer from CF, which contains two components, secondary trauma and job burnout. Developing a valid and reliable instrument to detect CF is also a prelude to devising intervention strategies designed to mitigate its negative effects on the caregiving practitioner. As Figley (2002a) has noted, “it is, therefore, up to all of us to elevate these issues to a greater level of awareness in the helping professions. Otherwise, we will lose clients and compassionate psychotherapists” (p. 1440).