Understanding & Mitigating Compassion Fatigue for Legal

Understanding & Mitigating Compassion Fatigue for Legal Professionals

Speakers: Linda Albert, LCSW, CSAC

Compassion fatigue is defined as the cumulative physical, emotional and psychological effects of being continually exposed to traumatic stories or events when working in a helping capacity. It has been studied extensively in social workers, nurses, doctors and therapists who work with victims of trauma. The State Bar of Wisconsin holds the largest set of research data examining the impact upon legal professionals. This seminar will look at which legal professionals are most at risk, the development of compassion fatigue, the interface between attorney impairment and discipline and what individual and organizational measures can prevent and mitigate compassion fatigue.

Presentation Outline

I. Defining compassion fatigue. What is it? How does it play out? II. What puts legal professionals at risk? III. Development of compassion fatigue; the brain keeps the score. IV. Interface between impairment and ethical violations. V. Mitigating compassion fatigue; what is the formula for staying fit to practice.

Linda Albert is a Licensed Clinical Social Worker and a Certified Alcohol and Drug Counselor. She received her Master’s Degree from UW-Madison in Social Work. She has professional assessment/treatment/referral competencies in the areas of addictions, eating disorders, depression, anxiety, trauma and illness impacted by stress. Linda has worked over the past 30 years as an administrator, consultant, trainer and psychotherapist in a variety of settings including providing services to impaired professionals. Linda has co-facilitated a research project on compassion fatigue and legal professionals resulting in two recent publications. She has done multiple presentations for conferences at the local, state and national level. Currently Linda is employed by the State Bar of Wisconsin as the WisLAP Coordinator

Seminar Outline Keeping Legal Minds Intact: Mitigating Compassion Fatigue Among Legal Professionals

I. DISCIPLINARY RULES

A. SCR 20:1.1. Competence

A lawyer shall provide competent representation to a client. Competent representation requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation.

B. SCR 20:1.3. Diligence

A lawyer shall act with reasonable diligence and promptness in representing a client.

C. SCR 20:1.4 Communication

(a) A lawyer shall:

(1) Promptly inform the client of any decision or circumstance with respect to which the client's informed consent, as defined in SCR 20:1.0(f), is required by these rules; (2) reasonably consult with the client about the means by which the client's objectives are to be accomplished; (3) keep the client reasonably informed about the status of the matter; (4) promptly comply with reasonable requests by the client for information; and (5) consult with the client about any relevant limitation on the lawyer's conduct when the lawyer knows that the client expects assistance not permitted by the Rules of Professional Conduct or other law. (b) A lawyer shall explain a matter to the extent reasonably necessary to permit the client to make informed decisions regarding the representation.

C. SCR 20:1.16. Declining or terminating representation

(a) Except as stated in par. (c), a lawyer shall not represent a client or, where representation has commenced, shall withdraw from the representation of a client if:

(1) the representation will result in violation of the Rules of Professional Conduct or other law;

(2) the lawyer's physical or mental condition materially impairs the lawyer's ability to represent the client; or

D. SCR 20:8.4.

It is professional misconduct for a lawyer to:

(a) violate or attempt to violate the Rules of Professional Conduct, knowingly assist or induce another to do so, or do so through the acts of another; ….

(d) state or imply an ability to influence improperly a government agency or official or to achieve results by means that violate the Rules of Professional Conduct or other law;

…., or

(f) violate a statute, supreme court rule, supreme court order or supreme court decision 1 regulating the conduct of lawyers;

….

II. MEDICAL INCAPACITY

SCR 21.17. Medical incapacity suspension, conditions

The license of an attorney to practice law may be suspended indefinitely or conditions may be imposed on the attorney's practice of law with the attorney's consent or upon a finding that the attorney has a medical incapacity, pursuant to the procedure set forth in SCR chapter 22.

III. REPORTING SCR 20:8.3 Reporting Professional Misconduct

(a) A lawyer having knowledge that another lawyer has committed a violation of the Rules of Professional Conduct that raises a substantial question as to that lawyer’s honesty, trustworthiness or fitness as a lawyer in other respects, shall inform the appropriate professional authority.

(b) A lawyer having knowledge that a judge has committed a violation of applicable rules of judicial conduct that raises a substantial question as to the judge’s fitness for office shall inform the appropriate authority.

(2) Information acquired by one of the following: (i) A member of any committee or organization approved by any bar association to assist ill or disabled lawyers where such information is acquired in the course of assisting an ill or disabled lawyer.

(ii) Any person selected by a court or any bar association to mediate or arbitrate disputes between lawyers arising out of a professional or economic dispute involving law firm dissolutions, termination or departure of one or more lawyers from a law firm where such information is acquired in the course of mediating or arbitrating the dispute between lawyers.

IV. ETHICS CASES V. UNDERSTANDING COMPASSION FATIGUE AND MENTAL HEALTH CONDITIONS WHICH CORRELATE WITH ATTORNEY DISCIPLINE; BEST PRACTICES FOR STAYING FIT TO PRACTICE

VI. ASSISTANCE PROGRAMS

A. Wisconsin Lawyer’s Assistance Program (WISLAP)

WisLAP Manager: Linda Albert [email protected] (800) 444-9404 ext. 6172 WisLAP Coordinator: Mary Spranger [email protected] (800) 444-9404 Ex. 6159

24 hour helpline: (800) 543-2625

B. Ethics Hotline

State Bar's ethics counsel: Monday through Friday, 9 a.m. to 5 p.m.

Timothy Pierce [email protected] (608) 250-6168 or (800) 444-9404, ext. 6168, Aviva Kaiser [email protected] (800) 444-9404 ext. 6158

C. Law Office Management Assistance Program

State Bar’s LoMAP counsel: Monday through Friday, 9 a.m. to 5 p.m. Tison Rhine [email protected] (608) 250-6012 or (800)-444-9404 ext. 6012

CASE LAW

MEDICAL INCAPACITY CASES-GENERAL

In Matter Of Disciplinary Proceedings Against John Loew, 2010 WI 23 (60 day suspension for multiple counts of neglect. Attorney stated that after starting his own firm his practice mushroomed. Although he was offered assistance with office staff, he said he was too overwhelmed to take the time to train these individuals. Due to his inability to turn down new clients, Attorney Loew began working virtually around the clock, leading to a deterioration of his physical and mental health. He then began to practice from his home and began experiencing the onset of severe depression. Attorney Loew sought help from his primary physician in August 2007 and was diagnosed with anxiety and depression. He was given anti-depressant medication along with a recommendation for therapy. Attorney Loew consulted a psychiatrist in December 2007 and attended some therapy sessions beginning in early 2008. In August 2009 Dr. Gregory J. Van Rybroek, a licensed psychologist, conducted an independent psychological evaluation of Attorney Loew. Dr. Van Rybroek concluded, to a reasonable degree of professional certainty, that during the time Attorney Loew was retained in the misconduct cases he was in the throes of a major depressive episode primarily brought on by a workload he ultimately became unable to manage. Dr. Van Rybroek said it appeared the volume of Attorney Loew's workload, exacerbated by the death of a close friend, created overwhelming stress that led to serious depression. Dr. Van Rybroek opined that the situational stress led to the point where Attorney Loew became largely dysfunctional. Dr. Van Rybroek was of the opinion that Attorney Loew's depression symptoms were situational rather than driven by a genetic predisposition toward professional certainty, that Attorney Loew's current level of functioning was much improved. Dr.

Van Rybroek was of the opinion that Attorney Loew could return to the practice of law, subject to certain conditions, including entering into a therapy relationship to assist with better self- understanding, specifically in learning more about the internal triggers that make it difficult for him to limit his workload; remain under the care of a psychiatrist for purposes of monitoring his medication; and he should not work in the solo practice of law but rather should work in a structured legal environment where his work can be monitored. While the referee said the fact that Attorney Loew was suffering from depression in late 2006 and early 2007 was well documented in the record, his testimony at the hearing made it clear that his difficulty in running a solo law practice began well before the onset of his depression. The referee said at most Attorney Loew's depression was a contributing factor to the events outlined in the OLR's complaint, not a causal factor. The OLR had sought a six-month suspension of Attorney Loew's license. The court agreed with the referee’s recommendation of a 60 day suspension and that for a period of five years following his resumption of the practice of law, the following conditions shall be imposed upon John R. Loew's practice of law: A. enter into a therapy relationship to assist with better self-understanding, specifically in learning more about the internal triggers that make it difficult for him to limit his workload; B. remain under the care of a psychiatrist, particularly for monitoring of his medication needs; C. shall engage in the practice of law under the direct supervision of a licensed attorney acceptable to and approved by the Office of Lawyer Regulation. Attorney Loew's supervising attorney shall have all the duties generally held by a supervising attorney under SCR 20:5.1(b); D. submit to the OLR, on a semiannual basis, a report from his treating psychiatrist outlining his ongoing treatment. The reports should disclose any recurrences of the depression, and if any, whether such recurrences pose an unreasonable risk to the public if John R. Loew is permitted to continue the practice of law; E. submit to the OLR, on a semi-annual basis, a report from his supervising attorney.

In Matter Of Disciplinary Proceedings Against Nunnery, 2009 WI 89, 769 N.W.2d 858 (3 year suspension for neglect and misrepresentation in multiple matters. Although Nunnery did not specifically argue mitigating factors here, the record discloses the referee's consideration of Attorney Nunnery's significant personal and physical problems during the relevant time periods, including his hospitalization, as well as the illness and death of a member of his family. The referee noted Attorney Nunnery admitted that, as a sole practitioner, he had taken on more work than he could handle effectively. Nunnery also claimed stressors of being caught in Hurricane Katrina, as well as being subjected to a $2.5 million malpractice verdict which was ultimately reversed on appeal and handling complex litigation. The court found that these factors did not excuse the misconduct but were considered in mitigation.

In Matter Of Disciplinary Proceedings Against Scott E. Hansen, 2009 WI 56, 768 N.W.2d 1 (9 month suspension for multiple counts of misconduct and conditions on reinstatement. Hansen claimed the misconduct was due to depression. The OLR indicated it considered Attorney Hansen's depression a mitigating factor, but noted that there was insufficient medical evidence submitted in this proceeding to substantiate the scope of his condition. The OLR observed that progressive discipline was warranted because of Attorney Hansen's disciplinary history. It noted that sanctions in other cases ranged from revocation to a six-month suspension for comparable misconduct. The OLR noted further that Attorney Hansen's failure to return unearned fees and his failure to cooperate with the OLR were aggravating factors not easily attributable to depression. The OLR eventually recommended a one-year suspension of Attorney Hansen's for Attorney Hansen's claim that he has suffered from depression since 1994. Attorney Hansen did file a letter from a nurse that states that he has had a "major depressive disorder" for the past "several years." Moreover, the referee stated that he failed to "see how [Attorney Hansen]'s disorder can explain Attorney Hansen's attempts to deceive not only his clients but also the Court of Appeals and the Office of Lawyer Regulation. Nor can I see how this disorder would prevent Attorney Hansen from returning fees which he agrees were unearned and which

he promised to return." The referee then recommended a six-month suspension of Attorney Hansen's license describing his depression as "a partial mitigating factor." The court imposed a 9 motnh suspension and the following condition upon reinstatement: Hansen “shall undergo a medical evaluation by a medical professional selected by the Office of Lawyer Regulation and a copy of that evaluation shall be provided to the Office of Lawyer Regulation as a condition for reinstatement, with the understanding that Scott E. Hansen may also submit an evaluation performed by a medical professional of his choice. If this condition is not met, the license of Scott E. Hansen to practice law in Wisconsin shall remain suspended until further order of this court.”

In the Matter of Disciplinary Proceedings Against John E Raftery, 2007 WI 137, 306 Wis.2d 28, 742 N.W.2d 315 (6 month license suspension for multiple acts of neglect, failure to cooperate and failure to comply with rules regarding prior temporary suspension. Referee rejected assertion that all of the misconduct was the result of his longstanding chronic depression, for which he had received both inpatient and outpatient treatment. While the referee noted that Attorney Raftery's medical experts were of the opinion that Attorney Raftery had suffered from chronic depression for years, the referee said there was no evidence in the record that the medical experts related each and every count of misconduct that occurred in this case solely to the chronic depression. The referee pointed out that many of the counts of misconduct related to deliberate misrepresentations by Attorney Raftery to the OLR during its investigation of the various client grievances and to Attorney Raftery's failure to comply with supreme court rules following his temporary suspension. The referee concluded that this misconduct was not solely the result of Attorney Raftery's being "paralyzed" by depression, but rather was intentional misconduct undertaken in an effort to deceive the OLR and the public. Rather than viewing Attorney Raftery's chronic depression as an excuse for his misconduct, the referee found it to be a mitigating circumstance to be considered in recommending an appropriate discipline. Conditions imposed for a period of two years: A. law practice shall be monitored as set forth in the "Suggested plan for oversight and accountability filed by the attorney; B. shall be required to submit to OLR, on a quarterly basis, a report that he is in compliance with the "Suggested plan for oversight and accountability." The report shall disclose any patterns of neglect of client matters. Such quarterly reports shall include the written approval of the attorney; C. required to submit to OLR, on a semiannual basis, a report from his treating psychologist/psychiatrist outlining his ongoing treatment. The reports should disclose any recurrences of the depression, and if any, whether such recurrences pose an unreasonable risk to the public if he is then permitted to continue the practice of law; D. Should he, for any reason, leave the Ritger Law Office, he shall immediately inform OLR, and any continuance of his practice of law shall be subject to monitoring and a plan similar to the above.

Disciplinary Proceedings Against Shindell, 2002 WI 133, 258 Wis.2d 63, 654 N.W.2d 844 (1 year suspension added onto to existing suspension for various misconduct. The court noted that “the record demonstrates the existence of a number of mitigating factors, including the fact that Attorney Shindell suffered from serious personal and health problems and also had

administrative problems in her office which seemed to peak in early 1999. As the referee also noted, however, it would have been appropriate for Attorney Shindell to have obtained assistance from other attorneys in her office or brought in outside counsel during this time, and the absence of this assistance was not the fault of her clients, who were unaware of Attorney Shindell's problems.”

Alcohol & Drug Cases – Mitigation

In the Matter of Disciplinary Proceedings Against Compton 2010 WI 112 (2010) On May 6, 2010, the OLR filed a disciplinary complaint in this matter alleging that Attorney Compton committed criminal acts that reflect adversely on his honesty, trustworthiness, or fitness as a lawyer in other respects, in violation of SCR 20:8.4(b). Aggravating factors include Attorney Compton's prior disciplinary history, the pattern of misconduct, the potential vulnerability of K.L., and the harm to K.L. Mitigating factors include the fact that Attorney Compton has been "wholly cooperative in this matter," including entering into an agreement admitting his misconduct and agreeing to the level of discipline sought by the OLR. He acknowledged the wrongfulness of his conduct and offered to turn in his law license. He voluntarily entered into a drug treatment program and has expressed remorse for his conduct. The court is advised that Attorney Compton is undergoing voluntary monitoring and as of June 11, 2010, the monitor reported Attorney Compton's continued compliance with the requirements of his monitoring program. We approve the stipulation and adopt the stipulated facts and legal conclusions of professional misconduct. We agree that a two-year suspension is appropriate and consistent with this court's past practice. We further agree that it is appropriate to order the suspension to commence on March 16, 2010, the date Attorney Compton's law license was summarily suspended by this court. The imposition of a two-year suspension will require Attorney Compton to complete successfully the formal reinstatement process in order to regain his license to practice law in Wisconsin.

In re Brandt, 766 N.W.2d 194 (Wisconsin 2009) The Office of Lawyer Regulation brought a complaint against Attorney Brandt for failure to properly supervise an employee and for multiple convictions of operating a motor vehicle while intoxicated. The Supreme Court found that the lawyer’s multiple drunk driving convictions did constitute a violation Wis. Sup. Ct. R. 20:5.3(b) because his convictions did demonstrate a serious lack of respect for the law and, therefore relate to his fitness as a lawyer. The court imposed a public reprimand. In this case, a sanction beyond a public reprimand was not necessary because they found that Attorney Brandt was the victim of his employees embezzlement, accepted responsibility for his failure to supervise his employee, had entered pleas to the drunk driving charges and served a significant jail sentence, paid substantial fines, had taken remedial action to address his drinking problem and sought treatment.

Disciplinary Proceedings Against Verlin Peckham, 115 Wis.2d 494, 340 N.W.2d 198 (1983) (The appellant argues that any suspension of his license should be stayed because his misconduct was caused by his alcoholism. He distinguishes his case from Disciplinary

Proceedings Against Glasschroeder, 113 Wis.2d 672 (1983), where the court denied a request for a stay of suspension and revoked the attorney's license. Unlike the appellant, Glasschroeder misappropriated a large sum of client money over a long period of time and uttered a forged check, for both of which he was criminally convicted. Also, Glasschroeder did not present evidence that his misconduct was causally related to his alleged drug dependency. The appellant contends that his recognition of his alcoholism and his efforts at rehabilitation render this case susceptible of a "new" discipline, one which has been recognized in other jurisdictions. He argues that the public can be adequately protected if the court conditions his continued practice on supervision by a practicing attorney, complete abstinence from alcohol, weekly attendance at Alcoholics Anonymous or Lawyers Concerned for Lawyers meetings, post treatment by his psychologist, and his maintenance of complete records concerning client funds and property coming into his possession. These conditions were imposed by the Minnesota Supreme Court in a disciplinary proceeding, Petition of Johnson, 322 N.W.2d 616 (1982), after the disciplinary agency and the attorney had stipulated to them. The referee considered the appellant's misconduct in light of the alcoholism defense and concluded: ". . . His conduct does appear to have been precipitated by a drinking problem rather than an intent on his part to deceive his clients or to convert their funds. This, however, does not afford him a recognizable defense. The trust account violations which came to light as the result of a board audit and which were not denied renders his misconduct extremely serious and warrants the discipline sought by the Board based upon similar cases in the past." We agree, and we adopt the findings, conclusions and recommendation of the referee).

Law and Human Behavior

The Effect of Attorneys' Work With Trauma-Exposed Clients on PTSD Symptoms, Depression, and Functional Impairment: A Cross-Lagged Longitudinal Study Andrew Levin, Avi Besser, Linda Albert, Deborah Smith, and Yuval Neria Online First Publication, April 2, 2012. doi: 10.1037/h0093993

CITATION Levin, A., Besser, A., Albert, L., Smith, D., & Neria, Y. (2012, April 2). The Effect of Attorneys' Work With Trauma-Exposed Clients on PTSD Symptoms, Depression, and Functional Impairment: A Cross-Lagged Longitudinal Study. Law and Human Behavior. Advance online publication. doi: 10.1037/h0093993 Law and Human Behavior © 2012 American Psychological Association 2012, Vol. 00, No. 00, 000–000 0147-7307/12/$12.00 DOI: 10.1037/h0093993

The Effect of Attorneys’ Work With Trauma-Exposed Clients on PTSD Symptoms, Depression, and Functional Impairment: A Cross-Lagged Longitudinal Study

Andrew Levin Avi Besser Westchester Jewish Community Services, Sapir Academic College Hartsdale, New York and Columbia University

Linda Albert Deborah Smith State Bar of Wisconsin, Madison, Wisconsin Wisconsin State Public Defender’s Office, Madison, Wisconsin

Yuval Neria Columbia University and New York State Psychiatric Institute, New York, New York

To date, few studies have examined mental health consequences among attorneys exposed to clients’ traumatic experiences. A longitudinal, 2-wave, cross-lagged study was used in a cohort of attorneys (N ϭ 107) from the Wisconsin State Public Defender’s Office. We assessed changes in posttraumatic stress disorder (PTSD), depression, and functional impairment over a 10-month period and tested the effects of intensity of contact with trauma-exposed clients on symptom levels over time. Attorneys demonstrated strong and significant symptom stability over time in PTSD, depression, functional impairment, and levels of exposure. Analyses involving cross-lagged panel correlation structural equation modeling path models revealed that attorneys’ levels of exposure to trauma-exposed clients had significant positive effects, over time, on PTSD, depression, and functional impairment. Gender, age, years on the job, and office size did not predict any of the outcomes. Level of exposure to trauma-exposed clients predicted reduction of weekly working hours over time, but there was no reciprocal relationship between PTSD, depression, and functional impairment and level of exposure over time. These findings underscore the central role of exposure to trauma-exposed clients in predicting mental health outcomes and emphasize the need to support attorneys by managing the intensity of exposure as well as addressing emerging symptoms.

Keywords: attorneys, psychological trauma, posttraumatic stress disorder, depression, functional impairment

To date, few studies have examined the mental health conse- these findings were not related to work experiences. In a study of quences among attorneys exposed to clients who have experienced 23 Canadian prosecutors using semistructured interviews, Gomme or been directly involved in traumatic events (trauma-exposed and Hall (1995) reported symptoms of demoralization, anxiety, clients). In addition, available quantitative studies of distress in helplessness, exhaustion, and social withdrawal. They linked these attorneys have only been cross-sectional in nature. Focusing on symptoms to high caseloads of “sensitive cases” such as domestic depression, Benjamin, Darling, and Sales (1990) and Eaton, An- violence and incest as well as long work hours. Lynch (1997) thony, Mandel, and Garrison (1990) identified a 20% rate of reported that public defenders ranked work overload, the unpre- clinically significant depression in the attorneys surveyed, but dictability of trials, the frequent lack of a defense, harsh sentences,

This study was supported by the State Bar of Wisconsin and the Andrew Levin, Westchester Jewish Community Services, Hartsdale, Wisconsin State Public Defender’s Office. We thank Jeff Apotheker New York; Department of Psychiatry, College of Physicians and Surgeons, for his critique of the original design and support of the submission Columbia University. Avi Besser, Department of Behavioral Sciences and to the Westchester Jewish Community Services Research Committee, Center for Research in Personality, Life Transitions, and Stressful Life and Alex Zelenski for assistance in data gathering and management. Events, Sapir Academic College, D. N. Hof Ashkelon, Israel. Linda Albert, Grateful thanks are also extended to all of the participants in this Wisconsin Lawyers Assistance Program, State Bar of Wisconsin, Madison, study. Wisconsin. Deborah Smith, Assigned Counsel Division, Wisconsin State Correspondence concerning this article should be addressed to An- Public Defender’s Office, Madison, Wisconsin. Yuval Neria, Department drew P. Levin, Medical Director, Westchester Jewish Community Ser- of Psychiatry, College of Physicians and Surgeons, Columbia University; vices, 141 North Central Avenue, Hartsdale, NY 10530. E-mail: New York State Psychiatric Institute, New York, New York. [email protected]

1 2 LEVIN, BESSER, ALBERT, SMITH, AND NERIA arguing with prosecutors, and dealing with angry clients and Method families as the most frequent and intense sources of stress, but the study did not measure specific symptoms of stress. A pilot study Participants and Procedure by Levin and Greisberg (2003) found that attorneys working in family and criminal courts demonstrated higher levels of second- We conducted a longitudinal follow-up study on a sample of ary trauma and burnout compared with mental health professionals attorneys working in the 38 offices of the Wisconsin State Public and social service workers, and these measures of distress corre- Defender’s Office (Levin et al., 2011). In that study, we collected lated with caseload. Comparing 50 attorneys working in criminal data in March 2010 via the Wisconsin State Public Defender’s courts with 50 working in the civil arena, Vrklevski and Franklin Office intranet to 307 attorneys, with an initial response of 238 (2008) found more depressive symptoms, subjective stress, and attorneys (78%). The data for the current study were based on a changes in sense of safety and intimacy among the criminal follow-up survey that was distributed in December 2010 to all attorneys. A personal history of multiple traumas predicted higher attorneys working in the office. This resulted in 142 responses, of scores on measures of vicarious trauma, posttraumatic stress, and which 107 were attorneys who had also completed the original depression. In another study comparing criminal and civil attor- survey, representing 45% of the 238 who initially responded. The neys, Hasnain, Naz, and Bano (2010) also found that criminal 107 attorneys (51 men and 56 women) were in their mid-40s (M ϭ attorneys reported higher levels of stress than civil attorneys. This 45.72 years, SD ϭ 11.0), with almost 16 years’ experience on the ϭ ϭ difference was seen among attorneys with more than 10 years’ job (M 15.89, SD 11.03), working on average in local offices ϭ ϭ experience but was not observed in attorneys in training. Pi- (total staff) of more than 10–20 people (M 2.40, SD 1.0). wowarczyk et al. (2009) reported that among 57 attorneys special- Preliminary analyses indicated that the means for hours worked, ϭ ϭ izing in asylum cases, hours per week devoted to those cases t(236) 0.81, ns, caseload of trauma-exposed clients, t(236) ϭ correlated with trauma score. All of these studies of attorneys 0.20, ns, size of local office, t(236) 0.09, ns, and background ␹2 ϭ ϭ suggest a relationship between exposure to trauma and attorneys’ variables of gender ( 0.59 ns), age, t(236) 0.11, ns, years on ϭ symptoms but suffer from small sample size and, given their the job, t(236) 0.55, ns, as well as the outcome variables of ϭ ϭ cross-sectional design, do not elucidate the course of the symptoms intrusion, t(236) 0.44, ns, avoidance, t(236) 1.42, ns, hyper- ϭ ϭ or the direction of effects between exposure and symptomatology. arousal, t(236) 0.79, ns, depression, t(236) 0.22, ns, and ϭ Studies of other human service professionals working with trauma- functional impairment, t(236) 0.47, ns, did not differ at the initial survey in March 2010 between the subset that followed up exposed clients such as social workers (Kassam-Adams, 1999), (n ϭ 107) and the remaining 131 participants. law enforcement officers (Follette, Polusny, & Milbeck, 1994), Survey materials were made available online by the survey and psychotherapists (Pearlman & MacIan, 1995) are also limited office of the State Bar of Wisconsin. Potential participants re- by cross-sectional design. Some of these studies have linked in- ceived an e-mail providing the necessary link to the questionnaires tensity of work-related exposure (Creamer & Liddle, 2005; Er- and were encouraged to complete the survey from personal com- iksson, Kemp, Gorsuch, Hoke, & Foy, 2001; Kassam-Adams, puters on the job site. All participants received information regard- 1999) to secondary trauma symptoms, although other findings ing the study in the form of an informed consent cover letter at the have suggested the primary importance of organizational and start of the online survey packet. Proceeding to the questionnaire work-related factors (Baird & Jenkins, 2003; Devilly, Wright, & indicated consent. Participation was voluntary and anonymous and Varker, 2009; Regehr, Hemsworth, Leslie, Howe, & Chau, 2004) there was no remuneration for participation. The research proposal compared with exposure. was reviewed and approved by the Westchester Jewish Commu- Recently, in a large cross-sectional study, our group examined nity Services Research Committee as well as its board of directors ϭ indicators of secondary trauma among attorneys (n 238) and and chief executive officers. Leadership at both the Wisconsin ϭ their administrative support staff (n 109) and found that the Public Defender’s Office and the Wisconsin Bar also reviewed and attorneys demonstrated significantly higher levels of posttraumatic approved the study. stress disorder (PTSD) symptoms, depression, secondary traumatic stress, burnout, and functional impairment compared with administrative support staff (Levin et al., 2011). In addition, we found Measures that the difference in symptoms was mediated by attorneys’ longer Background and trauma exposure assessments. Demo- work hours and greater exposure to trauma-exposed clients and graphic and personal information included age, gender, number of was not related to other variables such as gender, years on the job, years on the job, average number of hours worked per week (for office size, or personal history of trauma. The present study used the prior 3 months), and size of local office (total staff) specified a longitudinal design in a subsample of attorneys from our on a 1–4 scale, ranging from 1 (fewer than 10),2(10–20), 3 previously reported cohort of the Wisconsin State Public De- (21–40), and 4 (more than 40). Because participants expressed a fender’s Office (Levin et al., 2011) to assess changes in symp- strong need to protect their anonymity, information regarding the toms of PTSD, depression, and functional impairment over a specific office where the participant worked as well as ethnic 10-month period. In addition, our design sought to measure the origin were omitted. relative contributions of caseload of trauma-exposed clients and The attorneys routinely interact closely with defendants in local hours worked to symptom and functional impairment levels jails, prisons, courthouses, and in their own offices. Cases run the over time and the direction of effects between caseload of gamut from mild violence or substance abuse to homicide and trauma-exposed clients, hours worked, and symptoms and func- sexual offenses such as rape or child abuse allegedly perpetrated tional impairment. by the attorneys’ clients. In addition to hearing first-hand accounts, ATTORNEYS WORKING WITH TRAUMA-EXPOSED CLIENTS 3 the attorneys review reports and photographs and have contact with functioning in three spheres at Time 1 and Time 2. Partici- with physical evidence (e.g., bloody clothing). Exposure to client pants rated the following question (in three forms): “My feelings trauma was assessed at baseline (Time 1) and 10 months later about the clients and cases at work have disrupted my (work, social (Time 2) by asking, “How many clients have you worked with life/leisure, or family life/home responsibilities)” on a visual ana- within the last three months who had experienced or been directly logue scale ranging from 0 (none), 1–3 (mild), 4–6 (moderate), involved with trauma such as death, physical assault or abuse, 7–9 (severe), to 10 (very severe). In the present sample, the domestic violence, rape, violence or fire?” Participants were in- estimates of internal consistency Cronbach’s reliability coeffi- structed to select the closest number on a 0–5 scale: 0 (none), 1 cients were ␣ϭ.91 and ␣ϭ.90 at Time 1 and Time 2, respec- (1–20),2(21–40),3(41–60),4(61–80), and 5 (81 or more). We tively. elected to use six categories rather than a precise number because our pilot study indicated that attorneys were not able to report an Data Analysis exact number based on their recollection of the prior 3 months. However, it is important to note that five response categories are Mean scores for exposure to traumatic clients and hours at work believed to represent an interval level of measurement. The use of as well as for IES–R, CES–D, and SDS scores were calculated and the six categories in our study does not violate the axiom of compared between times (repeated measure) using t tests and transitivity for the ordinal scale; the intervals between the scale stability of symptoms was assessed using Pearson correlation points (number of clients represented by each category) corre- among same assessments over time. We then performed a bivariate spond to empirical observations in our pilot study (see, e.g., analysis correlating demographics, work variables, and exposure Dawes, 2008). with the symptoms scales at each time point. Outcome variables. Following these initial tests, we tested our hypotheses regarding PTSD symptoms. The Impact of Event Scale—Revised the role of work-related exposure (exposure to client trauma and (IES–R; Weiss & Marmar, 1997) was used to assess symptoms of hours at work) for the outcome variables using multivariate anal- PTSD at Time 1 and Time 2. This instrument comprises 22 items yses. We used cross-lagged panel correlation path models to ex- derived from the PTSD criteria of the Diagnostic and Statistical plore the causal sequence between exposure to traumatic clients Manual of Mental Disorders (4th ed.; DSM–IV; American Psychi- and work hours at Time 1 and symptomatology at Time 2 (PTSD atric Association, 1994). Respondents were asked to rate each item or CES–D or SDS), using structural equation modeling (SEM) that on a scale of 0 (not at all), 1 (a little bit), 2 (moderately), 3 (quite assessed measurement errors for the dependent and independent a bit), and 4 (extremely), according to how distressed they had variables (Hoyle & Smith, 1994) with AMOS software (Version been by symptoms of intrusion, hyperarousal, and avoidance over 18.0.0; Arbuckle, 2009) and the maximum likelihood method. the past 7 days. All participants were asked to specifically link the Several components of these models are noteworthy. First, they symptoms to traumatic material related to a case or cases they had include two time points, and the effects of exposure and hours at encountered as part of their work. No timeframe was specified work on PTSD, depression, and functional impairment are esti- regarding when the material was encountered. The IES–R has mated. These aspects of the models are referred to as cross-lagged good psychometric properties (Creamer, Bell, & Failla, 2003) and effects. Second, the model also includes the influence of exposure has good convergent validity with other measures of PTSD and hours at work at the first time point on exposure and hours at (Ljubotina & Muslic, 2003). In the present study, we obtained work at the later time point. The same is true for PTSD, depres- internal consistency Cronbach’s reliability coefficients of ␣ϭ.79, sion, and functional impairment. These aspects of the model, .80, and .85, and ␣ϭ.80, .78, and .86, for avoidance, hyper- called autoregressive effects, can be thought of as indicators of the arousal, and intrusion subscales, at Time 1 and Time 2, respec- temporal stability of the measures. Estimations of these parameters tively. in the model control for the stability of the variables. Thus, any Depressive symptoms. The Center for Epidemiological Stud- cross-lagged effects can be considered effects that add predictive ies Depression Scale (CES–D; Radloff, 1977) is a 20-item scale power over and above that which can be simply obtained from the designed to measure severity of current depression in the general stability of the measures. Finally, note that exposure, hours at population and was used at Time 1 and Time 2. The items, each of work, PTSD, depression, and functional impairment are each al- which is assessed on a scale from 0 to 3, measure depressed mood, lowed to intercorrelate within each time point. These aspects of the feelings of guilt and worthlessness, feelings of helplessness and model are called synchronous correlations. Estimating these errors hopelessness, psychomotor retardation, loss of appetite, and sleep in the model allows for correlations between variances in PTSD or disturbances (Radloff, 1977). All participants were asked to report depression or functional impairment and exposure and hours at symptoms they had experienced in the past week. The CES–D is work that are not already explained by the influences of the in wide use and has acceptable levels of internal consistency variables from earlier time points. (Radloff, 1977). Extensive evidence from a variety of samples A nonsignificant chi-square has traditionally been used as a attests to the reliability and validity of the CES–D (Eaton, Mun- criterion for not rejecting an SEM; a nonsignificant chi-square taner, Smith, Tien, & Ybarra, 2004). In the present sample, the indicates that the discrepancy of the matrix of the parameters estimates of internal consistency Cronbach’s reliability coeffi- estimated based on the model being evaluated is not different from cients were ␣ϭ.91 and ␣ϭ.93 at Time 1 and Time 2, respec- the one based on the empirical data. Given the restrictiveness of tively. the chi-square approach for assessing model fit (Jöreskog & Sör- Functional impairment. The Sheehan Disability Scale (SDS; bom, 1993; Kenny & McCoach, 2003; Landry, Smith, Swank, & Sheehan, Harnett-Sheehan, & Raj, 1996) was used to assess the Miller-Loncar, 2000), we also used alternative criteria that reflect extent to which exposure to clients’ traumatic material interfered the real-world conditions of clinical research in addition to the 4 LEVIN, BESSER, ALBERT, SMITH, AND NERIA overall chi-square test of exact fit to evaluate the proposed models: also demonstrated for average number of hours worked as well (a) the ␹2/df ratio, (b) the root mean square error of approximation as for level of exposure to trauma-exposed clients. (RMSEA), (c) the comparative fit index (CFI), and (d) the non- Fifteen percent and 9% of the sample met screening criteria for normed fit index (NNFI). A model in which ␹2/df was Յ 2, CFI PTSD at Time 1 and Time 2, respectively (p Ͼ .18). A cutoff of and NNFI were greater than 0.95, and the RMSEA index was 1.5 (equivalent to a total score of 33) was found to provide the between 0.00 and 0.08 (Hu & Bentler, 1999) was deemed accept- highest levels of sensitivity/specificity when comparing the IES–R able. These moderately stringent acceptance criteria clearly reject with the PTSD Checklist (Creamer et al., 2003) and was used as a inadequate or poorly specified models, but accept for consideration cutoff for preliminary diagnosis of PTSD (see, e.g., Weiss, 2007). models that meet real-world criteria for reasonable fit and repre- Forty-three percent and 40.2% of the sample met screening criteria sentation of the data (Kelloway, 1998). Effect sizes were computed for depression at Time 1 and Time 2, respectively. A score of Ն16 using Cohen’s d (Cohen, 1992). has been used as the cutoff point for high likelihood of clinically significant depression (Radloff, 1977). Finally, 74.8% and 73.8% Results of the sample met screening criteria for functional impairment at Time 1 and Time 2, respectively. A score of Ն5 for any of the Descriptive Statistics three questions is associated with significant functional impairment (Sheehan et al., 1996). On average, participants had almost 16 years on the job (M ϭ ϭ 15.89 years, SD 11.03), were working more than 46 hr/week Bivariate Associations (M ϭ 46.07, SD ϭ 6.61), were from offices of (total staff) more than 10–20 people (M ϭ 2.40, SD ϭ 1), and were exposed to Table 2 provides a summary of the zero-order correlations for 41–60 clients within the past 3 months who had experienced or the study variables. Gender, age, years on the job, and size of been directly involved with trauma such as death, physical assault local office did not significantly correlate with any of the or abuse, domestic violence, rape, violence, or fire (M ϭ 3.16, outcome variables at either time point. Work-related exposure SD ϭ 1.23). was significantly correlated with depression (r ϭ .24, d ϭ 0.49, and r ϭ .22, d ϭ 0.45) and impairment (r ϭ .27, d ϭ 0.56, and Baseline to Follow-up Differences in PTSD, r ϭ .33, d ϭ 0.70) at both time points and at Time 2 with Depression, and Functional Impairment intrusion (r ϭ .24, d ϭ 0.49) and hyperarousal (r ϭ .27, d ϭ 0.56) symptoms. Average number of hours worked per week As shown in Table 1, no significant changes were found in correlated with depression (r ϭ .27, d ϭ 0.56), functional mean scores of baseline and follow-up for average number of impairment (r ϭ .31, d ϭ 0.65), intrusion (r ϭ .34, d ϭ 0.72), hours worked per week, depression, functional impairment, and and hyperarousal (r ϭ .30, d ϭ 0.63) at Time 1 but not with any hyperarousal symptoms. However, participants reported signif- of the outcome variables at Time 2. icantly lower mean scores of work-related exposure, intrusion, and avoidance at Time 2. As can also be seen from Table 1, Multivariable Analyses: Cross-Lagged Models correlations indicate that all symptom scores reported at Time 1 were significantly and strongly associated with the correspond- Prediction of PTSD symptoms (IES–R). At each time point, ing symptom scores reported at Time 2, indicating strong and we defined the latent PTSD construct (factor) using participants’ significant stability. Moreover, significant strong stability was intrusion, avoidance, and hyperarousal scores as its indicators

Table 1 Number of Working Hours, Caseload of Trauma-Exposed Clients, and Outcome Variables at Time 1 and Time 2

Time 1 Time 2 95% CI

Variable MSDMSDt(106) LL UL Cohen’s da r (Time 1 and 2)

ءءءAverage number of hours working 46.06 6.61 46.01 6.97 Ϫ0.22, ns Ϫ1.33 1.06 .58 ءءء59. 0.27 0.49 0.080 ءءWork-related exposureb 3.17 1.23 2.89 1.10 2.76 PTSD ءءء58. 0.63 0.43 0.23 ءءءIES–R Intrusion 0.76 0.59 0.43 0.55 6.30 ءءء52. 0.34 0.34 0.093 ءءءIES–R Avoidance 0.76 0.70 0.55 0.57 3.44 ءءءIES–R Hyperarousal 0.61 0.65 0.61 0.61 Ϫ0.06, ns Ϫ0.11 0.10 .62 ءءءCES–D 14.54 10.51 15.67 9.59 Ϫ1.43, ns Ϫ2.67 0.43 .68 ءءءSDS 10.17 6.95 9.82 6.78 0.57, ns Ϫ0.84 1.53 .62

Note. N ϭ 107. PTSD ϭ posttraumatic stress disorder; IES–R ϭ Impact of Event Scale—Revised; CES–D ϭ Center for Epidemiological Studies Depression Scale; SDS ϭ Sheehan Disability Scale; CI ϭ confidence interval; LL ϭ lower limit; UL ϭ upper limit. a Cohen’s d has been corrected for dependence between means using Morris and DeShon’s (2002) Equation 8. b “How many clients have you worked with, within the last three months, who had experienced or been directly involved with trauma such as death, physical assault or abuse, domestic violence, rape, violence or fire?” Participants were instructed to select the closest number on a 0–5 scale, where 0 ϭ none,1ϭ 1–20,2ϭ 21–40,3ϭ 41–60,4ϭ 61–80, and 5 ϭ 81 or more. .(p Ͻ .001 (two-tailed ءءء .p Ͻ .01 ءء ATTORNEYS WORKING WITH TRAUMA-EXPOSED CLIENTS 5

Table 2 Correlations Between Predictors and Outcome Variables at Time 1 and Time 2

PTSD subscale

Predictor Intrusion Avoidance Hyperarousal CES–D SDS

Time 1 Gendera Ϫ.14 Ϫ.26 Ϫ.19 Ϫ.23 Ϫ.23 Age .05 Ϫ.06 Ϫ.00 Ϫ.03 .05 Years on the job .00 Ϫ.03 Ϫ.04 Ϫ.07 .00 Size of local office .16 Ϫ.03 .19 Ϫ.03 .06 ءءء31. ءءء27. ءءء30. 16. ءءءAverage number of hours working .34 ءءء27. ءءءWork-related exposure .13 .07 .20 .24 Time 2 Gendera Ϫ.15 Ϫ.07 Ϫ.17 Ϫ.11 Ϫ.21 Age .08 .12 .05 .11 .00 Years on the job .00 .08 Ϫ.00 .07 Ϫ.08 Size of local office .03 .10 Ϫ.12 Ϫ.001 Ϫ.03 Average number of hours working .05 .11 .07 .18 .18 ءءء33. ء22. ءء27. 16. ءءWork-related exposure .24

Note. N ϭ 107. PTSD ϭ posttraumatic stress disorder; IES–R ϭ Impact of Event Scale—Revised; CES–D ϭ Center for Epidemiological Studies Depression Scale; SDS ϭ Sheehan Disability Scale. a Gender is a binary-coded variable (0 ϭ women, 1 ϭ men). To ensure that the overall chance of a Type I error remained less than .05, we applied a full Bonferroni correction. .(p Ͻ .01 (two-tailed ءءء while controlling for the autocorrelations among same measures effects of Time 1 hours at work on Time 2 exposure or CES–D errors (within-subject repeated measures). This cross-lagged SEM symptoms, ␤ϭ.07, t ϭ 0.76, ns, and ␤ϭ.10, t ϭ 1.37, ns, (see Figure 1) fit the observed data well, ␹2(21) ϭ 14.90, p ϭ .83, respectively). In contrast, Time 1 exposure had a noteworthy and ␹2/df ϭ 0.71; NNFI ϭ 1.0; CFI ϭ 1.0; RMSEA ϭ 0.0001, 95% CI statistically significant follow-up effect on CES–D symptoms, [0.000, 0.05]. This model showed a nonsignificant effect of Time exposure, and hours at work, such that higher levels of exposure at 1 PTSD symptoms on Time 2 exposure or hours at work, ␤ϭ–.06, one time point were related to an increased level of PTSD symp- t ϭ –0.65, ns, and ␤ϭ–.06, t ϭϪ1.39, ns, respectively, as well toms, ␤ϭ.20, t ϭ 2.70, p Ͻ .01, d ϭ 0.53, and a decreased level as nonsignificant effects of Time 1 hours at work on Time 2 of hours at work, ␤ϭ–.19, t ϭϪ2.26, p Ͻ .02. d ϭ 0.44, at the exposure or PTSD symptoms, ␤ϭ.07, t ϭ 0.82, ns, and ␤ϭ–.04, subsequent time point, as evidenced by the statistically significant t ϭ –0.46, ns, respectively. In contrast, Time 1 exposure had a cross-lagged parameters. These findings indicate that exposure noteworthy and statistically significant follow-up effect on PTSD significantly predicted or affected attorneys’ CES–D symptom- symptoms, exposure, and hours at work, such that higher levels of atology and hours spent at work at Time 2, and that attorneys’ exposure at one time point were related to an increased level of CED–D symptomatology or hours spent at work at Time 1 did not PTSD symptoms, ␤ϭ.28, t ϭ 3.43, p Ͻ .001, d ϭ 0.67, and a predict or affect levels of exposure at Time 2. Moreover, hours at decreased level of hours at work, ␤ϭ–.18, t ϭϪ2.17, p Ͻ .03, work at Time 1 affected exposure and CES–D symptoms at Time d ϭ 0.42, at the subsequent time point, as evidenced by the 2 indirectly through its association with exposure at Time 1. These statistically significant cross-lagged parameters. These findings associations were not altered when we controlled for gender, age, indicate that exposure significantly predicted or affected attor- years on the job, and size of local office and their associations with neys’ PTSD symptomatology and hours spent at work Time 2, predictors and outcomes. and that attorneys’ PTSD symptomatology or hours spent at To obtain the most parsimonious model and allow the evalua- work at Time 1 did not predict or affect levels of exposure at tion of the overall goodness of fit of the path model, we calculated Time 2. Moreover, hours at work at Time 1 affected exposure the final model in which we removed the nonsignificant paths and PTSD symptoms at Time 2 indirectly through its associa- found in the full model (i.e., of Time 1 hours at work on Time 2 tion with exposure at Time 1. These associations were not CES–D and exposure and of Time 1 CES–D on Time 2 exposure altered when we controlled for gender, age, years on the job, and hours at work). This model fit the observed data well, ␹2(4) ϭ and size of local office and their associations with predictors 3.14, p ϭ .54, ␹2/df ϭ 0.79; NNFI ϭ 1.0; CFI ϭ 1.0; RMSEA ϭ and outcomes. 0.0001, 95% CI [0.000, 0.08]. Prediction of depressive symptoms (CES–D). At each time Prediction of functional impairment (SDS). At each time point, we defined the observed variable overall CES–D scores. point, we defined the observed variable overall SDS scores. This This cross-lagged path model had zero degrees of freedom; thus, cross-lagged path model had zero degrees of freedom; thus, fit fit indices could not be estimated (see Figure 2). This model indices could not be estimated (see Figure 3). This model showed showed a nonsignificant effect of Time 1 CES–D symptoms on a nonsignificant effect of Time 1 SDS symptoms on Time 2 Time 2 exposure or hours at work, ␤ϭ–.07, t ϭ –0.82, ns, and exposure or hours at work, ␤ϭ–.09, t ϭϪ1.08, ns, and ␤ϭ–.04, ␤ϭ.03, t ϭ 0.36, ns, respectively, as well as nonsignificant t ϭ –0.53, ns, respectively, as well as nonsignificant effects of 6 LEVIN, BESSER, ALBERT, SMITH, AND NERIA

Figure 1. The cross-lagged model for the prediction of posttraumatic stress disorder (PTSD) symptoms. Rectangles indicate measured variables and large circles represent latent constructs. Small circles reflect residuals (e) or disturbances (d); bold numbers above or near endogenous variables represent the amount of variance explained (R2). Bidirectional arrows depict correlations and unidirectional arrows depict hypothesized directional or “causal” links. Standardized maximum likelihood parameters are used. Bold estimates are .p Ͻ .05 ء .statistically significant. The dotted paths indicate nonsignificant, “causal” links/associations. N ϭ 107 p Ͻ .001 (two-tailed). When we controlled for the effects of gender, age, years on the job, and ءءء .p Ͻ .01 ءء size of local office, the significant and nonsignificant effects, as presented in this figure, were not altered. These effects were removed to simplify the figure.

Time 1 hours at work on Time 2 exposure or SDS symptoms, ␤ϭ SDS and exposure and of Time 1 SDS on Time 2 exposure and .07, t ϭ 0.83, ns, and ␤ϭ.05, t ϭ .54, ns, respectively. In contrast, hours at work). This model fit the observed data well, ␹2(4) ϭ Time 1 exposure had a noteworthy and statistically significant 1.67, p ϭ .80, ␹2/df ϭ 0.42; NNFI ϭ 1.0; CFI ϭ 1.0; RMSEA ϭ follow-up effect on SDS symptoms, exposure, and hours at work, 0.0001, 95% CI [0.000, 0.09]. such that higher levels of exposure at one time point were related ␤ϭ ϭ Ͻ to an increased level of SDS symptoms, .20, t 2.39, p .01, Discussion d ϭ 0.47, and a decreased level of hours at work, ␤ϭ–.18, t ϭ Ϫ2.15, p Ͻ .03, d ϭ 0.42, at the subsequent time point, as To our knowledge, this study reports one of the first investiga- evidenced by the statistically significant cross-lagged parameters. tions in attorneys (or any helping professionals) examining longi- These findings indicate that exposure significantly predicted or tudinal changes in mental health outcome measures including affected attorneys’ SDS symptomatology and hours spent at work PTSD, depression, and functional impairment and the relationship at Time 2, and that attorneys’ SDS symptomatology or hours spent of these symptoms to work with trauma-exposed clients. The at work at Time 1 did not predict or affect levels of exposure at participants, 107 attorneys working in the Wisconsin State Public Time 2. Moreover, hours at work at Time 1 affected exposure and Defender’s Office, experienced continued stress over a 10-month SDS symptoms at Time 2 indirectly through its association with period as demonstrated by similar levels of depression, functional exposure at Time 1. These associations were not altered when we impairment, and PTSD hyperarousal at both time points. Further- controlled for gender, age, years on the job, and size of local office more, the percentage of attorneys who exceeded clinical thresholds and their associations with predictors and outcomes. for depression and functional impairment was unchanged over the To obtain the most parsimonious model and allow the evalua- period of the study. Although there was a modest but significant tion of the overall goodness of fit of the path model, we calculated decrease in the PTSD symptoms of intrusion and avoidance over the final model in which we removed the nonsignificant paths the 10-month period, there was no significant change in the num- found in the full model (i.e., of Time 1 hours at work on Time 2 ber of attorneys who scored above the threshold of clinically ATTORNEYS WORKING WITH TRAUMA-EXPOSED CLIENTS 7

Figure 2. The cross-lagged model for the prediction of depressive symptoms (Center for Epidemiological Studies Depression Scale [CES–D]). Rectangles indicate measured variables. Small circles reflect residuals (e). Bold numbers above or near endogenous variables represent the amount of variance explained (R2). Bidirectional arrows depict correlations and unidirectional arrows depict hypothesized directional or “causal” links. Stan- p Ͻ ء .dardized maximum likelihood parameters are used. Bold estimates are statistically significant. N ϭ 107 ,p Ͻ .001 (two-tailed). When we controlled for the effects of gender, age, years on the job ءءء .p Ͻ .01 ءء .05. and size of local office, the significant and nonsignificant effects, as presented in this figure, were not altered. These effects were removed to simplify the figure. significant PTSD. Total hours worked per week were unchanged, followed up with those who did not. Overall, these findings indi- but caseload of trauma-exposed clients did show a small but cate significant stability in levels of symptomatology over a 10- significant decrease. These decreases in the PTSD symptoms and month period. caseload of trauma-exposed clients may suggest that the attorneys Bivariate analysis revealed that average caseload of trauma- who participated in the follow-up survey were initially in less exposed clients significantly correlated with depression and func- distress, but there were no statistical differences found on any of tional impairment measures at both time points, whereas hours the symptom measures at baseline between the participants who worked per week only correlated with these measures at baseline.

Figure 3. The cross-lagged model for the prediction of functional impairment symptoms (Sheehan Disability Scale [SDS]). Rectangles indicate measured variables. Small circles reflect residuals (e). Bold numbers above or near endogenous variables represent the amount of variance explained (R2). Bidirectional arrows depict correlations and unidirectional arrows depict hypothesized directional or “causal” links. Standardized maximum .p Ͻ .01 ءء .p Ͻ .05 ء .likelihood parameters are used. Bold estimates are statistically significant. N ϭ 107 p Ͻ .001 (two-tailed). When we controlled for the effects of gender, age, years on the job, and size of local ءءء office, the significant and nonsignificant effects, as presented in this figure, were not altered. These effects were removed to simplify the figure. 8 LEVIN, BESSER, ALBERT, SMITH, AND NERIA

Furthermore, average caseload correlated with intrusion and hy- therapists reported similar levels of symptoms in both groups perarousal at follow-up but not at baseline. These inconsistent (Way, VanDeusen, Martin, Applegate, & Jandle, 2004). Moulden correlations between outcome variables and exposure measured by and Firestone (2007) concluded that work with perpetrators pre- hours worked and trauma-exposed client caseload mirror the cipitates symptoms in therapists via the same mechanisms (e.g., equivocal findings in the general secondary trauma literature (e.g., constructivist self-development theory) thought to cause symp- Kassam-Adams, 1999, and Creamer & Liddle, 2005, vs. Bosca- toms in any professional exposed to traumatic material. In light of rino, Figley, & Adams, 2004) and contrast with our earlier report these findings, it appears that exposure to traumatic material, (Levin et al., 2011) in which both factors correlated with symp- regardless of the relationship with the client, precipitates symptoms in the larger sample surveyed at the initial time point. toms. It should be noted that our questionnaire asked the attorneys Piwowarczyk et al. (2009), in a small sample, also found a rela- to quantify the number of clients who “had experienced or been tionship between caseload of trauma-exposed clients and increased directly involved with trauma.” Given that criminal defendants are symptoms in asylum attorneys. Gender, age, years on the job, and themselves often victims of trauma, the clients were most likely to size of office were not correlated with any of the outcome mea- be both perpetrators and victims. The attorneys in our study sures at either time point. commented that they frequently experienced negative feelings Consistent with our previous findings (Levin et al., 2011), these toward the people they were assigned to defend. Future research results on balance show that attorneys were more likely to exhibit should attempt to tease out the effects of sympathy versus revul- increased levels of symptomatology when working with trauma- sion toward the client as well as the effects of perpetrator versus exposed clients. However, the current study’s follow-up findings victim status on the development of symptoms. indicate cross-lagged effects in which exposure had a significant Although our prior report (Levin et al., 2011) identified the need effect over time on both hours worked and symptomatology such to support attorneys by addressing their work hours and their that higher levels of exposure at Time 1 were related to increased caseloads, this study suggests that given limited resources to effect symptoms and decreased hours worked 10 months later. Stated change, the focus should be on the attorneys with the largest differently, the cross-lagged findings indicate that over and above caseloads of trauma-exposed clients. In addition to developing the continued levels PTSD, depression, and functional impairment, strategies to decrease the size of these caseloads, perhaps by there was an additional unique effect of exposure on these outcome rotation of attorneys who receive these cases, the present study measures over time. Furthermore, no reciprocal effects were suggests that resources such as counseling and education should be found, that is, whereas higher levels of exposure predicted in- concentrated on supporting these attorneys. Although the efficacy creased levels of symptoms and decreased hours at work 10 of traditional approaches for assisting professionals who experi- months later, PTSD, depression, and functional impairment did not ence secondary trauma exposure (e.g., Gentry, Baranowsky, & have any effects on exposure over time. Dunning, 2002) including education about trauma and develop- Our findings suggest several possible conclusions. First, they ment of personal resilience have been challenged (Bober & Re- provide strong evidence that exposure to trauma-exposed clients gehr, 2006), the current findings again tilt toward more specific may be a vulnerability factor, given that higher levels of exposure emphasis on the work with the trauma-exposed clients rather than resulted in increased severity of symptoms and reduced time spent simply addressing general working conditions. Longitudinal stud- at work 10 months later. Furthermore, the lack of reciprocal effects ies of primary victims of trauma suggest that social support (Galea suggests that the attorneys surveyed may have ignored their symp- et al., 2002; Neria, Besser, Kiper, & Westphal, 2010) and coping toms of distress in making decisions about working hours and strategies (Mayou, Ehlers, & Bryant, 2002) affect long-term out- caseload. We might speculate that attorneys decreased their work come in victims beyond the intensity of exposure, highlighting the hours over the 10-month period in response to study participation need to examine these factors in future studies of outcomes in and the accompanying increased awareness of the phenomenon of helping professionals working with trauma-exposed populations. secondary trauma and not because of how they felt. At the same The present study has several limitations. First, the study used a time, they did not (or were unable to) decrease their caseloads of sample confined to attorneys who work as public defenders who trauma-exposed cases. In this regard, a number of participants of were demographically homogeneous, thus limiting generalizability the study stated (sometimes quite emphatically) in a comments to other attorneys and other helping professionals. Second, al- field at the end of the survey that they felt “powerless” to manage though this is the only study we are aware of that has surveyed their caseloads. Thus, alternatively, we might speculate that con- attorneys longitudinally, the sample of attorneys who repeated the tinued high caseloads of trauma-exposed clients were unavoidable. survey (N ϭ 107) was relatively small. One factor contributing to Our results and the results of other studies of professionals the follow-up rate of 45% may have been the lack of remuneration. working with perpetrators suggest a need to expand Figley’s The validity of our findings in this limited sample is bolstered by (1995) formulation that secondary trauma is “the stress resulting the lack of difference on any variables between the participants from helping or wanting to help a traumatized or suffering person” and the attorneys who did not repeat the study. A future study (p. 7). The present study, as well as our earlier study (Levin et al., should examine responses in a larger, more diverse sample across 2011), and the studies of Gomme and Hall (1995) with prosecutors a range of attorney types, that is, defense, prosecution, civil, and and Vrklevski and Franklin (2008) with criminal defense attor- even corporate, and if possible, compare them with other profes- neys, illustrate that work with perpetrators precipitates secondary sionals with different levels of exposure. A further limitation is the traumatic stress responses. Likewise, studies of sexual offender lack of a precise characterization of the specific types and frequen- therapists document secondary traumatic responses similar to cies of trauma (assault, homicide, rape, fire, etc.) encountered by those seen in therapists treating victims (see review by Moulden & the public defenders. This is a general limitation in the legal field; Firestone, 2007). The only study comparing these two groups of for example, Gomme and Hall (1995) characterized the impact of ATTORNEYS WORKING WITH TRAUMA-EXPOSED CLIENTS 9 work with domestic violence and incest on prosecutors, but they Eaton, W. W., Muntaner, C., Smith, C., Tien, A., & Ybarra, M. (2004). The did not quantify this caseload nor did they have a comparison with Revised Center for Epidemiologic Studies Depression Scale (RCES–D). prosecutors working, for instance, in homicide. This is another rich In M. E. Maruish (Ed.), The use of psychological assessment for treat- area for future exploration. ment planning and outcome assessment (3rd ed.) Retrieved from http:// Despite these limitations, our study investigated a unique phe- www.mdlogix.com/cesdrpaper.pdf nomenon, focusing on longitudinal exposure and symptoms that Eriksson, C. B., Kemp, H. V., Gorsuch, R. K., Hoke, S., & Foy, D. W. may well have significant ecological validity. The study focused (2001). Trauma exposure and PTSD symptoms in international relief and development personnel. Journal of Traumatic Stress, 14, 205–212. doi: on participants who reported on their experiences as they were 10.1023/A:1007804119319 occurring over a 10-month period. Moreover, to our knowledge, Figley, C. R. (1995). Compassion fatigue as secondary traumatic stress the present study represents the first attempt toward understanding disorder: An overview. In C. R. Figley (Ed.), Compassion fatigue: the relationships between attorneys’ exposure to trauma-exposed Coping with secondary traumatic stress disorder in those who treat the clients and symptomatology, over time, through the use of a traumatized (pp. 1–20). Levittown, PA: Brunner/Mazel. cross-lagged design. An important next step will be to use longi- Follette, V. M., Polusny, M. M., & Milbeck, K. (1994). Mental health and tudinal designs to explore the underlying mechanisms of attorneys’ law enforcement professionals: Trauma history, psychological symp- exposure-related symptoms. For example, one possible direction toms, and impact of providing services to child sexual abuse survivors. would be to examine the longitudinal role of various affect regu- Professional Psychology: Research and Practice, 25, 275–282. doi: lation strategies, coping mechanisms, and social support as poten- 10.1037/0735-7028.25.3.275 tial mediators and/or moderators of the effects of exposure over Galea, S., Ahern, J., Resnick, H., Kilpatrick, D., Bucuvalas, M., Gold, J., time. Taken as a whole, the present study points to the central role & Vlahov, D. (2002). Psychological sequelae of the September 11 of attorneys’ individual differences in exposure to trauma-exposed terrorist attacks in New York City. New England Journal of Medicine, clients in the development of symptoms of PTSD, depression, and 346, 982–987. doi:10.1056/NEJMsa013404 Gentry, E. J., Baranowsky, A. B., & Dunning, K. (2002). The accelerated functional impairment. recovery program (ARP) for compassion fatigue. In C. R. Figley (Ed.), Treating compassion fatigue (pp. 123–138). New York: Brunner- References Routledge. American Psychiatric Association. (1994). Diagnostic and statistical man- Gomme, I. M., & Hall, M. P. (1995). Prosecutors at work: Role overload ual of mental disorders (4th ed.). Washington, DC: Author. and strain. Journal of Criminal Justice, 15, 191–200. doi:10.1016/0047- Arbuckle, J. L. (2009). AMOS: A structural equation modeling program. 2352(95)00006-C Chicago, IL: Small Waters. Hasnain, N., Naz, I., & Bano, S. (2010). Stress and well-being of lawyers. Baird, S., & Jenkins, S. R. (2003). Vicarious traumatization, secondary Journal of the Indian Academy of Applied Psychology, 36, 165–168. traumatic stress, and burnout in sexual assault and domestic violence Hoyle, R. H., & Smith, G. T. (1994). Formulating clinical research hy- agency staff. Violence and Victims, 18, 71–86. doi:10.1891/vivi potheses as structural equation models: A conceptual overview. Journal .2003.18.1.71 of Consulting and Clinical Psychology, 62, 429–440. doi:10.1037/0022- Benjamin, G. A., Darling, E. J., & Sales, B. (1990). The prevalence of 006X.62.3.429 depression, alcohol abuse, and cocaine abuse among United States Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance lawyers. International Journal of Law and Psychiatry, 13, 233–246. structure analysis: Conventional criteria versus new alternatives. Struc- doi:10.1016/0160-2527(90)90019-Y tural Equation Modeling, 6, 1–55. doi:10.1080/10705519909540118 Bober, T., & Regehr, C. (2006). Strategies for reducing secondary or Jöreskog, K. G., & Sörbom, D. (1993). LISREL 8: Structural equation vicarious trauma: Do they work? Brief Treatment and Crisis Interven- modeling with the SIMPLIS command language. Chicago, IL: Scientific tion, 6, 1–9. doi:10.1093/brief-treatment/mhj001 Software International. Boscarino, J. A., Figley, C. R., & Adams, R. E. (2004). Compassion fatigue Kassam-Adams, N. (1999). The risks of treating sexual trauma: Stress and following the September 11 terrorist attacks: A study of secondary secondary trauma in psychotherapists. In B. H. Stamm (Ed.), Secondary trauma among New York City social workers. International Journal of traumatic stress (pp. 37–48). Lutherville, ME: Sidran Press. Emergency Mental Health, 6, 57–66. doi:10.1176/appi.ps.55.3.274 Kelloway, E. K. (1998). Using LISREL for structural equation modeling: Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159. A researcher’s guide. Newbury Park, CA: Sage. doi:10.1037/0033-2909.112.1.155 Kenny, D. A., & McCoach, D. B. (2003). Effect of the number of variables Creamer, M., Bell, R., & Failla, S. (2003). Psychometric properties of the on measures of fit in structural equation modeling. Structural Equation Impact of Event Scale—Revised. Behaviour Research and Therapy, 41, Modeling, 10, 333–351. doi:10.1207/S15328007SEM1003_1 1489–1496. doi:10.1016/j.brat.2003.07.010 Creamer, T. L., & Liddle, B. J. (2005). Secondary traumatic stress among Landry, S. H., Smith, K. E., Swank, P. R., & Miller–Loncar, C. L. (2000). disaster mental health workers responding to the September 11 attacks. Early maternal and child influences on children’s later independent Journal of Traumatic Stress, 18, 89–96. doi:10.1002/jts.20008 cognitive and social functioning. Child Development, 71, 358–375. Dawes, J. (2008). Do data characteristics change according to the number doi:10.1111/1467-8624.00150 of scale points used? An experiment using 5-point, 7-point and 10-point Levin, A. P., Albert, L., Besser, A., Smith, D., Zelenski, A., Rosenkranz, scales. International Journal of Market Research, 50, 61–77. S., & Neria, Y. (2011). Secondary trauma in attorneys and their admin- Devilly, G. J., Wright, R., & Varker, T. (2009). Vicarious trauma, second- istrative support staff working with trauma-exposed clients. Journal of ary traumatic stress or simply burnout? Effect of trauma therapy on Nervous and Mental Disease, 199, 946–955. doi:10.1097/NMD. mental health professionals. Australia and New Zealand Journal of 0b013e3182392c26 Psychiatry, 43, 373–385. doi:10.1080/00048670902721079 Levin, A. P., & Greisberg, S. (2003). Vicarious trauma in attorneys. Pace Eaton, W. W., Anthony, J. C., Mandel, W., & Garrison, R. (1990). Law Review, 2, 245–257. Occupations and the prevalence of major depressive disorder. Journal of Ljubotina, D., & Muslic, L. (2003). Convergent validity of four instru- Occupational Medicine, 32, 1079–1087. doi:10.1097/00043764- ments for measuring posttraumatic stress disorder. Review of Psychol- 199011000-00006 ogy, 10, 11–21. 10 LEVIN, BESSER, ALBERT, SMITH, AND NERIA

Lynch, D. R. (1997). The nature of occupational stress among public Regehr, C., Hemsworth, D., Leslie, B., Howe, P., & Chau, S. (2004). defenders. Justice System Journal, 19, 17–35. Predictors of post-traumatic distress in child welfare workers: A linear Mayou, R. A., Ehlers, A., & Bryant, B. (2002). Posttraumatic stress structural equation model. Children and Youth Services Review, 26, disorder after motor vehicle accidents: 3-year follow-up of a prospective 331–346. doi:10.1016/j.childyouth.2004.02.003 longitudinal study. Behaviour Research and Therapy, 40, 665–675. Sheehan, D. V., Harnett–Sheehan, K., & Raj, B. A. (1996). The measure- doi:10.1016/S0005-7967(01)00069-9 ment of psychopharmacology. International Clinical Psychopharmacol- Morris, S. B., & DeShon, R. P. (2002). Combining effect size estimates in ogy, 11, 89–95. doi:10.1097/00004850-199606003-00015 meta-analysis with repeated measures and independent-groups designs. Vrklevski, L., & Franklin, J. (2008). Vicarious trauma: The impact on Psychological Methods, 7, 105–125. doi:10.1037//1082-989X.7.1.105 solicitors of exposure to traumatic material. Traumatology, 14, 106–118. Moulden, H. M., & Firestone, P. (2007). Vicarious traumatization: The doi:10.1177/1534765607309961 impact on therapists who work with sexual offenders. Trauma, Violence, Way, I., VanDeusen, K. M., Martin, G., Applegate, B., & Jandle, D. & Abuse, 8, 67–83. doi:10.1177/1524838006297729 (2004). Vicarious trauma: A comparison of clinicians who treat survi- Neria, Y., Besser, A., Kipper, D., & Westphal, M. (2010). A longitudinal vors of sexual abuse and sexual offenders. Journal of Interpersonal study of posttraumatic stress disorder, depression and generalized anx- Violence, 19, 49–71. doi:10.1177/0886260503259050 iety disorder in civilians exposed to war trauma. Journal of Traumatic Weiss, D. S. (2007). The Impact of Event Scale: Revised. In J. P. Wilson Stress, 23, 322–330. doi:10.1002/jts.20522 Pearlman, L. A., & MacIan, P. S. (1995). Vicarious traumatization: An & C. S. Tang (Eds.), Cross–cultural assessment of psychological trauma empirical study of the effects of trauma work on trauma therapists. and PTSD (pp. 219–238). New York, NY: Springer. Professional Psychology: Research and Practice, 26, 558–565. doi: Weiss, D., & Marmar, C. (1997). The Impact of Event Scale—Revised. In 10.1037/0735-7028.26.6.558 J. Wilson & T. Keane (Eds.), Assessing psychological trauma and PTSD Piwowarczyk, L., Ignatius, S., Crosby, S., Grodin, M., Heeren, T., & (pp. 399–411). New York, NY: Guilford Press. Sharma, A. (2009). Secondary trauma in asylum lawyers. Bender’s Immigration Bulletin, 14, 263–269. Radloff, L. S. (1977). The CES–D scale: A self-report depression scale for Received October 13, 2011 research in the general population. Applied Psychological Measurement, Revision received December 23, 2011 1, 385–401. doi:10.1177/014662167700100306 Accepted February 16, 2012 Ⅲ Reprint permission granted by theWisconsin Defender Magazine

The

WISCONSIN Winter/Spring 2009 DEFENDER Volume 17, Issue 1 Keeping Legal Minds Intact: Mitigating Compassion Fatigue among Government Lawyers By Linda Albert*, LCSW, CSAC WisLAP Coordinator

What is compassion fatigue? There are several different terms often used to refer to the same phenomenon; to name a few: compassion fatigue, vicarious trauma, secondary traumatic stress, second hand shock and secondary stress reaction. Compassion fatigue is defined as the cumulative physical, emotional and psychological effects of being continually exposed to traumatic stories or events when working in a helping capacity. It has been studied extensively in social workers, nurses, doctors and therapists who work with victims of trauma. Recently researchers have begun to examine the impact upon legal professionals including lawyers doing criminal law or family law and judges. Compassion fatigue involves a cluster of symptoms such as, but not limited to, sleep disturbance, anxiety, intrusive thoughts, a sense of futility or pessimism about people, lethargy, isolation and irritability. The development of compassion fatigue involves neurophysiology and is best addressed from both the neurobiological and the social psychological research and perspectives.

Who is most at risk? Levin et al (2003) found that attorneys and judges who work in the field of criminal or family law are considered at higher risk of developing compassion fatigue compared to those who work in other areas of the law. These legal professionals listen day after day to stories of human induced violence. They read and re-read detailed documentation of the traumatic material within cases. Attorneys are often times in long term relationships with their clients thereby witnessing the impact of the trauma upon their client or their clients’ victim. They observe domestic violence victims re-entering into risky environments without regard for safety and throughout their work with victims, offenders and the system are expected to perform at the top of their game without being impacted by the traumatic material. After all, lawyers are taught not to show weakness, to deny, defend and deflect vulnerability, while staying emotionally detached at all times.

The reality is that government lawyers are human beings. Any person regardless of professional competence can develop compassion fatigue. The struggle for government lawyers is the assumption (both their own and that of others) that they will not be impacted by the work that they do. The reality can be quite different. Lawyers that are exposed to traumatic stories and events may have physiological reactions

*Linda Albert is a Licensed Clinical Social Worker and a Certified Alcohol and Drug Counselor. She received her Master’s Degree from UW-Madison in Social Work. She has professional assessment/ treatment/referral competencies in the areas of addictions, eating disorders, depression, anxiety, trauma and illness impacted by stress. Linda has worked over the past 25 years as an administrator, consultant, trainer and psychotherapist in a variety of settings including providing services to impaired professionals. She has done multiple presentations for conferences at the local, state and national level. Currently Linda is employed by the State Bar of Wisconsin as the WisLap Coordinator. such as increased heart rate, breathing rate and muscle tension. They can have emotional responses such as sadness, anger or fear. They may also experience changes in their assumptions about life, other people and issues of safety. Often lawyers will be unaware of these reactions or ignore or dismiss them as unimportant. These reactions are indicative of the physiological and psychological changes occurring within the mind/body due to the processes of empathy or identification, reactions of the autonomic nervous system and patterns of thinking. If left unchecked and unattended to these reactions wear on the mind and the body resulting in the above mentioned cluster of symptoms known as compassion fatigue. The results can be varying degrees of impairment for the attorney.

What places government lawyers at increased risk? Levin et al. (2003) found that compared to mental health providers and social service workers attorneys who worked with domestic violence and criminal defendants had “significantly higher levels of secondary traumatic stress and burnout”. Researchers went on to state that this is likely due to higher case loads, lack of supervision or support and lack of education in regards to the impact of ongoing exposure to traumatic material and events. Osofsky et al. (2008) also identified similar organizational and job issues which contribute to the development of compassion fatigue. Factors included high caseloads, minimal support from supervisors, lack of peer support, excessive paperwork, inadequate resources to meet demands and limited job recognition. These researchers also reported the impact of compassion fatigue upon the work environment listing such issues as increased absenteeism, impaired judgment, low motivation, lower productivity and high staff turnover.

These factors coupled with the culture of practicing law may discourage government lawyers from recognizing the signs of distress, disclosing if they are struggling or prevent them from seeking assistance. In contrast social service and mental health workers are educated about the potential impact of the work upon their mental and physical health and are encouraged to talk about it and address how the work affects them in order to lessen the impact. This is often done in a safe, confidential and supportive environment. Government lawyers and their office managers universally state they do not have this provision built in to their work environment, that they are bound by confidentiality and would lack the resources, time or energy to create this environment for themselves. However, some recognize the need for it.

Those working as public defenders or prosecutors may identify with some of the above. For example, prosecutors or public defenders involved in a long, arduous trial are seldom afforded the time to replenish and restore themselves following the trial. Instead they are likely to go forward the next day into another formidable case without the ability to take pause and reflect upon how the work is impacting them physically, emotionally or mentally. One lawyer stated, “I am expected to operate like a machine, often getting notices to be at four places at the same time and go from trial to trial with no regard for what I can reasonable do or what the impact might be on myself as a professional or a person”. Another lawyer expressed, “I am supposed to take it all in and not be affected by it; it’s like mental battering”.

What can legal organizations do? A review of the literature suggests that organizations that employ government lawyers first and foremost need to recognize and acknowledge that compassion fatigue exists and identify how it impacts the lawyer and the organization. Prevention strategies include reducing caseloads due to the correlation between high caseloads and the prevalence of compassion fatigue and educating government lawyers about what compassion fatigue is and how a person may be impacted while working with traumatic stories and events. Supervisors and managers would be astute to address this issue, educate their legal staff and encourage staff to debrief their high trauma cases on a regular basis in a supportive atmosphere. With the current culture of budget deficits, limited space and resources and increasing caseloads it is imperative (albeit difficult) for managers and organizations to adopt a strategy of how offices can address and mitigate this versus why they cannot.

What can legal professionals do? Whether an attorney, judge, doctor or a mental health professional the recommendations to mitigate or treat compassion fatigue are similar.

Awareness. It is important for government lawyers to understand what compassion fatigue is, be assessing for it through utilizing a survey, checklist or other instrument on a regular basis.

Debriefing. Talking on a regular basis with another government lawyer who understands and is supportive is seen as helpful. This involves talking about the traumatic material, how one thinks and feels about it, acknowledging how one is personally affected by it and putting a plan in place for balance.

Balance. Working on balance in all areas of one’s life is emphasized throughout the research on mitigating compassion fatigue. Because of the physiological changes that occur a holistic approach is best. Yes, this means establishing a healthy diet, sleep and exercise program (argh) which we all talk about but few of us actually attend to. Exercise and relaxation work can be beneficial in counteracting the impact on the autonomic nervous system. Working on healthy interpersonal relationships is also a good idea (even if we have been married or divorced a zillion years and live with small children, adolescents or have aging parents giving us the excuse to say “balance is impossible”). Most of us give up on finding balance as work and personal life just keeps pouring it on but the truth is there are probably steps we can take to simplify, to do less of, to ask for help or just plain stop trying to be all things to all people, including our clients. Sound familiar? Start thinking about how you can work on balance versus why you cannot.

Be intentional. If your life is out of whack, you have compassion fatigue, depression, anxiety, substance abuse problems or are just plain overwhelmed, put a plan in place for change. Work with your thoughts. Recognize and acknowledge that the skills you possess which contribute to your success as an attorney (motivated, perfectionist, achievement oriented, driven, fixer,) and the environment in which you work in may contribute to an imbalance in your life. Seeking balance encompasses a change in lifestyle which requires hard work addressing thoughts, emotions and behaviors. Intentionally seek assistance to help yourself implement change and redirect the thoughts that tell you, “I should be able to do this by myself”. Your new mantra can become, “I don’t have to do it all by myself”.

The good news: WisLAP can be a resource specifically for you. If you want to consult with a mental health professional or work with a trained attorney consider calling the Wisconsin Lawyers Assistance Program (WisLAP). WisLAP specializes in understanding and addressing the issues which face today’s legal professionals. The program offers free in house educational sessions or one on one consultation or assistance for problems like compassion fatigue, depression, anxiety, addictions or other challenges.

What is WisLAP?

WisLAP is a member service of the State Bar of Wisconsin. The program utilizes trained Wisconsin judges and attorneys who provide confidential assistance to judges, lawyers, law students and their families. Each request for help is treated with the same confidentiality as the lawyer-client relationship. WisLAP is exempt from reporting professional misconduct to the Office of Lawyer Regulation (OLR) or to the Judicial Commission. WisLAP does not ask callers to disclose their identity and does not keep case records. The program is designed to help members build on their strengths and provide support through the enhancement of physical, mental and emotional health. Confidential support is available 24/7 by calling 800-543-2625. Or contact Linda Albert, WisLAP Coordinator directly at 800-444-9404 ext 6172 or email [email protected].

References Figley, C.R., Compassion Fatigue: Coping with Secondary Traumatic Stress Disorder in Those Who Treat the Traumatized. New York: Brunner/Mazel, (1995)

Jaffe, P., Crooks, C., Dunford-Jackson, B., & Town, M. Vicarious Trauma in Judges: The Personal Challenge of Dispensing Justice. 54 Juv. & Fam. Ct. J. 1-9 (2003)

Levin, A.P., & Greisberg, S. Vicarious Trauma in Attorneys. Pace Law Review, 24, 245-252, (2003).

Murry, D., & Royer, J. Vicarious Traumatization: The Corrosive Consequences of Law Practise for Criminal Justice and Family law Practitioners, (2003).

Osofsky, J., Putnam, F., & Lederman, C. How to Maintain Emotional Health When Working with Trauma. Juv and Fam Court J. 58, no. 4 (2008).

Rothschild, B. Help for the Helper: The Psychophysiology of Compassion Fatigue and Vicarious Trauma. New York: W.W. Norton & Company, (2006)

Trippany, R., White Kress, V., & Wilcoxon, S.A. Preventing Vicarious Trauma: What Counselors Should Know When Working with Trauma Survivors. J of Counsel. & Devel. Vol 82, Winter (2004).

Vrklevski, L., & Franklin, J., Vicarious trauma: The Impact on Solicitors of Exposure to Traumatic Material. Traumatology, 14; 106 (2008) Online version found at http://tmt.sagepub.com/cgi/content/ abstract/14/1/106. ORIGINAL ARTICLE

Secondary Traumatic Stress in Attorneys and Their Administrative Support Staff Working With Trauma-Exposed Clients

Andrew P. Levin, MD,*Þ Linda Albert, LCSW,þ Avi Besser, PhD,§ Deborah Smith, JD,|| Alex Zelenski, MBA,þ Stacey Rosenkranz, PhD,¶ and Yuval Neria, PhDÞ#

STS and VT including female sex (Kassam-Adams, 1999), intensity Abstract: Although secondary trauma has been assessed in various groups of of the exposure (Creamer and Liddle, 2005; Erikson et al., 2001; mental health professionals, few studies, to date, have examined secondary Kassam-Adams, 1999), history of previous trauma (Brady et al., 1999; trauma among attorneys exposed to clients’ traumatic experiences. This study Bride et al., 2007; Kassam-Adams, 1999), and less experience on the examined indicators of secondary trauma among attorneys (N = 238) and their job (Pearlman and Mac Ian, 1995). Subsequent studies have suggested administrative support staff (N = 109) in the Wisconsin State Public Defender the primary importance of organizational and work-related factors Office. Attorney participants demonstrated significantly higher levels of post- compared with exposure (Baird and Jenkins, 2003; Devilly et al., 2009; traumatic stress disorder symptoms, depression, secondary traumatic stress, burn- Regehr et al., 2004) and have found no relationship with personal out, and functional impairment compared with the administrative support staff. trauma history (Boscarino et al., 2004; Ortlepp and Friedman, 2002; This difference was mediated by attorneys’ longer work hours and greater con- Schnaube and Frazier, 1995). Risk factors for BO include female sex, tact with clients who had experienced or had been directly involved with trauma. overwork, the slow and erratic pace of the work, lack of success, and Sex, age, years on the job, office size, and personal history of trauma did not the tendency of the work to raise personal issues (Jenkins and Baird, predict symptoms. These findings suggest a need to support attorneys experi- 2002; Maslach et al., 2001). encing these symptoms and to address high workloads as well as the intensity Drawing on the concepts of STS and VT, the ‘‘clinical’’ of contact with trauma-exposed clients. (practice-related) law literature was the first to address the impact of Key Words: Attorneys, secondary traumatic stress, PTSD, depression, lawyer-client relationship on the attorney (Meier, 1993; Silver, 1999) functional impairment, burnout. and the need for increased training of attorneys in managing the (J Nerv Ment Dis 2011;199: 946Y955) ‘‘face-to-face, long-term, and intensely personal relationship’’ that develops between client and attorney (Allegretti, 1993, p. 7). Early quantitative studies of attorneys focused on rates of depression, identifying a 20% rate of clinically significant depression in the attorneys he phenomenon of Secondary Traumatic Stress (STS; Figley, who were surveyed (Benjamin et al., 1990; Eaton et al., 1990). T 1995) or Vicarious Traumatization (VT; McCann and Pearlman, Only a handful of studies have attempted to characterize and 1990) have been described since the mid-1980s, roughly coinciding quantify secondary trauma and BO symptoms experienced by attorneys with the growth in treatments focused on clients who were victims of and delineate their relationship to risk factors. Using semistructured trauma. Originally described in therapists, secondary trauma occurs interviews of 23 Canadian prosecutors working with ‘‘sensitive cases’’ when the professional develops intrusive thoughts, avoidance and involving domestic violence and incest, Gomme and Hall (1995) withdrawal, and symptoms of tension and disturbed sleep related to found symptoms of demoralization, anxiety, helplessness, exhaustion, exposure to traumatic material presented by the client (Figley, 1995). and social withdrawal that were qualitatively linked to high caseloads In addition, the professional may develop alterations in ‘‘basic assump- and long work hours. Lynch (1997) reported that public defenders tions’’ about themselves, people, society, and safety (McCann and ranked work overload, the unpredictability of trials, the frequent lack Pearlman, 1990). In addition to STS and VT, professionals working of a defense, harsh sentences, arguing with prosecutors, and interac- intensely with clients develop Burnout (BO), an accumulation of tions with angry clients and families as the most frequent and intense stress and the erosion of idealism characterized by fatigue, poor sleep, sources of job stress but did not relate these to any symptom measures. headaches, anxiety, irritability, depression, hopelessness, aggression, More recently, Levin and Greisberg (2003) compared 55 attorneys cynicism, and substance abuse (Farber and Heifetz, 1982). In this working in family and criminal court with 87 mental health profes- study, we examined the impact of work with clients who have expe- sionals and 25 social service workers. Their results indicated that rienced or have been directly involved in trauma on attorneys and their compared with the other groups, attorneys demonstrated higher levels administrative support staff in the Wisconsin State Public Defender of secondary trauma and BO that were correlated with caseload. Office. Comparing 50 attorneys working in criminal courts with 50 working Available research among mental health and social service in the civil arena, Vrlevski and Franklin (2008) found more depres- providers has identified several risk factors for the development of sive symptoms, subjective stress, and changes in sense of safety and intimacy among the criminal attorneys. A personal history of multiple traumas predicted higher scores on measures of vicarious trauma, post- *Westchester Jewish Community Services, Hartsdale, NY; †Department of Psychia- try, College of Physicians and Surgeons, Columbia University, New York, NY; traumatic stress, and depression. Piwowarcyzy et al. (2009) reported ‡Wisconsin Lawyers Assistance Program, State Bar of Wisconsin, Madison, WI; that among 57 attorneys specializing in asylum cases, the hours per §Department of Behavioral Sciences and Center for Research in Personality, week devoted to those cases correlated with trauma score. All three of Life Transitions, and Stressful Life Events, Sapir Academic College, D. N. Hof these studies of distress in attorneys suggest a relationship between Ashkelon, Israel; ||Assigned Counsel Division, Wisconsin State Public Defender Office, Madison, WI; ¶Weill Medical College, Cornell University, New York, NY; exposure to trauma and distress but suffer from small sample size, and #New York State Psychiatric Institute, New York, NY. selection bias involving convenience samples, and relatively low per- Send reprint requests to Andrew P. Levin, MD, Westchester Jewish centage responses from the pool of possible subjects (Levin and Community Services, 141 North Central Avenue, Hartsdale, NY 10530. Greisberg, 2003; Piwowarcyzy et al., 2009; Vrlevski and Franklin, E-mail: [email protected]. Copyright * 2011 by Lippincott Williams & Wilkins 2008). ISSN: 0022-3018/11/19912<0946 The current study sought to address those limitations in a rel- DOI: 10.1097/NMD.0b013e3182392c26 atively larger study, assessing the relationships between exposure to

946 www.jonmd.com The Journal of Nervous and Mental Disease & Volume 199, Number 12, December 2011

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. The Journal of Nervous and Mental Disease & Volume 199, Number 12, December 2011 Secondary Traumatic Stress in Attorneys

clients’ traumatic experiences and a range of outcomes including as its board of directors and chief executive officers. Leadership at posttraumatic stress disorder (PTSD) symptoms, depression, func- both the Wisconsin Public Defenders Office and the Wisconsin Bar tional impairment, and STS and BO symptoms in attorneys and also reviewed and approved the study. The final sample contains 347 administrative support staff working at the Wisconsin State Public participants (an overall response rate of 73.20%) including 238 at- Defender Office. In light of previous studies, we hypothesized that torneys (response rate of 77.52%) and 109 administrative support staff a) the average number of hours working and the caseload of trauma- (response rate of 65.27%). exposed clients would predict higher symptom load and b) attorneys would experience greater symptoms than would administrative sup- Measures port staff because of their greater client involvement. Moreover, we conceptualized attorneys’ work-related exposure (hours per week work- Background and Trauma Exposure Assessments ing and number of trauma-exposed clients) as mediating variables1 Demographic and personal information included age, sex, job based on our interpretation of the literature on both exposure and description (attorney versus administrative support staff), number of STS. As such and consistent with the literature on exposure, our years on the job, average number of hours worked per week (for the primary hypothesis was that c) work-related exposure would serve as a previous 3 months), and size of local office (total staff) specified on a vehicle through which being directly versus indirectly exposed to 1-to-4 scale: less than 10 (1), 10 to 20 (2), 21 to 40 (3), and greater clients who had experienced or had been directly involved in trauma is than 40 (4). Because participants expressed a strong need to protect associated with psychological symptoms. Specifically, attorneys, in their anonymity, information regarding the specific office where the comparison with administrative support staff, were expected to report participant worked as well as ethnic origin was omitted. high levels of exposure, which, in turn, would be associated with their Personal history of trauma was gathered by asking, ‘‘Have you significantly higher levels of PTSD symptoms, depression, functional been a victim of any of the following types of trauma? Please estimate impairment, STS, and BO symptoms. Lastly, the study explores the numbers of incidents from childhood/adolescence (up to age 15).’’ relationship between personal characteristics such as age, sex, years Types of trauma were divided into six groups: a) physical assault or on the job, office size, and personal trauma history and the outcome abuse, b) sexual assault or abuse, c) witness to violence, d) other crime variables. Given that the findings have varied for these factors in victim, e) fire, and f) natural disaster. The question was repeated for previous studies, we did not predict specific effects for these inde- age 16 years and older. Sum scores were generated for each of the age pendent variables. periods for the a) total number of physical and sexual assault or abuse incidents and b) total number of nonphysical/nonsexual trauma incidents. METHODS Exposure to client trauma was assessed by asking, ‘‘How many clients have you worked with, within the last three months who had Participants and Procedures experienced or been directly involved with trauma such as death, We sampled participants for this study from the Wisconsin physical assault or abuse, domestic violence, rape, violence or fire?’’ State Public Defender Office. At the time of the study (in early 2010) Participants were instructed to select the closest number on a 0-to-5 there were a total of 474 potential participants, including 307 attor- scale: none (0), 1 to 20 (1), 21 to 40 (2), 41 to 60 (3), 61 to 80 (4), and neys and 167 administrative support staff, in the 38 offices across the 81 or more (5). state. The attorneys routinely interact closely with defendants in local jails, prisons, courthouses, and in their own offices. Cases run the gamut from mild violence or substance abuse to homicide and sexual Outcome Variables offenses such as rape or child abuse. In addition to hearing first-hand PTSD symptoms accounts, the attorneys review reports and photographs and have The Impact of Events ScaleYRevised (IES-R; Weiss and Marmar, contact with physical evidence such as bloody clothing. Administra- 1997) was used to assess the symptoms of PTSD. This instrument tive support staff typically performs brief financial eligibility eva- is composed of 22 items derived from the PTSD criteria according to luations in their offices and at times, at the jail. On occasion, the DSM-IV (American Psychiatric Association, 1994). Respondents defendants spontaneously relate details of their offense to the support were asked to rate each item on a scale of 0 (not at all), 1 (a little bit), staff, who also have contact with reports and photographs. 2 (moderately), 3 (quite a bit), and 4 (extremely), according to how Potential participants received encouragement to participate in distressed they had been by symptoms of intrusion, hyperarousal, and the study from the Wisconsin State Public Defender Office and the avoidance over the past 7 days. All participants were asked to specifi- State Bar of Wisconsin as part of a program to raise awareness about cally link the symptoms to traumatic material related to a case or cases stress. Survey materials were made available online by the survey they had encountered as part of their work. No time frame was speci- office of the State Bar of Wisconsin. Potential participants received an fied regarding when the material was encountered. The IES-R has good email providing the necessary link to the questionnaires and were psychometric properties (Creamer et al., 2003) and has good conver- encouraged to complete the survey from personal computers on the gent validity with other measures of PTSD (Ljubotina and Muslic, job site. All subjects were provided with informed consent in the form 2003). In the present study, we obtained internal consistency Cronbach’s of a cover letter at the start of the online survey packet. Proceeding to alpha reliability coefficients of > = 0.80, 0.82, and 0.87, for avoidance, the questionnaire indicated consent. Participation was voluntary and hyperarousal, and intrusion, respectively. The maximum score for the anonymous. The research proposal was reviewed and approved by the scale is 88; a cutoff of 1.5 (equivalent to a total score of 33) was found Westchester Jewish Community Services Research Committee as well to provide the best diagnostic accuracy (Creamer et al., 2003).

Depressive symptoms 1Baron and Kenny (1986) characterize mediation as a case in which a variable, such as exposure, functions as a Bgenerative mechanism through which a focal The Center for Epidemiological Studies Depression Scale independent variable [such as attorney vs. support staff] is able to influence the (CES-D; Radloff, 1977) is a 20-item scale designed to measure the dependent variable of interest[ (p. 1173; see also Frazier et al., 2004). Mediation severity of current depression in the general population. The items, occurs when an external variable such as exposure better explains a relationship between a predictor, such as being directly (attorneys) versus indirectly (adminis- each of which is assessed on a scale from 0 to 3, measure depressed trative support staff) exposed to trauma-exposed clients, and an outcome, such mood, feelings of guilt and worthlessness, feelings of helplessness and as various symptoms (Frazier et al., 2004). hopelessness, psychomotor retardation, loss of appetite, and sleep

* 2011 Lippincott Williams & Wilkins www.jonmd.com 947

disturbances (Radloff, 1977). All participants were asked to report used as a criterion for not rejecting an SEM model; a nonsignificant symptoms they had felt in the past week. The CES-D is in wide use chi-square value indicates that the discrepancy of the matrix of the and has acceptable levels of internal consistency (Radloff, 1977). parameters estimated based on the model being evaluated is not Extensive evidence from a variety of samples attests to the reliability different from the one based on empirical data. Because of the re- and validity of the CES-D (Eaton et al., 2004). In the present sample, strictiveness of the chi-square approach for assessing model fit (Bentler the estimate of internal consistency Cronbach’s alpha reliability co- and Bonnet, 1980; Jo¨reskog and So¨rbom, 1993; Kenny and McCoach, efficient was 0.90. A score of 16 or higher (of a possible maximum 2003; Landry et al., 2000), we also used alternate criteria that reflect the of 60) has been used as the cutoff point for high likelihood of clini- real-world conditions of clinical research, in addition to the overall chi- cally significant depression (Radloff, 1977). square test of exact fit to evaluate the proposed models: a) the chi- Functional impairment levels square/df ratio, b) the root mean square error of approximation (RMSEA), c) the comparative fit index (CFI), and d) the nonnormed fit The Sheehan Disability Scale (SDS; Sheehan et al., 1996) was index (NNFI). A model in which the chi-square/df was 2 or less, CFI used to assess the extent to which exposure to clients’ traumatic and NNFI were greater than 0.95, and the RMSEA index was between material interfered with functioning in three spheres. Participants 0.00 and 0.06 with confidence intervals between 0.00 and 0.08 (Hu and rated the following question (in three forms): ‘‘My feelings about the Bentler, 1999) was deemed acceptable. These moderately stringent clients and cases at work have disrupted my (work, social life/leisure, acceptance criteria clearly reject inadequate or poorly specified models or family life/home responsibilities)’’ on a 0-to-10 visual analogue while accepting for consideration models that meet real-world criteria scale with the following descriptions: none (0), mild (1 to 3), moderate for reasonable fit and representation of the data (Kelloway, 1998). (4 to 6), severe (7 to 9), and very severe (10). In the present sample, the estimate of internal consistency Cronbach’s alpha reliability co- efficient was 0.92. According to the scale’s authors, a score of 5 or RESULTS higher for any of the three questions is associated with significant functional impairment (Sheehan et al., 1996). Group Differences Levels of STS and BO We first compared the attorneys and the administrative support staff groups on background and work characteristics (Table 1), work- The Professional Quality Of Life Scale Version 5 (ProQOL5; related exposure and personal history of previous trauma (Table 1), Stamm, 2010) is a 30-item questionnaire broken into three 10-item and the study outcome variables (Table 2). No significant differences groups measuring Compassion Satisfaction (CS), STS, and BO. The were found with regard to age and size of local office. However, as CS dimension (CS) ‘‘is about the pleasure you derive from being shown in Table 1, the administrative support staff group has signifi- able to do your work well’’ (Stamm, 2010, p. 12), with higher scores cantly fewer men than the attorneys group, and participants in the indicating greater work satisfaction. STS items measure fear, sleep attorneys’ group reported significantly more years on the job and of difficulties, intrusive images, or avoiding reminders of the person’s hours per week working compared with the administrative support traumatic experiences. BO items measure feelings of hopelessness staff group. No significant differences were found for childhood or and difficulties in dealing with work. Higher scores on these dimen- adulthood-related exposure variables. However, as shown in Table 1, sions indicate more distress. Participants were instructed to answer participants in the attorneys group reported working with significantly questions with respect to their reactions and symptoms in the previous more clients who experienced or were directly involved in trauma 30 days as related to work at the Wisconsin State Public Defender compared with the administrative support staff group. Office. Responses were scored on a 1-to-5 visual analogue scale, with Comparing attorneys and support staff on outcome variables never (1), rarely (2), sometimes (3), often (4), and very often (5). In the (Table 2), attorneys had significantly higher mean scores on all present sample, the estimates of internal consistency Cronbach’s alpha measures except CS, the latter being lower among attorneys than reliability coefficients were 0.90, 0.85, and 0.83 for CS, STS, and BO, among administrative support staff. Furthermore, significantly more respectively. These are similar to alpha coefficients reported by Stamm participants in the attorney group met screening criteria for PTSD (2010): 0.88, 0.81, and 0.75 for CS, STS, and BO, respectively. (11% vs. 1%), depression (39.5% vs. 19.3%), functional impairment Analysis of the scale produces Z scores that are then converted to (74.8% vs. 27.5%), BO (37.4% vs. 8.3%), and STS (34% vs. 10.1%) T-scores, with a mean (SD) of 50 (10). A T-score greater than 57 for compared with the administrative support group. Only a minority of CS or greater than 56 for STS and BO are above the 75th percentile attorneys (19.3%) and administrative supports staff (25.7%) reported in samples used in the development of the scale (Stamm, 2010). CS above the 75th percentile level (the groups did not differ) com- Data Analysis pared with norms for the ProQOL5 CS. Descriptive statistics were first calculated for demographics, trauma history, and work and exposure variables and compared be- Bivariate Associations tween groups using t-tests. Groups were compared regarding sex Table 3 provides a summary of the zero-order correlations for differences using chi-square analyses. Mean scores for the IES-R, all of the study variables. Sex, age, years on the job, size of local CES-D, SDS, and the three subscales of the ProQOL5 were calculated office, and a personal history of childhood or adult trauma did not and compared between groups using t-tests. In addition, the cutoff significantly correlate with any of the outcome variables. Group scores for each of the measures for the two groups were compared membership (attorneys vs. administrative support staff) was signifi- using chi-square analyses. We then performed a bivariate analysis cantly associated with all outcomes, except with the ProQOL5 CS correlating demographics, work variables, exposure and trauma his- scale, with attorneys reporting higher scores for symptoms and im- tory with the symptoms scales. pairment. In addition, work-related exposure as measured by the av- After these initial tests, we tested our hypotheses regarding the erage number of hours working and the number of clients worked with mediating role of work-related exposure for the outcome variables in the last 3 months who experienced or were directly involved with using multivariate analyses with an structural equation modeling trauma were both significantly and positively correlated with symp- (SEM; Hoyle and Smith, 1994) strategy that assessed measurement tom measures, again with the exception of the ProQOL5 CS scale. For errors for the dependent and independent variables using AMOS each of the three variables with significant correlations to outcome software (Version 18.0.0; Arbuckle, 2009) and the maximum likeli- variables, the strongest correlations were consistently seen with BO hood method. A nonsignificant chi-square value has traditionally been and functional impairment.

948 www.jonmd.com * 2011 Lippincott Williams & Wilkins

TABLE 1. Background and Trauma Exposure Variables Among Attorneys and Administrative Support Staff Administrative Background Variables Attorney (N = 238) Support Staff (N = 109) Statistic Sex Female 132 94 Male 106 13 W2 = 34.29** Mean SD Mean SD t (df = 343) Effect Size (d) Age 45.72 11.00 45.07 11.32 0.50, ns 0.06 Years on the job 15.22 10.26 12.11 8.47 2.65* 0.31 Average number of hours working 46.43 9.08 34.73 9.74 10.72** 1.24 Size of local office 2.39 1.02 2.53 1.00 j1.17, ns j0.14 Trauma exposure variables Childhood trauma Physical and sexual abuse 3.16 15.37 3.27 15.13 j0.60, ns j0.07 Not physical and sexual abuse 4.02 16.37 1.50 5.90 1.54, ns 0.18 Adulthood trauma Physical and sexual abuse 4.90 15.43 3.96 11.26 0.57, ns 0.07 Not physical and sexual abuse 3.59 13.79 1.78 5.88 1.31, ns 0.15 Work-related trauma Number of clients working with in the last 3 mos who 3.20 1.299 1.98 1.455 7.70** 0.89 experienced or were directly involved with trauma *p G 0.01 (two-tailed). **p G 0.001 (two-tailed). ns indicates not significant.

Multivariable Analyses PTSD symptoms (A = 0.26, t = 4.833, p G 0.0001). This model significantly explained 7% of the variance in PTSD symptoms. The Mediating Models In testing our primary hypothesis that work-related exposure Mediational association model variables mediate the relationships between groups and PTSD symp- We tested whether work-related exposure (the mediators) sig- toms (IES-R), depressive symptoms (CES-D), functional impartment nificantly reduced (accounted for) the direct relation between groups (SDS), and levels of STS and BO (ProQOL5), we followed Baron and and PTSD symptoms (the outcome). To do this, we specified a model Kenny’s (1986) criteria for mediation, according to which, a) there in which groups had a direct path to PTSD symptoms, as well as an must be a significant association between the predictor and criterion indirect path through work-related exposure variables (controlling for variables; b) in an equation including both the mediator and the cri- the shared variance among mediators). The mediational model fit the 2 2 terion variable, there must be a significant association between the observed data well (W [6] = 6.346, p = 0.386, W /df = 1.06, NNFI = 1.0, predictor and the mediator, and the mediator must be a significant CFI = 1.0, RMSEA = 0.01 [CI, 0.000 to 0.07]). As noted earlier, the predictor of the criterion variable; and c) there must be a decrease in direct path from groups to PTSD symptoms was significant. However, the direct relationship between the independent and the dependent this path became significantly weaker (A =0.09,t = 1.46, not signifi- variables (Baron and Kenny, 1986; Kenny et al., 1998). If the sig- cant [ns]) when hours at work (z¶ =3.06,p G 0.01) and exposure to nificant direct relationship between the predictor and the criterion trauma-exposed clients (z¶ = 3.003, p G 0.01) were included in the variables decreases when both the mediator and the predictor variable model. As shown in Figure 1, attorneys were significantly associated are included in the equation, then the obtained pattern is consistent with higher hours at work (A =0.50;t =10.71,p G 0.0001), which, in with the mediation hypothesis. If the direct association approaches turn, was associated with PTSD symptoms (A =0.20;t =3.18,p G zero, the mediator fully (although not necessarily exclusively) 0.001); moreover, attorneys were significantly associated with higher accounts for the relation between the predictor and the criterion exposure to trauma-exposed clients (A =0.39;t = 7.76, p G 0.0001), (Baron and Kenny, 1986). As a further test of mediation, MacKinnon which, in turn, was associated with PTSD symptoms (A =0.19;t =3.23, et al.’s (2002) z¶ test was used to examine the significance of the p G 0.001). Therefore, the work-related exposure variables mediated indirect relationship between the independent variable and the de- (albeit not exclusively) the attorneys’ vulnerability to PTSD symptoms. pendent variable via the hypothesized mediator. This model significantly explained 14% of the variance in PTSD symptoms. Therefore, when work-related exposure (the mediators) was Models for the Prediction of PTSD symptoms (IES-R) included in the model, it added a significant 7% to the explained var- Direct association model iance in PTSD symptoms. We first confirmed the existence of a significant direct relation between groups and PTSD symptoms. We defined the latent PTSD Models for the Prediction of Functional Impairment construct (factor) using participants’ intrusion, avoidance, and hy- Levels (SDS) perarousal scores as its indicators. This model fit the observed data Direct association model well (W2[2] = 2.081, p = 0.35, W2/df = 1.04, NNFI = 1.0, CFI = 1.00, We first confirmed the existence of a significant direct relation RMSEA = 0.01 [confidence interval (CI), 0.000 to 0.08]). As pre- between groups and functional impairment. We defined the latent dicted, attorneys were significantly associated with high levels of SDS construct (factor) using the participants’ SDS scales scores as its

* 2011 Lippincott Williams & Wilkins www.jonmd.com 949

TABLE 2. Means, SDs, and Prevalence of Cutoff Scores for Outcome Variables Attorney Administrative Support (N = 238; 69%) Staff (N = 109; 31%) Mean SD Mean SD t (df = 343) Effect Size (d) PTSD IES-R intrusion 0.73 0.58 0.46 0.34 4.52** 0.52 IES-R avoidance 0.65 0.65 0.33 0.49 4.42** 0.51 IES-R hyperarousal 0.55 0.65 0.25 0.44 4.31** 0.50 G33 212 108 W2 = 10.21** 933 26 (11%) 1 (0.92%) CES-D 14.08 10.27 8.91 7.68 4.66** 0.54 G16 144 88 W2 = 13.11** 916 94 (39.5%) 21 (19.3%) SDS 9.80 6.77 3.61 4.57 8.58** 0.99 G56079W2 = 66.30** 95 178 (74.8%) 30 (27.5%) ProQOL5a-CS 34.92 6.53 36.62 6.46 2.24* 0.26 G57 192 81 W2 = 2.05, ns 957 46 (19.3%) 28 (25.7%) ProQOL5a-BO 27.36 6.09 21.57 5.36 8.47** 0.98 G56 149 100 W2 = 34.99** 956 89 (37.4%) 9 (8.3%) ProQOL5a-STS 21.20 5.91 16.82 4.80 6.73** 0.78 G56 157 98 W2 = 21.30** 956 81 (34%) 11 (10.1%) aAt the 75th percentile. *p G 0.05 (two-tailed). **p G 0.001 (two-tailed). PTSD indicates posttraumatic stress disorder; IES-R, Impact of Events ScaleYRevised; CES-D, Center for Epidemiological Studies Depression Scale; SDS, Sheehan Disability Scale; ProQOL5, Professional Quality of Life Scale version 5; CS, Compassion Satisfaction; BO, Burnout; STS, Secondary Traumatic Stress; ns, not significant.

indicators. This model fit the observed data well (W2[2] = 0.70, p = with high levels of functional impairment symptoms (A = 0.44, t = 0.71, W2/df = 0.35, NNFI = 1.0, CFI = 1.00, RMSEA = 0.000 [CI, 8.370, p G 0.0001). This model significantly explained 20% of the 0.000 to 0.07]). As predicted, attorneys were significantly associated variance in SDS.

TABLE 3. Correlations Between Predictors and Outcome Variables PTSD (IES-R) ProQOL5 Intrusion Avoidance Hyper-Arousal CS BO STS CES-D SDS Groupa 0.24* 0.24* 0.23* j0.12 0.42* 0.34* 0.24* 0.42* Sexb j0.00 j0.07 j0.04 0.07 0.02 j0.09 j0.04 j0.05 Age 0.01 j0.06 j0.05 0.07 j0.03 j0.05 j0.09 j0.07 Years on the job 0.10 0.04 0.05 0.00 0.09 0.09 0.03 0.08 Average number of hours working 0.29* 0.26* 0.25* j0.05 0.38* 0.37* 0.26* 0.40* Size of local office 0.12 0.01 0.10 0.10 j0.02 j0.00 0.04 0.07 Childhood physical and sexual abuse 0.12 0.11 0.16 j0.04 0.10 0.15 0.13 0.15 Childhood not physical and sexual abuse j0.01 j0.01 0.06 j0.03 0.08 0.07 0.05 0.04 Adulthood physical and sexual abuse 0.02 j0.02 0.04 j0.03 0.05 0.05 0.03 j0.00 Adulthood not physical and sexual abuse j0.00 j0.04 0.02 j0.02 0.04 0.02 0.01 j0.02 Work-related exposure 0.24* 0.28* 0.27* j0.17 0.38* 0.31* 0.30* 0.37* To ensure that the overall chance of a type I error remained less than 0.05; we applied a full Bonferroni correction. aGroup is a binary-coded variable (0, administrative support staff; 1, attorney). bSex is a binary-coded variable (0, women; 1, men). *p G 0.001 (two-tailed). PTSD indicates posttraumatic stress disorder; IES-R, Impact of Events ScaleYRevised; CES-D, Center for Epidemiological Studies Depression Scale; SDS, Sheehan Disability Scale; ProQOL5, Professional Quality of Life Scale version 5; CS, Compassion Satisfaction; BO, Burnout; STS, Secondary Traumatic Stress.

950 www.jonmd.com * 2011 Lippincott Williams & Wilkins

Models for the Prediction of Levels of STS and BO (ProQOL5) Direct association model We first confirmed the existence of a significant direct relation between groups and ProQOL5 STS and BO. We defined the latent ProQOL5 construct (factor) using participants’ STS and BO scales scores as its indicators. This model has zero degrees of freedom; thus, fit indices could not be estimated. As predicted, attorneys were significantly associated with high levels of STS and BO symptoms (A = 0.45, t = 6.74, p G 0.0001). This model significantly explained 20% of the variance in ProQOL5 STS and BO.

Mediational association model We tested whether work-related exposure (the mediators) significantly reduced (accounted for) the direct relation between groups and STS and BO symptoms (the outcome). To do this, we specified a model in which groups had a direct path to ProQOL5 symptoms, as well as an indirect path through work-related exposure variables (controlling for the shared variance among mediators). The mediational model fit the observed data well (W2[2] = 2.939, p =0.23,W2/df =1.47, NNFI = 1.0, CFI = 1.0, RMSEA = 0.004 [CI, 0.000 to 0.08]). As noted earlier, the direct path from groups to ProQOL5 symptoms was significant. However, this path became significantly weaker (A =0.24, t = 3.866, p G 0.0001) when hours at work (z¶ =3.60,p G 0.001) and FIGURE 1. Mediational model for PTSD symptom levels (IES-R). Rectangles indicate measured variables and large circles represent latent constructs. Small circles reflect residuals (e) or disturbances (d); bold numbers above or near endogenous variables represent the amount of variance explained (R2). Unidirectional arrows depict hypothesized directional or ‘‘causal’’ links. Standardized maximum likelihood parameters are used. Bold estimates are statistically significant. GROUPS is a binary-coded variable (0, administrative support staff; 1, attorney). IES-R indicates Impact of Events ScaleYRevised.

Mediational association model We tested whether work-related exposure (the mediators) significantly reduced (accounted for) the direct relation between groups and functional impairment symptoms (the outcome). To do this, we specified a model in which groups had a direct path to SDS symptoms, as well as an indirect path through work-related exposure variables (controlling for the shared variance among mediators). The mediational model fit the observed data well (W2[6] = 6.103, p = 0.412, W2/df = 1.02, NNFI = 1.0, CFI = 1.0, RMSEA = 0.007 [CI, 0.000 to 0.07]). As noted earlier, the direct path from groups to SDS symptoms was significant. However, this path became significantly weaker (A = 0.25, t =4.24,p G 0.0001) when hours at work (z¶ = 3.83, p G 0.001) and exposure to clients’ traumatic events (z¶ =3.60, p G 0.001) were included in the model. As shown in Figure 2, attorneys were significantly associated with higher hours at work (A =0.50;t = 10.69, p G 0.0001), which, in turn, was associated with FIGURE 2. Mediational model for functional impairment levels SDS symptoms (A =0.21;t =4.09,p G 0.0001); moreover, attorneys (SDS). Rectangles indicate measured variables and large circles were significantly associated with higher exposure to trauma-exposed represent latent constructs. Small circles reflect residuals (e) clients (A =0.39;t =7.78,p G 0.0001), which, in turn, was associated or disturbances (d); bold numbers above or near endogenous with SDS symptoms (A =0.19;t = 4.03, p G 0.0001). Therefore, work- variables represent the amount of variance explained (R2). related exposure variables mediated (albeit not exclusively) the attor- Unidirectional arrows depict hypothesized directional or neys’ vulnerability to functional impairment symptoms. This model ‘‘causal’’ links. Standardized maximum likelihood parameters significantly explained 29% of the variance in SDS. Therefore, when are used. Bold estimates are statistically significant. GROUPS work-related exposure (the mediators) was included in the model, it is a binary-coded variable (0, administrative support staff; added a significant 9% to the explained variance in SDS. 1, attorney). SDS indicates Sheehan Disability Scale.

* 2011 Lippincott Williams & Wilkins www.jonmd.com 951

t =4.67,p G 0.0001). This model significantly explained 6% of the variance in CES-D symptoms.

Mediational association model We tested whether work-related exposure (the mediators) significantly reduced (accounted for) the direct relation between groups and depressive symptoms (the outcome). To do this, we specified a model in which groups had a direct path to CES-D symptoms, as well as an indirect path through work-related exposure variables (controlling for the shared variance among mediators). This model (Fig. 4) has zero degrees of freedom; thus, fit indices could not be estimated. As noted earlier, the direct path from groups to CES-D symptoms was significant. However, this path became significantly weaker (A = 0.08, t = 1.39, ns) when hours at work (z¶ = 2.45, p G 0.05) and exposure to trauma-exposed clients (z¶ = 3.20, p G 0.01) were included in the model. As shown in Figure 4, attorneys’ were significantly associated with higher hours at work (A = 0.50; t = 10.72, p G 0.0001), which, in turn, was associated with CES-D symptoms (A = 0.15; t = 2.51, p G 0.05); moreover, attorneys were significantly associated with higher exposure to trauma-exposed clients (A = 0.39; t = 7.76, p G 0.0001), which, in turn, was associated with CES-D symptoms (A = 0.22; t =3.90,p G 0.0001). Therefore, work-related exposure variables mediated (albeit not exclusively) the attorneys’ vulnerability to depressive symptoms. This model significantly explained 12% of the variance in CES-D symptoms. Therefore, when work-related exposure

FIGURE 3. Mediational model for secondary traumatic stress and burnout levels (ProQOL5). Rectangles indicate measured variables and large circles represent latent constructs. Small circles reflect residuals (e) or disturbances (d); bold numbers above or near endogenous variables represent the amount of variance explained (R2). Unidirectional arrows depict hypothesized directional or ‘‘causal’’ links. Standardized maximum likelihood parameters are used. Bold estimates are statistically significant. GROUPS is a binary-coded variable (0, administrative support staff; 1, attorney). ProQOL5 indicates Professional Quality of Life Scale version 5. exposure to trauma-exposed clients (z¶ = 3.85, p G 0.001) were included in the model. As shown in Figure 3, attorneys were significantly associated with higher hours at work (A = 0.50; t = 10.70, p G 0.0001), which, in turn, was associated with ProQOL5 symptoms (A = 0.23; t =3.81,p G 0.0001); moreover, attorneys were significantly associated with higher exposure to trauma-exposed clients (A =0.39;t =7.80,p G 0.0001), which, in turn, was associated with ProQOL5 symptoms (A =0.25;t =4.38,p G 0.0001). Therefore, work- related exposure variables mediated (albeit not exclusively) the attorneys’ vulnerability to STS and BO symptoms. This model significantly explained 32% of the variance in ProQOL5 STS and BO. Therefore, when work-related exposure (the mediators) was included in the model, it added a significant 12% to the explained variance in ProQOL5 STS and BO. FIGURE 4. Mediational Model for Depressive Symptoms Levels (CES-D). Rectangles indicate measured variables and large circles represent latent constructs. Small circles reflect Models for the Prediction of Depressive residuals (e) or disturbances (d); bold numbers above or near Symptoms (CES-D) endogenous variables represent the amount of variance Direct association model explained (R2). Unidirectional arrows depict hypothesized We first confirmed the existence of a significant direct relation directional or ‘‘causal’’ links. Standardized maximum likelihood between groups and depressive symptoms. We defined the observed parameters are used. Bold estimates are statistically significant. variable CES-D scores. This model has zero degrees of freedom; thus, GROUPS is a binary-coded variable (0, administrative support fit indices could not be estimated. As predicted, attorneys were signif- staff; 1, attorney). CES-D indicates Center for Epidemiological icantly associated with high levels of depressive symptoms (A =0.24, Studies Depression Scale.

952 www.jonmd.com * 2011 Lippincott Williams & Wilkins

(the mediators) was included in the model, it added a significant 6% high caseloads, are often not valued by clients, the justice system, or to the explained variance in CES-D symptoms. society and generally lack sufficient resources appears to make at least an equal contribution to overall distress (see also Lynch, 1997). Future studies are needed to better characterize the relationships between DISCUSSION these stressors and attorneys’ symptoms and functioning. To our knowledge, this is the largest study of attorneys’ emo- In contrast with the previous study by Vrlevski and Franklin tional responses to work with clients who have experienced or have (2008), no relationships were found between personal trauma and been directly involved with trauma. Our data, collected from 238 attor- distress variables. Given that the literature for mental health providers neys and 109 administrative support staff of the Wisconsin State Public is inconsistent (Brady et al., 1999 versus Boscarino et al., 2004), our Defender Office, indicated a significant level of distress among the finding is expectable. The disparate findings across studies may be attorneys compared with administrative support staff. Measures of related to the challenges of accurately measuring past trauma, that is, PTSD symptoms, depression, functional impairment, BO, and STS the subjects’ hesitancy to record this information and their widely were consistently higher among attorneys compared with adminis- varying interpretations of this type of question. The two other findings trative support staff, which was predicted given the longer work hours were the lack of impact of sex or years on the job. Because previous and higher level of exposure to clients with a history of trauma among literature studying therapists has found female sex predictive of STS the attorneys. Bivariate analysis demonstrated that these measures of (Kassam-Adams, 1999) our finding raises questions about differences distress were, in fact, significantly correlated with hours worked per between attorneys and their administrative support staff and mental week and the number of trauma-exposed clients. Subsequent SEM health professionals. Regarding years on the job, available results are modeling illustrated that work-related exposure variables (hours at contradictory, at times indicating greater risk of symptoms of STS and work and number of trauma-exposed clients) were significant, albeit BO with increasing years on the job (e.g., Jaffe et al., 2003) versus a not exclusive, mediators of the differences of group membership on protective effect of greater experience (Maslach et al., 2001; Pearlman symptoms. Therefore, although both attorneys and administrative sup- and Mac Ian, 1995), suggesting that this variable is multidimensional port staff were exposed to trauma-exposed clients, the attorneys’ longer and that its effects vary in different settings. work hours and greater direct contact with these clients associated with What emerges is that similar to mental health professionals, their vulnerability to PTSD symptoms, depression, functional impair- attorneys working as public defenders with clients who have experiment, STS, and BO compared with the administrative support staff’s enced or have been directly involved in trauma are at high risk of indirect exposure to these trauma-exposed clients. developing clinically significant symptoms of secondary trauma and The findings of this study confirmed the results of earlier small BO as well as depression and functional impairment. Our study adds studies (Levin and Greisberg, 2003; Vrlevski and Franklin, 2008) and a potential mechanism by which this high vulnerability is a result of also demonstrated a significant relationship between work and expo- the intensity of their exposure and the length of work hours. These sure variables and depression and functional impairment. Specifically, findings point to the need to support attorneys in identifying the de- we found significant impairment in the attorney group, with 74.8% velopment of these symptoms and to implement interventions to re- scoring above threshold on the SDS, 39.5% demonstrating significant duce them. The current trend is to encourage professionals with STS symptoms of depression (compared with the earlier findings of a 20% and BO to seek peer and supervisory support, increase leisure and rate of depression in attorneys [Benjamin et al., 1990; Eaton et al., physical activity, seek counseling and psychiatric treatment as needed, 1990]), more than a third scoring above the 75th percentile on STS and and develop a variety of resiliency skills (e.g., Gentry et al., 2002). BO, and 11% with clinically significant PTSD symptoms. In a recent However, Bober and Regher (2006) found that these individual ap- review of secondary trauma, Elwood et al. (2011) pointed out that the proaches did not reduce traumatic stress scores. Instead, they recom- secondary trauma literature has largely failed to characterize impair- mended institutional interventions. Our findings reinforce this more ment in professionals experiencing secondary trauma. It appears that nuanced picture and suggest that emphasis must be placed on reducing at least for attorneys working in the public defender setting, PTSD, long work hours as well as on the extent of client exposure such as the secondary trauma, and BO symptoms are accompanied by significant rotation of attorneys between different types of services. Given that impairment and rates of depression (Kessler et al., 1994) and PTSD public defender services are underfunded and overloaded, these types (Kessler et al., 1995) greater than those reported in community samples. of institutional changes remain a significant challenge. In addition, the attorneys reported less compassion satisfaction There are several limitations to this study. Our study’s cross- on the ProQOL5 compared with administrative support staff, and only sectional nature limits any assignment of causality; our model cannot a minority in both groups reported high levels of satisfaction with provide a definitive answer to the question of the direction of the their work. Linley and Joseph (2007), also using the ProQOL, found observed effects. One might argue that mediation variables may have the therapeutic bond was the best predictor of compassion satisfaction been affected by the outcome variables, that is, attorneys with more in a sample of therapists. This suggests a need to better characterize symptoms and impairment may have worked longer hours because of the relationship between public defenders and their clients and its low efficiency or may have been attracted to work with clients who impact on work satisfaction, particularly given Lynch’s (1997) finding had experienced trauma. Second, the administrative support staff may that public defenders felt stressed by encountering angry clients and not have represented the best comparison group. Although this group families. did provide a good comparator because of differences in work and Our SEM analysis raises a question concerning the relative exposure variables, another group of attorneys working with clients contribution of general workload as measured in hours per week with no trauma exposure (e.g., corporate attorneys) may have been a compared with that of exposure to traumatized clients, given that each better comparison, particularly given that attorneys and support staff made nearly equal contributions to the outcome measures. Although differ in education and responsibilities. The administrative support Figley (1995) proposed that secondary trauma is ‘‘the stress resulting staff group also had significantly fewer men than the attorney group, from helping or wanting to help a traumatized or suffering person’’ although the absence of a relationship between sex and outcomes (p. 7), Regehr et al. (2004) found that work load stressors such as suggests that this difference did not affect the study’s findings. documentation and lack of resources, as well as public scrutiny and Despite these limitations, our naturalistic study investigated a organizational issues, played a stronger role in mediating STS and unique phenomenon that may well have significant ecological valid- depression compared with client exposure. The stress of the work ity. To the best of our knowledge, the present study represents the first setting itself, particularly a public legal setting where attorneys have attempt to apply SEM analysis to the association between indicators

* 2011 Lippincott Williams & Wilkins www.jonmd.com 953

of STS symptoms and to examine the mediating role of work-related Creamer T, Liddle B (2005) Secondary traumatic stress among disaster mental health workers responding to the September 11 attacks. J Trauma Stress. exposure in attorneys and administrative support staff. Our findings 18:89Y96. highlight the importance of theoretical models that include job-related Devilly GJ, Wright R, Varker T (2009) Vicarious trauma, secondary traumatic stress description (direct versus indirect exposure to clients’ traumatic events) or simply burnout? Effect of trauma therapy on mental health professionals. and related job exposure (intensity and amount of exposure) and their Aust N Z J Psychiatry. 43:373Y385. role in the development of symptoms and impairment. Eaton WW, Anthony JC, Mandel W, Garrison R (1990) Occupations and the prevalence of major depressive disorder. J Occup Med. 32:1079Y1087. Eaton WW, Smith C, Ybarra M, Muntaner C, Tien A (2004) Center for Epide- CONCLUSIONS miologic Studies Depression Scale: Review and revision (CESD and CESD-R). In Maruish ME (Ed), The use of psychological testing for treatment planning Attorneys working in the Wisconsin State Public Defender and outcomes assessment: Vol. 3: Instruments for adults (pp 363Y377). Office demonstrated significantly higher levels of PTSD symptoms, Princeton, NJ: Lawrence Erlbaum Associates. depression, STS, BO, and functional impairment compared with ad- Elwood LS, Mott J, Lohr JM, Galovski TE (2011) Secondary trauma symptoms in ministrative support staff. This difference was mediated by attorneys’ clinicians: A critical review of the construct, specificity, and implications for longer work hours and greater contact with clients who had experi- trauma-focused treatment. Clin Psychol Rev. 31:25Y36. enced trauma. These findings suggest a need to support attorneys and Erikson CB, Kemp HV,Gorsuch R, Hoke S, Foy DW (2001) Trauma exposure and administrative support staff experiencing these symptoms and to ad- PTSD symptoms in international relief and development personnel. J Trauma Y dress high workloads as well as the intensity of contact with trauma- Stress. 14:205 212. exposed clients. Farber BA, Heifetz L (1982) The process and dimensions of burnout in psychotherapists. Prof Psychol.13:293Y301. Figley CR (1995) Compassion fatigue as secondary traumatic stress disorder: an overview. In Figley CR (Ed), Compassion fatigue: Coping with secondary trau- ACKNOWLEDGMENTS matic stress disorder in those who treat the traumatized (pp 1Y20). Levittown, The authors thank Beth Stamm, PhD, for her input on the use PA: Brunner/Mazel. and scoring of the ProQOL5 as well as the overall design and Jeff Frazier PA, Tix AP, Barron KE (2004) Testing moderator and mediator effects in Apotheker, PhD, for his critique of the original design and support counseling psychology research. J Couns Psychol. 51:115Y134. of the submission to the research committee. They also thank all of Gentry EJ, Baranowsky AB, Dunning K (2002) The accelerated recovery program the participants in this study. Finally, they thank the reviewers for (ARP) for compassion fatigue. In Figley CR (Ed), Treating compassion fatigue Y their constructive suggestions and comments on an earlier draft of (pp 123 138). New York: Brunner-Routledge. this paper. Gomme IM, Hall MP (1995) Prosecutors at work: role overload and strain. J Crim Justice. 15:191Y200. Hoyle RH, Smith GT (1994) Formulating clinical research hypotheses as structural Y DISCLOSURE equation models: A conceptual overview. J Consult Clin Psychol.62:429 440. Hu LT, Bentler PM (1999) Cutoff criteria for fit indexes in covariance structure This study was supported by the State Bar of Wisconsin and analysis: Conventional criteria versus new alternatives. Struct Equ Modeling. the Wisconsin State Public Defender Office. 6:1Y55. The authors have nothing to disclose. Jaffe PG, Crooks CV, Dunford-Jackson BL, Town M (2003) Vicarious trauma in judges: The personal challenge of dispensing justice. Juv Fam Court J. 54:1Y9. Jenkins SH, Baird S (2002) Secondary traumatic stress and vicarious trauma: A REFERENCES validation study. J Trauma Stress. 15:423Y432. Allegretti J (1993) Shooting elephants, serving clients: An essay on George Orwell Jo¨reskog KG, So¨rbom D (1993) LISREL 8: Structural equation modeling with the and the lawyer-client relationship. Creighton Law Rev.27:1Y24. SIMPLIS command language. Chicago: Scientific Software International. American Psychiatric Association (1994) Diagnostic and statistical manual of Kassam-Adams N (1999) The risks of treating sexual trauma: Stress and secondary mental disorders (4th ed). Washington, DC: American Psychiatric Association. trauma in psychotherapists. In Stamm BH (Ed), Secondary traumatic stress (pp 37Y48). Lutherville, ME: Sidran Press. Arbuckle JL (2009) Amos: A structural equation modeling program. Chicago, IL: Small Waters. Kelloway E K (1998) Using lisrel for structural equation modeling: A researcher’s guide. Newbury Park, CA: Sage. Baird S, Jenkins SR (2003) Vicarious traumatization, secondary traumatic stress, and burnout in sexual assault and domestic violence agency staff. Violence Vict. Kenny DA, Kashy DA, Bolger N (1998) Data analysis in social psychology. In 18:71Y86. Gilbert D, Fiske S, Lindzey G (Eds), The handbook of social psychology (4th ed, pp 233Y265). New York: McGraw-Hill. Baron RM, Kenny DA (1986) The moderator-mediator variable distinction in so- Kenny DA, McCoach DB (2003) Effect of the number of variables on measures cial psychological research: Conceptual, strategic, and statistical considerations. Y J Pers Soc Psychol. 51:1173Y1182. of fit in structural equation modeling. Struct Equ Modeling. 10:333 351. Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Eshleman S, Benjamin GA, Darling EJ, Sales B (1990) The prevalence of depression, alcohol Wittchen H, Kendler KS (1994) Lifetime prevalence and 12-month prevalence abuse, and cocaine abuse among United States lawyers. Int J Law Psychiatry. of DSM-III-R psychiatric disorders in the United States: Results from the 13:233Y246. national comorbidity survey. Arch Gen Psychiatry. 51:8Y19. Bentler PM, Bonnet DC (1980) Significance tests and goodness of fit in the analysis Y Kessler RC, Sonnega A, Bromet E, Hughes M, Nelson CB (1995) Posttraumatic of covariance structures. Psychol Bull.88:588 606. stress disorder in the national comorbidity survey. Arch Gen Psychiatry. Bober T, Regher C (2006) Strategies for reducing secondary or vicarious trauma: 52:1048Y1060. Y Do they work? Brief Treat Crisis Interv. 6:1 9. Landry SH, Smith KE, Swank PR, Miller-Loncar CL (2000) Early maternal and Boscarino JA, Figley CR, Adams RE (2004) Compassion fatigue following the child influences on children’s later independent cognitive and social function- September 11 terrorist attacks: A study of secondary trauma among New York ing. Child Dev. 71:358Y375. City social workers. Int J Emerg Ment Health. 6:57Y66. Levin AP, Greisberg S (2003) Vicarious trauma in attorneys. Pace Law Rev. Y Brady JL, Guy JD, Poelstra PL, Brokaw BF (1999) Vicarious traumatization, 2:245 257. spirituality, and the treatment of sexual abuse survivors: a national survey of Linley PA, Joseph S (2007) Therapy work and the therapist’s positive and negative women therapists. Prof Psychol Res Pract. 30:586Y393. well-being. J Soc Clin Psychol. 26:385Y403. Bride BE, Jones JL, Macmaster SA (2007) Correlates of secondary traumatic stress Ljubotina D, Muslic L (2003) Convergent validity of four instruments for measuring in child protective services workers. J Evid-Based Soc Work. 4:69Y80. posttraumatic stress disorder. Rev Psychol.10:11Y21. Creamer M, Bell R, Failla S (2003) Psychometric properties of the impact of event Lynch DR (1997) The nature of occupational stress among public defenders. Justice scaleVrevised. Behav Res Ther. 41:1489Y1496. Syst J. 19:17Y35.

954 www.jonmd.com * 2011 Lippincott Williams & Wilkins

MacKinnon DP, Lockwood CM, Hoffman J M, West SG, Sheets V (2002) A Regehr C, Hemsworth D, Leslie B, Howe P, Chau S (2004) Predictors of post- comparison of methods to test mediation and other intervening variable effects. traumatic distress in child welfare workers: A linear structural equation model. Psychol Methods. 7:83Y104. Child Youth Serv Rev. 26:331Y346. Maslach C, Schaufeli WB, Leiter MP (2001) Job burnout. Annu Rev Psychol. Schnaube LJ, Frazier PA (1995) Vicarious trauma: the effects of female counselors 52:397Y422. of working with sexual violence survivors. Psychol Women Q. 19:49Y64. McCann IL, Pearlman LA (1990) Vicarious traumatization: A framework for understan- Y Sheehan DV, Harnett-Sheehan K, Raj BA (1996) The measurement of psycho- ding the psychological effects of working with victims. JTraumaStress.3:131 149. pharmacology. Int Clin Psychopharm. 11:89Y95. Meier JS (1993) Notes from the underground: Integrating psychological and legal Silver MA (1999) Love, hate, and other emotional interference in the lawyer/client perspectives on domestic violence in theory and practice. Hofstra Law Rev. Y 21:1295Y1366. relationship. Clin Law Rev. 6:259 313. Ortlepp K, Friedman M (2002) Prevalence and correlates of secondary traumatic Stamm BH (2010) The ProQOL 5 (Professional Quality of Life Scale: Compassion stress in workplace lay trauma counselors. J Trauma Stress. 15:213Y222. Satisfaction and Compassion Fatigue, version 5). Pocatello, ID: ProQOL.org. Available at: http://www.proqol.org/. Accessed on June 26, 2011. Pearlman LA, Mac Ian P S (1995) Vicarious traumatization: An empirical study of the effects of trauma work on trauma therapists. Prof Psychol Res Pract. 26:558Y565. Vrlevski LP, Franklin J (2008) Vicarious trauma: The impact on solicitors of expo- Y Piwowarcyzy L, Ignatius S, Crosby S, Grodin M, Heeren T, Sharma A (2009) sure to traumatic material. Traumatology.14:106 118. Y Secondary trauma in asylum lawyers. Bender’s Immigr Bull. 14:263 269. Weiss D, Marmar C (1997) The Impact of Event Scale-Revised. In Wilson J, Radloff LS (1977) The CES-D scale: A self-report depression scale for research Keane T (Eds), Assessing psychological trauma and PTSD (pp 399Y411). in the general population. Appl Psychol Meas. 1:385Y401. New York: Guilford.

* 2011 Lippincott Williams & Wilkins www.jonmd.com 955