1. ORIGINAL CONTRIBUTION
Treatment of Complicated Grief
A Randomized Controlled Trial
Katherine Shear, MD Context Complicated grief is a debilitating disorder associated with important nega-
Ellen Frank, PhD tive health consequences, but the results of existing treatments for it have been dis-
appointing.
Patricia R. Houck, MSH
Objective To compare the efficacy of a novel approach, complicated grief treat-
Charles F. Reynolds III, MD
ment, with a standard psychotherapy (interpersonal psychotherapy).
M
ANY PHYSICIANS ARE UN- Design Two-cell, prospective, randomized controlled clinical trial, stratified by man-
certain about how to ner of death of loved one and treatment site.
identify bereaved indi- Setting A university-based psychiatric research clinic as well as a satellite clinic in a
viduals who need treat- low-income African American community between April 2001 and April 2004.
ment, and what treatments work for be-
Participants A total of 83 women and 12 men aged 18 to 85 years recruited through
reavement-related mental health professional referral, self-referral, and media announcements who met criteria for com-
problems.1 Bereavement-related ma- plicated grief.
jor depressive disorder is a well-
Interventions Participants were randomly assigned to receive interpersonal psy-
recognized consequence of loss.2,3 Com- chotherapy (n=46) or complicated grief treatment (n=49); both were administered
plicated grief also occurs in the in 16 sessions during an average interval of 19 weeks per participant.
aftermath of loss but needs to be dif-
Main Outcome Measure Treatment response, defined either as independent evalu-
ferentiated from depression. Compli- ator-rated Clinical Global Improvement score of 1 or 2 or as time to a 20-point or bet-
cated grief can be reliably identified by ter improvement in the self-reported Inventory of Complicated Grief.
administering the Inventory of Com-
Results Both treatments produced improvement in complicated grief symptoms. The
plicated Grief (ICG) 4 more than 6
response rate was greater for complicated grief treatment (51%) than for interper-
months after the death of a loved one. sonal psychotherapy (28%; P=.02) and time to response was faster for complicated
Key features of complicated grief5,6 in- grief treatment (P=.02). The number needed to treat was 4.3.
clude (1) a sense of disbelief regard-
Conclusion Complicated grief treatment is an improved treatment over interper-
ing the death; (2) anger and bitterness sonal psychotherapy, showing higher response rates and faster time to response.
over the death; (3) recurrent pangs of JAMA. 2005;293:2601-2608 www.jama.com
painful emotions, with intense yearn-
ing and longing for the deceased; and
(4) preoccupation with thoughts of the symptoms of complicated grief load associated with a range of negative
loved one, often including distressing separately from both depression and health consequences.14-16 Prevalence
intrusive thoughts related to the death. anxiety.10,11 Comparisons of compli- rates are estimated at approximately
Avoidance behavior is also frequent cated grief, major depression, and PTSD 10% to 20% of bereaved persons.17,18
and entails a range of situations and ac- are listed in TABLE 1. Co-occurrence of Approximately 2.5 million people die
tivities that serve as reminders of the complicated grief with major depres- yearly in the United States. 19 Esti-
painful loss. Studies indicate that treat- sive disorder and PTSD is also com- mates suggest each death leaves an av-
ments for bereavement-related depres- mon. Prior studies indicate that rates erage of 5 people bereaved, suggesting
sion show minimal effects on compli- of complicated grief co-occurring with that more than 1 million people per year
cated grief symptoms.7,8 Complicated major depressive disorder range from are expected to develop complicated
grief bears some resemblance to post- 21%5 to 54%4 and co-occurring with grief in the United States.
traumatic stress disorder (PTSD), al- PTSD range from 30%12 to 50%.13
Although it is not included in the Di- Author Affiliations: Department of Psychiatry, Uni-
though again, there are important dif- versity of Pittsburgh School of Medicine, Pittsburgh,
ferences.9 Factor analysis shows that agnostic and Statistical Manual of Men- Pa.
tal Disorders, Fourth Edition (DSM-IV), Corresponding Author: Katherine Shear, MD, De-
partment of Psychiatry, University of Pittsburgh School
complicated grief is a source of signifi- of Medicine, 3811 O’Hara St, Room E-1116, Pitts-
See also p 2658 and Patient Page.
cant distress and impairment and is burgh, PA 15213 (shearmk@upmc.edu).
©2005 American Medical Association. All rights reserved. (Reprinted) JAMA, June 1, 2005—Vol 293, No. 21 2601
Downloaded from www.jama.com on February 21, 2008

2. TREATMENT OF COMPLICATED GRIEF
Given observations regarding the CGT,21 we report here the results of a site, by violent (accident, homicide, or
specificity and clinical significance of randomized controlled trial compar- suicide) vs nonviolent death of a loved
complicated grief symptoms, includ- ing CGT with standard IPT. We hy- one. A blinded randomization num-
ing the lack of response to standard pothesized that CGT would be supe- ber was assigned using a computer-
treatments for depression,20,21 we de- rior to IPT with respect to overall ized random number generator with-
veloped a targeted complicated grief response rates and time to response, out blocking. Decisions regarding
treatment (CGT). Since complicated with CGT producing a more rapid and eligibility were based primarily on in-
grief includes depressive symptoms greater resolution of complicated grief dependent evaluator assessment and al-
such as sadness, guilt, and social with- symptoms than IPT. ways made by the study team prior to
drawal, we used a framework for the disclosure of the treatment assign-
treatment based on previous research METHODS ment (FIGURE 1).
with interpersonal psychotherapy (IPT) Study Design Treatment was provided in approxi-
for grief-related depression.22 In view Patients who met criteria for compli- mately 16 sessions over a 16- to 20-
of the presence of PTSD symptoms of cated grief, defined as score on the ICG week period. Three additional ses-
disbelief, intrusive images, and avoid- of at least 30, were recruited to a uni- sions could be added in the event of a
ance behaviors, as well as unique symp- versity-based clinic. To include a broad second death. Time could be ex-
toms related to the loss (eg, yearning range of participants, we also enrolled tended if there was a serious life event
and longing for the deceased), we modi- study participants at a clinic attended (eg, hospitalization for medical ill-
fied IPT techniques to include cogni- by primarily low-income African ness, severe stressor.) Patients whose
tive-behavioral therapy–based tech- American patients. Race was assessed treatment coincided with the attacks of
niques for addressing trauma. We used by self-report. We obtained this infor- September 11, 2001, were offered an ex-
cognitive strategies for working with mation as part of a concerted effort to tra session to discuss their reaction.
loss-specific distress. As suggested by include low-income minorities in our Treatment could be shorter if both
results of a comparison of the 2 meth- study. The originally proposed sample therapist and patient agreed that the pa-
ods,23 we have previously found that size was 60. Participants were ran- tient had successfully completed the
IPT and cognitive-behavioral therapy domly assigned to receive CGT or IPT course of treatment. Posttreatment as-
lend themselves to integration.24,25 Fol- in a ratio of 1:1. Randomization was sessment was obtained on completion
lowing completion of an open trial of stratified by treatment site and, within of treatment by evaluators blinded to
randomized treatment assignment. For
early dropouts, therapists provided an
Table 1. Similarities and Differences Between Complicated Grief and DSM-IV Disorders estimated global improvement score
Similarities Between Complicated Grief and DSM-IV Disorders and a written paragraph justifying their
Major Depression Posttraumatic Stress Disorder rating. The independent evaluator re-
Sadness, loss of interest Triggered by traumatic event viewed this information with all avail-
Loss of self-esteem Sense of shock, helplessness
Guilt Intrusive images able ratings prior to finalizing the re-
Avoidance behavior sponse rating. Study nonresponders
Differences Between Complicated Grief and DSM-IV Disorders were either treated openly or referred
Major Depression Complicated Grief to geographically convenient or pre-
Pervasive sad mood Sadness related to missing the deceased ferred outside treatment. The study was
Loss of interest or pleasure Interest in memories of the deceased approved by the University of Pitts-
maintained; longing and yearning for
contact; pleasurable reveries burgh Institutional Review Board. Par-
Pervasive sense of guilt Guilt focused on interactions with the deceased ticipants were enrolled between April
Rumination about past failures or misdeeds Preoccupation with positive thoughts of the 2001 and April 2004.
deceased
Intrusive images of the person dying Participants
Avoidance of situations and people related to Bereaved individuals recruited via pro-
reminders of the loss
fessional referral, media advertisement,
Posttraumatic Stress Disorder Complicated Grief
and self-referral gave oral informed con-
Triggered by physical threat Triggered by loss
sent for a brief screening interview by
Primary emotion is fear Primary emotion is sadness
Nightmares are very common Nightmares are rare
telephone (n=405) or in person (n=12).
Painful reminders linked to the traumatic event; Painful reminders more pervasive and
A subgroup (n=26) was recruited from
usually specific to the event unexpected the clinic with predominantly low-
Yearning and longing for the person who died income African American patients. In-
Pleasurable reveries dividuals who screened positive (n=329)
Abbreviation: DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. on the ICG and signed written in-
2602 JAMA, June 1, 2005—Vol 293, No. 21 (Reprinted) ©2005 American Medical Association. All rights reserved.
Downloaded from www.jama.com on February 21, 2008

3. TREATMENT OF COMPLICATED GRIEF
formed consent (n=218) were assessed terpersonal psychotherapy was deliv- also included an introductory, middle,
for eligibility, initial symptom ratings, ered as described in a published and termination phase. In the intro-
and drug stabilization for patients tak- manual, 2 8 using an introductory, ductory phase, the therapist provided
ing antidepressant medication (n=92). middle, and termination phase. Dur- information about normal and compli-
Patients were permitted to take medica- ing the introductory phase, symptoms cated grief and described the dual-
tion for depression during the study if were reviewed and identified and an in- process model of adaptive coping, en-
(1) medication management was trans- terpersonal inventory was completed. tailing both restoration of a satisfying
ferred to the study pharmacotherapist Interpersonal psychotherapists used a life and adjustment to the loss.29 This
and (2) medication use was stable for a grief focus, sometimes accompanied by model posits that grief proceeds opti-
minimum of 3 months, with at least 6 a secondary focus on role transition or mally when attention to loss and res-
weeks at the same dose. The study phar- interpersonal disputes. The relation- toration alternate, while coping with
macotherapist made a judgment about ship between symptoms and grief and both processes proceeds more or less
adequacy of pharmacotherapy and ad- other interpersonal problems was dis- in concert. Thus, in addition to discus-
justed medications as necessary, prior to cussed. The middle phase was used to sion of the loss, the introductory phase
randomization. address grief and other interpersonal of CGT included a focus on personal
Inclusion required a score of at least problems, as indicated. The IPT thera- life goals. In the middle phase, the
30 on the ICG at least 6 months after pist helped patients to arrive at a more therapist addressed both processes in
the death of a loved one and judgment realistic assessment of the relation- tandem. Similar to IPT, the termina-
by the independent evaluator that com- ship with the deceased, addressing both tion phase focused on review of
plicated grief was the most important its positive and negative aspects, and en- progress, plans for the future, and feel-
clinical problem. Individuals with cur- couraged the pursuit of satisfying re- ings about ending treatment.
rent substance abuse or dependence lationships and activities. In the termi- In contradistinction to IPT, how-
(past 3 months), history of psychotic nation phase, treatment gains were ever, traumalike symptoms were ad-
disorder or bipolar I disorder, suicid- reviewed, plans were made for the fu- dressed using procedures for retelling
ality requiring hospitalization, pend- ture, and feelings about ending treat- the story of the death and exercises en-
ing lawsuit or disability claim related ment were discussed. tailing confrontation with avoided situ-
to the death, or concurrent psycho- Complicated grief treatment, deliv- ations, modified from imaginal and in
therapy were excluded. ered according to a manual protocol, vivo exposure used for PTSD.30,31 We
Therapists
Figure 1. Flow of Participants Through the Trial
All therapists were master’s- or doctoral-
level clinicians who had at least 2 years 218 Patients Assessed for Eligibility
of psychotherapy experience and who
underwent extensive training and cer- 116 Excluded
tification in either IPT or CGT. Certifi- 80 Did Not Meet Inclusion Criteria
19 Refused to Participate
cation entailed completion of 2 treat- 17 Served as Training Cases
ment cases in a manner judged
competent by K.S. (for CGT) or E.F. (for 102 Randomized
IPT). Therapists received ongoing group
supervision, separately for IPT and CGT,
51 Assigned to Complicated Grief Treatment 51 Assigned to Interpersonal Psychotherapy
throughout the study period. Selected 49 Entered Treatment 46 Entered Treatment
audiotapes or videotapes were used in 1 Never Came to a Treatment Session 4 Never Came to a Treatment Session
1 Revealed Pending Lawsuit 1 Randomized Inadvertently Before
supervision sessions as a part of the dis- Eligibility Confirmed
cussion. Therapy sessions were audio-
taped for adherence and competence rat- 2 Lost to Follow-up 12 Discontinued Intervention
1 Reason Unknown 7 Dissatisfied With Treatment
ings, performed on a randomly selected 1 Could Not Be Scheduled in 2 Began Antidepressant Medication
subset of sessions. Required Time Window 1 Insurmountable Scheduling Problems
13 Discontinued Intervention 1 Hospitalized for Suicidality
6 Dissatisfied With Treatment 1 Therapist Protocol Violation
Treatment Conditions 3 Withdrew for Serious Physical Illness
2 Insurmountable Child Care Conflicts
Interpersonal psychotherapy is a proven 1 Another Death in the Family
efficacious treatment, well studied for 1 Improved and Believed Treatment
Completed
the treatment of depression.26,27 Our
group has done extensive research us-
49 Included in Analysis 46 Included in Analysis
ing this treatment, and therapists in this 2 Excluded (Never Entered Treatment) 5 Excluded (Never Entered Treatment)
study had a strong allegiance to IPT. In-
©2005 American Medical Association. All rights reserved. (Reprinted) JAMA, June 1, 2005—Vol 293, No. 21 2603
Downloaded from www.jama.com on February 21, 2008

4. TREATMENT OF COMPLICATED GRIEF
called the retelling procedure “revisit- tion of a set of memories question- using a technique derived from moti-
ing.” To conduct a revisiting exercise, naires, primarily focused on positive vational enhancement therapy.32 Pa-
the therapist asked patients to close memories, though also inviting remi- tients were encouraged to consider what
their eyes and tell the story of the death. niscence that was negative. The imagi- they would like for themselves if their
The therapist tape-recorded the story, nal conversation was conducted with grief was not so intense. The therapist
and periodically asked the patient to re- the patient’s eyes closed. The patient then helped patients identify ways to
port distress levels. The patient was was asked to imagine that he/she could know that they were working toward
given the tape to listen to at home dur- speak to the person who died and that their identified goals. Concrete plans
ing the week. Distress related to the loss the person could hear and respond. The were discussed and the therapist en-
(eg, yearning and longing, reveries, fears patient was invited to talk with the couraged the patient to put these into
of losing the deceased forever) was tar- loved one and then to take the role of action. Standard IPT procedures tar-
geted using techniques to promote a the deceased and answer. The thera- geting role transition and/or interper-
sense of connection to the deceased. pist guided this “conversation” for 10 sonal disputes were also used, as
These included an imaginal conversa- to 20 minutes. For the restoration fo- needed, to encourage patients to reen-
tion with the deceased and comple- cus, patients defined personal life goals gage in meaningful relationships. More
detailed information describing the
treatment is available from the au-
Table 2. Pretreatment Comparison of Treatment Groups thors.
Complicated Interpersonal
2
Grief Treatment Therapy or t P
(n = 49)* (n = 46)* Value df Value Assessment Procedures
Age, mean (SD), y 49.4 (13.9) 47.3 (11.3) 0.80 93 .42 Independent evaluators were experi-
Male 6 (12) 6 (13) 0.01 1 .91 enced master’s- or doctoral-level clini-
White 36 (75) 36 (78) 0.14 1 .71 cians trained for reliability on rating in-
Education struments and monitored throughout
12 y 12 (27) 7 (16) the study. Evaluators were blinded to
Partial college 17 (38) 16 (36) treatment assignment, and study staff
2.61 3 .46
4-y college 8 (18) 11 (25) closely monitored procedures to main-
Postgraduate 8 (18) 12 (27) tain the blinding. Independent evalu-
Marital status
Never married 9 (18) 9 (20)
ators conducted assessments prior to as
Married 12 (24) 18 (39)
well as after treatment. Additionally, for
3.25 3 .36 randomized participants who dropped
Separated/divorced 16 (33) 9 (20)
Widowed 12 (24) 10 (22) out after at least 1 treatment session, a
Relationship of deceased Clinical Global Improvement (CGI)
Spouse/partner 17 (35) 9 (20) Scale score was generated. To do this,
Parent 12 (24) 14 (30)
4.44 3 .22
therapists provided a global improve-
Child 10 (20) 16 (35) ment rating and a brief narrative justi-
Other 10 (20) 7 (16) fying their rating, without including in-
Violent death of deceased 16 (33) 15 (33) 0.11 1 .99 formation related to the treatment. The
Years since loss, median (range)† 2.1 (0.5-36.6) 2.5 (0.5-22.3) −0.08 93 .94 independent evaluator reviewed the rat-
Major depressive disorder ing and narrative as well as available
Current 22 (45) 19 (41) 0.13 1 .72
participant self-report assessments from
Lifetime 33 (67) 33 (72) 0.22 1 .64
the final session to finalize the CGI
Posttraumatic stress disorder
Current 24 (49) 21 (46) 0.11 1 .75 score. The CGI Scale33 is a single Likert-
Lifetime 27 (55) 24 (52) 0.08 1 .77 type rating from 1 to 7 where 1 through
Inventory of Complicated Grief score, 45.8 (8.0) 44.2 (9.9) 0.87 93 .39 3 indicate very much, much, and mini-
mean (SD)‡ mally improved, respectively; 4 indi-
Hamilton Rating Scale for Depression 24.5 (9.2) 22.3 (8.9) 1.18 93 .24 cates no change; and 5 through 7 in-
score, mean (SD)§
dicate minimally, much, and very much
Structured Interview Guide for Hamilton 19.7 (7.8) 18.9 (7.9) 0.49 93 .63
Rating Scale for Anxiety score, worse, respectively.
mean (SD) Pretreatment assessment included
*Data are expressed as No. (%) unless otherwise noted. the Structured Clinical Interview for the
†Natural log transformation prior to statistical comparison.
‡Mean (SD) scores for healthy controls are reported as 10.28 (6.6); mean (SD) score in a bereaved population is 17.74 DSM-IV,34 Hamilton Rating Scale for
(12.4).4
§Scores of 23 or greater indicate very severe depression; 19 to 22, severe; 14 to 18, moderate; 8 to 13, mild; and 7 or Depression,35 Hamilton Rating Scale for
less, none.42,43 Anxiety,36 structured clinical inter-
Scores in healthy controls are reported as less than 5; scores greater than 14 are considered clinically important.44,45
views for complicated grief and for sui-
2604 JAMA, June 1, 2005—Vol 293, No. 21 (Reprinted) ©2005 American Medical Association. All rights reserved.
Downloaded from www.jama.com on February 21, 2008

5. TREATMENT OF COMPLICATED GRIEF
cidality, and screening medical evalu-
Figure 2. Survival Analysis (Time to Response)
ation. We diagnosed major depression
without making an effort to discrimi- 1.0
nate grief from depression. Self-
Proportion Without Response
reported measures included the ICG4 0.8
and the Work and Social Adjustment
Scale.37 The Beck Depression38 and 0.6
Anxiety39 Inventory scales as well as the
0.4
ICG and the Work and Social Adjust-
ment Scale were completed at treat-
0.2
ment sessions. Responder status was IPT (n = 46)
CGT (n = 49)
determined in 2 different ways: inde- Wilcoxon χ2 = 5.65; P = .02
0
pendent evaluator score of 2 or lower 0 4 8 12 16 20 24
on the CGI and self-reported improve- Week
No. at Risk
ment of at least 20 points (2 SDs above IPT 46 45 41 35 31 24 9
baseline mean) on the ICG. CGT 49 46 39 32 22 18 7
Response was defined as a decrease in the Inventory of Complicated Grief score of 20 points or more. CGT
Statistical Analyses
indicates complicated grief treatment; IPT, interpersonal psychotherapy.
The study was designed to address the
question of whether CGT produced bet-
ter results than standard IPT for the ficacious treatments, the number cant stratum effect for site or type of
treatment of complicated grief. To an- needed to treat falls between 2 and 4.40 death was observed, we aggregated data
swer this question, we examined rate Continuous measures were evalu- across strata.
of response, defined using either an in- ated using end-point analysis with base- Treatment completion rates (73% for
terviewer (CGI) or a self-reported (ICG) line score as covariates in both modi- CGT and 74% for IPT) did not differ
measure for all randomized patients fied intention-to-treat and completer across groups. Mean number of CGT
who attended at least 1 treatment ses- analyses for the self-reported mea- sessions for completers was 16 (range,
sion (modified intention-to-treat study sures, ICG, Work and Social Adjust- 7-19). Mean number of IPT sessions
group, n=95). ment Scale, and Beck Depression and was 16 (range, 15-16). Mean time to
Data were first descriptively ana- Anxiety Inventory scales. Interview- completion of CGT was 19.4 weeks.
lyzed to check range and distribution rated Hamilton Depression and Anxi- Mean time to completion of IPT was
of all variables. We further checked to ety scores were obtained only at base- 18.4 weeks. Mean number of sessions
ensure equivalent distribution of scores line and posttreatment assessment prior to dropout for CGT was 5.9 (SD,
across study groups. Baseline compari- points and so are available only for 3.7; range, 1-12) and for IPT was 4.3
sons included all demographic and completers. (SD, 2.6; range, 1-8). Three patients in
clinical variables. To examine the possible difference CGT each had 3 additional sessions to
Cochran-Mantel-Haenszel general as- in response by baseline measures, a Co- deal with a second death and 2 had 1
sociation analyses were used to com- chran-Mantel-Haenszel test, stratified additional session. Two had extra ses-
pare CGI responder rates for IPT and by treatment, was used. To examine dif- sions to address an intercurrent medi-
CGT. Statistical significance was de- ferential treatment response in differ- cal problem (kidney stone and blepha-
fined as P .05 with a 2-tailed test. We ent subgroups, a Breslow-Day test41 was rospasm) and 1 had an extra session to
used a survival analytic strategy to com- used, stratifying by group. A signifi- discuss the September 11 attacks. Three
pare time to response using the ICG cri- cant result indicates that a differential patients ended treatment early with the
terion. Kaplan-Meier curves were used treatment group interaction exists. agreement of their therapists. A total of
to investigate time to response and pro- SAS software, version 8.2 (SAS Insti- 6 CGT and 3 IPT patients had treat-
portion surviving by treatment groups. tute Inc, Cary, NC) was used for all ment lasting more than 20 weeks.
Wilcoxon 2 tests were used to assess analyses. Twenty IPT (43%) and 23 CGT (47%)
differences in survival curves. We fur- patients continued to take antidepres-
ther calculated number needed to treat RESULTS sant medication begun prior to ran-
as 1 divided by the proportion respond- Baseline, Site, domization.
ing in CGT-IPT as an estimate of the and Stratum Analyses
number of patients who would need to There were no significant differences in Responder Analyses
be given CGT for 1 of them to achieve demographic measures or baseline ICG Using the independent evaluator cri-
a response outcome who would not scores between the 2 randomized terion of a CGI score of 2 (much im-
have achieved it with IPT. For most ef- groups (TABLE 2). Because no signifi- proved) or 1 (very much improved),
©2005 American Medical Association. All rights reserved. (Reprinted) JAMA, June 1, 2005—Vol 293, No. 21 2605
Downloaded from www.jama.com on February 21, 2008

6. TREATMENT OF COMPLICATED GRIEF
rate of response in the modified inten- pleters. Median time to response using Beck Depression Inventory, and Work
tion-to-treat sample was greater for the self-report (ICG) criterion was and Social Adjustment Scale.
CGT than for IPT among all random- shorter for CGT than for IPT (FIGURE 2)
ized participants; 51% (95% confi- 2
(Wilcoxon 1 =5.65; P=.02). Attrition
dence interval [CI], 37%-65%) treated Early treatment discontinuation oc-
with CGT responded compared with Results for Continuous Measures curred for 13 (27%) of 49 CGT and 12
28% (95% CI, 15%-41%) treated with TABLE 3 shows results for the ICG, Beck (26%) of 46 IPT participants. Reasons
2
IPT ( 1 = 5.07; P = .02; cohort relative Depression Inventory, Beck Anxiety for discontinuation differed; 6 CGT pa-
risk [RR], 1.69 [95% CI, 1.03-2.77]). Inventory, and Work and Social Ad- tients (12%) considered the treatment
Among completers, 66% (95% CI, 50%- justment Scale. In the modified inten- too difficult and/or did not believe that
82%) vs 32% (95% CI, 16%-48%) re- tion-to-treat analysis, outcome was mar- telling the highly painful story of the
2
sponded ( 1 =7.56; P=.006; cohort RR, ginally better for CGT than for IPT. death could help them. An additional
2.03 [95% CI, 1.16-3.49]). The num- Results for completers showed signifi- 7 participants (14%) discontinued CGT
ber needed to treat was 4.3 for modi- cantly better outcome for CGT with me- for serious medical illness (n=3; after
fied intention to treat and 2.9 for com- dium effect size differences on the ICG, sessions 4, 10, and 12), insurmount-
able child care conflicts (n=2; after ses-
sions 3 and 5), a death in the family
Table 3. Posttreatment Scores and Group Effect From End-point Analyses With Pretreatment
as a Covariate (n=1; after session 6), and sufficient im-
No. (%) provement (n=1; after session 12). Also
of note, 5 CGT patients who com-
Complicated Interpersonal P
Grief Treatment Psychotherapy F df Value pleted the treatment refused participa-
Modified Intention to Treat tion in the imaginal exposure exercise
Inventory of Complicated Grief n = 49 n = 46 because they considered it too difficult.
Pretreatment 45.8 (8.0) 44.2 (9.9) For IPT, 7 (15%) of 46 left treat-
Posttreatment 28.6 (16.2) 31.4 (12.9) ment dissatisfied because of perceived
Difference 17.2 (15.3) 12.8 (11.9) 1.86 92 .18 lack of effectiveness. Five additional IPT
Beck Anxiety Inventory n = 47 n = 45 patients (11%) discontinued treat-
Pretreatment 17.5 (12.0) 15.4 (10.2) ment. Reasons included scheduling
Posttreatment 9.3 (10.7) 9.4 (8.7) problems (n=1; after session 8), hos-
Difference 8.2 (8.7) 6.0 (9.3) 0.69 89 .41 pitalization for active suicidal ide-
Beck Depression Inventory n = 47 n = 45 ation (n = 1; after session 5), begin-
Pretreatment 23.9 (10.3) 22.4 (9.8) ning antidepressant medication (n=2;
Posttreatment 13.4 (10.0) 15.2 (10.8) after sessions 10 and 12), and with-
Difference 10.4 (9.6) 7.2 (7.2) 2.70 89 .10 drawal because of serious protocol vio-
Work and Social Adjustment Scale n = 49 n = 46 lation on the part of the therapist (n=1;
Pretreatment 20.3 (10.1) 20.5 (9.6) after session 9) related to insertion of
Posttreatment 12.5 (10.5) 16.2 (11.0) CGT into the IPT session.
Difference 7.8 (11.3) 4.2 (9.5) 3.59 92 .06
Treatment Completers Secondary Analyses
Inventory of Complicated Grief n = 35 n = 34 We found no statistically significant dif-
Pretreatment 46.4 (8.4) 43.4 (9.8) ferences in response based on race, age,
Posttreatment 25.8 (15.7) 30.6 (13.8) sex, time since the loss, or relation-
Difference 20.6 (15.0) 12.8 (10.7) 5.18 66 .03 ship to the deceased. Patients taking an-
Beck Anxiety Inventory n = 35 n = 34 tidepressant medication had margin-
Pretreatment 17.6 (12.5) 14.5 (9.8) ally better response rates: for CGT, 13
Posttreatment 8.1 (10.4) 8.7 (9.5) of 22 (59% [95% CI, 38%-80%]) vs 11
Difference 9.5 (7.3) 5.8 (9.5) 1.94 66 .17 of 26 (42% [95% CI, 23%-61%]) not
Beck Depression Inventory n = 35 n = 34 taking antidepressant medication and
Pretreatment 24.6 (10.8) 20.9 (9.8) for IPT, 8 of 20 (40% [95% CI, 19%-
Posttreatment 11.9 (10.0) 13.6 (11.4) 61%]) vs 5 of 26 (19% [95% CI, 4%-
Difference 12.7 (9.8) 7.3 (5.6) 5.92 66 .02 34%]) not taking antidepressant medi-
Work and Social Adjustment Scale n = 35 n = 34 cation. Patients who lost a loved one
Pretreatment 21.5 (10.9) 20.1 (10.0) through violent death (suicide, homi-
Posttreatment 11.0 (10.4) 15.1 (11.1) cide, or accident) had a 56% (95% CI,
Difference 10.4 (11.2) 5.0 (9.9) 4.47 66 .04 32%-80%) response rate with CGT and
2606 JAMA, June 1, 2005—Vol 293, No. 21 (Reprinted) ©2005 American Medical Association. All rights reserved.
Downloaded from www.jama.com on February 21, 2008

7. TREATMENT OF COMPLICATED GRIEF
13% (95% CI, 0%-30%) response rate generalizability of our findings if we ex- loss-focused cognitive-behavioral
with IPT, while for natural, nonacci- cluded such patients. There was no dif- therapy techniques and restoration-
dental death, there was a 47% (95% CI, ference in the rate of medication use in focused IPT strategies. Cognitive-
30%-64%) response to CGT and 35% CGT vs IPT. There was a marginally sig- behavioral therapy techniques in-
(95% CI, 18%-52%) response to IPT. nificant effect of medication on out- clude repeated retelling of the story of
Parents who lost a child had a low re- come, which was more pronounced for the death and work on confronting
sponse rate to CGT (17% [95% CI, 0%- IPT (2.1 times the response rate of those avoided situations. Cognitive tech-
52%]) compared with those who lost not taking medication) than CGT (1.4 niques include an imaginal conversa-
a spouse, parent, or other friend or rela- times the response of those not taking tion with the deceased and work on
tive (average, 60%), while this was not medication.) A similar proportion of pa- memories. Interpersonal psycho-
true for IPT, for which the response rate tients taking concurrent antidepres- therapy techniques enhance rapport
(28%) did not differ by type of loss. sant medication responded to IPT building, assistance in restoring effec-
While provocative, none of these com- (40%) as those who responded to CGT tive interpersonal functioning, and
parisons was statistically significant. without medication (42%). guided treatment termination.
Heterogeneity is another potential In summary, we conducted the first
COMMENT limitation. It is possible that sub- randomized controlled trial of therapy
This randomized controlled trial groups might respond differently to dif- targeting symptoms of complicated
showed better response to CGT than to ferent treatment approaches. We had grief. We found better response to CGT
IPT, with a number needed to treat of no prior hypotheses regarding these compared with IPT, which is a proven
4.3. Since this is the first such study in variables; however, we had insuffi- efficacious psychotherapy for depres-
this chronically ill population, this re- cient power to detect differences. For sion. Similarity of ICG scores across age,
sult is encouraging. Nevertheless, only example, we observed that patients ex- cultural, and death-related variables
51% responded to CGT, and it is clear periencing violent loss had a very low supports the diagnostic validity of the
that more work is needed. In other stud- response to IPT (13%). On the other syndrome. Our treatment findings sug-
ies,20 antidepressant medication alone hand, parents who lost a child showed gest that complicated grief is a specific
has shown small changes in compli- a much lower rate of response to CGT condition in need of a specific treat-
cated grief symptoms. However, pa- than patients with other losses (17% vs ment. More research is needed to con-
tients taking antidepressant medica- 60%). Our study was not large enough firm our findings, to test potential mod-
tion prior to starting this study did have to have confidence in these observa- erators of treatment response, and to
a marginally better outcome than those tions; thus, they should be considered improve treatment acceptance.
not taking medication. Systematic study preliminary. Our conclusions are also Author Contributions: Dr Shear and Ms Houck had
of combined medication and psycho- limited by the 26% dropout rate from full access to all of the data in the study and take re-
therapy is needed. both treatments and the additional 10% sponsibility for the integrity of the data and the ac-
curacy of the data analysis.
Participants in our study spanned the who refused to undergo key CGT pro- Study concept and design: Shear, Frank.
adult age range and included individu- cedures. Acquisition of data: Shear, Frank.
Analysis and interpretation of data: Shear, Frank,
als who lost parents, spouses, chil- Intervention studies for bereaved in- Houck, Reynolds.
dren, other relatives, or close friends dividuals often recruited participants Drafting of the manuscript: Shear, Houck.
Critical revision of the manuscript for important in-
through violent (33%) or natural (66%) without regard to symptom status and tellectual content: Shear, Frank, Reynolds.
deaths; 22% of participants were Afri- used supportive interventions.46,47 A re- Statistical analysis: Shear, Frank, Houck, Reynolds.
can American and 40% were older than cent meta-analysis of bereavement sup- Obtained funding: Shear.
Administrative, technical, or material support: Shear,
50 years. The heterogeneity of the port interventions showed an effect size Houck.
sample provides further evidence that of 0.15. 48 However, 2 earlier stud- Study supervision: Shear, Frank.
Financial Disclosures: Dr Shear has received finan-
complicated grief, like most DSM-IV ies49,50 examined efficacy of an expo- cial support from Pfizer and Forest Pharmaceuticals.
disorders, can be identified in differ- sure-based treatment for individuals Dr Frank has received financial support from Pfizer,
Pfizer Italia, Eli Lilly, Forest Research Institute, and the
ent adult populations and in different considered to have pathological grief Pittsburgh Foundation.
psychosocial contexts. and showed significant treatment ef- Funding/Support: This work was supported by grants
R01MH60783, P30MH30915, and P30MH52247
This study has several important fects on measures of anxiety and de- from the National Institute of Mental Health (NIMH).
limitations. Forty-five percent of study pression. There was no measure of com- Role of the Sponsor: The NIMH had no direct input
participants were taking psychotropic plicated grief in these studies. into the design or conduct of the study; collection, man-
agement, analysis, or interpretation of the data; or
medications. We considered it neces- Our treatment is the first to target preparation, review, or approval of the manuscript.
sary to permit continued use of medi- complicated grief symptoms directly. Acknowledgment: We acknowledge the contribu-
tions of the following individuals, without whose as-
cation for co-occurring DSM-IV Axis I The dual-process model of coping of sistance this project would not have been possible:
disorders for which CGT, and some- Stroebe and Schut29 forms the frame- Krissa Caroff, BS (study coordinator); Jacqueline Fury,
BS (study research associate); Russell Silowash, BS (data
times IPT, had not been studied. We be- work for our approach. Complicated manger); Mary Herschk (study administrative assis-
lieved we would unnecessarily limit the grief treatment is implemented using tant); Rose Zingrone, LCSW, and Randi Taylor, PhD
©2005 American Medical Association. All rights reserved. (Reprinted) JAMA, June 1, 2005—Vol 293, No. 21 2607
Downloaded from www.jama.com on February 21, 2008

8. TREATMENT OF COMPLICATED GRIEF
(independent evaluators); Andrea Fagiolini, MD (study ments associated with diagnostic criteria for trau- Rape: Cognitive-Behavioral Therapy for PTSD. New
pharmacotherapist); Bonnie Gorscak, PhD (CGT backup matic grief. Psychol Med. 2000;30:857-862. York, NY: Guilford Press; 1998:286.
supervisor); Allan Zuckoff, PhD (study psychothera- 14. Prigerson HG, Bierhals AJ, Kasl SV, et al. Trau- 32. Miller WR, Rollnick R. Motivational Interview-
pist and trainer in motivational enhancement therapy); matic grief as a risk factor for mental and physical ing: Preparing People for Change. 2nd ed. New York,
Daniel Ford, MD, Wayne Katon, MD, and Sidney morbidity. Am J Psychiatry. 1997;154:616-623. NY: Guilford Press; 2002.
Zisook, MD (data and safety monitoring board con- 15. Chen JH, Bierhals AJ, Prigerson HG, et al. Gen- 33. Guy W. Clinical Global Impressions: ECDEU As-
sultants); and David J. Kupfer, MD, Edna Foa, PhD, der differences in the effects of bereavement-related sessment Manual for Psychopharmacology, Revised.
Holly Prigerson, PhD, and Camille Wortman, PhD (con- psychological distress on health outcomes. Psychol Bethesda, Md: National Institute of Mental Health; 1976.
sultants). Med. 1999;29:367-380. 34. First MB, Spitzer RL, Gibbon M, Williams JB. Struc-
16. Ott CH. The impact of complicated grief on men- tured Clinical Interview for DSM-IV Axis I Disorders.
tal and physical health at various points in the be- New York: New York State Psychiatric Institute; 1995.
REFERENCES reavement process. Death Stud. 2003;27:249-272. 35. Hamilton M. A rating scale for depression. J Neu-
17. Middleton W, Burnett P, Raphael B, Martinek N. rol Neurosurg Psychiatry. 1960;23:56-62.
1. Prigerson HG, Jacobs SC. Caring for bereaved pa- The bereavement response: a cluster analysis. Br J 36. Shear MK, Vander Bilt J, Rucci P, et al. Reliability
tients: “all the doctors just suddenly go.” JAMA. 2001; Psychiatry. 1996;169:167-171. and validity of a structured interview guide for the
286:1369-1376. 18. Jacobs SC. Pathologic Grief: Maladaptation to Hamilton anxiety rating scale (SIGH-A). Depress
2. Clayton PJ, Halikas JA, Maurice WL. The depres- Loss. Washington DC: American Psychiatric Press; Anxiety. 2001;13:166-178.
sion of widowhood. Br J Psychiatry. 1972;120:71-77. 1993. 37. Mundt JC, Marks IM, Shear MK, Greist JH. The work
3. Jacobs SC. Measures of the psychological distress 19. National Center for Health Statistics. Births, mar- and social adjustment scale: a simple measure of impair-
of bereavement. In: Zisook S, ed. Biopsychosocial As- riages, divorces and deaths. Available at: http://www ment in functioning. Br J Psychiatry. 2002;180:461-464.
pects of Bereavement. Washington, DC: American Psy- .cdc.gov/nchs/data/nvsr/nvsr52/nvsr52_22.pdf. Ac- 38. Beck AT, Steer RA. Manual for the Revised Beck
chiatric Press; 1987:125-138. cessed May 3, 2005. Depression Inventory. San Antonio, Tex: Psychologi-
4. Prigerson HG, Maciejewski PK, Reynolds CF, et al. 20. Zygmont M, Prigerson HG, Houck PR, et al. A post cal Corp; 1987.
Inventory of Complicated Grief: a scale to measure mal- hoc comparison of paroxetine and nortriptyline for 39. Beck AT, Epstein N, Brown G, Steer RA. An in-
adaptive symptoms of loss. Psychiatry Res. 1995;59: symptoms of traumatic grief. J Clin Psychiatry. 1998; ventory for measuring clinical anxiety: psychometric
65-79. 59:241-245. properties. J Consult Clin Psychol. 1988;56:893-897.
5. Horowitz MJ, Siegel B, Holen A, Bonanno GA, Mil- 21. Shear MK, Frank E, Foa EB, et al. Traumatic grief 40. Chatellier G, Zapletal E, Lemaitre D, Menard J, De-
brath C, Stinson CH. Diagnostic criteria for compli- treatment: a pilot study. Am J Psychiatry. 2001;158: Goulet P. The number needed to treat: a clinically use-
cated grief disorder. Am J Psychiatry. 1997;154:904- 1506-1508. ful nomogram in its proper context. BMJ. 1996;312:
910. 22. Klerman GL, Weissman MM, Rounsaville B, Chev- 426-429.
6. Prigerson HG, Shear MK, Jacobs SC, et al. Con- ron ES. Interpersonal Psychotherapy of Depression. 41. Breslow N, Day N. Statistical methods in cancer
sensus criteria for traumatic grief: a rationale and pre- New York, NY: Basic Books; 1984. research, volume I: the analysis of case-control studies.
liminary empirical test. Br J Psychiatry. 1999;174:67- 23. Ablon JS, Jones EE. Validity of controlled clinical IARC Sci Publ. 198032:5-338.
73. trials of psychotherapy: findings from the NIMH treat- 42. Yonkers KA, Samson J. Mood disorder measures.
7. Zisook S, Shuchter SR, Pedrelli P, et al. Acute, open ment of depression collaborative research program. In: Handbook of Psychiatric Measures. Washington,
trial of bupropion SR therapy in bereavement. J Clin Am J Psychiatry. 2002;159:775-783. DC: American Psychiatric Association; 2000:527.
Psychiatry. 2001;62:227-230. 24. Frank E, Swartz HA, Kupfer DJ. Interpersonal and 43. Kearns NP, Cruickshank CA, McGuigan KJ, Riley
8. Reynolds CF, Miller MD, Pasternak RE, et al. Treat- social rhythm therapy: managing the chaos of bipo- SA, Shaw SP, Snaith RP. A comparison of depression
ment of bereavement-related major depressive epi- lar disorder. Biol Psychiatry. 2000;48:593-604. rating scales. Br J Psychiatry. 1982;141:45-49.
sodes in later life: a randomized, double-blind, placebo- 25. Cyranowski J, Frank E, Shear MK, Swartz HA, Fa- 44. Shear MK, Feske U, Brown C, Clark D, Mammen
controlled study of acute and continuation treatment giolin A, Scott J. Interpersonal psychotherapy for de- O, Scotti J. Anxiety disorders measures. In: Hand-
with nortriptyline and interpersonal psychotherapy. Am pression with panic spectrum features (IPT-PS): a pi- book of Psychiatric Measures. Washington, DC: Ameri-
J Psychiatry. 1999;156:202-208. lot study. Depress Anxiety. In press. can Psychiatric Association; 2000:555.
9. Prigerson HG, Shear MK, Jacobs S, et al. Grief and 26. Frank E. Interpersonal psychotherapy as a main- 45. Kobak KA, Reynolds WM, Greist JH. Develop-
its relation to posttraumatic stress disorder. In: Nutt tenance treatment for patients with recurrent ment and validation of a computer-administered ver-
D, Davidson JR, Zohar J, eds. Posttraumatic Stress Dis- depression. Psychotherapy. 1991;28:259-266. sion of the Hamilton Rating Scale. Psychol Assess. 1993;
orders: Diagnosis, Management and Treatment. New 27. Frank E, Grochocinski VJ, Spanier CA, et al. In- 5:487-492.
York, NY: Martin Dunitz Publishers; 2000:163-177. terpersonal psychotherapy and antidepressant medi- 46. Windholz MJ, Marmar CR, Horowitz MJ. A re-
10. Prigerson HG, Frank E, Kasl SV, et al. Compli- cation: evaluation of a sequential treatment strategy view of the research on conjugal bereavement: im-
cated grief and bereavement-related depression as dis- in women with recurrent major depression. J Clin pact on health and efficacy of intervention. Compr
tinct disorders: preliminary empirical validation in el- Psychiatry. 2000;61:51-57. Psychiatry. 1985;26:433-447.
derly bereaved spouses. Am J Psychiatry. 1995;152: 28. Weissman MM, Markowitz JC, Klerman GL. Com- 47. Marmar CR, Horowitz MJ, Weiss DS, Wilner NR,
22-30. prehensive Guide to Interpersonal Psychotherapy. New Kaltreider NB. A controlled trial of brief psycho-
11. Prigerson HG, Bierhals AJ, Kasl SV, et al. Com- York, NY: Basic Books; 2000. therapy and mutual-help group treatment of conju-
plicated grief as a disorder distinct from bereave- 29. Stroebe M, Schut H. The dual process model of gal bereavement. Am J Psychiatry. 1988;145:203-209.
ment- related depression and anxiety: a replication coping with bereavement: rationale and description. 48. Jordan JR, Neimeyer RA. Does grief counseling
study. Am J Psychiatry. 1996;153:1484-1486. Death Stud. 1999;23:197-224. work? Death Stud. 2003;27:765-786.
12. Melhem NM. Traumatic grief among adoles- 30. Hembree EA, Foa EB, Dorfan NM, Street GP, Kow- 49. Mawson D, Marks IM, Ramm L, Stern RS. Guided
cents exposed to their peer’s suicide. Am J Psychiatry. alski J, Tu X. Do patients drop out prematurely from mourning for morbid grief: a controlled study. Br J
2004;161:1411-1416. exposure therapy for PTSD? J Trauma Stress. 2003;16: Psychiatry. 1981;138:185-193.
13. Silverman GK, Jacobs SC, Kasl SV, Maciejewski 555-562. 50. Ramsay RW. Behavioural approaches to
PK, Noaghiul FS, Prigerson HG. Quality of life impair- 31. Foa EB, Rothbaum BO. Treating the Trauma of bereavement. Behav Res Ther. 1977;15:131-135.
2608 JAMA, June 1, 2005—Vol 293, No. 21 (Reprinted) ©2005 American Medical Association. All rights reserved.
Downloaded from www.jama.com on February 21, 2008