ORIGINAL CONTRIBUTION
An Empirical Examination
of the Stage Theory of Grief
Paul K. Maciejewski, PhD
Context The stage theory of grief remains a widely accepted model of bereavement
Baohui Zhang, MS
adjustment still taught in medical schools, espoused by physicians, and applied in di-
verse contexts. Nevertheless, the stage theory of grief has previously not been tested
Susan D. Block, MD
empirically.
Holly G. Prigerson, PhD
Objective To examine the relative magnitudes and patterns of change over time post-
HE NOTION THAT A NATURAL PSY- loss of 5 grief indicators for consistency with the stage theory of grief.
chological response to loss
Design, Setting, and Participants Longitudinal cohort study (Yale Bereavement
involves an orderly progres-
Study) of 233 bereaved individuals living in Connecticut, with data collected between
Tsion through distinct stages of
January 2000 and January 2003.
bereavement has been widely accepted
Main Outcome Measures Five rater-administered items assessing disbelief, yearn-
by clinicians and the general public.
ing, anger, depression, and acceptance of the death from 1 to 24 months postloss.
BowlbyandParkes1-4 werethefirsttopro-
Results Counter to stage theory, disbelief was not the initial, dominant grief indi-
pose a stage theory of grief for adjust-
cator. Acceptance was the most frequently endorsed item and yearning was the domi-
ment to bereavement that included 4
nant negative grief indicator from 1 to 24 months postloss. In models that take into
stages: shock-numbness, yearning-
account the rise and fall of psychological responses, once rescaled, disbelief de-
searching, disorganization-despair, and
creased from an initial high at 1 month postloss, yearning peaked at 4 months post-
reorganization. Kübler-Ross5 adapted loss, anger peaked at 5 months postloss, and depression peaked at 6 months postloss.
Acceptance increased throughout the study observation period. The 5 grief indicators
Bowlby and Parkes theory to describe a
achieved their respective maximum values in the sequence (disbelief, yearning, an-
5-stage response of terminally ill patients
ger, depression, and acceptance) predicted by the stage theory of grief.
to awareness of their impending death:
denial-dissociation-isolation, anger, bar- Conclusions Identification of the normal stages of grief following a death from natu-
ral causes enhances understanding of how the average person cognitively and emo-
gaining, depression, and acceptance. The
tionally processes the loss of a family member. Given that the negative grief indicators
stage theory of grief became well-
all peak within approximately 6 months postloss, those who score high on these in-
known and accepted, and has been gen-
dicators beyond 6 months postloss might benefit from further evaluation.
eralizedtoawidevarietyoflosses,includ-
JAMA. 2007;297:716-723 www.jama.com
ing children s reactions to parental
separation,3 adults reactions to marital
separation,6 and clinical staffs reactions time.10-14 Bonanno et al10 found 5 diver- (yearning-anger-anxiety), depression-
to the death of an inpatient.7 A 1997 sur- gent grieving trajectories from preloss to mourning, and recovery. To date, no
veyconductedbyDowne-Wamboldtand 18 months postloss (common grief, study has explicitly tested whether the
Tamlyn8 documented the heavy reli- chronic grief, chronic depression, im- normal course of adjustment to a natu-
ance of medical education on the Kübler- provement during bereavement, and re- ral death progresses through stages of
Ross model of grief. The National Can- silience). Wortman and Silver13 exam-
Author Affiliations: Department of Psychiatry, Wom-
cer Institute currently maintains a Web ined and disproved the necessity of 1
en s Health Research, and Magnetic Resonance Re-
search Center, Yale University School of Medicine, New
site on loss, grief, and bereavement that stage in the grief theory when they found
Haven, Conn (Dr Maciejewski); Center for Psycho-
describes the phases of grief.9 As that depression was not an inevitable re-
Oncology and Palliative Care Research, Dana-Farber
Cancer Institute, Boston, Mass (Ms Zhang and Drs
entrenchedasthenotionofphasesofgrief sponse to loss. Based on Bowbly and
Block and Prigerson); and Department of Psychiatry,
may be, the hypothesized sequence of Parkes 1-4 and Kübler-Ross 5 theories, Ja-
Brigham and Women s Hospital, and Harvard Medi-
grief reactions has previously not been cobs14 synthesized and illustrated the hy- cal School Center for Palliative Care, Boston, Mass (Drs
Block and Prigerson).
investigated empirically. pothesized stage theory of grief, in which
Corresponding Author: Holly G. Prigerson, PhD, Cen-
Several bereavement scholars have in- the normal response to loss progresses ter for Psycho-Oncology and Palliative Care Re-
search, Dana-Farber Cancer Institute, 44 Binney St SW,
vestigated particular aspects of, or dia- through the following grief stages:
G440A, Boston, MA 02115 (holly_prigerson@dfci
grammed changes in, grief reactions over numbness-disbelief, separation distress .harvard.edu).
716 JAMA, February 21, 2007 Vol 297, No. 7 (Reprinted) ©2007 American Medical Association. All rights reserved.
STAGE THEORY OF GRIEF
disbelief, yearning, anger, depression,
Figure 1. Hypothesized Stage Theory of Grief
and acceptance.
The identification of the patterns of
Disbelief Yearning Anger Depression
typical grief symptom trajectories is of
Acceptance
clinical interest because it enhances the
understanding of how individuals cog-
nitively and emotionally process the
death of someone close. Such knowl-
edge aids in the determination of
whether a specific pattern of bereave-
ment adjustment is normal or not. Once
the normal patterns of grief are known,
individuals with abnormal bereave-
ment adjustment can be identified and
referred for treatment when indicated.
This study used data from a sample of
Time From Loss
community-based bereaved individu-
als to examine the course of disbelief,
yearning, anger, depression, and accep-
tance as described by Jacobs14 from 1 to als, collected data between January 2000 ticipants from Bridgeport / Fairfield
24 months postloss. FIGURE 1 illus- and January 2003, and was funded by the (mean [SD], 63.2 [11.5] years) (P=.05).
trates the hypothesized sequence of National Institute of Mental Health. For The institutional review boards of all par-
stages of grief for this analysis. Be- greater Bridgeport/Fairfield, Conn, the ticipating sites approved the research
cause approximately 94% of US deaths names of the newly bereaved ( 6 protocol.
result from natural causes (eg, vehicle months) were obtained from the divi- Individuals were invitedto participate
crashes, suicide),15 deaths from unnatu- sion of the American Association of Re- inthestudyviaaletterthatdescribedhow
ral causes (eg, car crashes, suicide) were tired Persons Widowed Persons Ser- their nameswereobtained, identifiedthe
excluded thereby enabling the results vice, a community-based outreach investigators, outlined the aims and pro-
to be generalized to the most common program. For the New Haven, Conn, cedures, and noted that they would be
types of deaths. Individuals who met the metropolitan and surrounding areas, contacted by study staff in the following
criteria for complicated grief disor- names were obtained from obituaries weekunlesstheyinformedusoftheirwish
der16,17 also were excluded so that the listed in the New Haven Register, through not to be contacted. Of the 575 persons
results would represent normal be- newspaper advertisements, fliers, per- contacted,317 (55.1%) agreedto partici-
reavement reactions. Although the pro- sonal referrals, and referrals from the pate. Reasons for nonparticipation in-
posed stage theory of grief1-5,14 does not chaplain s office of the St Raphael Hos- cludedreluctancetoparticipateinresearch
specify the precise timing of the stages, pital. A comparison between greater (n=11; 4.3%); being too busy (n=46;
Jacobs14 described the normal griev- Bridgeport/Fairfield Bureau of Vital Rec- 17.8%); being too upset (n=27; 10.5%);
ing process and each of its stages as ords death certificates and the Wid-  doing fine (n=23; 8.9%); not being in-
being completed within 6 months fol- owed Persons Service list during the same terested or having no reason (n=145;
lowing the loss of a loved one. How- 3-month period revealed that the Wid- 56.2%); and having other reasons (n=6;
ever, in the absence of an established, owed Persons Service listings captured 2.3%).Comparedwithparticipants,non-
empirical foundation for the length of 95% of all deaths leaving behind a wid- participantsweresignificantlymorelikely
time associated with the normal griev- owed individual, suggesting that the list- tobemale(25.9%vs37.2%;P .001)and
ing process, the normal grieving pro- ing provided an unbiased and compre- older (mean [SD] age, 61.7 [13.1] years
cess was not assumed to be limited to hensive ascertainment of recently vs 68.8 [13.7] years) (P .001). Non
6 months postloss in this study. In- widowed individuals in the sampled re- English-speakingpersonsandthosecon-
stead, the grief indicators were exam- gion. Participants recruited from greater sideredtoofrailtocompletetheinterview
ined as functions of time up to 24 New Haven (37.0%) did not differ sig- wereineligible.The317participantswere
months postloss. nificantly from participants recruited interviewed at a mean (SD) of 6.3 (7.0)
from greater Bridgeport / Fairfield monthsafterthedeathofalovedone.The
METHODS
(63.0%) with respect to sex, income, edu- firstfollow-upinterview(n=296;93.4%)
Study Sample
cation, race/ethnicity, or quality of life. wascompletedatamean(SD)of10.9(6.1)
The Yale Bereavement Study, a longitu- Participants recruited from greater New monthspostloss;secondfollow-upinter-
dinal examination of grief in a commu- Haven were significantly younger (mean view (n=263; 83.0%) at a mean (SD) of
nity-based sample of bereaved individu- [SD] age, 59.7 [16.4] years) than par- 19.7 (5.8) months postloss. Written in-
©2007 American Medical Association. All rights reserved. (Reprinted) JAMA, February 21, 2007 Vol 297, No. 7 717
Indicator Rating
STAGE THEORY OF GRIEF
at baseline interview was 0.65 (P .001).
Table 1. Demographic Variables of the Yale Bereavement Study Sample Compared With
To be consistent with the scale levels
2005 Data From the US Census
of other grief indicators, all levels of
Yale Bereavement
Study Sample 2005 US depressed mood were increased by 1 so
(N = 233), No. (%) Census, %
that 1 indicated  absence of depressed
Age 65 y 125 (53.5) 76.1*
mood and 5 indicated  patient reports
Female sex 166 (71.2) 80.7*
virtually only these feeling states in his
White race 226 (97.0) 80.2*
spontaneous verbal and non-verbal
Education beyond high school 145 (62.2) 54.7
communication.
Median household income, $ 52 000 46 242!
Individuals self-identified their racial/
*US widowed population (n = 13.7 million).
ethnic status according to the racial/
US general population aged 25 years or older (n = 189.0 million).
! US general household population (N = 111.1 million households).
ethnic categories defined in the US Cen-
sus.18 They also reported the cause of
formedconsentwasobtained fromallin- Measures of Grief death for the family member or loved
dividuals enrolled in the study. The indicators of disbelief, yearning, one. For deaths due to a terminal ill-
Of the 317 individuals identified, anger, and acceptance of the death were ness, the date of the diagnosis was re-
58 were excluded because they met assessed using single items obtained corded. Diagnoses of the terminal ill-
criteria for complicated grief disor- from the rater-administered version of ness within 6 months (52/199; 26.1%)
der, 19 because they survived trau- the Inventory of Complicated Grief- were compared with those 6 months or
matic deaths, and 14 because they Revised, formerly known as the Trau- longer (147/199; 73.9%) prior to the
had missing data on examined mea- matic Grief Response to Loss.19 death. Six months was used as the
sures. The study sample (N = 233) Although it would have been prefer- threshold because terminal diagnoses
consisted of individuals who did not able to use separate scales for the assess- of less than 6 months resulted in
meet criteria for complicated grief ment of yearning, disbelief, anger, and smaller, less reliable groupings and else-
disorder16,17 during the study; had a acceptance of the death, no such scales where16,17 it has been determined that
family member or loved one who exist for each of these grief stages. To 6 months is the time after which nor-
died from natural not traumatic maximize consistency across mea- mal grief can be distinguished from
causes; and had at least 1 complete sures, single items were used for all grief complicated grief disorder.
assessment of the 5 grief indicators phase indicators. Single-item inter-
Statistical Analyses
included in the stage theory of grief view screenings have proven remark-
within 24 months postloss. The par- ably accurate in the prediction of depres- Statistical analyses were conducted to
ticipants were significantly older sion.20 The frequency, rather than test for significant differences in the
(mean [SD] age, 62.9 [13.1] years; severity, of each grief indicator was used magnitude of each of the 5 grief indi-
53.5% aged 65 years) and more as the response format in the Inven- cators within each of the 3 postloss pe-
likely to be white (97.0%) than the tory of Complicated Grief-Revised riods ( 6 months [1-6 months cat-
excluded individuals (mean [SD] because frequency has proven to be a egory], 6 to 12 months [6-12
age, 58.5 [15.0] years; 90.4% white) more effective means of evaluating the months category], and 12 to 24
but did not significantly differ with impact of events.21 Grief phase indica- months [12-24 months category]); to
respect to sex, income, education, tors were measured using a 5-point compare the pattern of changes in the
and relationship to the deceased. The Likert scale in which 1 equaled less than absolute levels of each of the 5 grief in-
vast majority of participants were once per month; 2, monthly; 3, weekly; dicators over time; and to determine
spouses of the deceased (83.8%). The 4, daily; and 5, several times per day. when each of the 5 grief indicators
remaining participants (16.2%) were These items showed moderately high achieved its maximum value.
adult children, parents, or siblings of correlations with the total Inventory of Specifically, single-sample t tests and
the deceased. Complicated Grief-Revised score at nonlinear, ordinary least squares re-
The data from the participants were baseline interview, which ranged in gression analyses were used to exam-
compared with data from the 2005 US magnitude from 0.47 to 0.57 (all com- ine the differences in magnitude be-
Census (TABLE 1).18 Compared with the parisons yielded P .001). To enhance tween grief indicators at a given time
US widowed population, the study par- comparability in the measurement of postloss and changes in grief indica-
ticipants were younger, more likely to be each indicator, depression was assessed tors as a function of time postloss.
male,andahigherproportionwerewhite. using the single-item depressed mood Single-sample t tests were used to ex-
Compared with the US general popula- in the Hamilton Rating Scale for Depres- amine within-person differences in
tionaged25years orolder, thestudy par- sion.22 The correlation between magnitude between the 5 grief indica-
ticipants were better educated and had a depressed mood and the total Hamil- tors postloss at 1 to 6 months, 6 to 12
higher median household income. ton Rating Scale for Depression score months, and 12 to 24 months and
718 JAMA, February 21, 2007 Vol 297, No. 7 (Reprinted) ©2007 American Medical Association. All rights reserved.
STAGE THEORY OF GRIEF
Table 2. Grief Indicators Assessed During 3 Postloss Periods
Period of Postloss Assessment, mo
1-6 6-12 12-24
Grief No. of Mean No. of Mean No. of Mean
Indicator* Participants (SD) Participants (SD) Participants (SD)
Disbelief 173 2.27 (1.30) 211 1.80 (1.03) 205 1.61 (0.94)
Yearning 171 3.77 (1.14) 212 3.18 (1.19) 205 2.64 (1.20)
Anger 174 1.87 (1.16) 210 1.80 (1.04) 205 1.55 (0.87)
Depression 174 2.29 (1.25) 213 2.29 (1.21) 205 1.80 (1.07)
Acceptance 143 4.11 (1.21) 209 4.49 (0.93) 205 4.70 (0.68)
*Indicators are measured on a scale of 1 to 5.
within-person temporal changes in Institute Inc, Cary, NC). P .05 was postloss: t204=35.24, P .001), and de-
magnitude of each grief indicator post- considered significant. pression (1-6 months postloss:
loss between 1 to 6 months and 6 to 12 A series of multivariable analyses of t142=11.64, P .001; 6-12 months post-
months and between 6 to 12 months variance were conducted to evaluate loss: t208=18.84, P .001; 12-24 months
and 12 to 24 months. whether demographic variables and re- postloss: t204=29.97, P .001).
Nonlinear, ordinary least squares re- port of diagnosis of terminal illness Yearning is significantly greater than
gression analyses were used to model within 6 months of the death were sig- disbelief (1-6 months postloss:
the trajectory of each grief indicator as nificantly related to the 5 grief indica- t169=13.57, P .001; 6-12 months post-
a function of time postloss. Because the tors or to the within-person differ- loss: t210=15.57, P .001; 12-24 months
stage theory of grief predicts the se- ences between or temporal changes in postloss: t204=12.49, P .001), anger
quential rise and fall of each of the grief the 5 grief indictors. (1-6 months postloss: t170= 16.43,
indicators as a function of time post- P .001; 6-12 months postloss:
RESULTS
loss (ie, phase), we chose the follow- t209=15.10, P .001; 12-24 months post-
ing parametric functional form that The means and SDs for the 5 grief in- loss: t204=12.43, P .001), and depres-
would capture such phases: dicators of disbelief, yearning, anger, de- sion (1-6 months postloss: t170=14.40,
pression, and acceptance postloss at 1 P .001; 6-12 months postloss:
Y=[A B (-t/Ä 1)] exp(-1D 2 t/Ä) C
to 6 months, 6 to 12 months, and 12 t211=9.75, P .001; 12-24 months post-
where Y represents the value of the grief to 24 months appear in TABLE 2. Within loss: t204=9.41, P .001).
indicator and the term t/Ä represents each period, acceptance is greater than Depression is significantly greater
time postloss with t scaled by the model disbelief, yearning, anger, and depres- than anger (1-6 months postloss:
parameter Ä. The expression [A B sion; yearning is greater than disbe- t173=3.61, P .001; 6-12 months post-
(-t/Ä 1)] exp(-1D 2 t/Ä) represents a lin- lief, anger, and depression; and depres- loss: t209=5.32, P .001; 12-24 months
ear combination of normalized sion is greater than anger. Between 1 postloss: t204=3.16, P=.002).
(weighted) zero-order and first-order and 6 months postloss and 6 and 12 Disbelief is significantly greater than
Laguarre polynomials, scaled by the months postloss, disbelief and yearn- anger at 1 to 6 months postloss
model parameters A and B, respec- ing decline and acceptance increases. (t172=3.22, P=.002); depression is sig-
tively, included to capture the antici- From 6 to 12 months postloss and 12 nificantly greater than disbelief at 6 to
pated rise and fall in the data. Model to 24 months postloss, disbelief, yearn- 12 months postloss (t210=5.22, P .001)
parameter C represents the asymp- ing, anger, and depression decline and and at 12 to 24 months postloss
totic value that the grief indicator ap- acceptance increases. (t204=2.19, P=.03).
proaches as time postloss increases to More specifically, acceptance is sig- Between 1 and 6 months postloss and
infinity. One observation per person nificantly greater than disbelief (1-6 6 and 12 months postloss, disbelief
(N=233), selected randomly among months postloss: t142=10.79, P .001; (t157= 4.78, P .001) and yearning
those observations that contained com- 6-12 months postloss: t208= 23.16, (t156=7.89, P .001) decline and accep-
plete data for each of the 5 grief indi- P .001; 12-24 months postloss: tance increases (t130=3.91, P .001). Be-
cators within 24 months postloss, was t204= 31.88, P .001), yearning (1-6 tween 6 and 12 months postloss and 12
used to fit these regression models. For months postloss: t142=2.11, P=.04; 6-12 and 24 months postloss, disbelief
each grief indicator, the model para- months postloss: t208=10.80, P .001; (t190=2.84, P=.005), yearning (t191=5.96,
meters Ä, A, B, and C were estimated by 12-24 months postloss: t204= 19.39, P .001), anger (t189=3.91, P .001),
means of nonlinear, ordinary least P .001), anger (1-6 months postloss: and depression (t192= 5.60, P .001)
squares regression implemented using t142=12.66, P .001; 6-12 months post- decline and acceptance increases
PROC MODEL in SAS version 9.1 (SAS loss: t208=23.14, P .001; 12-24 months (t188=3.37, P .001).
©2007 American Medical Association. All rights reserved. (Reprinted) JAMA, February 21, 2007 Vol 297, No. 7 719
STAGE THEORY OF GRIEF
FIGURE 2 displays the results of the tween 1 and 4 months postloss, de- than depression between 1 and 4
nonlinear regression analyses. Accord- creases between 4 and 24 months post- months postloss. Depression in-
ing to the models displayed in the top loss, and is greater than disbelief, anger, creases between 1 and 6 months post-
part of Figure 2, acceptance increases and depression between 1 and 24 loss, decreases between 6 and 24
monotonically (uniformly in 1 direc- months postloss. Disbelief decreases months, is greater than disbelief post-
tion), and is greater than each of the monotonically between 1 and 24 loss between 4 and 24 months, and is
other grief indicators between 1 and 24 months, is greater than anger between greater than anger between 1 and 24
months postloss. Yearning increases be- 1 and 6 months postloss, and is greater months postloss. Anger increases be-
tween 1 and 5 months postloss and de-
creases between 5 and 24 months post-
Figure 2. Empirical and Rescaled Models for Grief Indicators as Functions of Time
loss. The close agreement between the
models and the data in the top part of
Grief Indicators
5.0
Figure 2 indicates that the phasic func-
Acceptance
tional form specified in the regression
models adequately represent the data.
4.0
The bottom part of Figure 2 dis-
plays the regression models following a
rescaling procedure that constrains each
3.0
grief indicator to fall within the inter-
Yearning
val of 0 through 1. In the top part of
Figure 2, the relative locations in time
2.0
of the peaks of the grief indicators are
Depression
obscured because the curves are not side
Anger
by side, thereby making comparisons
Disbelief
1.0
difficult. Those comparisons are facili-
0 6 12 18 24
tated in the bottom part of Figure 2 by
Time From Loss, mo
placing all of the indicators on the same
Rescaled Grief Indicators
scale. The 5 grief indicators achieved
Disbelief Yearning Anger Depression Acceptance their respective maximum values in the
Maximum
1.0
sequence (disbelief, yearning, anger, de-
Value
pression, and acceptance) predicted by
the stage theory of grief. Given that there
0.8
are 120 possible sequences of these 5 in-
dicators, the probability that the ob-
0.6
served sequence is exactly the se-
quence predicted by the stage theory of
0.4
grief by chance alone is P=.008.
Based on the results of the multivari-
able analyses of variance, the demo-
0.2
graphic factors of age, sex, race/
ethnicity (white/nonwhite), education,
Minimum
0
Value
and income and a terminal illness diag-
0 6 12 18 24
Time From Loss, mo nosis reported within 6 months of the
death were largely unrelated to within-
Top, The curves represent grief indicators as functions of time based on nonlinear regression models estimated
person differences and temporal changes
from the data (N=233). The data markers along the x-axis are determined by the mean value of time from
in the grief indictors throughout the
loss for the individuals included in the 10 groups of observations (n=23 observations per group for 9 groups;
n=26 observations for 1 group). The corresponding error bars indicate SDs. These 10 groups of observations
study observation period (1-6, 6-12, and
were formed by ordering all of the observations used in the regression analyses (N=233) by increasing time
12-24 months postloss). Education be-
postloss (observations that occurred at the same time postloss were randomly assigned a position in the or-
dered sequence of observations for that time), and then assigning the first 23 observations on this ordered list
yond high school was significantly as-
to the first group, the next 23 observations to the second group, etc. The regression curves are based on the
sociated with grief indicators 12 to 24
analysis of individual data points (N=233) for which time from loss varies from 1.5 to 23 months. Bottom, The
curves represent grief indicators as functions of time based on nonlinear regression models after the following months postloss (Wilks = 0.94,
rescaling procedure: (t)=[Y(t) -Ymin]/Ymax - Ymin], where Y(t) is the model value for the grief indicator at time
F5,199=2.52; P=.03), due to its signifi-
t, and Ymin and Ymax are the minimum and maximum model values of the grief indicator, respectively, between
1 and 24 months postloss. The 5 grief indicators achieve their respective maximum values in the exact se- cant associations with lesser disbelief
quence (disbelief, yearning, anger, depression, and acceptance) predicted by the hypothesized theory of grief
(P=.05) and depression (P=.003), and
presented in Figure 1.
with greater acceptance (P=.02) dur-
720 JAMA, February 21, 2007 Vol 297, No. 7 (Reprinted) ©2007 American Medical Association. All rights reserved.
Indicator Rating
Rescaled Indicator Rating
STAGE THEORY OF GRIEF
ing that period. Education beyond high als who survived a family member s trau- cal Manual of Mental Disorders, Fourth
school was also significantly associated matic death and those who met criteria Edition with respect to bereavement.
with within-person differences in grief for complicated grief disorder,16 both Models that tested for phasic epi-
indicators 6 to 12 months postloss groups of whom were found in prelimi- sodes of each grief indicator revealed that
(Wilks =0.95, F4,204=2.51; P=.04) and nary analyses to have significantly lower disbelief about the death is highest ini-
12 to 24 months postloss (Wilks =0.94, levels of acceptance relative to the study tially. As disbelief declined from the first
F4,200=3.11; P=.02) due to its signifi- sample. The lower frequency of accep- month postloss, yearning rose until 4
cant associations with greater differ- tance of the death among participants months postloss and then declined. An-
ences between acceptance and each of who reported that the patient s termi- ger over the death was fully expressed
the other grief indicators during each of nal illness diagnosis was within 6 months at 5 months postloss. After anger de-
those periods. Widowhood (compared compared with 6 months or longer prior clines, severity of depressive mood peaks
with loss of a parent, child, or sibling in to the death suggests that prognostic at approximately 6 months postloss and
this study group) was significantly as- awareness may promote acceptance of thereafter diminishes in intensity
sociated with within-person differ- the death. This result is consistent with through 24 months postloss. Accep-
ences in grief indicators 1 to 6 months findings reported elsewhere indicating tance increased steadily through the
postloss (Wilks =0.93, F4,135=2.51; that preparation for the death is associ- study observation period ending at 24
P=.05), due to its significant associa- ated with better psychological adjust- months postloss. Because of the minus-
tions with a greater difference between ment to the loss.23 Future research that cule probability that by chance alone
yearning and depression (P=.02) and a examines the effects of prospective rather these 5 grief indicators would achieve
lesser difference between acceptance and than retrospective reports of prognostic their respective maximum values in the
yearning (P=.01) during that period. Re- awareness on the bereaved survivor s ac- precise hypothesized sequence,14 these
port of a terminal illness diagnosis within ceptance are needed before definitive results provide at least partial support
6 months of the death was significantly conclusions can be drawn. for the stage theory of grief.
associated with grief indicators 12 to 24 Yearning was the most frequent nega- The results also offer a point of ref-
months postloss (Wilks = 0.93, tive psychological response reported erence for distinguishing between nor-
F5,172=2.62; P=.03), due to its signifi- throughout the study observation pe- mal and abnormal reactions to loss.
cant association with lower acceptance riod (1-6, 6-12, and 12-24 months post- Given that the negative grief indica-
of the death (P = .008) during that loss). Yearning was significantly more tors all peak within 6 months, those in-
period. common than depressed mood despite dividuals who experience any of the in-
the exclusive focus in the Diagnostic and dicators beyond 6 months postloss
COMMENT
Statistical Manual of Mental Disorders, would appear to deviate from the nor-
Results of this study identify normal Fourth Edition24 bereavement section on mal response to loss. These findings also
patterns of grief processing over time depressive symptomatology:  As part of support the duration criterion of 6
following the natural death of a loved their reaction to the loss, some griev- months postloss for diagnosing com-
one. Given that the vast majority ing individuals present with symptoms plicated grief disorder,16,17,19,25 or what
(94%) of deaths in the United States characteristic of a Major Depressive Epi- is now referred to as prolonged grief dis-
are the result of natural causes,15 the sode (e.g., feelings of sadness . . . The be- order.26 Unlike the term complicated,
findings reflect how the average per- reaved individual typically regards the which is defined as  difficult to ana-
son psychologically processes a typical depressed mood as  normal, . . . The di- lyze, understand, explain, 27 pro-
death of a close family member. agnosis of Major Depressive Disorder is longed grief disorder accurately de-
Although the temporal course of the generally not given unless the symp- scribes a bereavement-specific mental
absolute levels of the 5 grief indicators toms are still present 2 months after the disorder based on symptoms of grief
did not follow that proposed by the loss. 24(p684) Findings from this report that persist longer than is normally the
stage theory of grief,14 when rescaled demonstrate that yearning, not depres- case (ie, 6 months postloss based on
and examined for each indicator s sive mood, is the salient psychological the results of the present study). Fur-
peak, the data fit the hypothesized response to natural death. They indi- thermore, prolonged grief disorder per-
sequence exactly. cate that depressive mood in normally mits the recognition of other psychiat-
In terms of absolute frequency, and bereaved individuals tends to peak at ap- ric complications of bereavement, such
counter to the stage theory, disbelief was proximately 6 months postloss and does as major depressive disorder and post-
not the initial, dominant grief indica- not occur prior to 2 months postloss. traumatic stress disorder. Additional
tor. Acceptance was the most often en- Findings elsewhere25,26 indicate that analyses are needed to examine grief tra-
dorsed item. Evidently, a high degree of chronically elevated levels of yearning jectories among those meeting criteria
acceptance, even in the initial month are a cause for clinical concern. Taken for prolonged grief disorder.
postloss, is the norm in the case of natu- together, these results imply a need for The mode of death may be an im-
ral deaths. This contrasts with individu- revision of the Diagnostic and Statisti- portant factor that influences the course
©2007 American Medical Association. All rights reserved. (Reprinted) JAMA, February 21, 2007 Vol 297, No. 7 721
STAGE THEORY OF GRIEF
of bereavement adjustment. In the pre- markably similar patterns to those pre- months is therefore likely to reflect a
sent study, individuals bereaved by sented herein. We chose to present the more difficult than average adjust-
traumatic deaths (eg, vehicle crashes, items that fit most closely with the stage ment and suggests the need for fur-
suicide) were removed. Bereavement indicators illustrated in the literature.14 ther evaluation of the bereaved survi-
adjustment following deaths from trau- It should be noted that participants vor and potential referral for treatment.
matic causes may be more difficult to were younger and less likely to be male The results provide an evidence base
process and demonstrate higher de- compared with the study nonpartici- from which to educate clinicians (eg,
grees of disbelief and anger and lower pants, and that the study sample may primary and palliative care physi-
levels of acceptance than those re- be more resilient than is typically the cians, geriatricians, psychiatrists, on-
ported herein. A recent study found that case given the low prevalence of de- cologists, related hospital and hospice
those bereaved by traumatic vs natu- pression (8.9% of the individuals had staff, bereavement counselors) and lay-
ral deaths had greater difficulty in mak- a Hamilton Rating Scale for Depres- persons (eg, patients, family mem-
ing sense of the loss.28 Participants who sion summary score of 17) com- bers, friends) about what to expect fol-
reported that the family member or pared with other samples of bereaved lowing the death of a family member
loved one s terminal illness was diag- individuals.29-31 Samples with more or loved one.
nosed within 6 months of the death did males or with older and more dis-
Author Contributions: Drs Maciejewski and Prigerson
not differ significantly from other par- tressed individuals might reveal a dif- had full access to all of the data in the study and take
responsibility for the integrity of the data and the ac-
ticipants with respect to their level of ferent pattern of grief trajectories than
curacy of the data analysis.
grief indicators. However, the partici- those presented herein. Although the
Study concept and design: Maciejewski, Prigerson.
Acquisition of data: Prigerson.
pants who reported the diagnosis within study sample does show some gross
Analysis and interpretation of data: Maciejewski,
6 months of the death did report ac- similarities with the US widowed popu-
Zhang, Block, Prigerson.
Drafting of the manuscript: Maciejewski, Zhang,
ceptance of the death significantly less lation in terms of age, sex, and race/
Prigerson.
often. Subanalyses revealed that disbe- ethnicity, and with other comparable
Critical revision of the manuscript for important intel-
lief within 6 months postloss was also groups in terms of education and me- lectual content: Maciejewski, Zhang, Block, Prigerson.
Statistical analysis: Maciejewski, Zhang.
significantly higher in those for whom dian household income, it is not di-
Obtained funding: Maciejewski, Prigerson.
the patient s terminal illness diagnosis rectly representative of either the US
Administrative, technical, or material support: Zhang,
Prigerson.
was reported to be within 6 months widowed or US general population.
Study supervision: Maciejewski, Block, Prigerson.
prior to death. Thus, the manner and Nevertheless, age, income, race /
Financial Disclosures: None reported.
Funding/Support: This work was supported by grants
forewarning of the death appear to affect ethnicity, and sex were not signifi-
MH56529 (awarded to Dr Prigerson) and MH63892
the processing of grief. Studies are cantly associated with the magnitude
(awarded to Dr Prigerson) from the National Institute of
needed to explore the pattern of grief or course of grief and the representa- Mental Health and grant CA106370 (awarded to Dr
Prigerson) from the National Cancer Institute; and grant
trajectories among the survivors be- tiveness of the Yale Bereavement Study
NS044316 (awarded to Dr Maciejewski) from the Na-
reaved by traumatic causes of death. would not appear to restrict the gen- tionalInstituteofNeurologicalDisordersandStroke.Fund-
ingalsowasprovidedbytheCenterforPsycho-Oncology
The results should be understood in eralizability of the results to the US wid-
and Palliative Care Research, Dana-Farber Cancer Insti-
light of several study limitations. Ide- owed population. Despite these limi- tute, and Women s Health Research at Yale University.
Role of the Sponsors: The National Institute of Men-
ally, all individuals would have been as- tations, given that the Yale Bereavement
tal Health, National Cancer Institute, National Insti-
sessed immediately after the loss rather Study provides one of the most com-
tute of Neurological Disorders and Stroke, Center for
Psycho-Oncology and Palliative Care Research, Dana-
than beginning at month 1 postloss. Due prehensive longitudinal assessments of
Farber Cancer Institute, and Women s Health Re-
to respect for the initial mourning pe- grief, these data are as adequate as any
search at Yale University had no direct input into the
riod and institutional review board con- available for testing the stages of grief design or conduct of the study; collection, manage-
ment, analysis, or interpretation of the data; or prepa-
cerns about harm to participants, we did over time.
ration, review, or approval of the manuscript.
not interview individuals within a month In conclusion, the results of this
of the death. In addition, it would have study provide what appears to be the
REFERENCES
been better to analyze data that reas- first empirical examination of the stage
1. Bowlby J. Processes of mourning. Int J Psychoanal.
sessed individuals each month from 0 theory of grief. They indicate that in the
1961;42:317-339.
to 24 months postloss. However, no such circumstance of natural death, the nor- 2. Parkes C. Bereavement: Studies in Grief in Adult
Life. London, England: Tavistock; 1972.
data exist nor does the stage theory1-5 mal response involves primarily accep-
3. Bowlby J. Attachment and Loss. New York, NY: Ba-
specify in what month postloss each tance and yearning for the deceased. sic Books; 1980.
4. Parkes CM, Weiss RS. Recovery From Bereavement.
stage would predominate. And, al- Each grief indicator appears to peak in
New York, NY: Basic Books; 1983.
though we acknowledge that other grief the sequence proposed by the stage
5. Kübler-Ross E. On Death and Dying. New York,
NY: Macmillan; 1969.
indicators might have been used, the theory. Regardless of how the data are
6. Gray C, Koopman E, Hunt J. The emotional phases
various proxy measures (eg, stunned for analyzed, all of the negative grief indi-
of marital separation: an empirical investigation. Am
J Orthopsychiatry. 1991;61:138-143.
disbelief, bitterness for anger, hopeless- cators are in decline by approximately
7. Leibenluft E, Green SA, Giese AA. Mourning and
ness for depression, quality of life scores 6 months postloss. The persistence of
milieu: staff reaction to the death of an inpatient. Psy-
for acceptance/recovery) all revealed re- these negative emotions beyond 6 chiatr Hosp. 1988;19:169-173.
722 JAMA, February 21, 2007 Vol 297, No. 7 (Reprinted) ©2007 American Medical Association. All rights reserved.
STAGE THEORY OF GRIEF
8. Downe-Wamboldt B, Tamlyn D. An international 17. Zhang B, El-Jawahri A, Prigerson HG. Update on In: Diagnostic and Statistical Manual of Mental Dis-
survey of death education trends in faculties of nurs- bereavement research: evidence-based guidelines orders, Fourth Edition. Washington, DC: American Psy-
ing and medicine. Death Stud. 1997;21:177-188. for the diagnosis and treatment of complicated chiatric Association; 1994:684.
9. National Cancer Institute. Cancer topics: loss, grief bereavement. J Palliat Med. 2006;9:1188-1203. 25. Prigerson HG, Jacobs SC. Caring for bereaved pa-
and bereavement. http://www.nci.nih.gov 18. US Census Bureau. Population, housing, eco- tients:  all the doctors just suddenly go. JAMA. 2001;
/cancertopics/pdq/supportivecare/bereavement nomic, and geographic data. http://factfinder.census 286:1369-1376.
/Patient/page6. Accessed November 28, 2006. .gov/servlet/NPTable?_bm=y&-geo_id=01000US& 26. Prigerson HG, Vanderwerker LC, Maciejewski PK.
10. Bonanno GA, Wortman CB, Lehman DR, et al. -qr_name=ACS_2005_EST_G00_NP01&-ds_name= Prolonged grief disorder as a mental disorder: inclu-
Resilience to loss and chronic grief: a prospective study &-redoLog=false. Accessed November 28, 2006. sion in DSM. In: Stroebe MS, Hansson RO, Stroebe
from preloss to 18-months postloss. J Pers Soc Psychol. 19. Prigerson HG, Jacobs SC. Traumatic grief as a dis- W, Schut HA, eds. Handbook of Bereavement Re-
2002;83:1150-1164. tinct disorder: a rationale, consensus criteria, and a pre- search and Practice: 21st Century Perspectives. Wash-
11. Middleton W, Raphael B, Burnett P, Martinek N. liminary empirical test. In: Stroebe MS, Hansson RO, ington, DC: American Psychological Association Press;
A longitudinal study comparing bereavement phe- Stroebe W, Schut H, eds. Handbook of Bereavement 2007.
nomena in recently bereaved spouses, adult children Research: Consequences, Coping, and Care. Wash- 27. Dictionary.com Web site. Definitions for the word
and parents. Aust NZ J Psychiatry. 1998;32:235- ington, DC: American Psychological Association; 2001: complicated. http://dictionary.reference.com/browse
241. 588-613. /complicated. Accessibility verified January 18,
12. Ott CH, Lueger RJ. Patterns of change in mental 20. Chochinov HM, Wilson KG, Enns M.  Are you 2007.
health status during the first two years of spousal depressed? screening for depression in the termi- 28. Currier JM, Holland JM, Neimeyer RA. Sense-
bereavement. Death Stud. 2002;26:387-411. nally ill. Am J Psychiatry. 1997;154:674-676. making, grief, and the experience of violent loss: to-
13. Wortman CB, Silver RC. The myths of coping with 21. Horowitz M, Wilner N, Alvarez W. Impact of event ward a mediational model. Death Stud. 2006;30:403-
loss. J Consult Clin Psychol. 1989;57:349-357. scale: a measure of subjective stress. Psychosom Med. 428.
14. Jacobs S. Pathologic Grief: Maladaptation to Loss. 1979;41:209-218. 29. Clayton PJ, Halikas JA, Maurice WL. The depres-
Washington, DC: American Psychiatric Press; 1993. 22. Hamilton M. A rating scale for depression. J Neu- sion of widowhood. Br J Psychiatry. 1972;120:71-77.
15. Hoyert DL, Kung HC, Smith BL. Deaths: preliminary rol Neurosurg Psychiatry. 1960;23:56-62. 30. Zisook S, Shuchter SR. Depression through the first
data for 2003. Vital Health Stat 15. 2005;(53):16-18. 23. Barry LC, Kasl SV, Prigerson HG. Psychiatric dis- year after the death of a spouse. Am J Psychiatry. 1991;
16. Prigerson HG, Maciejewski PK. A call for sound orders among bereaved persons: the role of per- 148:1346-1352.
empirical testing and evaluation of criteria for com- ceived circumstances of death and preparedness for 31. Zisook S, Shuchter SR, Sledge PA, et al. The spec-
plicated grief proposed for DSM-V. Omega J Death death. Am J Geriatr Psychiatry. 2002;10:447-457. trum of depressive phenomena after spousal
Dying. 2005;52:9. 24. American Psychiatric Association. Bereavement. bereavement. J Clin Psychiatry. 1994;55(suppl):29-36.
I find that a great part of the information I have was
acquired by looking up something and finding some-
thing else along the way.
 Franklin P. Adams (1881-1960)
©2007 American Medical Association. All rights reserved. (Reprinted) JAMA, February 21, 2007 Vol 297, No. 7 723
LETTERS
Drafting of the manuscript: Croft, Darbyshire, van Thiel. 4. Ciminera P, Brundage J. Malaria in US military forces: a description of deploy-
Critical revision of the manuscript for important intellectual content: Croft, Dar- ment exposures from 2003 through 2005. Am J Trop Med Hyg. 2007;76:275-279.
byshire, Jackson, van Thiel. 5. Croft AM. Malaria: prevention in travelers. In: Tovey D, ed. Clinical Evidence.
Financial Disclosures: None reported. London, England: BMJ Publishing Group; 2005:954-972.
Funding/Support: No outside funding or support was received for this study. 6. Centers for Disease Control and Prevention. The yellow book, travelers health,
Disclaimer: The views, opinions, and/or findings contained in this study are those regional malaria information. http://www.cdc.gov/travel/regionalmalaria/indianrg
of the authors and should not be construed as the official North Atlantic Treaty .htm#malariarisk. Accessed February 26, 2007.
Organization or non-North Atlantic Treaty Organization military position, policy, 7. Lawrance CE, Croft AM. Do mosquito coils prevent malaria? a systematic re-
or decision, unless so designated by other official documentation. view of trials. J Travel Med. 2004;11:92-96.
Acknowledgment: We thank Ratimir Ben%0ńić, MD (ISAF Croatia), Roman Jantoa
,
MD (ISAF Slovakia), Serdor Kavak, MD (ISAF Turkey), Lopez Poves, MD (ISAF
Spain), and Carl Gustav Schultz, MD (ISAF Sweden), for their help in facilitating
this survey. They received no compensation for participation in this study.
CORRECTION
1. Funk-Baumann M. Geographic distribution of malaria at traveler destinations.
In: Schlagenhauf P, ed. Travelers Malaria. Hamilton, Ontario: BC Decker, Inc; 2001: Incorrect Example: In the Original Contribution entitled  An Empirical Examina-
56-93. tion of the Stage Theory of Grief published in the February 21, 2007, issue of
2. Malaria in military personnel returning from Afghanistan. Commun Dis Rep CDR JAMA (2007;297:716-723), an incorrect example was provided for natural causes.
Wkly. 2002;12:4-5. http://www.hpa.org.uk/cdr/archives/2002/cdr2702.pdf. Ac- On page 717, column 1, second full paragraph, the third sentence should be  Be-
cessed February 26, 2007. cause approximately 94% of US deaths result from natural causes (eg, heart dis-
3. Boecken GH. Pathogenesis and management of a late manifestation of vivax ease, cancer),15 deaths from unnatural causes (eg, car crashes, suicide) were ex-
malaria after deployment to Afghanistan: conclusions for NATO armed forces medi- cluded thereby enabling the results to be generalized to the most common types
cal services. Mil Med. 2005;170:488-491. of death.
Medicine requires not only the intellectual cultiva-
tion of a science, but the patience and the practical
skill of an art. At the bedside we must be animated
by the feeling of faithful artisans, of men whose ob-
ject and duty is practical work; for when the art of
medicine is needed by the suffering and the dying it
is no question of mere theoretical knowledge and ex-
traneous acquirement. But skill in the commonest art
it is not to be attained without much practice, far less
in the complicated and difficult art of healing, where
every case presents some peculiarities. To practice it
successfully, we must have made our home at the bed-
side, and, if I may say so, have lived with disease, ob-
serving it in all its forms and changes.
 Sir William Withey Gull (1816-1890)
2200 JAMA, May 23/30, 2007 Vol 297, No. 20 (Reprinted) ©2007 American Medical Association. All rights reserved.