Defining Grief in Childhood: Test-Retest
Reliability of the W.T. Grant Consortium Grief
Linda N. Freeman, M.D., Prudence Fisher, M.S.W., Robert Abramovitz,
Evidence suggests that loss in childhood or adolescence can result in disruptive
and enduring symptoms after the death (Brown & Harris, 1978), but little
information is available about the grief process and grief recovery in
bereaved inner city children. Information about developmental and gender
differences in the grief process which lead to children's vulnerability
or resilience to subsequent problem behavior is needed to determine optimal
interventions with children suffering the sudden loss of a significant
other. Social work practitioners on the front line receiving the casualties
of the violence and AIDS epidemics have an important role to play in developing
understanding of the youthful grief process, intervening in pathological
grief, and prevention planning to reduce the risk of dangerous sequelae
to losses due to violence and AIDS. This study is an attempt to develop
one of the important tools necessary to study the phenomenon of childhood
and adolescent grief.
The study of child and adolescent bereavement has been hampered by the
lack of clinically useful standardized instruments to measure children's
grief reactions. After extensive search, the investigators have been able
to identify two grief measures that are specific for children and adolescents,
the Hogan Sibling Inventory of Bereavement (1990) and the W.T. Grant Consortium
Grief Inventory (1994). Some adolescent bereavement studies have used an
adult measure of grief, the Texas Inventory of Grief (Brent, 1993; Faschingbauer
et al., 1987), but this measure does not tap many of the domains of the
bereavement experience that are of clinical interest in younger children
and adolescents. (For brevity these will be referred to as the Hogan, Grant,
and Texas Inventories.) In addition, researchers who used the Texas Inventory
to study adult bereavement due to sudden unexpected loss have not found
it useful to discriminate those with pathological grief from those with
uncomplicated bereavement (E. K. Rynearson, personal communication, October
19, 1994). This limitation is attributable to the measure’s not being comprehensive
enough to identify intrusive recollections of the lost person, a symptom
that seems to differentiate the two groups.
The Hogan Inventory was designed for use with young adolescents who lost
a sibling. It has been used for young people whose siblings' deaths were
expected after a long illness. It has not been used in inner city child
populations or with children who suffer sudden and/or unexpected losses.
The measure has not been widely used.
The Grant Inventory (Clark et al.,1994) shows promise of being a comprehensive
bereavement measure that would be appropriate for different childhood and
adolescent study populations. It was developed in 1990 by a consortium
of researchers who were interested in loss in childhood including Madelyn
Gould, Ph.D., Robert Pynoos, M.D., David Clark, Ph.D., David Brent, M.D.
and David Balk, Ph.D. Members of the consortium each added items to the
scale that they thought represented important domains of grieving among
children and adolescents. A factor analysis was performed which identified
four factors. Six tapes were rated by an average of 8 raters per tape to
determine reliability. The factor analysis was performed on 48 response
sets of 142 questionnaire items each. Items which showed no or inadequate
variance were removed. A four-factor solution with factors of Avoidance,
Identification, Affective Compensation, and Anticipated Neediness was recommended
but no further reliability or validity studies have been done on this measure.
The significance of the current study lies in the major public health and
social work practice concern about children's exposure to untimely death
and the loss of significant others to violence and AIDS. Unfortunately,
these losses are becoming more common, particularly in inner city areas.
Because grief is considered to be a normal phenomenon, few social workers
become involved in understanding or intervening in the grief process. But,
children who have lost significant others due to violence can have responses
similar to those who are direct victims or witnesses to violence (Freeman,
et al., 1996). These symptoms can disrupt the child's functioning and influence
future development (Pynoos & Nader, 1990). For example, children can
suffer long-term consequences in their ability to regulate aggression and
can develop a propensity toward violence or victimization. In addition,
sudden unexpected loss is associated with the highest risk of persistent
grief (Pynoos & Nader, 1990). Reminiscing about the deceased, a necessary
part of grief resolution, can be aborted because the child cannot think
about the deceased without recalling mutilation or violence.
The purpose of this study is to examine the test-retest reliability of
the Grant Inventory and to compare it with the Texas Inventory (a measure
that is commonly used in adult and in some adolescent bereavement studies).
If the Grant Inventory is found to be reliable, it will be used to evaluate
the practice outcome of bereavement groups. Further, a clearer description
of grief in childhood, obtained by using this instrument, will contribute
to the definition of grief and the construct called complicated grief in
childhood. If the Grant Inventory is found to be unreliable, further modifications
will be made so that the authors can develop an acceptable instrument to
measure grief among children and adolescents. The correlation of the Grant
Inventory and the Texas Inventory would allow workers who use the latter
to make comparisons.
This study aims to demonstrate the test-retest reliability of the Grant
Inventory when used by raters examining inner city children and adolescents
and to determine its correlation with a standard bereavement scale used
in adult populations, the Texas Inventory. Study questions include: Is
the Grant Inventory reliable (correlation of 0.8 or above) when administered
twice over the course of one week? Can the Grant Inventory distinguish
between children and adolescents who are bereaved and who are not bereaved?
What is the correlation between the Grant Inventory and the Texas Inventory?
The study is being conducted by the New York State Psychiatric Institute
and the Center for the Study of Social Work Practice. Bereaved subjects
are 9-18 years old inner city boys and girls who have been bereaved by
the loss of a significant other within the two years prior to the interview.
The subjects fall within an age group for which the vocabulary in the Inventory
seems appropriate. The up-to-2 year interval between loss and interview
was selected because a loss occurring within 2 years is likely to be still
actively grieved but the subjects will be at different stages of grief.
This time span will thus increase variability in our responses. Bereaved
subjects are being recruited through the Jewish Board of Family and Children's
Services. Subjects who incurred losses due to any type of death will be
included to maximize variability in response. Announcements of the study
and subject inclusion criteria have been sent to administrators at JBFCS
sites. Consecutive referrals of any children who meet the criteria will
be accepted regardless of their treatment status, i.e., whether they are
referred at initial intake, or while they are in treatment for bereavement
related or other difficulties. For the purposes of the study, the definition
of significant others who have died will be limited to parents or primary
caretaking parent figures, siblings (full, half, or step siblings) or other
relatives with whom the subject has had at least biweekly contact over
the year prior to their death. Non-bereaved subjects are 9-18 year old
inner city boys and girls who have not suffered the loss of a significant
other as defined above. They are recruited from a Boys and Girls Club in
the Bronx, New York City. Twenty subjects between the ages of 9 and 18
years are being recruited. Ten of the subjects will be bereaved and ten
will be non-bereaved. After signing consent, each subject is given the
Grant Inventory by a research assistant. The assistant reads the 67 item
Inventory to the subject and the subject places the response on an answer
sheet. At the first interview only, the 23 item Texas Inventory is also
read to the subject and coded on the individual answer sheet. The assistant
audiotapes the session so that intra-rater reliability can be examined.
All the sessions are timed and are expected to last approximately one to
one-and-a-half hours. Before the subject leaves, the assistant reviews
the answer sheet for completeness. Any incomplete answers are re-asked
and completed. Seven days after the administration, the same subjects are
given the Grant Inventory by the same interviewer. The procedure outlined
above is followed. The non-bereaved subjects are asked the same questions
which appear in the inventories, but about a living significant other.
Questions that refer to the time period since a death are rephrased to
"in the past month, how often have you...".
Responses will be analyzed to determine test-retest reliability of the
Grant Inventory and correlation between the responses on the Grant Inventory
and the Texas Inventory. Sums of the factor scores Avoidance, Identification,
Affective Compensation, and Anticipated Neediness and a total score will
be computed. The test-retest paradigm will be a one-way random effects
analysis of variance which will estimate the components of variance due
to subject and to error variance. An inter-class coefficient of reliability
will be calculated from these components of variance (Bartko, 1966) for
the four subscale scores and for the total score. Correlations between
the two inventories will be determined with the Pearson correlation coefficient.
At this writing, data have been collected from ten non-bereaved subjects
and from five bereaved subjects. After data are collected from 5 more bereaved
subjects, the analysis will be performed to determine the reliability of
the measure. If it proves reliable, the three Co-Investigators who are
currently studying different aspects of bereavement in children and adolescents
(i.e., the grief of children who have lost significant others to homicide
- Freeman, suicide - Fisher, and AIDS - Abramovitz) will utilize it in
their respective studies of treatment outcome, identification of pathological
grief in childhood, and the description of grief recovery.
Bartko, J.J. (1966). The intraclass correlation coefficient as a measure
of reliability. Psychological Reports 19, 3-11.
Brent, D., Perper, J., Moritz, G., et al. (1993). Psychiatric sequelae
of the loss of an adolescent peer to suicide. American Journal of Child
and Adolescent Psychiatry, 32, 509-517.
Clark, D.C., Pynoos, R.S., & Goebel, A.E. (1994). Mechanisms and processes
of adolescent bereavement. In R.J. Haggerty, L.R. Sherrod, N. Garmezy,
& M. Rutter (Eds.), Stress, risk and resilence in children and adolescents:
Processes, mechanisms, and interventions (pp. 100-146). New York: Cambridge
Faschingbauer, T.R., Zisook, S., & DeVaul, R. (1989). The Texas revised
inventory of grief. In S. Zisook (Ed.), Biopsychosocial aspects of
bereavement. Washington, D.C. : American Psychiatric Association.
Freeman, L.N., Shaffer, D., & Smith, H. I. (1996). Neglected victims
of homicide: The needs of young siblings of murder victims. American
Journal of Orthopsychiatry, 66, 337-345.
Hogan, N. (1990). Hogan sibling inventory of bereavement. In J. Touliatos,
B. Permutter, & M. Strauss (Eds.), Handbook of Family Measurement
Techniques (p. 524). Newbury Park, CA: Sage Press.
Pynoos, R.S. (1990). Children’s exposure to violence and traumatic death.
Psychiatric Annals, 20, 334-344.