Abstract
In the late 1990s, a panel of experts proposed criteria for complicated grief (CG) and found these to have satisfactory operating characteristics. The present study aimed to replicate that finding in 4 groups of mourners divided by cause of loss and time from loss. Data were available from 1,052 bereaved individuals. All had completed the revised Inventory of Complicated Grief through the Internet. Receiver operator characteristic analysis was used to analyze data. Results showed that, in all 4 groups, the proposed criteria for CG performed well at distinguishing between people who did and did not meet criteria for caseness of CG.
Notes
Note. Best threshold refers to level that provided optimal balance of sensitivity and specificity. Coding was 1 = never, 2 = seldom, 3 = sometimes, 4 = frequently, 5 = all the time. Sen = Sensitivity. Spe = Specificity. N/A, indicates no available data for this term. AUC = Area Under the Curve.
Sample 1 = victims of violent loss in the second half year of bereavement; Sample 2 = victims of violent loss > 1 yr removed from their loss; Sample 3 = victims of non-violent loss in the second half year of bereavement; Sample 4 = victims of non-violent loss > 1 yr removed from their loss.
∗p < 0.05. ∗∗p < 0.01. ∗∗∗p < 0.001.
1Examples of data that were deemed unreliable were data from respondents who filled in that they lost a child older than they were or a parent younger than they were. We looked at demographic variables age of respondent, sex, education, kinship to the deceased, age of deceased, cause of death and time from loss to judge if the same respondents possibly had filled in the questionnaires repeatedly. That is, successive sets of responses in which at least five of the seven key demographic variables had exactly the same values were considered to be possibly filled in by the same person and were excluded from the analyses.
2The mean total score on the ICG–R of the mourners scoring in the upper 20% of the summed score of the refined criteria was M = 105.2. This score was closely similar to the ICG–R total score of M = 103.9 of 22 mourners who met criteria for CG according to a structured clinical interview (the Traumatic Grief Evaluation of Response to Loss) in one of our previous studies (Boelen et al., Citation2003). That ICG–R scores (i.e., overall level of CG severity) of both groups were so similar, supports the appropriateness of defining “caseness” for CG as a score in the upper quintile of the summed refined criteria.