Complicated Grief (CG): DSM-V Diagnostic Criteria & Empirical Evaluation – Study Notes

Overview of the Paper

• Authors: Holly G. Prigerson, Ph.D. (Harvard) & Paul K. Maciejewski, Ph.D. (Yale)
• Publication: OMEGA – Journal of Death & Dying, Vol. 52(1), 2005-2006, pp. 9-19
• Purpose of article
  • Issue a call for rigorous, impartial, empirically-based testing of proposed DSM-V diagnostic criteria for Complicated Grief (CG).
  • Respond to and critique Hogan, Worden, & Schmidt (2003-2004) who attempted an empirical test of CG criteria but, according to the authors, asked the wrong research questions and drew questionable conclusions.
• Funding Acknowledgement: Multiple NIH grants (NIMH $MH56529, MH63892$; NCI $CA106370$) and foundations (Soros, RAND/Hartford, Fetzer).

Key Concepts & Definitions

• Complicated Grief (CG)
  • A chronic, debilitating bereavement reaction marked by persistent yearning/longing plus associated symptoms that cause significant dysfunction.
  • Distinguished from normal (integrated) grief by intensity, duration (≥ 6 months), and functional impairment.
• Separation Distress vs. Traumatic Distress
  • Conceptual sub-clusters originally used to organize symptoms.
  • Authors emphasize these are not separate latent dimensions; empirically CG is unidimensional.
• Diagnostic Psychometrics
  • Reliability, validity, sensitivity, specificity, diagnostic efficiency.
  • Crucial metrics when proposing inclusion in DSM-V.
• Cross-Validation
  • Replicating findings across independent samples to avoid criterion-set biases.

Proposed DSM-V Criteria for Complicated Grief (Table 1)

• Criterion A (Core Symptom)
  • Chronic, disruptive yearning / pining / longing for the deceased.
  • Sample probe: “Do you feel yourself yearning and longing for the person who is gone?”
• Criterion B (Associated Symptoms) – ≥ 4 of 8, present several times per day or intensely distressing/disruptive:
  1. Trouble accepting the death
  2. Inability to trust others
  3. Excessive bitterness/anger about the death
  4. Uneasiness about moving on (e.g., new friends/interests)
  5. Emotional numbness / detachment
  6. Sense that life is empty/meaningless without the deceased
  7. Bleak view of the future
  8. Feeling agitated / on edge since the death
• Criterion C – Marked, persistent social, occupational, or other functional impairment.
• Criterion D – Duration ≥ 6 months post-loss.
• Diagnosis – Must meet A + B + C + D.

Empirical Evidence Supporting CG as a Distinct, Unidimensional Syndrome

• Initial Discovery: 82 widowed older adults (Prigerson et al., 1995a)
  • CG symptoms formed a single factor distinct from depression/anxiety.
• Independent Replications (select list; diverse populations)
  • $n=97$ conjugally bereaved elders (Prigerson et al., 1995b)
  • $n=150$ community widowed adults (Prigerson et al., 1996)
  • $n=146$ adolescents (Melhem et al., 2004)
  • $n=76$ friends of a peer’s suicide (Prigerson et al., 1999a)
  • $n=151$ psychiatric out-patients, Karachi (Prigerson et al., 2002)
  • $n=103$ Dutch mental-health patients (Boelen et al., 2003a)
  • $n=250$ Dutch bereaved relatives (Boelen et al., 2003b)
  • $n=304$ women ≤ 6 mo post-miscarriage (Ritsher & Neugebauer, 2002)
  • $n=398$ Canadian psychiatric out-patients (Ogrodniczuk et al., 2003)
  • $n=232$ Norwegian parents (suicide/SIDS/accident) (Dyregrov et al., 2003)
• Item Response Theory (IRT) & Combinatorics used in DSM-V field trials to select final symptom set with invariance across age, sex, race/ethnicity, kinship.

Critical Appraisal of Hogan et al. (2003-2004)

1 | Misplaced Research Questions ("Red Herrings")

• Tested independence of separation vs traumatic distress; authors argue this is conceptually mistaken because CG is meant to be unidimensional.
• Tested whether CG is orthogonal to depression & “normal” grief; the important clinical question is functional impairment & morbidity prediction, not statistical independence.

2 | Factor-Structure Findings Actually Support Unidimensionality

• Hogan et al. report correlation $r=.96$ between separation & traumatic distress factors; CFI $= .89$ for single-factor model.
  • Authors note combining anger + bitterness items or allowing correlated residuals would likely raise CFI > .90 (threshold for “good fit”).

3 | CG vs. Major Depressive Disorder (MDD)

• Hogan et al. correlations with Beck Depression Inventory (BDI):
  • Shared variance Separation Distress ↔ BDI = $46.5\%$
  • Shared variance Traumatic Distress ↔ BDI = $53.1\%$
• Interpretation: ≈ 50 % of CG variance is not explained by depression → MDD alone would miss half of CG cases.
• Comorbidity ≠ Redundancy: analogous to high overlap of anxiety & depression but separate DSM entities.
• Structural Equation Model (Hogan) yielded CFI $=.94$ yet authors claimed “poor fit” – contradicts SEM conventions (Byrne, 2001).

4 | CG vs. Normal Grief

• Pathology determined by severity + duration + impairment (aligned with DSM-IV definition of mental disorder).
• CG symptoms at high intensity & > 6 months predict:
  • Psychiatric morbidity (depression, suicidality)
  • Physical morbidity (elevated blood pressure)
  • Maladaptive health behaviors (↑ smoking/alcohol, poor sleep)
  • Quality of life impairments.

5 | “Stigmatization” Argument Rebutted

• Survey (DSM-V field trial, $n=135$, mean 19 mo post-loss):
  • $98.5\%$ would welcome help if diagnosed.
  • $96.3\%$ would feel relieved & better understood with CG label.
• Potential benefits: focused treatments, insurance reimbursement, employer accommodations, destigmatization through recognition.

Distinctive Features of CG (vs MDD, GAD, PTSD)

• Unique Symptomatology (e.g., persistent yearning, disbelief, difficulty moving on, detachment specific to attachment loss).
• Distinct Risk Factors
  • Security-enhancing relationship with deceased (van Doorn et al., 1998).
  • Absence of depression biomarker (normal REM latency) (McDermott et al., 1997).
  • Preference for high lifestyle regularity (Beery et al., 1997).
• Course & Outcome
  • CG predicts suicidality (Prigerson et al., 1999a) and somatic disease independent of depression.
• Treatment Response
  • Interpersonal Psychotherapy & TCAs help bereavement-related depression but not CG (Reynolds et al., 1999).
  • Necessitates CG-specific interventions.

Methodological Principles for Future Research

• Cross-Validation across cultures, age groups, relationship types.
• Metrics to Examine:
  • Sensitivity, specificity, positive predictive value, negative predictive value.
  • Longitudinal predictive validity for morbidity & functional outcomes.
• Avoid “Red Herrings”: focus on clinical utility rather than purely statistical orthogonality.

Authors’ Conclusions & Call to Action

• Existing evidence supports CG as a valid, reliable, unidimensional, and clinically significant disorder distinct from depression/anxiety/PTSD.
• Nonetheless, continued empirical work is vital to refine criteria, optimize cut-points, and evaluate diagnostic efficiency.
• Researchers should prioritize:
  1. Identifying bereaved individuals at greatest risk of chronic impairment.
  2. Testing diagnostic algorithms in diverse populations.
  3. Developing & validating CG-specific treatments.

Selected Numerical / Statistical Highlights

• Correlation Separation vs Traumatic Distress: $r = .96$
• Goodness-of-fit threshold (SEM): CFI > .90 regarded as acceptable.
• Hogan et al. single-factor model: CFI ≈ .89–.94 (borderline to good).
• Comorbidity statistic example: Phi $\phi \le .50$ between CG & MDD (multiple studies).
• Field-trial acceptability: > 95\% bereaved respondents welcome CG diagnosis/help.

Study & Exam Tips

• Memorize Criterion A (Yearning) + B’s 8 associated symptoms; remember “4 of 8”.
• Understand why CG is unidimensional → know the $r=.96$ figure.
• Be able to articulate half-variance argument: $\approx 50\%$ of CG unexplained by depression.
• Differentiate CG from MDD, GAD, PTSD via hallmark symptoms & risk factors.
• Cite at least 3 replication studies (e.g., adolescents, Dutch samples, Karachi clinic) to demonstrate cross-validation.
• Recall the DSM-IV definition of mental disorder (distress, disability, risk) and apply it to justify CG as a diagnosis.
• Recognize common misconceptions (stigmatization, “normal” vs CG, dimensional debates).