Content overlap analyses of ICD-11 and DSM-5 prolonged grief disorder and prior criteria-sets

ABSTRACT

Background

The International Classification of Diseases eleventh edition (ICD-11) has recently included prolonged grief disorder (PGD), a diagnosis characterized by severe, persistent, and disabling grief. The text revision of the Diagnostic and Statistical Manual of Mental Disorders 5 (DSM-5-TR) is scheduled to include a similar but distinct diagnosis, also termed PGD. Concerns have been raised that these new diagnoses are qualitatively different from both prior proposed diagnoses for pathological grief and each other, which may affect the generalizability of findings obtained with different criteria sets.

Objective

We conducted a content overlap analysis of PGD_ICD-11, PGD_DSM-5-TR, and previous proposals for pathological grief diagnoses (i.e. PGD 2009; complicated grief (CG), PGD ICD-11 beta draft, persistent complex bereavement disorder (PCBD) per DSM-5).

Methods

Using the Jaccard’s Index, we established the degree of content overlap between core and accessory symptoms of PGD_ICD-11, PGD_DSM-5-TR, and prior proposals for pathological grief diagnoses.

Results

Main findings are that PGD_ICD-11 and PGD_DSM-5-TR showed moderate content overlap with each other and with most prior proposed diagnoses for pathological grief. PGD_ICD-11 and PGD_DSM-5-TR showed the strongest content overlap with their direct predecessors, PGD_{ICD-11 beta draft} and PCBD, respectively.

Conclusions

Limited content overlap between PGD_ICD-11 and PGD_DSM-5-TR and preceding criteria sets may threaten generalizability of past research on phenomenological characteristics of pathological grief to current criteria sets. Similarly, findings obtained with instruments to assess PGD_ICD-11 may not generalize to PGD_DSM-5-TR and vice versa. Researchers should aim to determine under which circumstances criteria sets for PGD yield similar or distinct characteristics. Convergence of criteria sets for PGD remains an important goal for the future.

HIGHLIGHTS

• Content overlap analyses showed moderate overlap between symptoms of PGD per DSM-5-TR and ICD-11 and between these diagnoses and prior criteria sets.

• We should establish when new criteria sets for PGD behave similarly or differently.

• Convergence of PGD criteria sets is needed.

Dear Editor,

Over the past decades, there have been multiple attempts to define a diagnosis characterized by severe, persistent, and disabling grief, i.e. pathological grief. These proposed diagnoses have received different names, including complicated grief (CG: Shear et al., 2011), persistent complex bereavement disorder (PCBD: American Psychiatric Association [APA], 2013) and, most commonly, prolonged grief disorder (PGD, e.g. Prigerson et al., 2009, PGD₂₀₀₉; Maercker et al., 2013; PGD_{ICD-11 beta draft}). In 2018, a diagnosis termed PGD was formally added to the International Classification of Diseases eleventh edition (PGD_ICD-11, ICD-11: World Health Organization [WHO], 2018). A different diagnosis named PGD will be added to the text revision of the Diagnostic and Statistical Manual of Mental Disorders 5 in 2022 (PGD_DSM-5-TR, DSM-5-TR: Boelen, Eisma, Smid, & Lenferink, 2020; Prigerson, Kakarala, Gang, & Maciejewski, 2021). Figure 1 displays core and accessory symptoms of all mentioned criteria-sets.

Figure 1.

Similarities and differences between five diagnostic criteria sets of pathological grief.

A concern regarding the development of new criteria-sets for pathological grief is that they are, as a rule, qualitatively different from preceding criteria-sets (e.g. Boelen & Prigerson, 2012; Djelantik et al., 2021; Eisma & Lenferink, 2017; Stelzer, Zhou, Maercker, O’Connor, & Killikelly, 2020). Criteria-sets differ in number of included symptoms, symptom content, and diagnostic algorithms (Eisma, Rosner, & Comtesse, 2020; Lenferink, Boelen, Smid, & Paap, 2021). Consequently, the phenomenological characteristics of different pathological grief criteria-sets vary. For example, PGD_ICD-11 has limited diagnostic agreement with prior proposed criteria-sets, such as PCBD and PGD₂₀₀₉ (e.g. Boelen & Lenferink, 2020; Boelen, Lenferink, Nickerson, & Smid, 2018; Comtesse et al., 2020; Cozza et al., 2020), although the extent of agreement partially depends on the chosen diagnostic algorithm (Eisma et al., 2020). Therefore, previous findings on important clinical issues, ranging from dimensionality of diagnoses to treatment efficacy, may not generalize to newer criteria-sets. Additionally, since PGD_ICD-11 and PGD_DSM-5-TR also differ in symptom count, content, and diagnostic algorithms, findings obtained with one version of PGD may not generalize to the other.

Therefore, clarifying to what extent criteria-sets capture the same content and whether and when criteria-sets of pathological grief yield similar or different results appears important. The aim of the present contribution is to assess the comparability of different criteria-sets using a mathematical approach. Specifically, we will establish the extent to which the content of core and accessory symptoms in PGD_ICD-11, PGD_DSM-5-TR, and preceding criteria-sets overlap. We derive these methods from Fried (2017), who used a similar approach to illustrate the limited content overlap between items from seven frequently used self-report measures of depression.

We estimated content overlap between criteria-sets using the Jaccard Index, a similarity coefficient for binary data ranging from 0 (no overlap among criteria-sets) to 1 (complete overlap). It is calculated with the following formula: J =s/(u1+u2+s), where J is the Jaccard Index, s is the number of items that two criteria-sets share, and u1 and u2 are the number of symptoms unique to each criteria set. Since there is no established guideline on categorizing the strength of overlap using the Jaccard Index, we will apply the rule by Evans (1996) for the correlation coefficient as an indicator: very weak 0.00–0.19, weak 0.20–0.39, moderate 0.40–0.59, strong 0.60–0.79, and very strong 0.80–1.0 (Fried, 2017).

Supplemental Table S1 shows the results. A first main finding is that there is moderate overlap between the most recent criteria-sets PGD_ICD-11 and PGD_DSM-5-TR (J =0.47). PGD_ICD-11 shows the strongest overlap with the PGD_{ICD-11 beta draft} (J =0.58), whereas PGD_DSM-5-TR shows the strongest overlap with PCBD (J = 0.63), illustrating that they most closely resemble their direct predecessors. Both PGD_ICD-11 and PGD_DSM-5-TR show least overlap with CG (J =0.22 and 0.37, respectively). Overall, the mean overlap between PGD_ICD-11 and PGD_DSM-5-TR with all other criteria-sets is moderate (J =0.41 and 0.48, respectively). CG stands out as the diagnosis showing the least overlap with all other criteria-sets (J =0.35), whereas PCBD shows most content overlap with other criteria-sets (J =0.49).

Overall, our analysis demonstrated modest content overlap between prior proposed criteria-sets and both PGD_ICD-11 and PGD_DSM-5-TR. Moreover, the two newest criteria-sets showed limited content overlap with each other. These findings complement prior empirical research demonstrating differences between the characteristics of different criteria-sets for pathological grief (e.g. Boelen et al., 2020; Boelen & Lenferink, 2020; Comtesse et al., 2020; Cozza et al., 2020). Using a single validated instrument, such as the recently developed Traumatic Grief Inventory–Self Report Plus, to assess symptoms of different criteria-sets, may be instrumental to further elucidate when criteria-sets behave similarly or differently (Lenferink, Eisma, Smid, de Keijser, & Boelen, Lenferink, et al., 2022).

Together, results suggest that limited content overlap could partly explain differences in findings across different criteria-sets. Two courses of action may help reduce this problem of generalizability in the future. First, we should strive for greater convergence of future diagnostic criteria-sets with presently used criteria-sets (Lenferink et al., 2021). Second, since PGD_ICD-11 uses a descriptive diagnosis without a formal diagnostic algorithm, we could investigate which PGD_ICD-11 algorithm yields to the greatest convergence with past criteria-sets, and, more importantly, with PGD_DSM-5-TR criteria (Eisma et al., 2020).

Some limitations warrant mention. First, this work is a mathematical exercise that complements but does not substitute empirical studies of similarities and differences between pathological grief criteria-sets. Second, we only compared core and accessory symptoms of criteria-sets. For example, we did not take into account differences in time criteria or diagnostic algorithms between proposed diagnoses. A third limitation is that for CG we split up some compound symptoms (e.g. ‘Frequent intense feeling of loneliness or like life is empty or meaningless without the person who died’ was separated into ‘loneliness’ and ‘feeling life is empty/meaningless’) because these symptoms were also separated in other criteria-sets (see Lenferink et al., 2021 for details). This may have led us to overestimate content overlap between CG and other criteria-sets. Fourth, one grief researcher assessed overlap between criteria sets (cf. Lenferink et al., 2021). Multiple assessors may have yielded more reliable and replicable classifications of symptoms.

Notwithstanding these limitations, our analyses have demonstrated suboptimal comparability in the content of past and current pathological grief criteria-sets. We have highlighted how this may result in problems of generalizability of findings obtained with past and current criteria-sets. This work illustrates the need for further convergence of diagnoses and empirical investigations of the similarities and differences between pathological grief diagnoses and related phenomenological characteristics.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Supplementary material

Supplemental data for this article can be accessed here.