Prevalence rates of prolonged grief disorder are overestimated

ABSTRACT

Prolonged grief disorder (PGD) has been included in the International Classification of Diseases 11th edition (ICD-11) and the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR). Accurate assessment of the prevalence of PGD is imperative to healthcare policy and practice. Systematic reviews and meta-analyses of at-risk bereaved subgroups (e.g. those bereaved by natural disasters, unnatural causes, and COVID-19) yield very high prevalence rates (24–87%). These findings appear at odds with the well-established finding that most people respond to major negative life events resiliently. This letter explores two potential explanations for overestimated prevalence rates of PGD. First, non-probability sampling, typical to research in vulnerable bereaved subgroups, leads to overestimation of prevalence rates of PGD relative to probability sampling. Secondly, based on epidemiological research in depression and post-traumatic stress disorder, we consider the risk that the use of self-report scales, typical to grief research, may lead to overestimation of prevalence rates compared to clinical interviewing. We recommend international research, based on probability sampling, that systematically assesses loss characteristics and circumstances as well as ICD-11 and DSM-5-TR PGD symptoms, using validated scales, and, ideally, clinical interviewing in a subset of participants, to enable more accurate assessment of prevalence rates of PGD in different bereaved populations.

HIGHLIGHTS

• Systematic reviews and meta-analyses of at-risk bereaved samples report exceptionally high prevalence rates of prolonged grief.

• These differences can partly be explained by overreliance on voluntary response samples and self-report scales in research on at-risk bereaved samples.

• International research is needed, using probability sampling, self-report, and interview assessment of ICD-11 and DSM-TR prolonged grief symptoms and loss-related characteristics and circumstances.

A minority of bereaved individuals experiences severe, persistent, and disabling grief, termed prolonged grief. In 2019, prolonged grief disorder (PGD) was included to the International Classification of Diseases, 11th edition (ICD-11); World Health Organization, 2019) and 4 years later a similar, but distinct diagnosis with the same name was included in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR) (American Psychiatric Association, 2022; Eisma et al., 2022). Establishing the prevalence of PGD is an important scientific goal as it may help clarify how many people experience this condition in the general population, and whether, and to what extent, specific groups of bereaved people may be at risk for PGD, which can help guide mental healthcare policy and practice (Stroebe et al., 2024).

Accordingly, research in this area has been increasing; a search in Scopus showed that the combination of ‘prolonged grief’ OR ‘complicated grief’ AND ‘prevalence’ OR ‘rate’ in an article title occurred approximately four times more often from 2019 to 2024 than from 2013 to 2019 (38 vs nine times, respectively). Recent systematic reviews and meta-analyses of prevalence rates of prolonged grief in at-risk populations of bereaved people report very high estimates. For example, elevated prevalence rates of prolonged grief are reported for people who experienced deaths due to unnatural causes (33–65%) (Djelantik et al., 2020), natural disasters (24–53%) (Zareiyan et al., 2024), and coronavirus disease 2019 (COVID-19) (30–87%) (Reitsma et al., 2025). This suggests that experiencing particular types of death greatly enhances the risk of PGD. While there is substantive evidence for the notion that unexpected deaths and violent/unnatural deaths are associated with a higher risk of prolonged grief (for a review, see Buur et al., 2024), reported prevalence rates are probably too high. They are, for instance, at odds with a large body of trajectory research showing that, following major stressful life events, a minority experiences chronically high distress, while a majority shows resilient responses (Galatzer-Levy et al., 2018). This finding is mirrored in the only two published trajectory studies among people who experienced sudden, violent deaths: a minority experiences chronic, high levels of prolonged grief symptoms (18% in the study by Lenferink et al., 2020; 11% in Sveen et al., 2018), but a majority recovers or is resilient.

There are multiple reasons why the true prevalence of PGD is likely to be significantly lower than the reported estimates in systematic reviews and meta-analyses. First, most studies that aim to establish prevalence rates of prolonged grief use non-probability sampling methods, such as voluntary response sampling (Comtesse et al., 2024). This is problematic, because people who have been more severely affected by the death of a loved one appear to be more likely to voluntarily participate in bereavement research. A systematic review and meta-analysis illustrated that ICD-11 and DSM-5-TR PGD prevalence rates in the general bereaved population were 5% in probability samples versus 16% in non-probability samples (Comtesse et al., 2024). Since voluntary response sampling is particularly likely to be used when investigating difficult-to-recruit at-risk populations of bereaved people who have experienced types of death that are uncommon, reported prevalence rates in such samples typically overestimate the true prevalence of prolonged grief. Unfortunately, detailed information about loss characteristics and circumstances is often lacking in the few studies that draw probability samples of bereaved people (e.g. Ben-Ezra et al., 2020; Shevlin et al., 2023; for notable exceptions, see e.g. Doering et al., 2022; Treml et al., 2022). This further limits our ability to draw valid conclusions about the prevalence rates of these conditions in at-risk bereaved populations.

Secondly, prevalence estimates of PGD in reviews may be elevated because they summarize studies relying almost exclusively on self-report questionnaires (e.g. Reitsma et al., 2025). Diagnostic handbooks specify that establishing diagnoses requires the involvement of a clinician. For example, the DSM-5-TR (American Psychiatric Association, 2022) holds that ‘Clinical training and experience are needed to use DSM for determining a clinical diagnosis’ (p. 94) and ‘Diagnostic criteria are offered as guidelines for making diagnoses, and their use should be informed by clinical judgment’ (p. 116). The recommended way to establish a diagnosis is by clinical interviewing, i.e. a personal exchange between clinician and client designed to systematically gather information about social, medical, education, familial, psychological, and developmental characteristics and circumstances from the client and, sometimes, from their close others (e.g. a parent). Self-report assessment forms an important source of information in establishing diagnoses, but cannot form the only basis for a diagnosis (Stroebe et al., 2024).

Research on a related condition, major depression, has shown that self-report scales markedly overestimate prevalence rates compared to clinical interviews (Thombs et al., 2018). Systematic reviews and meta-analyses demonstrate that prevalence rates of depression derived from screening scales can be three times higher than rates derived from clinical interviews (e.g. Dawes et al., 2016). Similar overestimation through survey research relative to clinical interviewing is observed in research on the prevalence of post-traumatic stress disorder (PTSD) (for a brief discussion, see Hyland & Shevlin, 2024). It seems likely that comparable differences will eventually be found between self-report scales and newly developed clinical interviews for PGD, such as the Traumatic Grief Inventory – Clinician Administered (TGI-CA) (Lenferink et al., 2023) and the International Interview for Prolonged Grief Disorder according to ICD-11 (I-PGD-11) (Rueger et al., 2024). However, in the only empirical study on the topic to date, rates of ICD-11 PGD were higher (not lower) when using the I-PGD-11 than when using a self-report scale in a sample of 101 German psychiatric inpatients (Rueger et al., 2024). If, and to what extent, such results generalize when using a more elaborate clinical interviewing procedure and to the general bereaved population remains to be established. An additional concern in this context is that prevalence estimates of DSM-5-TR PGD (American Psychiatric Association, 2022) will need to account for the fact that symptoms should not be better explained by conditions with which it often co-occurs and with which it shows some content overlap, e.g. depression and PTSD (Komischke-Konnerup et al., 2021).

A key consideration in the discussion on the use of different instruments to assess PGD is that the ‘gold standard’ to determine diagnoses, clinical interviewing, has not gone without criticism (Hyland & Shevlin, 2024). For example, it has been argued that there is no way of knowing the ‘true’ prevalence of a mental health condition as it is not directly observable. Moreover, clinical interviews may be subject to similar biases to self-reporting; and in clinical interviews there may be two sources of potential bias (client and clinician) while in self-reporting there is only one source of potential bias (Hyland & Shevlin, 2024). Moreover, administering a clinical interview requires more time and resources than administering a self-report questionnaire. Triangulation of research using different methods, rather than exclusively relying on self-report or clinical interviews, may therefore be the most viable way to reduce bias in prevalence estimates.

In summary, reported rates of prolonged grief derived from surveys, and summarized in systematic reviews and meta-analyses, are likely to be too high, particularly in at-risk populations of bereaved people. This effect may be strongest in research most typical to this field, i.e. research that relies on voluntary response sampling and self-report scales. There is a need for international research, based on robust probability sampling, that systematically assesses loss characteristics and circumstances as well as ICD-11 and DSM-5-TR PGD symptoms, using validated scales, and, ideally, clinical interviewing in a subset of participants, for systematic comparisons of prevalence rate estimates. Such research will help to establish which bereaved people are most at risk of experiencing PGD, and to what extent, which can guide efforts by policymakers, researchers, and clinicians to provide appropriate care for specific groups of bereaved people.

Funding Statement

This work was supported by Nederlandse Organisatie voor Wetenschappelijk Onderzoek [grant number 016.veni195.113].

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

Data sharing is not applicable to this article as no new data were created or analysed in this study.