Prevalence of prolonged grief disorder and its symptoms among bereaved individuals in China: a systematic review and meta-analysis

Literature retrieval

We conducted a literature search across major Chinese and English bibliographic databases from their inception until 4 October 2023. The databases searched included China National Knowledge Infrastructure, Wanfang Data, SinoMed, VIP Information, PubMed, Embase and PsycInfo. The search terms used were ‘(grief*) AND (bereave* OR mourn* OR sorrow OR surviv* OR Shidu OR los* OR death OR die* OR suicid*) AND (Chin* OR Taiwan OR Hong Kong OR Macau)’. Additionally, we manually searched the reference lists of related literature to ensure that no potential studies were overlooked. Details of the literature search strategies are provided in online supplemental material appendix 2.

Inclusion and exclusion criteria

The inclusion criteria for the present study were as follows: (a) cross-sectional surveys or baseline data of cohort and interventional studies that reported the prevalence of PGD and/or its symptoms; (b) participants must be Chinese individuals who have experienced bereavement due to various reasons, including suicide, earthquake and illness; (c) standardised tools, such as the 13-item Prolonged Grief Disorder and the Inventory of Complicated Grief, should be used to assess the presence of PGD and its symptoms. PGD is a mental disorder characterised by an intense and prolonged grieving process that significantly impairs functioning and well-being. Diagnosis can be made using existing diagnostic criteria like ICD-11 and DSM-5 or criteria proposed by researchers, such as the Prigerson et al’s criteria and the Maciejewski et al’s criteria. Symptoms of PGD typically include intense yearning and longing for the deceased, emotional distress, difficulty accepting the death and an inability to move forward in life. These symptoms should be assessed using validated scales, such as the Inventory of Complicated Grief. If multiple studies using the same data were available, only the study with the most complete data was included. Conference papers, reviews and case reports were excluded as well as studies focusing on anticipatory grief or bereavement, resulting from the ending of a significant relationship, abortion, stillbirth or the loss of a pet.

Data extraction

Information extraction was performed using a predesigned spreadsheet, which captured the following data: (a) basic study details, such as first author, year of publication, study site and survey date; (b) study characteristics, including sampling method, sample size, mean age of participants and outcome assessment; (c) data related to the bereaved individuals, such as PRs or symptom severity scores based on sex and marital status; (d) data related to the deceased individuals, such as PRs or symptom severity scores based on the cause of death; (e) outcome data such as PRs of PGD and its symptoms.

RoB assessment of included studies

The RoB of the included studies was assessed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Studies Reporting Prevalence Data (hereafter abbreviated as the ‘JBI checklist’).25 Nine items covering four domains of RoB in a cross-sectional study were evaluated in the JBI checklist: sampling methods, study population, data collection, and analysis methods.26 The assessment was carried out by authors who had undergone extensive training by experienced epidemiologists and used a four-choice response scale (yes, no, unclear and not applicable). A score of one was assigned to an ‘yes’ response, while a score of zero was given for all other responses. A higher total score on the JBI checklist indicates a lower RoB.

Certainty of evidence

We used the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) method to determine the certainty of evidence.27 The GRADE method examines five domains: RoB, imprecision, inconsistency, indirectness and publication bias.28 By applying the rules of downgrading and upgrading within the GRADE method, we were able to rate the certainty levels of evidence for the pooled prevalence of PGD and its symptoms separately. The rating levels used were high, moderate, low or very low.

Statistical analysis

Meta-analysis of prevalence data was conducted using the ‘metaprop’ package in R V.4.1.2. This analysis employed one-step generalised linear mixed models with the Logit link function, as recommended by Lin and Chu.29 Heterogeneity among studies was evaluated using the I² statistic. When I² was less than 50%, a fixed-effects model was used to obtain combined estimates. If I² was 50% or greater, a random-effects model was employed. Subgroup analysis was carried out to identify potential sources of heterogeneity in the prevalence estimates. The Q-value test was employed to assess the significance of differences in estimated PRs across various subgroups. These subgroups were defined based on the study site, type of bereavement, average age, proportion of men in the study sample, survey method and the assessment of PGD or its symptoms. Publication bias was evaluated using funnel plots and Egger’s test. A two-sided p value of less than 0.05 was considered statistically significant.

As some of the included studies provided and directly compared subgroup-specific data—for instance, comparing PRs of PGD between males and females, and comparing time since loss (expressed as mean and SD) between grieving and non-grieving bereaved individuals—we performed a meta-analysis of these comparative data. This allowed us to better characterise individuals with PGD or its symptoms. For example, if the meta-analysis identified significantly higher PGD prevalence in females than in males, it could indicate that bereaved women are more likely to experience PGD than bereaved men. We conducted meta-analyses of both dichotomous and continuous outcomes using the ‘metabin’ and ‘metacont’ packages in R software V.4.1.2. Prevalence ratios and standardised mean differences (SMDs) were used as corresponding effect size measures.30

Characteristics of the included studies

The flowchart of study inclusion is shown in figure 1. Twenty-eight studies involving 10 994 bereaved persons were included: 14 focused solely on PGD,8 17 19 20 31–40 12 focused solely on PGD symptoms,15 16 41–50 and two on both PGD and its symptoms.18 51 Detailed characteristics of the included studies are displayed in online supplemental material appendix 3.

Request permission

RoB of the included studies

The JBI checklist scores of the included studies ranged from 3 to 7, with a median score of 6. The two most common methodological problems among the included studies were a problematic sample frame (n=23) and an inappropriate sampling method (n=23).

Meta-analysis of the prevalence of PGD and its symptoms

The combined PRs of PGD and its symptoms were 8.9% (95% confidence interval (CI): 4.2% to 17.6%) and 32.4% (95% CI: 18.2% to 50.8%), respectively (figure 2).

Request permission

In comparison to men, women exhibited significantly higher PRs of both PGD (12.6% vs 8.4%, PR: 1.84, p<0.001) and its symptoms (29.6% vs 20.1%, PR: 1.69, p<0.001) (table 1). Additionally, women had higher PGD symptom scores compared with men (SMD: 0.24, p=0.031). Individuals diagnosed with PGD had a significantly shorter duration since the loss compared with those without PGD (SMD: −0.39, p<0.001). Moreover, individuals with religious beliefs had statistically higher PGD symptom scores than those without (SMD: 0.25, p=0.016), while people who lost loved ones due to violent causes scored higher on PGD symptom scales compared with those whose deceased loved ones died due to illnesses (SMD: 0.45, p=0.010) (online supplemental appendix 4).

Publication bias among the included studies

As depicted in online supplemental material appendix 5, the funnel plot for the meta-analysis of PGD was visually asymmetrical, while the funnel plot for the meta-analysis of PGD symptoms was visually symmetrical. Nonetheless, the results of Egger’s test indicated no statistically significant publication bias across the 16 studies regarding PGD (z=1.50, p=0.067) or the 14 studies regarding PGD symptoms (z=0.50, p=0.308).

Sources of heterogeneity in the meta-analysis of the prevalence of PGD and its symptoms

As displayed in table 2 and table 3, study site, type of bereavement and diagnostic criteria for PGD were significant sources of heterogeneity in the meta-analysis of the prevalence of PGD, while study site, type of bereavement and mean age of the study sample were significant sources of heterogeneity in the meta-analysis of the prevalence of PGD symptoms. Specifically, surveys conducted solely online (18.1%) and offline (15.2%) had a significantly higher PGD prevalence than those conducted both online and offline (1.0%). Among bereaved participants, those who had deceased loved ones as their only children had the highest prevalence of PGD (22.7%), followed by those who experienced loss due to COVID-19 (16.8%) and earthquakes (8.5%). Bereaved individuals who experienced loss due to illnesses had the lowest prevalence of PGD (1.7%). In addition, surveys using the ICD-11 diagnostic criteria for PGD yielded a significantly higher PGD prevalence than those using the Prigerson et al’s 52 criteria (22.9% vs 5.8%, p=0.010).

Similarly, surveys conducted solely online (22.0%) and offline (50.9%) had a significantly higher prevalence of PGD symptoms than those conducted both online and offline (3.8%). The highest prevalence of PGD symptoms was observed among bereaved participants whose loved ones died due to earthquakes (80.4%), followed by those who experienced loss due to AIDS (35.1%) and COVID-19 (30.7%). Additionally, individuals whose deceased loved ones were the only children (28.9%) and those whose loved ones died of illnesses (13.6%) also exhibited notable rates of PGD symptoms. Studies predominantly involving middle-aged participants (49.7%) and older adults (23.5%) demonstrated significantly higher PRs of PGD symptoms compared with studies predominantly involving young adult participants (6.9%).

Certainty of the evidence for the pooled prevalence of PGD and its symptoms

All our included studies were cross-sectional studies, so we initially classified the certainty levels for both the pooled prevalence of PGD and its symptoms as ‘low’. However, given the considerable RoB, wide 95% CIs, high levels of heterogeneity and clinical diversity among the samples from the included studies in this meta-analysis, we ultimately rated the certainty levels for both outcomes as ‘very low’.

Main findings

Characterising the prevalence of PGD and its symptoms in the context of Chinese culture has important nosological, clinical and management implications. The present study systematically reviewed and quantitatively analysed the PRs of PGD and its symptoms among bereaved individuals in China. First, we found a prevalence of 8.9% for PGD and a prevalence of 32.4% for its symptoms. The rates of both PGD and its symptoms were statistically higher in women compared with men. Second, we found that individuals whose deceased loved ones died from illnesses had the lowest prevalence of PGD and its symptoms. On the other hand, individuals whose deceased loved ones were the only child had the highest prevalence of PGD, and those whose deceased loved ones died due to earthquakes had the highest prevalence of PGD symptoms. Third, the study observed that individuals with PGD had a shorter duration since the loss compared with those without PGD. Finally, several study-level factors were found to be associated with the prevalence of PGD or its symptoms including the survey site, the diagnostic criteria used to define PGD and the mean age of the study sample.

The 8.9% prevalence of PGD suggests that a significant proportion of bereaved individuals in China meet the diagnostic criteria for PGD; additionally, the 32.4% prevalence of its symptoms indicates that a larger portion of this population experiences symptoms of complex grief, even though they may not meet the full diagnostic criteria for PGD. The higher prevalence of PGD and its symptoms, as well as the greater severity of PGD symptoms in women compared with men among the bereaved Chinese, is consistent with findings from previous international studies.53 54 This difference may be attributed to females’ vulnerability, their involvement in caregiving and nurturing roles and their tendency to express emotions more openly.21 55 56

The literature indicates that the risk of PGD or its symptoms is higher when the death of a loved one occurs unexpectedly, suddenly, violently or under traumatic circumstances.57–61 Previous meta-analyses of international studies have reported PRs of PGD following non-violent bereavement, loss of family members due to cancer, unnatural deaths and loss during the COVID-19 pandemic at 9.8%, 14.2%, 49.0% and 46.4%, respectively.61–64 The PRs of complicated grief among bereaved family members of patients with cancer in Japan and earthquake survivors in Iran were 14% and 76%, respectively.65 66 Similarly, the present study found that the nature of bereavement is a significant determinant of the prevalence of PGD and its symptoms among the bereaved Chinese population.

In general, deaths due to illnesses are often anticipated and not violent, which may result in lower PRs of PGD and its symptoms as well as lower levels of PGD symptoms in Chinese bereaved individuals. However, the sudden and violent loss of lives and significant damage to physical assets during natural disasters like earthquakes intensify feelings of grief in bereaved individuals. This heightened intensity can persist for an extended period of time, leading to a higher risk of PGD and its symptoms among those who have experienced earthquakes.

Due to the stigma, discrimination and misconceptions associated with HIV/AIDS, the loss associated with AIDS can lead to additional complications in the grief process.46 67 This may explain the high prevalence of PGD symptoms, as high as 35.1%, observed among Chinese individuals bereaved by AIDS-related loss in this study. Additionally, due to the one-child policy in China, the loss of the only child can be particularly devastating for the bereaved parents. These parents may experience self-blame, guilt and a sense of not fulfilling their filial duties to their ancestors due to their inability to continue the family line.13 Moreover, since adult children traditionally serve as the primary home carers for their elderly parents in China, losing the only child also means losing the major source of financial and emotional support.21 These sociocultural factors could contribute to the increased risk of PGD and its symptoms in individuals who experience the loss of their only child.

Our study found that PGD and its symptoms were highly prevalent among those who have lost loved ones to COVID-19, with rates as high as 16.8% and 30.7%, respectively. These findings highlight the critical need for postpandemic grief counselling services in China, as patients COVID-19 often passed away in isolated settings without the opportunity for loved ones to say goodbye. This is supported by previous findings that a lack of meaningful communication before or during death can increase the risk of PGD among bereaved family members.68

Consistent with the universal phenomenon that grief alleviates over time,61 we found that individuals with PGD were more likely to have experienced a more recent loss. The positive association between the severity of PGD symptoms and religious beliefs among bereaved Chinese aligns with the higher prevalence of mental health challenges observed in Chinese individuals with religious beliefs, compared with those without.69 70 This could be due to the fact that, in China, the world’s most atheistic country, people often turn to religions for assistance when facing mental health challenges such as complicated grief.21

Compared to the Prigerson et al’s criteria,52 the ICD-11 diagnostic criteria are more lenient,36 which might explain the lower prevalence of PGD when conducting a meta-analysis of studies using the Prigerson et al’s criteria compared with those using the ICD-11 criteria. The findings that the study site and mean age of the study sample were associated with the pooled prevalence of PGD and its symptoms highlight the importance of considering an appropriate sample frame when examining the epidemiology of PGD.

Limitations

We acknowledge that this study has certain limitations. First, none of the included studies was rated as having a low RoB. The dynamic nature of the population of bereaved individuals and the challenges of reaching them through household-based surveys led to the majority of the included studies being conducted solely online or through a combination of online and offline approaches. This makes it challenging to ascertain the representativeness of the online samples, warranting caution when generalising the findings to the broader population of bereaved individuals in China. Second, due to the limited number of included studies focusing on bereavement due to COVID-19, AIDS and earthquakes, the ranking of PRs for PGD and its symptoms by type of bereavement in this study may not be robust enough. Third, the identification of high-risk subgroups for PGD and its symptoms in this study was based solely on the results of direct comparisons between subgroups, without adjusting for potential confounders. As a result, the findings regarding high-risk subgroups are preliminary and require further validation through additional studies.

Implications

In China, mental health service resources remain insufficient to meet the growing needs of the general population.71 72 This inadequacy constitutes a significant barrier to accessing grief counselling and other mental health services.73 74 Specifically, grief counselling services in China are still in their early developmental stages, which results in an insufficient and uneven distribution of these services in hospitals and communities. Concerns have also been raised regarding the quality of grief counselling services offered within China’s medical institutions.21 Additionally, the challenges of providing culturally adapted services further complicate the task of addressing the needs of the bereaved. Notably, no established community-based system for screening, managing or referring PGD currently exists in China.75

Considering the sheer size of the bereaved population in China, along with an 8.9% prevalence of PGD and a 32.4% prevalence of PGD symptoms, the demand for grief counselling services could be exceptionally high. Our study emphasises the urgency of strengthening these services for the bereaved population in China. Strategies could include integrating services into primary care, developing a two-way referral system between mental health institutions and clinics and providing training for community mental health workers.

The shorter duration of grief observed in bereaved individuals with PGD implies that the early provision of grief counselling services could be more effective in preventing PGD. Services could include psychosocial support, regular PGD screening and facilitating psychiatric referrals when necessary. If these services targeted high-risk groups such as women, individuals with religious beliefs, those who lost their only child and those grieving severe losses, they could be more cost-effective.

Finally, the very low level of evidence certainty suggests that further large-scale, representative studies employing stringent methodology are necessary. These would yield more accurate prevalence data on PGD and its symptoms in the bereaved Chinese population.