Abstract
With the COVID-19 pandemic prompting predictions of a “grief pandemic,” rates and risks for Prolonged Grief Disorder (PGD) warrant further investigation. Data were collected online from 1470 respondents between October 2020 and July 2021. Shorter time since death, deaths of siblings and “others,” and deaths from accidents and homicides were positively associated with potential risk of probable PGD; deaths of extended family and from dementia were negatively associated with probable PGD. When compared directly to deaths from COVID-19, natural causes of death were associated with lower potential risk of probable PGD, while deaths from unnatural causes were associated with higher potential risk.
Introduction
An unfortunate consequence of the COVID-19 pandemic, and the more than 5 million deaths worldwide (World Health Organization, 2021) that occurred as a result, is the grief experienced by bereaved survivors. Recent meta-analyses report a prevalence rate of 10% of prolonged grief disorder (PGD)1 in bereaved samples that lost a loved one due to natural (i.e., non-violent) causes (Lundorff et al., 2017) and 49% in bereaved samples that lost a loved one to unnatural (i.e., violent) causes (Djelantik et al., 2020). PGD, a new mental disorder recently added to the ICD-11 and approved for inclusion in the DSM-5-TR (Prigerson et al., 2021), is characterized by persistent yearning for and/or preoccupation with thoughts of the deceased and associated distressing and disabling grief symptoms (Prigerson et al., 2021). However, the characteristics associated with COVID-19 (e.g., the patient’s rapid demise and increased likelihood of death in the intensive care unit (ICU) which is associated with heightened risk of psychiatric morbidity for the bereaved (Wright et al., 2010) and the circumstances surrounding the pandemic (e.g., social isolation and inability to be present at the time of death or have proper funerals) may exacerbate levels of psychological distress among mourners, promoting a setting ripe for development of PGD (Eisma et al., 2020; Lichtenthal et al., 2020).
Research that has examined psychological consequences of natural disasters, which share similarities to pandemics including high death toll and disruptions to daily life, suggests that PGD prevalence may rise dramatically as a result of the pandemic (Eisma et al., 2020). Despite these predictions, there has been little empirical investigation in the context of the pandemic to substantiate these claims. In the Netherlands, it has been shown that being recently bereaved during the pandemic elicited more severe acute grief (a significant predictor of PGD) than before the pandemic (Eisma & Tamminga, 2020). When comparing individuals bereaved specifically by deaths from COVID-19 versus other causes of death, those who have lost loved ones to COVID-19 exhibited greater grief severity when compared to those bereaved due to natural, but not unnatural, causes (Eisma et al., 2021). In both these studies, however, the amount of time that had elapsed since the death was insufficient to meet the time-criterion of PGD (i.e., 12 months) and, thus, precluded conclusions on the development of PGD (Eisma et al., 2021; Eisma & Tamminga, 2020). A recent study in China found elevated rates of PGD (~27–38%) among those bereaved due to COVID-19 or a COVID-19 related complications as well as differences by kinship relationship to the deceased, but no significant differences in prevalence before and after six months post-loss (Tang & Xiang, 2021).
Determining those at risk for PGD is an important public health priority given the association of PGD with both physical and mental health problems (e.g., suicidal thoughts, impaired quality of life; Prigerson et al., 1997, 2021). This need may be particularly severe in the context of the pandemic, as the risk for and severity of adverse outcomes associated with PGD may themselves be elevated due to the pandemic. Given the limited research on this topic, further research is needed to confirm results of the few early studies in this area (e.g., Eisma et al., 2021; Tang & Xiang, 2021). In addition, while past studies have used measures based upon the DSM-5 criteria for persistent complex bereavement disorder (PCBD) or the ICD-11 criteria for PGD, there is a need for findings based upon the new DSM-5-TR criteria for PGD, which has replaced PCBD. The present study aims to fill these research gaps by using data from a self-reported online survey to determine the rates of (probable) PGD associated with COVID-19 versus other causes of death (e.g., other natural and unnatural causes of death), both with and without adjustment for time since loss and respondents’ relationship to the deceased.
Materials and methods
Participants and procedure
Data were collected from 2,040 participants between October 2020 and July 2021. Participants were eligible to participate in the study if they were 18 years or older, experienced the death of a person they considered a significant other, and were able to access and complete an online survey. The online survey was hosted on the Cornell Center for Research on End-of-Life Care website. Participants were not actively recruited to complete the survey; instead, participants found and voluntarily completed our survey online. This study was deemed exempt by the local ethics committee (record number #21-02023366), which necessitated de-identification of study participants. As a result, sociodemographic information and other identifying characteristics were not obtained from study participants. In addition, participants (n = 570) were excluded from the analytic sample if data were missing on key analytic variables. No significant differences in study variables emerged between participants who did and those who did not have missing data (i.e., participants and those excluded from the analysis). The final analytic sample consisted of 1,470 participants.
Measures
In addition to completing measures of grief intensity (described below), participants were also asked to report on the length of time (in months) since the death, the respondent’s relationship to the deceased, the deceased’s cause of death, and whether grief-related symptoms caused significant impairment in participant functioning. Respondents’ relationship to the deceased was coded as parent, offspring, sibling, extended family (a composite variable formed from grandparent, aunt, uncle, cousin, and friend responses), and “other” which served as the reference group.
Grief intensity scale
The Grief Intensity Scale (GIS) was created to serve as an online version of the PG-13-R, a self-rated scale consisting of 10 Likert-scale questions pertaining to symptoms of grief (Prigerson et al., 2021). The GIS uses the same 5-point Likert response format to evaluate the intensity and severity of the assessed PGD symptoms. The PG-13-R has demonstrated good reliability in three independent bereaved study samples (Cronbach’s α = .83–.93; Prigerson et al., 2021). In the current study, the GIS exhibited excellent reliability, with Cronbach’s α = .93. A symptom threshold score of 30 or greater was shown to correspond well to a probable diagnosis of PGD using the DSM criteria set (kappa ≥0.70 across the datasets; Prigerson et al., 2021). We acknowledge that scores of this online assessment are not equivalent to clinical diagnosis made by a trained mental health professional. Furthermore, the DSM-5-TR diagnosis requires 12 months or more to have elapsed since the death, and therefore diagnoses consistent with the DSM-5-TR criteria cannot be made prior to a year after the death. For these reasons, we refer to the primary outcome variable as “probable PGD.” Throughout the survey, participants were instructed to seek evaluation and support from a mental health professional if they met criteria for probable PGD.
Statistical analyses
Prior to conducting data analyses, all variables were examined for normality and homogeneity. Descriptive statistics (means, standard deviations, frequency distributions) were examined to characterize the sample based on the limited background data available in this IRB exempt study. These variables included cause of death, the respondent’s relationship to the deceased, and length of time since the death.
Chi-square tests were conducted to examine differences in participant characteristics for categorical variables, and independent samples t-tests were used to examine differences in continuous variables. Variables that exhibited significant differences (i.e., type of respondent’s relationship to the deceased and time since loss) at p < .10 were retained for subsequent “adjusted” analyses that included these variables. Although diagnostic criteria for a PGD diagnosis requires symptoms to persist for one year, we include those bereaved for less than 12 months to examine correlates of meeting symptom threshold for probable PGD because acute grief has been shown to be a strong predictor for development of PGD (Boelen & Lenferink, 2020). Using the DSM-5-TR symptom threshold of 30 for meeting diagnostic criteria for probable PGD, a dichotomous GIS variable was created in which GIS scores ≥30 were coded as 1 = probable PGD, and GIS scores <30 were coded as 0 = no probable PGD. We also calculated relative risk (RR) ratios with 95% confidence intervals using the Mantal-Haenszel formula for variance.
To assess whether length of time since the loss and/or respondents’ relationship to the deceased were associated with probable PGD, we calculated RR ratios with 95% CIs and crude and adjusted odds ratios (ORs). Further, to assess whether cause of death (COVID-19 versus each cause of death) was associated with probable PGD scores independent of respondents’ relationship to the deceased and time since loss, we used binary logistic regression models to calculate adjusted ORs. Finally, we replicated the above analyses and calculated RR ratios with 95% CIs and crude and adjusted ORs to assess the relationships among cause of death and probable PGD scores in a subsample of participants (n = 599) bereaved ≥12 months, consistent with the DSM’s time criterion for PGD. A significance value of p < .05 was used for all analyses.
For both sets of analyses, G*Power Software version 3.1 was used to conduct post-hoc power analyses to calculate achieved power for our sample. Power was calculated in two ways: Cohen’s ω (0.1=small; 0.3=medium) was used for the unadjusted analysis and f2 for adjusted analyses (f2: 0.1=small; 0.3=medium). Statistical analyses were conducted using SPSS Version 25.
Results
Overall, the prevalence rate of probable PGD in this self-selected sample was 66.53% (978/1470). Table 1 presents descriptive statistics for the full study sample and subgroup comparisons based on PGD status (i.e., meets criteria for probable PGD by scoring ≥30 on the GIS [n=922] versus not [i.e., GIS scores <30; n=548]). Independent samples t-tests indicate that those in the probable PGD group reported a significantly shorter amount of time since the death (t[1468]=2.31, p = .021). Chi-square tests were then performed to examine the associations between respondents’ relationship to the deceased, cause of death, and meeting criteria for probable PGD. Results (Table 1) indicate that those who were siblings (χ2 [1, N=1470]=4.53, p = .03), extended family members (χ2 [1, N=1470]=24.75, p < .001), and those who reported their relationship to the deceased as “other” (χ2 [1, N=1470]=24.75, p < .001) were significantly associated with probable PGD scores. When examining differences in probable PGD by cause of death, we found significant associations among dementia (χ2 [1, N=1470]=11.32, p < .001) and accidents (χ2 [1, N=1470]=7.39, p = .007), while homicide was trending significance (p = .079). When examining time since loss and each category of respondents’ relationship to the deceased as covariates, only variables significantly associated with probable PGD (e.g., length of time since death and the categories of sibling, extended family, and “other” as respondents’ relationship to the deceased) were included.
Table 1.
Descriptive Statistics for Study Sample Examining Probable PGDa (n=1470).
| Sample (n = 1470) | No PGD (n = 548) | Probable PGD (n = 922) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Variable | n | % | M (SD) | n | % | M (SD) | n | % | M (SD) | p * | Probable PGD Prevalence□ |
| Time Since Death (Months) | 24.5 (84.4) | 31.1 (81.9) | 20.5 (85.7) | .021 | |||||||
| GIS Score | 25.1 (16.4) | 21.5 (5.73) | 39.3 (6.03) | <.001 | |||||||
| Respondent’s Relationship to Deceased | |||||||||||
| Parent | 477 | 32.4 | 184 | 33.6 | 293 | 31.8 | .490 | 61.4% | |||
| Offspring | 203 | 13.8 | 74 | 13.5 | 129 | 14.0 | .815 | 63.5% | |||
| Sibling | 102 | 6.90 | 28 | 5.10 | 74 | 8.00 | .034 | 72.5% | |||
| Extended Family | 186 | 12.7 | 100 | 18.2 | 86 | 9.30 | <.001 | 46.2% | |||
| Friend | 111 | 7.60 | 45 | 8.20 | 66 | 7.20 | .476 | 59.4% | |||
| Other | 391 | 26.6 | 117 | 21.4 | 274 | 29.7 | <.001 | 70.0% | |||
| Cause of Death | |||||||||||
| Cancer | 363 | 24.8 | 148 | 27.1 | 215 | 23.4 | .118 | 59.5% | |||
| Heart Attack/Heart Failure | 204 | 13.9 | 81 | 14.8 | 123 | 13.4 | .483 | 60.1% | |||
| COVID-19 | 118 | 8.10 | 42 | 7.70 | 76 | 8.30 | .766 | 64.1% | |||
| Dementia | 48 | 3.30 | 29 | 5.30 | 19 | 2.1 | <.001 | 38.0% | |||
| Accident | 169 | 11.6 | 47 | 8.60 | 122 | 13.3 | .007 | 71.7% | |||
| Natural/Man-Made Disaster | 26 | 1.80 | 7 | 1.30 | 19 | 2.10 | .312 | 73.0% | |||
| Suicide | 130 | 8.90 | 44 | 8.10 | 86 | 9.40 | .447 | 66.1% | |||
| Homicide | 52 | 3.50 | 13 | 2.40 | 39 | 4.3 | .079 | 75.0% | |||
| Other | 353 | 24.1 | 135 | 24.7 | 218 | 23.8 | .705 | 61.6% | |||
Uses v2 for categorical variables and independent samples t-tests for continuous variables.
Note: Sample included participants bereaved ≤12-months and thus do not meet the time criterion for PGD.
Adjusted for time since loss and respondent’s relationship (Sibling, Extended Family, and ‘Other’) to the deceased.
Table 2 summarizes unadjusted and adjusted associations comparing COVID-19 versus other causes of death as correlates of probable PGD in the full sample. Each death cause category was used as the reference group in each analysis. In general, when comparing COVID-19 to dementia, those who reported COVID-19 as cause of death were more likely to score above the threshold for probable PGD. When adjusting for time since loss and kinship to the deceased, these findings remained significant (B = .93, p = .015). Additionally, although no significant differences were identified in the unadjusted associations comparing COVID-19 to homicide as cause of death, the adjusted model revealed that those who reported COVID-19 as cause of death were less likely to meet criteria for probable PGD compared to deaths from homicide (B = −1.29, p = .010).
Table 2.
Analyses of Cause of Death and Kinship to the Deceased as Associations with Probable PGDa (n = 1470).
| Unadjusted | Adjusted□ | ||||||
|---|---|---|---|---|---|---|---|
| COVID-19 versus Other Causes of Death | No. Scoring ≥ 30 on GIS | % | RR | 95% CI | OR | OR | 95% CI |
| COVID-19 versus Dementia | 1.63* | 1.12 to 2.37 | 2.76 | 2.52* | 1.19 to 5.31 | ||
| Dementia | 19 | 39.6 | |||||
| COVID | 76 | 64.4 | |||||
| COVID-19 versus Cancer | 1.09 | .928 to 1.28 | 1.25 | 1.41 | .898 to 2.21 | ||
| Cancer | 215 | 59.2 | |||||
| COVID | 76 | 64.4 | |||||
| COVID-19 versus Heart Attack/Heart Failure | 1.07 | .897 to 1.27 | 1.19 | 1.09 | .667 to 1.80 | ||
| Heart Attack/Heart Failure | 123 | 60.3 | |||||
| COVID-19 | 76 | 64.4 | |||||
| COVID-19 versus Accident | 0.89 | .758 to 1.05 | 0.697 | 0.724 | .409 to 1.28 | ||
| Accident | 122 | 72.2 | |||||
| COVID-19 | 76 | 64.4 | |||||
| COVID-19 versus Natural/Man-Made Disaster | 0.88 | .673 to 1.15 | 0.667 | 0.525 | .181 to 1.53 | ||
| Natural/Man-Made Disaster | 19 | 73.1 | |||||
| COVID-19 | 76 | 64.4 | |||||
| COVID-19 versus Suicide | 0.97 | .812 to 1.17 | 0.926 | 1.08 | .607 to 1.91 | ||
| Suicide | 86 | 66.2 | |||||
| COVID-19 | 76 | 64.4 | |||||
| COVID-19 versus Homicide | 0.86 | .699 to 1.06 | 0.603 | 0.276* | .104 to .736 | ||
| Homicide | 39 | 75.0 | |||||
| COVID-19 | 76 | 64.4 | |||||
| COVID-19 versus Other | 1.04 | .891 to 1.22 | 1.12 | 0.964 | .611 to 1.52 | ||
| Other | 218 | 61.8 | |||||
| COVID-19 | 76 | 64.2 | |||||
Note: Sample included participants bereaved ≤ 12-months and thus do not meet the time criterion for PGD. RR=Relative Risk; OR ≕ Odds Ratio.
Adjusted for time since loss and respondent’s relationship (Sibling, Extended Family, and ‘Other’) to the deceased.
Statistically significant results p < .05.
A post-hoc analysis using G*Power software version 3.1 was conducted to calculate power for our sample. In the full sample, power to detect small and medium effects (Cohen’s ω = 0.1; 0.5 for small and medium effects, respectively) at the p < .05 level for the unadjusted analyses indicated that power ranged from 22% to detect small effects when comparing COVID-19 versus natural/manmade disasters to 59% for COVID-19 versus cancer; while power to detect a medium effect size was 94% and 99% for COVID-19 versus natural disasters and COVID-19 versus cancer, respectively. While accounting for covariates in the logistic regression model, the power to detect a small effect (Cohen’s f2 = 0.1) at p < .05 ranged from 28% to detect a small effect comparing COVID-19 to natural/manmade disasters, to 63.4% when comparing COVID-19 versus cancer. When examining power to detect a medium effect size, power was 87.6% and 99% for the comparisons of COVID-19 versus natural/manmade disaster and COVID-19 versus cancer, respectively.
Subgroup analysis
Finally, we conducted a set of exploratory analyses in a subsample of participants (n=599) who were bereaved ≥ 12-months.When compared to the full sample, subsample participants had significantly lower mean GIS scores compared to those bereaved for fewer than 12months t[1456]=3.54, p < .001. Additionally, subsample participants were less likely to report cancer (χ2 [1, N=1470]=11.26, p < .001), COVID-19 (χ2 [1, N=1470]=42.29, p < .001), and dementia (χ2 [1, N=1470]=7.35, p = .007) as death causes compared to those bereaved for fewer than 12-months. However, results also indicate that subsample participants were more likely to report suicide (53.2%) as cause of death compared to those bereaved for fewer than 12months (46.8%); (χ2 (1, N=1470) = 8.19, p = .006).
We then examined the unadjusted and adjusted associations comparing COVID-19 versus other causes of death as correlates of probable PGD among participants (n=599) bereaved ≥12 months (see Table 3). Each cause of death category was used as the reference group in adjusted models. Results only showed a significant association when comparing COVID-19 versus dementia in the unadjusted analysis, such that those bereaved from COVID-19 were significantly more likely to score above the threshold for probable PGD. When adjusting for covariates, this association was no longer significant. Further, although the comparison of COVID-19 versus cancer was not significant in the unadjusted analysis, the adjusted model was trending significance: those bereaved from COVID-19 were more likely to score above the threshold for probable PGD compared to those bereaved from cancer (B=1.14, p = .069).
Table 3.
Analyses examining cause of death and kinship to the deceased as associations with probable PGD in participants bereaved ≥ 12-Months (n = 599).
| Unadjusted | Adjusted□ | |||||||
|---|---|---|---|---|---|---|---|---|
| COVID-19 versus Other Causes of Death | N | No. Scoring ≥ 30 on GIS | % | RR | 95% CI | OR | OR | 95% CI |
| COVID-19 versus Dementia | 4.03* | 1.11 to 14.67 | 12.38 | 2.70 | .195 to 37.16 | |||
| Dementia | 11 | 2 | 18.2 | |||||
| COVID-19 | 15 | 11 | 73.3 | |||||
| COVID-19 versus Cancer | 1.37 | .977 to 1.91 | 2.37 | 3.12í | .917 to 10.60 | |||
| Cancer | 175 | 94 | 53.7 | |||||
| COVID-19 | 15 | 11 | 73.3 | |||||
| COVID-19 versus Heart Attack/Heart Failure | 1.41 | .974 to 2.05 | 2.55 | 2.21 | .604 to 8.08 | |||
| Heart Attack/Heart Failure | 79 | 41 | 51.9 | |||||
| COVID-19 | 15 | 11 | 73.3 | |||||
| COVID-19 versus Accident | 1.06 | .753 to 1.49 | 1.22 | 1.96 | .442 to 8.70 | |||
| Accident | 75 | 52 | 69.3 | |||||
| COVID-19 | 15 | 11 | 73.3 | |||||
| COVID-19 versus Natural/Man-Made Disaster | 1.03 | .587 to 1.80 | 1.10 | 1.28 | .080 to 20.65 | |||
| Natural/Man-Made Disaster | 7 | 5 | 71.4 | |||||
| COVID-19 | 15 | 11 | 73.3 | |||||
| COVID-19 versus Suicide | 1.07 | .756 to 1.51 | 1.263 | 1.64 | .417 to 6.45 | |||
| Suicide | 67 | 46 | 68.7 | |||||
| COVID-19 | 15 | 11 | 73.3 | |||||
| COVID-19 versus Homicide | 1.05 | .689 to 1.59 | 1.18 | .164 | .013 to 2.02 | |||
| Homicide | 20 | 14 | 70.0 | |||||
| COVID-19 | 15 | 11 | 73.3 | |||||
| COVID-19 versus Other | 1.40 | .995 to 1.97 | 2.50 | 2.12 | .604 to 7.43 | |||
| Other | 147 | 77 | 52.4 | |||||
| COVID-19 | 15 | 11 | 73.3 | |||||
Notes. RR = Relative Risk. OR = Odds Ratio.
Adjusted for time since loss and respondent’s relationship (Sibling, Extended Family, and ‘Other’) to the deceased.
Statistically significant results.
Trending significance at p = .069.
A post-hoc analysis was conducted to calculate achieved power for the subsample bereaved ≥12-months, power to detect small and medium effects (Cohen’s ω=0.1; 0.5 for small and medium effects, respectively) at the p < .05 level for the unadjusted analyses indicated that power ranged from 7.5% to detect small effects when comparing COVID-19 versus natural/manmade disasters to 28% for COVID-19 versus cancer; while power to detect a medium effect size was 29% and 98% for COVID-19 versus natural disasters and COVID-19 versus cancer, respectively. When accounting for covariates in the logistic regression model, the power to detect a small effect (Cohen’s f2 = 0.1) at p < .05 ranged from 9.9% to detect a small effect when comparing COVID-19 to natural/manmade disasters, and 33.9% when comparing COVID-19 versus cancer. When examining power to detect a medium effect size, power was 24% and 94.5% for the comparisons of COVID-19 versus natural/manmade disasters and COVID-19 versus cancer, respectively.
Discussion
This cross-sectional study uses the new DSM-5-TR diagnostic criteria for PGD (Prigerson et al., 2021) to determine rates and correlates of probable PGD in the context of the COVID-19 pandemic. Comparing COVID-19 versus other causes of death, we found COVID-19 deaths to be associated with probable PGD when compared to dementia and trended likewise with other natural causes of death. Compared to unnatural causes of death such as deaths from homicide, those who reported COVID-19 as cause of death were less likely to meet criteria for PGD. Shorter time since death, either a sibling or “other” relationship with the deceased, and accidental deaths were positively associated with likelihood of meeting diagnostic criteria for PGD, while extended family member decedents and deaths attributed to dementia were negatively associated with likelihood of meeting diagnostic criteria for PGD.
Our results revealed a substantially higher probable PGD prevalence rate of 66.5% compared to both the 9.8% reported in the Lundorff et al. (2017) meta-analysis of the naturally bereaved and the 49% reported in the Djelantik et al. (2020) meta-analysis of the unnaturally bereaved. It is important to acknowledge our sample was self-selected; participants found our Center for Research on End-of-Life Care website on their own and voluntarily completed a survey about their loss and reactions to it. Thus, respondents were online users seeking information about grief which would be expected to represent inflated rates compared to more representative community-based samples (see 6.3%−16.6% in Prigerson et al., 2021). A decade earlier, Lichtenthal et al. (2011) showed that bereaved caregivers experiencing PGD did not readily access mental health services, with the most common reason being not having mental health concerns. However, Google Analytics of the study website revealed that 54.3% of our site traffic comes from organic searches using keywords such as “grief assessment,” “grief intensity scale,” or “grief scale” and 32.86% from a direct link suggesting referrals from mental health professionals or recurrent use. This suggests that most of our site visits are self-referrals from bereaved people who want and seek information about their grief, as well among those who express an interest in and willingness to complete a voluntary online survey about their grief intensity. Although certainly not conclusive, it does suggest an interest in and receptivity to disordered grief and the use of online bereavement resources.
Contrary to our expectations, our findings failed to support the predictions that COVID-19 deaths would be associated with a higher prevalence of PGD (Eisma et al., 2020). In the context of disproportionate sampling of unnaturally bereaved in our sample (Kersting et al., 2011), this is consistent with Eisma et al. (2021) finding that the COVID-19 bereaved do not experience greater grief than the unnaturally bereaved. One explanation for this is that the 2–8 week period between COVID-19 symptom onset to death (Baud et al., 2020) perhaps allows mourners to prepare for the loss of their loved one more than would be the case in deaths due to accidents or homicides, but less so than in deaths attributed to a protracted illness such as dementia. This explanation is consistent with a prior finding that the unexpectedness of the death explains greater PGD symptom levels among the COVID-19 bereaved compared to those bereaved by deaths from natural causes (Eisma et al., 2021). We here found that those bereaved by a COVID-19 death were more likely to meet criteria for probable PGD when directly compared to dementia as a cause of death. Another explanation is that there may be some comfort found in the high death toll, a sense of community among those affected fostering empathy and support, and a reduced sense of guilt given that the virus has indiscriminately infected people from all walks of life (Mein, 2020). Further research is needed on loss characteristics associated with COVID-19 and coping strategies used during this pandemic (Breen et al., 2022).
With regard to other causes of death, our findings indicate accidents and homicides were positively associated with likelihood of probable PGD overall, which aligns with previous research on the propensity of unnatural causes of death to precipitate mental health disorders (Kristensen et al., 2012). Likewise, our results suggest that dementia may be a negative correlate of probable PGD, which also aligns with previous research demonstrating that the confrontation of “serial losses” (e.g., as a result of cognitive and physical declines of the person with dementia) (Chan et al., 2013; Holley & Mast, 2009; Kiely et al., 2008; Schulz et al., 2003; Singer & Papa, 2021) prior to the death, and the stress reduction as a result from a reduction in caregiving responsibilities after the death (Ghesquiere et al., 2011), may alleviate grief severity during bereavement and facilitate positive bereavement adjustment. When considering these results by death types, they are consistent with Eisma et al. (2021) finding that only unnatural bereavement yields higher grief levels than COVID-19 bereavement.
This study also found that being a sibling of the deceased is positively correlated with probable PGD, while being an extended family member of the deceased is negatively correlated with probable PGD. These findings are consistent with previous research demonstrating that closer kinship to the deceased is consistently related to higher grief intensity across different age and cultures (Glickman, 2020; He et al., 2014; Stammel et al., 2013; Tang & Xiang, 2021). “Other” being a positive correlate of probable PGD is difficult to explain given the lack of specificity, but one possible explanation is that this option may have captured the loss of a spouse or pet, which have been shown to precipitate PGD (Lee, 2020; Lundorff et al., 2017), but was not explicitly measured in the current study. Lastly, we found that less time elapsed since the death was positively correlated with probable PGD. This result may be explained by the phenomenon of adjusting to the loss over time contributing to declining grief severity (Guldin et al., 2012; Latham & Prigerson, 2004; Prigerson et al., 1997), which is substantiated by our finding of lower mean GIS scores for subjects bereaved ≥12 months compared to those bereaved <12 months. Given similarities in the risks for PGD before and after 12 months post-loss, and prior studies showing that a 6-month assessment predicts a 12-month assessment of PGD as well as other forms of future morbidity (Prigerson et al., 2009; 2021), this suggests the need for more empirical work to examine the harms and benefits of PGD diagnosis before and after one year post-loss.
Although the current study provides timely insight into potential risk factors and intensity of grief experienced among those bereaved during the COVID-19 pandemic, the results should be considered in the context of its limitations. First, while the DSM criteria for PGD requires at least 12 months to have elapsed since the death, our overall sample did not exclude those who did not meet this temporal requirement, which only further contributes to the caution that participants in this study were not formally diagnosed; a subanalysis exploring only those bereaved ≥12 months was performed to provide additional information about differences in rates of probable PGD based on time from loss. Second, the cross-sectional nature of our survey does not allow us to examine changes in grief over time nor make inferences of causality, which restricts interpretation of the study variables as risk factors. It should also be noted that our analyses in those bereaved greater than or equal to 12 months had a low number of reported COVID-19 as cause of death (n = 15), thus these analyses are considered exploratory and were not adequately powered to detect significant effects. Relatedly, a third limitation is the correlational nature of the study, and thus future research should contribute to the sparse literature examining unassessed variables, such as social support and relationship quality with the deceased, that may influence grief intensity following the death of a loved one due to COVID-19 (Eisma et al., 2021; Eisma & Tamminga, 2020). Lastly, the use of an online survey limits formal diagnosis of PGD, as well as the assessment of other demographic factors (i.e., gender, race/ethnicity, economic vulnerability) that we could examine for their associations with the respondent’s grief intensity. Future research should contribute to the sparse literature examining the impact of demographic and cultural (i.e., importance and use of mourning rituals) factors on bereavement adjustment (Eisma et al., 2021; Eisma & Tamminga, 2020) and seek to replicate our current findings in larger samples with more than a single assessment time point to capture more granular changes in grief intensity during bereavement in the wake of the COVID-19 pandemic.
In conclusion, this cross-sectional study found that probable PGD onset was associated with lower rates than unnatural deaths (e.g., accidents, homicide) and higher rates than natural deaths (e.g., dementia) compared to deaths from COVID-19. Less time since death, closeness of the respondent’s relationship to the deceased, and violent causes of death were positively associated with probable PGD, while extended family kinship and dementia was negatively associated. More focused research on the prevalence of PGD among the COVID-19 bereaved is needed in light of the number of grieving individuals, availability of a new psychiatric diagnosis, and online tools and resources to identify and minimize the public health risks posed by PGD.
Acknowledgements
All sponsors had no direct input into the design or conduct of the study; collection, management, analysis, or interpretation of the data; or preparation, review, or approval of the manuscript.
Funding
This work was supported by grants from the National Cancer Institute [CA197730; Prigerson; CA218313; Prigerson/Lichtenthal], the National Institute of Minority Health and Health Disparities [MD007652; Maciejewski/Prigerson], the National Institute of Nursing Research [NR018693; Prigerson/Epstein]; the National Institute on Aging [AG049666; Reid/Prigerson; K99 AG073509: Falzarano; T32 AG049666; Prigerson/Falzarano], the National Institute of Mental Health [MH121886; Maciejewski/Prigerson].
Footnotes
Disclosure statement
No potential conflict of interest was reported by the authors.
References to PGD throughout this study refer to the psychometrically validated diagnostic criteria for PGD that will appear in DSM-5-TR. However, the citations in this introduction may refer to/have investigated PGD predecessors such as complicated grief and traumatic grief. For the purposes of this introduction, these can all be assumed to be roughly equivalent.
References
- Baud D, Qi X, Nielsen-Saines K, Musso D, Pomar L, & Favre G (2020). Real estimates of mortality following COVID-19 infection. The Lancet. Infectious Diseases, 20(7), 773–773. 10.1016/S1473-3099(20)30195-X [DOI] [PMC free article] [PubMed] [Google Scholar]
- Boelen PA, & Lenferink L (2020). Symptoms of prolonged grief, post-traumatic stress, and depression in recently bereaved people: Symptom profiles, predictive value, and cognitive behavioural correlates. Social Psychiatry and Psychiatric Epidemiology, 55(6), 765–777. 10.1007/s00127-019-01776-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- Breen LJ, Mancini VO, Lee SA, Pappalardo EA, & Neimeyer RA (2022). Risk factors for dysfunctional grief and functional impairment for all causes of death during the COVID-19 pandemic: The mediating role of meaning. Death Studies, 46(1), 43–52. 10.1080/07481187.2021.1974666 [DOI] [PubMed] [Google Scholar]
- Chan D, Livingston G, Jones L, & Sampson EL (2013). Grief reactions in dementia carers: A systematic review. International Journal of Geriatric Psychiatry, 28(1), 1–17. 10.1002/gps.3795 [DOI] [PubMed] [Google Scholar]
- Djelantik A, Smid GE, Mroz A, Kleber RJ, & Boelen PA (2020). T The prevalence of prolonged grief disorder in bereaved individuals following unnatural losses: Systematic review and meta regression analysis. Journal of Affective Disorders, 265, 146–156. 10.1016/j.jad.2020.01.034 [DOI] [PubMed] [Google Scholar]
- Eisma MC, Boelen PA, & Lenferink LIM (2020). Prolonged grief disorder following the Coronavirus (COVID-19) pandemic. Psychiatry Research, 288, 113031. 10.1016/j.psychres.2020.113031 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Eisma MC, & Tamminga A (2020). Grief before and during the COVID-19 pandemic: Multiple group comparisons. Journal of Pain and Symptom Management, 60(6), e1–e4. 10.1016/j.jpainsymman.2020.10.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Eisma MC, Tamminga A, Smid GE, & Boelen PA (2021). Acute grief after deaths due to COVID-19, natural causes and unnatural causes: An empirical comparison. Journal of Affective Disorders, 278, 54–56. 10.1016/j.jad.2020.09.049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ghesquiere A, Haidar YM, & Shear MK (2011). Risks for complicated grief in family caregivers. Journal of Social Work in End-of-Life & Palliative Care, 7(2–3), 216–240. 10.1080/15524256.2011.593158 [DOI] [PubMed] [Google Scholar]
- Glickman K (2020). Prolonged grief disorder in a diverse college student sample. Frontiers in Psychology, 11, 604573. 10.3389/fpsyg.2020.604573 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Guldin MB, Vedsted P, Zachariae R, Olesen F, & Jensen AB (2012). Complicated grief and need for professional support in family caregivers of cancer patients in palliative care: A longitudinal cohort study. Supportive Care in Cancer, 20(8), 1679–1685. 10.1007/s00520-011-1260-3 [DOI] [PubMed] [Google Scholar]
- He L, Tang S, Yu W, Xu W, Xie Q, & Wang J (2014). The prevalence, comorbidity and risks of prolonged grief disorder among bereaved Chinese adults. Psychiatry Research, 219(2), 347–352. 10.1016/j.psychres.2014.05.022 [DOI] [PubMed] [Google Scholar]
- Holley CK, & Mast BT (2009). The impact of anticipatory grief on caregiver burden in dementia caregivers. The Gerontologist, 49(3), 388–396. 10.1093/geront/gnp061 [DOI] [PubMed] [Google Scholar]
- Kersting A, Brähler E, Glaesmer H, & Wagner B (2011). Prevalence of complicated grief in a representative population-based sample. Journal of Affective Disorders, 131(1–3), 339–343. 10.1016/j.jad.2010.11.032 [DOI] [PubMed] [Google Scholar]
- Kiely DK, Prigerson H, & Mitchell SL (2008). Health care proxy grief symptoms before the death of nursing home residents with advanced dementia. The American Journal of Geriatric Psychiatry, 16(8), 664–673. 10.1097/JGP.0b013e3181784143 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kristensen P, Weisaeth L, & Heir T (2012). Spring). Bereavement and mental health after sudden and violent losses: A review. Psychiatry, 75(1), 76–97. 10.1521/psyc.2012.75.1.76 [DOI] [PubMed] [Google Scholar]
- Latham AE, & Prigerson HG (2004). Suicidality and bereavement: Complicated grief as psychiatric disorder presenting greatest risk for suicidality. Suicide & Life-Threatening Behavior, 34(4), 350–362. 10.1521/suli.34.4.350.53737 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lee SA (2020). Does the DSM-5 grief disorder apply to owners of deceased pets? A psychometric study of impairment during pet loss. Psychiatry Research, 285, 112800. 10.1016/j.psychres.2020.112800 [DOI] [PubMed] [Google Scholar]
- Lichtenthal WG, Roberts KE, & Prigerson HG (2020). Bereavement care in the wake of COVID-19: Offering condolences and referrals. Annals of Internal Medicine, 173(10), 833–835. 10.7326/M20-2526 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lichtenthal WG, Nilsson M, Kissane DW, Breitbart W, Kacel E, Jones EC, & Prigerson HG (2011). Underutilization of mental health services among bereaved caregivers with prolonged grief disorder. Psychiatric Services, 62(10), 1225–1229. 10.1176/ps.62.10.pss6210_1225 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lundorff M, Holmgren H, Zachariae R, Farver-Vestergaard I, & O’Connor M (2017). Prevalence of prolonged grief disorder in adult bereavement: A systematic review and meta-analysis. Journal of Affective Disorders, 212, 138–149. 10.1016/j.jad.2017.01.030 [DOI] [PubMed] [Google Scholar]
- Mein SA (2020). COVID-19 and health disparities: The reality of “The Great Equalizer”. Journal of General Internal Medicine, 35(8), 2439–2440. 10.1007/s11606-020-05880-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Prigerson HG, Bierhals AJ, Kasl SV, Reynolds CF 3rd, Shear MK, Day N, Beery LC, Newsom JT, & Jacobs S (1997). Traumatic grief as a risk factor for mental and physical morbidity. The American Journal of Psychiatry, 154(5), 616–623. May). 10.1176/ajp.154.5.616 [DOI] [PubMed] [Google Scholar]
- Prigerson HG, Boelen PA, Xu J, Smith KV, & Maciejewski PK (2021). Validation of the new DSM-5-TR criteria for prolonged grief disorder and the PG-13-Revised (PG-13-R) scale. World Psychiatry, 20(1), 96–106. 10.1002/wps.20823 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Prigerson HG, Horowitz MJ, Jacobs SC, Parkes CM, Aslan M, Goodkin K, Raphael B, Marwit SJ, Wortman C, Neimeyer RA, Bonanno GA, Bonanno G, Block SD, Kissane D, Boelen P, Maercker A, Litz BT, Johnson JG, First MB, & Maciejewski PK (2009). Prolonged grief disorder: Psychometric validation of criteria proposed for DSM-V and ICD-11. PLoS Medicine, 6(8), e1000121. 10.1371/journal.pmed.1000121 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Prigerson HG, Kakarala S, Gang J, & Maciejewski PK (2021). History and status of prolonged grief disorder as a psychiatric diagnosis. Annual Review of Clinical Psychology, 17(1), 109–126. 10.1146/annurev-clinpsy-081219-093600 [DOI] [PubMed] [Google Scholar]
- Schulz R, Mendelsohn AB, Haley WE, Mahoney D, Allen RS, Zhang S, Thompson L, & Belle SH (2003). End-of-life care and the effects of bereavement on family caregivers of persons with dementia. The New England Journal of Medicine, 349(20), 1936–1942. 10.1056/NEJMsa035373 [DOI] [PubMed] [Google Scholar]
- Singer J, & Papa A (2021). Preparedness for the death of an elderly family member: A possible protective factor for pre-loss grief in informal caregivers. Archives of Gerontology and Geriatrics, 94, 104353. 10.1016/j.archger.2021.104353 [DOI] [PubMed] [Google Scholar]
- Stammel N, Heeke C, Bockers E, Chhim S, Taing S, Wagner B, & Knaevelsrud C (2013). Prolonged grief disorder three decades post loss in survivors of the Khmer Rouge regime in Cambodia. Journal of Affective Disorders, 144(1–2), 87–93. 10.1016/j.jad.2012.05.063 [DOI] [PubMed] [Google Scholar]
- Tang S, & Xiang Z (2021). Who suffered most after deaths due to COVID-19? Prevalence and correlates of prolonged grief disorder in COVID-19 related bereaved adults. Globalization and Health, 17(1), 19. 10.1186/s12992-021-00669-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- World Health Organization. (2021). WHO Coronavirus (COVID-19) Dashboard. https://covid19.who.int/ [Google Scholar]
- Wright AA, Keating NL, Balboni TA, Matulonis UA, Block SD, & Prigerson HG (2010). Place of death: Correlations with quality of life of patients with cancer and predictors of bereaved caregivers’ mental health. Journal of Clinical Oncology, 28(29), 4457–4464. 10.1200/JCO.2009.26.3863 [DOI] [PMC free article] [PubMed] [Google Scholar]