Quantitative peer-reviewed academic journal articles written in English were included. The publication period was set from January 2020 until July 15, 2022. The studies needed to report on positive and/or negative psychological outcomes in adults who had lost a family member, spouse, or friend due to COVID-19. Additional inclusion criteria were that the main outcome was PG and any secondary outcomes were for instance posttraumatic stress, depression, well-being, and/or posttraumatic growth. Lastly, all outcomes should have been measured using validated instruments. An article was excluded when it 1) was a qualitative study, 2) was an intervention study, 3) did not include data of participants (e.g., a literature review), or 4) was conducted with children or adolescents (i.e., <18 years of age).

The three following electronic databases were searched: PsycInfo ( https://psycnet.apa.org), Web of Science ( https://webofscience.com), and Medline ( https://www.nlm.nih.gov). Three topics structured the search terms: 1) bereavement, 2) positive and/or negative psychological outcomes, and 3) COVID-19. To be as comprehensive as possible, we inserted at least six synonyms for each topic. We limited the search to 1) scholarly peer-reviewed journal articles, 2) date of publication (i.e., January 2020–present), and 3) English language. Figure 2 displays the search string in Medline.

The first search was performed on March 3, 2021 and was updated on July 15, 2022. The LSR process deviates from the process described in the study protocol ( Reitsma et al., 2021) with regard to anticipated starting and completion date. Due to practical reasons we started later with the study than was expected. The anticipated latest search will be performed in January, 2024 (instead of June 1 ^st, 2022 as described in the study protocol). Several steps were performed to select studies based on eligibility criteria. Each of the steps were executed independently by two reviewers. First, two reviewers (LR and HM) selected relevant studies by screening titles and abstracts based on the inclusion and exclusion criteria. Second, LR and LHL assessed the relevance of the remaining studies by screening the full texts. The remaining eligible studies were included in the LSR. The level of interrater agreement was calculated with Cohen’s Kappa (κ) for both the title/abstract and full text screening procedures ( McHugh, 2012). Kappa can be interpreted as follows: values ≤ 0 as indicating no agreement and 0.01–0.20 as none to slight, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.80 as substantial, and 0.81–1.00 as almost perfect agreement. Disagreements between the raters were resolved through discussion.

Data of the included studies were extracted by two reviewers (LR and CK). One reviewer (LR) extracted data from the articles using an Excel spreadsheet, and the extracted data were then checked by the other reviewer (CK). Any differences between individual judgements were resolved through discussion. The following data were collected of the included studies: 1) characteristics of study design, 2) characteristics of participants, 3) findings on indicators of negative and positive psychological outcomes after a COVID-19 loss, and 4) findings on correlates of negative and positive psychological outcomes after a COVID-19 loss.

In accord with the PRISMA guidelines, we assessed the risk of bias of included studies. The quality of the included studies was evaluated using the Systematic Assessment of Quality in Observational Research (SAQOR; Ross et al., 2011), which has been used in prior systematic reviews (cf. Dubreucq et al., 2021; Lenferink et al., 2019). The SAQOR examines the risk of bias in six domains: 1) sample, 2) control/comparison group, 3) outcome measurements, 4) follow-up, 5) confounders, and 6) reporting of data. Every domain is comprised of several criteria. All domains are scored as ‘adequate’, ‘inadequate’, ‘unclear’ or ‘not applicable’ based on the number of criteria of the specific domain. Subsequently, for each study an overall quality assessment (high, moderate, low, or very low) is defined according to the frequency of adequate domains. Following the example of prior research ( Lenferink et al., 2019), studies with very low-quality ratings were not further described in results of this review. In line with previous systematic reviews (e.g., Lenferink et al., 2019) and recommendations proposed by Ross et al. (2011), we have adapted the SAQOR to accommodate our specific target group. See Table 1 for a detailed explanation of the SAQOR. The risk of bias assessment was conducted independently by LR and CK. Discrepancies between researchers were resolved through discussion.

We implemented a narrative synthesis on the results of the studies. For each included study, a descriptive summary is provided. We incorporated a description of the design of the studies, characteristics of participants, and results on prevalence rates, symptom-levels, and correlates of negative and positive psychological outcomes. LR extracted the information from the eligible studies using an Excel spreadsheet, and synthesis was then checked by the other reviewer (CK). In case of disagreement between individual judgements, consensus was reached via discussion. No sensitivity analyses were performed.

The exact same electronic database searches will be run every six months up until January 2024, once the first version of the LSR has been published, resulting in two additional literature search updates. After each literature search, any new eligible studies will be incorporated in an update of the review. More specifically, we will update the methods section with the new search date, update the results, tables and figures, and revise the conclusion and discussion (if needed) as new evidence becomes available.

It is anticipated that this review will cease to be living one year after submission of the first version. In case relevant literature is still regularly published, the living method will be extended by six months, which will result in one additional update of the review.

In total, 9871 articles (published by July 15, 2022) resulted from the systematic literature search. After removing duplicates, the remaining 7707 studies were screened by title and abstract. This resulted in 21 studies being included for full text screening. Finally, twelve studies remained for inclusion in this LSR. Lastly, the reference lists of the twelve included articles were screened for additional eligible studies, however this did not result in any additional eligible studies. The levels of agreement between the reviewers for title/abstract and full text screening were moderate (i.e., 0.43 and 0.46, respectively). Interrater reliability was relatively low because LR was stricter in adhering to the defined inclusion and exclusion criteria, while HM and LHL were more lenient in the screening procedure. See Figure 3 for the results of the study selection.

Based on the SAQOR criteria, six studies were assessed as high quality ( Breen et al., 2021; Downar et al., 2022; Eisma & Tamminga, 2022; Gang et al., 2022; Tang et al., 2021; Yaghoubi et al., 2021), and six studies as moderate quality ( Bovero et al., 2022; Breen et al., 2022a, 2022b; Chen & Tang, 2021; Eisma et al., 2021; Tang & Xiang, 2021). In Table 1 more detailed information is presented with respect to the quality assessment of the twelve studies.

In Table 2 an overview of the characteristics of the studies is provided. All twelve studies used a cross-sectional design. Sample sizes ranged from 30 to 422 people bereaved due to COVID-19 (M = 224.08; SD = 162.33). The studies used different definitions of PG and varied in instruments used to assess PG (see Table 2 for more detailed information).

Three studies relied on the same sample (n = 422) of Chinese people who lost a loved one due to COVID-19 on average 5.10 months (SD = 1.72) ago ( Chen & Tang, 2021; Tang et al., 2021; Tang & Xiang, 2021). Three studies were conducted among people from the United States ( Breen et al., 2021, 2022a; Gang et al., 2022). Breen et al. (2021) included COVID-19 related bereaved adults (n = 307) in which most deaths (95%) occurred < 6 months ago. The study by Breen et al. (2022a) was based on 409 people bereaved due to COVID-19 (n = 206), natural losses (n = 111), and unnatural losses (n = 92) during the pandemic with a median time since loss of 4 months. Gang et al. (2022) included 1470 people bereaved due to COVID-19 (n = 118) and from other causes of death (n = 1352) (e.g., dementia, cancer) on average 24.50 months (SD = 84.40) ago. Two studies ( Eisma et al., 2021; n = 1441; Eisma & Tamminga, 2022; n = 1266) were executed in the Netherlands including people bereaved due to COVID-19 (n = 49; n = 99), natural losses (n = 1182; n = 1006), and unnatural losses (n = 210; n = 161). The mean time since loss in COVID-19 bereaved people was 1.95 (SD = 1.17) months ( Eisma et al., 2021) and 4.31 (SD = 3.50) months ( Eisma & Tamminga, 2022).

Another study, executed in Iran, focused on 400 COVID-19 bereaved people in which most deaths (64%) occurred more than 4 months ago ( Yaghoubi et al., 2021). In Italy, Bovero et al. (2022) studied a small sample of 31 bereaved caregivers who lost a family member through COVID-19 at least 6 months ago. The authors did not provide information about the time since death. Another study focused on a UK sample (n = 183), all bereaved through COVID-19 with most deaths (56%) occurring over 6 months ago ( Breen et al., 2022b). Lastly, people (n = 121) bereaved through COVID-19 (n = 30), non-COVID-19 illness during the pandemic (n = 46) and pre-pandemic (n = 45) at least 6 months ago were the subjects of a study in Canada ( Downar et al., 2022). No details were provided about the time since death.

All studies reported prevalence rates and/or symptom-levels of negative psychological outcomes of bereavement through COVID-19. Since different instruments were used to assess psychological outcomes, prevalence rates and/or severity levels of negative psychological symptoms are described separately for every study (see Table 2 for the main findings of the studies). In eight out of twelve studies, prevalence rates were reported of negative psychological symptoms of people who lost a loved one due to COVID-19. Most studies reported on (acute) PG (k = 4), followed by pandemic grief (k = 3), depression (k = 3), anxiety (k = 3), functional impairment (k = 3), PTSD (k = 2), and grief experience (k =1). Details regarding severity levels of negative psychological outcomes of COVID-19 bereaved people are depicted in Table 2, but are not further explained in text.

PG: In two studies acute PG reactions, namely within six months post-loss, were examined. Tang et al. (2021) showed self-rated prevalence rates of 49% for acute PG as per PGD ICD-11 criteria. Another study demonstrated self-reported prevalence rates of 29% as per PCBD DSM-5 criteria ( Tang & Xiang, 2021). Three studies reported on PG rates at least six months post-loss. Two of these studies demonstrated self-rated prevalence rates of 48% and 30% for PG in line with PG criteria as per Prigerson et al. (1995) ( Bovero et al., 2022; Downar et al., 2022, respectively). A third study included self-rated prevalence rates of 38% for PG according to PGD ICD-11 criteria ( Tang & Xiang, 2021).

Pandemic grief: Three studies examined grief reactions following a COVID-19 loss based on the Pandemic Grief Scale ( Lee & Neimeyer, 2022) ( Breen et al., 2021, 2022a, 2022b). Self-rated prevalence rates of 66%, 69%, and 40% for pandemic grief symptoms were reported, respectively.

Depression: One study showed self-rated prevalence rates of 62% for depression ( Breen et al., 2022b). Breen et al. (2021) reported self-rated prevalence rates of 74% for depression. A third study indicated self-rated prevalence rates of 70% for depression ( Tang et al., 2021).

Anxiety: Two studies reported self-rated prevalence rates of 70% for anxiety ( Breen et al., 2021; Tang et al., 2021). A third study demonstrated self-reported prevalence rates of 59% for anxiety ( Breen et al., 2022b).

Functional impairment: Self-rated prevalence rates of 63% were found for functional impairment in Breen et al. (2021). Another study demonstrated self-rated prevalence rates of 75% for functional impairment ( Breen et al., 2022a). A third study indicated self-reported prevalence rates of 56% for functional impairment ( Breen et al., 2022b).

PTSD: One study showed self-reported prevalence rates of 22% for PTSD ( Tang et al., 2021), whereas a second study indicated self-rated prevalence rates of 83% for PTSD ( Breen et al., 2022b).

Grief experience: Yaghoubi et al. (2021) showed that 45% reported interview-based moderate to high grief experience levels, based on the Grief Experience Questionnaire ( Mehdipour et al., 2009).

Merely one out of twelve studies included indicators of positive psychological outcomes of bereavement due to COVID-19 ( Chen & Tang, 2021). This study reported on posttraumatic growth levels (PTG; M = 66.81; SD = 16.71) based on the Posttraumatic Growth Inventory ( Tedeschi & Calhoun, 1996) (see Table 2). Following an earlier study that used a cutoff of ≤ 45 as none to low PTG levels, and > 45 representing moderate to very high PTG levels ( Mazor et al., 2016), PTG levels in Chen and Tang (2021) can be considered moderately high. The objective of the study was to examine profiles of PTG, PG, and PTSD, in COVID-19 bereaved people using latent profile analysis. Prevalence rates of positive psychological outcomes of bereavement due to COVID-19 were not reported.

Eleven out of twelve studies reported on correlates of negative psychological outcomes of bereavement due to COVID-19. Most studies reported on COVID-19 stressors (k = 6) and concurrent psychopathology levels (k = 6), followed by cause of death (k = 5), gender (k = 4), time since death (k = 4), kinship and closeness to the deceased (k = 4), age (k = 3), expectedness of the death (k = 2), and educational level (k = 1). No studies reported on correlates of positive psychological outcomes due to COVID-19 bereavement.

COVID-19 stressors: Differences in symptom-levels and/or probable caseness of negative psychological outcomes according to COVID-19 stressors were evaluated in six studies ( Bovero et al., 2022; Breen et al., 2021, 2022a, 2022b; Downar et al., 2022; Eisma & Tamminga, 2022). Bovero et al. (2022) found that being single or widowed during lockdown was associated with probable PG caseness. Funeral attendance and more social support were associated with lower probable PG caseness. Breen et al.’s (2021) study showed that being diagnosed with COVID-19 was associated with higher functional impairment levels. Two studies showed that COVID-19 stressors (such as feeling upset that the deceased was not given a proper funeral or memorial service or feeling alone in the grieving process because of social distancing policies) were associated with elevated pandemic grief, functional impairment ( Breen et al., 2022a), depression, anxiety, and PTSD symptom-levels ( Breen et al., 2022b). Another study revealed that the inability to say goodbye appropriately explained differences in higher PG levels between people bereaved by COVID-19 and natural causes ( Eisma & Tamminga, 2022). Lastly, Downar et al. (2022) reported that physical presence in the final 48 hours of life of the deceased and intubation of the deceased were not associated with PG symptom-levels.

Concurrent psychopathology levels: Six studies examined the association between concurrent psychopathology levels and probable caseness and/or symptom-levels of negative psychological outcomes ( Bovero et al., 2022; Breen et al., 2021, 2022a, 2022b; Tang et al., 2021; Tang & Xiang, 2021). Concurrent depression levels were positively correlated with probable PG caseness in a study by Bovero et al. (2022). On the contrary, they found that attachment style, anxiety, and stress levels were not associated with probable PG caseness. Breen et al. (2021) reported that functional impairment increased by 25% for elevated pandemic grief levels, and 13% for elevated PTSD levels. Two studies showed that disrupted meaning partially explained the association between COVID-19 stressors with pandemic grief, functional impairment ( Breen et al., 2022a), depression, anxiety, and PTSD symptom-levels ( Breen et al., 2022b). Lastly, two studies found that feeling traumatized by the loss (measured by one item: “How traumatized do you feel by the loss?” on a five-point Likert scale) was associated with elevated (acute) PG ( Tang & Xiang, 2021), anxiety, and depression levels ( Tang et al., 2021).

Cause of death: The difference in symptom-levels and/or probable caseness of negative psychological outcomes according to cause of death was examined in five studies ( Breen et al., 2022a; Downar et al., 2022; Eisma et al., 2021; Eisma & Tamminga, 2022; Gang et al., 2022). Three studies found that COVID-19 bereavement was associated with higher acute PG symptom-levels ( Eisma et al., 2021; Eisma & Tamminga, 2022) and probable PG caseness ( Gang et al., 2022) than natural bereavement, but not to unnatural bereavement. However, another study found no differences between people bereaved from COVID-19, natural, or unnatural causes on pandemic grief and functional impairment levels ( Breen et al., 2022a). Lastly, no differences in cause of death (COVID-19, non-COVID-19 illness, pre-COVID-19 illness) and PG levels were found by Downar et al. (2022).

Gender: Four studies examined the association between gender and negative psychological symptom-levels ( Breen et al., 2021, 2022a, 2022b; Yaghoubi et al., 2021). In one study, males showed higher pandemic grief symptoms than females ( Breen et al., 2022a). Yet, another study found that females reported higher grief experience levels compared to males ( Yaghoubi et al., 2021). Two studies demonstrated that gender was unrelated to functional impairment ( Breen et al., 2021), pandemic grief, depression, anxiety, and PTSD levels ( Breen et al., 2022b).

Time since death: In four samples, the association between symptom-levels of negative psychological outcomes and time since death was studied ( Breen et al., 2021, 2022a, 2022b; Tang et al., 2021). One study found that time since death was negatively correlated with levels of pandemic grief, depression, anxiety, functional impairment, and PTSD ( Breen et al., 2022b). Time since death was also negatively associated with anxiety and depression symptoms, but not with acute PG or PTSD, in a study by Tang et al. (2021). Another study found that time since death was positively correlated with pandemic grief and functional impairment levels ( Breen et al., 2022a). Breen et al. (2021) reported that functional impairment levels were not correlated with time since death.

Kinship and closeness to the deceased: The difference in negative psychological outcomes according to type of kinship and closeness to the deceased was examined in four studies ( Breen et al., 2021; Downar et al., 2022; Tang et al., 2021; Tang & Xiang, 2021). The loss of a spouse or immediate family member was significantly associated with higher functional impairment levels compared to other losses (i.e., an extended family member, acquaintance, close friend, and other) ( Breen et al., 2021). Another study showed that the death of a spouse or child (vs. other relationship) related to higher symptoms of acute PG, anxiety, and depression ( Tang et al., 2021). They also found that subjective closeness to the deceased (assessed with a single item on a 5-point Likert scale ranging from 1 (not at all) to 5 (very much)) was positively correlated with (acute) PG and PTSD levels. Tang & Xiang (2021) demonstrated that losing a spouse, child or (grand) parent (vs. other relationship) and more subjective closeness to the deceased were associated with elevated acute PG symptoms. However, Downar et al. (2022) found that kinship to the deceased was not correlated with severity of PG.

Age: The association between age of the participant and negative psychological symptom-levels was evaluated in three studies ( Breen et al., 2021, 2022a, 2022b). One study reported that age was negatively associated with PTSD symptom-levels, but not with pandemic grief, anxiety or depression levels ( Breen et al., 2022b). Two studies showed that age was unrelated to functional impairment ( Breen et al., 2021) and pandemic grief levels ( Breen et al., 2022a).

Expectedness of the death: The association between negative psychological symptom-levels and expectedness of the death was assessed twice ( Eisma et al., 2021; Eisma & Tamminga, 2022). In both studies it was found that expectedness of the death was associated with differences in acute PG levels between people bereaved by COVID-19 and natural causes.

Educational level: The association between negative psychological symptoms and educational level was examined in one study ( Yaghoubi et al., 2021). They showed that primary education (vs. > primary education) was associated with elevated grief experience levels.

Summary of correlational findings: Studies showed that a closer relationship to the deceased was associated with elevated psychological symptom-levels (except for Downar et al., 2022). Furthermore, findings suggest that unexpectedness of the death was related to higher acute PG levels. Findings indicated that being exposed to various COVID-19 stressors is associated with elevated psychological symptoms (except for Downar et al., 2022). Based on one study, lower educational level was associated with elevated grief levels. No clear inferences can be made regarding the associations between psychological symptoms and gender, age, time since death, cause of death, or concurrent psychopathology levels.