Bereavement interventions to support informal caregivers in the intensive care unit: a systematic review

We searched MEDLINE, EMBASE, CINAHL and PsycINFO from inception to October 01, 2020. A medical librarian (D.L.L.) assisted with the development, piloting, and execution of searches (Supplemental Table 2). No language or date restrictions were applied. Reference lists of included papers were reviewed to identify potentially missed studies.

Two reviewers (S.J.M. and T.G.P.) independently evaluated all records for eligibility in two stages. In the title and abstract stage, any record selected by either reviewer as meeting one (or more) eligibility criteria progressed to the full-text review stage. Studies were eligible for inclusion if both reviewers agreed that the study met all eligibility criteria following review of the full-text. Disagreements were resolved through consensus with another author (K.W.).

We included quantitative, experimental studies reporting randomized controlled trials of bereavement interventions to support informal caregivers of adult patients who died in ICU. We excluded interventions for health care professionals and interventions that were conducted prior to patient death (i.e., at end-of-life). We included studies where the intervention was performed outside ICU (e.g., home follow-up). For the purposes of our review, we defined ICU bereavement interventions as services healthcare professionals provide or coordinate for informal caregivers of critically ill patients after patient death (including brain death) in the ICU [16, 17]. We defined an informal caregiver as any informal (i.e., non-clinical) person who regularly provides patient support and is in some way implicated in patient care or directly affected by patient health (e.g., family, friend) [18], and critically ill patients as any persons currently or previously admitted to ICU [18]. In addition to studies that reported on patient death, we included studies where brain death was considered as patient death [19], since previous work indicates informal caregivers accept brain death as patient death [20] and caregiver grieving processes are similar [21]. The trials investigated bereavement interventions for informal caregivers of adult patients (> 17 years) that applied at least one (or more) support category from the UK National Institute for Health and Clinical Excellence [22, 23] three-tiered model of bereavement support, that includes: (1) Universal information provided to all bereaved; (2) Selected or targeted non-specialist support provided to those with a medium level of need who are at-risk of developing complex needs; and/or (3) Indicated professional specialist interventions provided to those with a high level of complex needs. Any one bereavement intervention could have applied multiple categories (i.e., types) of support, but needed to report at least category to be included in our review. Finally, to be eligible for inclusion, we required that the interventions reported on at least one of 21 caregiver-level outcomes within at least one of nine categories from the core set of outcomes for evaluating bereavement support for adult caregivers in adult care settings [13] (Supplemental Table 3). References were managed in Endnote X9 (Clarivate Analytics, Philadelphia, PA, USA).

Two reviewers (S.J.M., K.W.) used structured forms developed by the study team to extract information independently and in duplicate for each included study. Information on document characteristics (e.g., year of publication, geographic location), study characteristics (e.g., setting, sites), patient and caregiver characteristics (e.g., age, relationship), intervention characteristics (e.g., type of support and target population level of need, follow-up), core outcomes (e.g., negative mental and emotional state, participation in work and/or other regular activities), statistical significance (e.g., p-values, measures of variance), and authors’ conclusions were collected. Risk of bias for objective (i.e., measurement-based) outcomes was independently assessed by two reviewers (S.J.M. and K.W.) using the Cochrane Collaboration’s tool to rate studies at low, high or unclear risk of bias [24]. We (S.J.M. and K.W.) independently assessed quality of core outcomes using the BMJ Best Practice GRADE of Evidence Tool [25].

A narrative synthesis was conducted for all trials. We mapped primary and secondary outcomes reported in the included studies to the standardized core outcomes set for bereavement interventions [13] to help guide future research in ICU bereavement care [26]. To account for variability in the timing of study end-points, we used common clinically relevant follow-up periods of 1- to 3-month and 4- to 6-month follow-up. To facilitate comparison among the different instruments used to evaluate core outcomes, standardized mean differences (SMDs) with corresponding 95% confidence intervals (CIs) were calculated using a Hedges adjusted g estimator to correct for small sample bias [24]. For studies that reported medians or proportions, we contacted the primary author to obtain the corresponding mean and standard deviation (SD). For studies in which authors did not provide additional data, we estimated the mean and SD using validated estimations [27]. If necessary, individual study results were corrected for directionality such that higher coping scores represented better ability to cope, and lower wellbeing scores indicated worse wellbeing. Heterogeneity across a small number of trials precluded pooling SMDs in meta-analysis.

Searches identified 1960 unique records, of which 14 were potentially relevant based on initial title and abstract screening (Fig. 1). Following full-text review, three of the 14 articles were included, describing three unique adult ICU bereavement interventions for 296 informal caregivers from 31 ICUs across 22 hospitals in France [28] and 4 in the United States [29, 30] (Table 1). Reasons for exclusion of 11 articles reviewed in full-text are provided in Supplemental Table 3. One trial conducted the ICU bereavement intervention after patient death with a single follow-up time point within 1- to 3-months [30]. The other two trials conducted interventions at 2 weeks following patient death with 1- and 6-month follow-up [28], and at 4 weeks following patient death with 3- and 6-month follow-up [29]. One trial integrated all three categories of support (i.e., universal information, selected or targeted non-specialist support, and an indicated professional specialist intervention) within their bereavement intervention [30]. The other two trials incorporated either selected or targeted non-specialist support with an indicated professional specialist intervention [28], or an indicated professional specialist intervention alone [29]. Included studies assessed in sum four (of nine) core outcomes categories [14 of 21 individual outcomes] that included: (1) negative and overwhelming grief; (3) understanding, accepting and finding meaning in grief; and (3) accessing appropriate support [relating to caregiver ability to cope with grief] (Supplemental Tables 5 and 6), as well as (4) participation in work and/or other regular activities and (5) negative mental and emotional state [relating to caregiver quality of life and mental wellbeing] (Supplemental Tables 7 and 8). All trials were judged as having low risk of bias (Table 2). Overall, the quantity of evidence on bereavement interventions for informal caregivers in adult ICUs is low (Table 3).

Table 2 provides a summary of findings among RCTs of ICU bereavement interventions whilst SDMs with corresponding interpretation of effect of intervention are in Supplemental Tables 5, 6, 7, and 8. Among the three included trials, no significant effect of intervention was determined for negative and overwhelming grief (assessed by the Decision Regret Scale in one study); accessing appropriate support (assessed by yes/no questions in two studies); and participation in work and/or other regular activities (assessed using the General Health Questionnaire-12 in one study). Mixed results (both negative and non-significant) were determined for understanding, accepting and finding meaning in grief (assessed by the Inventory of Complicated Grief in two studies) and negative mental and emotional state (assessed by the Impact of Event Scale; Impact of Event Scale-Revised; Hospital Anxiety and Depression-Anxiety; Hospital Anxiety and Depression-Depression; Hospital Anxiety and Depression-Total; Patient Health Questionnaire-9; and Prevalence of Post-Intensive Care Syndrome-Family in all three studies).

Two trials integrated multiple categories of bereavement support [28, 30] proposed by the UK National Institute for Health and Clinical Excellence [22] to be made available according to the level of need. Effect of family presence compared to absence at brain death evaluation investigated by Tawil et al. [30] across four ICUs within a single academic hospital, incorporated universal information [for all levels of need], selected or targeted non-specialist support [for medium level of need], and an indicated professional specialist intervention [for high and complex level of need] for 38 informal caregivers of adult patients in ICU whom the treating intensivist suspected had suffered brain death. Multiple caregivers per patient joined the evaluating physician at the patient’s bedside to observe the brain death evaluation. Caregivers were accompanied by a chaperone who explained the process and were available to answer questions during the evaluation. Caregivers randomized to be absent during the evaluation waited in an adjacent room, accompanied by a chaperone. After the brain death evaluation was complete, informal caregivers were notified of the results and given an opportunity to ask questions. All informal caregivers were sent infographics and hard copies of assessment surveys, then telephoned by a trained research nurse who administered the surveys and recorded responses within 1- to 3-months after patient death. Investigators found no significant improvement in emotional or psychological distress, or participation in work and/or other regular activities (e.g., daily tasks, social activities) up to 3-months. Investigators concluded that informal caregiver presence during brain death evaluation is feasible and safe.

Kentish-Barnes et al. [28] conducted a randomized parallel-group trial across 22 ICUs within 22 hospitals (11 academic and 11 non-academic). They facilitated two categories of bereavement support (i.e., selected or targeted non-specialist support [for medium level of need] with an indicated professional specialist intervention [for high and complex level of need]) to contribute evidence on two core outcomes (i.e., understanding, accepting and finding meaning in grief and negative mental and emotional state). They prepared a hand-written condolence letter for 123 informal caregivers sent within 15 days after patient death that included: (1) recognition of the death; (2) name of the deceased; (3) mention of a personal impression; (4) recognition of the informal caregiver; (5) offer to help; and (6) expression of sympathy. Their 1- and 6-month telephone follow-up assessments were conducted by any one of a psychologist, sociologist or research nurse. Condolence letters had no effect on any core outcome at 1-month follow-up. At 6-month follow-up they reported that among informal caregivers of patients who died in ICU, a condolence letter failed to alleviate grief symptoms and increased depression and and post-traumatic stress disorder-related symptoms.

A pilot single-blind trial across five ICUs within three hospitals (one academic) incorporated one category of bereavement support providing one professional specialist intervention for a high and complex level of need [29]. The storytelling intervention delivered by Barnato et al. [29] to 18 caregivers via home visit or telephone approximately 4 weeks following patient death consisted of: (1) non-judgmental elicitation of the story of the events leading up to the patient’s ICU admission; (2) description of the ICU experience and decision process; and (3) summary of the aftermath of the patient’s death. Follow-up was performed at 3- and 6-months, with rationale to assess selection bias rather than to provide additional opportunity for support. The authors reported that their storytelling intervention met all a priori feasibility, tolerability and acceptability targets and there were no significant improvements in anxiety, depression, post-traumatic stress or complicated grief at any time-point of follow-up. The authors noted that their sample size was too small to make any inferences about the effect of the storytelling intervention on individual psychological symptoms.

No trial reported on the effect of an ICU bereavement intervention on four of nine core outcomes categories at any time of follow-up that included: (1) communication and connectedness; (3) finding balance between grief and life going forwards; and (4) relationships and social functioning; and (5) positive mental wellbeing.

Though we used a broad definition to identify trials that employed bereavement interventions consistent with any one of the support categories from the three-tiered model of bereavement support, [22] bereavement interventions in adult ICU have been evaluated by few studies. In the absence of more extensive global research on this topic, findings on bereavement interventions need to be considered within the context of the individual studies in which they were performed. For example, in France, physicians have final authority regarding decisions to forgo life-sustaining treatments [2, 34], meaning that effectiveness of bereavement interventions may vary according to sociocultural circumstances, including the degree of family involvement and sense of responsibility in shared decision making [35, 36].

This review has several limitations and our results should be interpreted cautiously. First, we excluded quasi-experimental, observational and qualitative studies, many of which have suggested or demonstrated benefits of bereavement care and provided insights from different perspectives [37‐39]. Second, this review did not include conventional palliative care interventions (neither early nor at end-of-life) despite prevailing views that palliative care interventions are appropriate throughout the critical illness trajectory [14]. Third, quantitative results from these studies should be interpreted cautiously given that they were underpowered; furthermore, the design of these three RCTs precluded a quantitative meta-analysis. Fourth, we used the Cochrane Risk of Bias tool to perform risk of bias assessments, which is not designed to consider multi-component behavior change interventions [24]. Fifth, though we did not employ language or date restrictions, our review is restricted to countries represented by the included studies and should not be taken to represent a generalizable, global report on the state of bereavement interventions for adult ICU caregivers. Sixth, we did not include preliminary findings from conference proceedings or planned interventions from protocols [40, 41].

Bereavement interventions in adult ICU remains an active area of research with many gaps in our understanding. Given the complexities and inconsistencies in limited bereavement interventions, we were unable to determine effective components of bereavement interventions and how specific categories of support targeted to the caregivers’ level of need might affect their ability to cope. Single-component bereavement interventions offer a potentially resource-efficient means of preparing informal caregivers to cope with grief and sustain mental wellbeing following death of a loved one in ICU [42, 43]. However, it is possible that multi-component bereavement interventions bundles might be more effective [44].

From an intervention perspective, future research is needed to identify efficacious components of bereavement care. Informal caregivers may require different bereavement intervention components at different timepoints along the critically illness trajectory. Future trials should consider active comparisons of different components of ICU bereavement care strategies. From an outcomes perspective, to consolidate robust evidence corresponding to core outcomes, future studies should use standardized and validated measures appropriate for informal caregivers at clinically relevant follow-up time points. Four of nine categories from the core outcomes set for evaluating bereavement interventions in palliative care remain to be investigated in adult ICUs, including three coping-related outcomes and two mental wellbeing outcomes. From a process evaluation perspective, future studies should describe intervention fidelity, dose, and reach, to ensure interventions are consistently provided as intended and reliably adhered to as required for optimized impact of bereavement care.