Abstract
Objective: It is uncertain whether psychological distress in the family members of patients who die during an intensive care unit (ICU) admission may be improved by bereavement interventions. In this trial, relatives’ symptoms of anxiety and depression after 6 months were measured when allocated to three commonly used bereavement follow-up strategies.
Design: Single-centre, randomised, three parallel-group trial.
Setting: A tertiary ICU in Australia.
Participants: Relatives of patients who died in the ICU.
Interventions: Relatives received bereavement follow-up 4 weeks after the death using a condolence letter, short telephone call or no contact.
Main outcome measures: The primary outcome was the total Hospital Anxiety and Depression Scale (HADS-T) score. Secondary outcomes estimated anxiety, depression, complicated grief, post-traumatic stress, and satisfaction with ICU care.
Results: Seventy-one relatives participated (24 had no contact, 19 were contacted by letter and 28 by telephone 4 weeks after the death). The mean HADS-T score for no contact was 16.1 (95% CI, 12.4-19.8). Receipt of a letter was associated with a mean HADS-T increase of 1.4 (4.0 decrease to 6.8 increase), and a condolence call was accompanied by a mean decrease of 1.6 (6.6 decrease to 3.4 increase; P > 0.5). Non-significant differences were observed for all secondary outcomes.
Conclusions: Anxiety and depression at 6 months in the relatives of patients who died in the ICU was not meaningfully alleviated by receipt of either a condolence letter or telephone call.
Trial registration: Australia New Zealand Clinical Trials Registry (ACTRN12619000917134).
Family members of critically ill patients who die during intensive care unit (ICU) treatment may be at elevated risk of anxiety, depression, complicated grief, and post-traumatic stress-like symptoms compared with the relatives of patients dying outside of the critical care setting.1, 2, 3, 4 Such psychological morbidity may be associated with decreased quality of life, functional impairment, and possibly greater mortality within the group of affected relatives.5, 6 Accordingly, bereavement services have been proposed7, 8, 9, 10 to support the emotional needs of family members. Interventions such as viewing of the body, mementoes, storytelling, ICU diaries, sympathy letters, phone calls and follow-up meetings have all been reported in the literature;11, 12, 13, 14, 15 however, there is no evidence from randomised clinical trials to support any intervention.8, 10, 11, 16
Given this, a randomised trial was conducted to gather estimates of the effect of family-centred bereavement services, particularly to evaluate the impact of a telephone call from an ICU staff member. The primary hypothesis was that some bereavement support (either a semi-structured, short telephone call or a condolence letter) compared with no communication would alleviate symptoms of anxiety and depression at 6 months in relatives of patients who have died in the ICU.
Participants and methods
Study design and oversight
This was a single-centre, randomised, three parallel-group trial within a sample of patient relatives identified at the Royal Melbourne Hospital, Melbourne, Australia, which has a 42-bed, university-associated, adult ICU supporting several state referral services, including trauma and oncology. The protocol was approved by the Melbourne Health’s Human Research Ethics Committee before commencement of the study (HREC Ref. No. HREC/52855/MH-2019; HREC approval date, 23 May 2019), and the trial was registered before enrolment (Australia New Zealand Clinical Trials Registry: ACTRN12619000917134).
Informed verbal consent was obtained directly from all participants before any data collection. The contacted participant was first asked to provide informed verbal consent to trial questionnaire involvement. Apart from full agreement or complete withdrawal, participants could decline answers regarding themselves but allow the researchers to use patient and participant descriptive data already collected to that 6-month contact point.
Study participants
ICU patient deaths between 5 August 2019 and 24 March 2020 were identified using an automated notification system if aged at least 18 years, were not pregnant, and were admitted to the ICU for at least 24 hours. Under applicable Australian state law, a medical treatment decision maker (MTDM) becomes the patient representative when a patient loses capacity to decide about clinical care for themselves (Online Appendix, figure S1).17 Each patient’s MTDM was eligible for trial participation if, at the time of the patient’s death, the MTDM was also aged at least 18 years and had attended at least one ICU-conducted meeting where the patient’s clinical information and prognosis had been discussed. The exclusion criteria were previous inclusion in the study, multiple relatives in the trial ICU, anticipated unfamiliarity with the English language, and anticipated lack of mail or telephone access at 6 months following the ICU death. Baseline clinical and demographic data for each patient and participating MTDM were collected and managed using REDCap electronic data capture tools.18, 19 These data included demographic characteristics, relationship to the patient, the number of pre-mortem information meetings attended in the ICU, and whether the MTDM was present when the patient died.
Randomisation
Participants were randomly assigned from within REDCap in a 1:1:1 allocation to the control group (no post mortem contact from ICU staff) or one of two intervention groups (condolence letter or telephone call), using a sequence with variable block sizes generated within the Stata ralloc command (Stata version 15.1). Selection bias was minimised by maintaining allocation concealment until after randomisation.
Interventions
Trial interventions commenced 4 weeks after each patient’s death in the ICU. Participants allocated to the control group were not provided with any formal bereavement follow-up from the ICU. Study participants allocated to the condolence letter group were sent a short generic letter that was personalised with the participant’s and patient’s names but otherwise followed the ICU’s pre-existing standard bereavement letter template. The key components of the letter were an expression of sympathy and contact details for an ICU staff member if the recipient wished to initiate further contact. The letter and a pamphlet on bereavement were sent to the postal address provided by the recipient during the patient’s admission (Online Appendix, figure S2). The condolence telephone call group received a semistructured call from one of two researchers who were ICU clinical nurse consultants. The structured component of the call contained an expression of sympathy, an enquiry into any concerns or questions the relative might have, and the offer to organise further follow-up (Online Appendix, figure S3). No time limit was placed on the duration of the call; however, if the key components and participant responses could not be completed within 10 minutes, it was recommended that a follow-up meeting should be organised. Participants were considered uncontactable if there was no response to three voice mail messages or if seven phone calls that were each allowed to ring out at different times of the day remained unanswered.
Outcome data collection
An experienced ICU doctor blinded to the study group allocation contacted all participants by telephone 6 months after each patient’s death. The contacted participant was first asked to provide informed verbal consent to trial questionnaire involvement. Apart from full agreement or complete withdrawal, participants could decline to provide answers regarding themselves but allow the researchers to use patient and participant descriptive data already collected to that 6-month contact point.
Study outcomes
Outcome assessment comprised four separate questionnaires that aimed to measure aspects of each participant’s anxiety, depression, post-traumatic stress disorder (PTSD), and complicated grief. These were separately quantified, as was their satisfaction with the patient’s ICU care. The trial primary outcome was the total score of the Hospital Anxiety and Depression Scale (HADS-T) in participants, assessed at 6 months following each patient’s death.1, 2, 20, 21 As previously described, a HADS-T threshold of ≥ 13 was deemed to identify clinically important symptoms of anxiety and depression.20, 21, 22 The main secondary outcome used HADS-T assessed within a multivariable linear model. Other secondary trial outcomes assessed:
-
•
anxiety and depression, identified respectively by a HADS anxiety (HADS-A) and depression (HADS-D) subscale score ≥ 8;20, 21
-
•
PTSD, assessed using the 22-item Impact of Event Scale – Revised (IES-R) tool and represented by an IES-R ≥ 26;2, 20, 23, 24
-
•
complicated grief, assessed using the 19-item Inventory of Complicated Grief (ICG) tool, with a threshold score ≥ 25;2, 20, 25 and
-
•
participants’ experience of dying and death in the ICU, assessed using the 15-item CAESAR questionnaire.20, 26
Other collected data included the count of attempted calls, the duration of calls with successful contact, the caller’s perception of the recipient’s response to the call (categorised as positive, neutral, uncertain and negative), whether any enquiries or requests were made by the participant, and whether any follow-up was arranged.
Statistical analyses
All data were first manually entered into the trial REDCap database. Selected data were exported to Excel (Microsoft Corporation, 2018) or Stata (release 16) for subsequent analyses.
Baseline patient demographic and clinical characteristics, and other data within categories including several secondary outcomes, were summarised with descriptive statistics, including mean with standard deviation (SD) or 95% confidence interval (CI), median with interquartile range (IQR) and count (proportion within relevant categories). Where appropriate, comparisons between groups used linear regression, one-way analysis of variance or Fisher exact tests. The primary outcome of the total HADS score for participating relatives was modelled as a normally distributed continuous random variable with comparison between groups using ordinary least squares linear regression. Unadjusted effect estimates, representing differences between intervention group means with accompanying 95% CIs for those differences, were derived from the linear model that included only the three-level factor variable representing the intervention, with the reference category being the control intervention. Subsequently, adjusted estimates of these between-group mean differences with 95% CIs were derived from a multivariable linear model including the intervention factor and purposefully selected covariates including patient and respondent characteristics. Standard errors for all linear model coefficients were derived using the bootstrap method with 1000 replications. Given the missing outcomes for some participants, a sensitivity analysis of the primary outcomes used a dataset expanded by multivariate imputation using the chained equations method, again with bootstrap standard error estimates using 1000 replicates.
Sample size
This trial was planned for a 12-month period, based on an anticipated count of 140 eligible deaths in the ICU. After 7 months of recruitment, it became apparent that the emerging coronavirus disease 2019 (COVID-19) pandemic would substantially alter ICU care and resources.27, 28 Additionally, changes in complicated grief have been reported following disasters.29, 30 Considering these factors, we ceased recruitment after 33 weeks on 24 March 2020. This coincided with the first wave of the COVID-19 pandemic in Melbourne, Australia.31
Results
During the study period, 167 patients died in the ICU. All eligibility criteria and no exclusion criteria were met by the MTDM of 109 patients, and 71 of these consented to participate in all questionnaires, thus forming the trial primary analysis cohort. A further 25 MTDMs limited their consent to the use of existing data without providing any participant anxiety and depression outcome information. The remaining 13 MTDMs declined any trial participation (Figure 1).
Figure 1.
CONSORT diagram
MTDM = medical treatment decision maker. Limited consent = the participant declined to participate in the study but consented to the use of descriptive data for the participant and patient. These data were used in a sensitivity analysis of the primary outcomes using a dataset expanded by multivariate imputation.
Patients and participants
Baseline characteristics of patients and their MTDM participants were broadly similar across the three trial intervention groups, with the least balance observed for the participant’s relationship to the patient (Table 1). There were no apparent large differences between the baseline characteristics of the 71 MTDMs who participated in the study and the 25 MTDMs who limited consent to existing data (Online Appendix, table S1).
Table 1.
Characteristics of patients and Participants
| Patient characteristics | No contact (n = 24) | Condolence letter (n = 19) | Telephone call (n = 28) | Total (n = 71) |
|---|---|---|---|---|
| Sex | ||||
| Male | 13 (54%) | 9 (47%) | 19 (68%) | 41 (58%) |
| Female | 11 (46%) | 10 (53%) | 9 (32%) | 30 (42%) |
| Age, mean (SD), years | 60 (17) | 64 (12) | 58 (15) | 62 (15) |
| ICU length of stay, days, mean (SD) | 6 (6) | 7 (11) | 4 (2) | 6 (8) |
| Patient category | ||||
| Non-operative | 18 (75%) | 18 (95%) | 26 (93%) | 62 (87%) |
| Emergency surgery | 5 (21%) | 1 (5%) | 2 (7%) | 8 (11%) |
| Elective surgery | 1 (4%) | 0 (0%) | 0 (0%) | 1 (1%) |
| Admission diagnosis* | ||||
| Cardiac arrest | 4 (17%) | 7 (39%) | 6 (21%) | 17 (23%) |
| Sepsis | 3 (13%) | 3 (17%) | 7 (25%) | 13 (17%) |
| Intracerebral haemorrhage | 3 (13%) | 4 (22%) | 7 (25%) | 14 (20%) |
| Trauma | 6 (25%) | 0 (0%) | 4 (14%) | 10 (14%) |
| Other | 8 (33%) | 5 (28%) | 4 (14%) | 17 (23%) |
| Comorbid conditions | ||||
| Diabetes | 7 (29%) | 3 (16%) | 5 (18%) | 15 (21%) |
| Chronic respiratory disease | 0 (0%) | 2 (11%) | 3 (11%) | 5 (7%) |
| Chronic renal disease | 2 (7%) | 0 (0%) | 1 (4%) | 3 (4%) |
| Immune disease | 5 (21%) | 1 (5%) | 6 (21%) | 12 (17%) |
| Immunosuppression | 1 (4%) | 2 (11%) | 4 (14%) | 7 (10%) |
| Lymphoma | 1 (4%) | 0 (0%) | 0 (0%) | 1 (1%) |
| Metastatic cancer | 3 (13%) | 1 (6%) | 1 (4%) | 5 (7%) |
| Leukaemia | 4 (17%) | 1 (6%) | 5 (18%) | 10 (7%) |
| APACHE III risk of death, mean (SD) | 0.55 (0.23) | 0.66 (0.23) | 0.49 (0.26) | 0.58 (0.24) |
| APACHE III score, mean (SD) | 90 (20) | 84 (22) | 105 (36) | 95 (29) |
| ICU interventions | ||||
| Mechanical ventilation | 23 (96%) | 16 (84%) | 26 (93%) | 65 (92%) |
| Vasoactive medication | 19 (79%) | 17 (90%) | 25 (89%) | 61 (86%) |
| Renal replacement therapy | 4 (17%) | 1 (5%) | 8 (29%) | 13 (18%) |
| Organ donation | 5 (21%) | 6 (32%) | 6 (21%) | 17 (24%) |
| Death referred to coroner | 12 (50%) | 6 (32%) | 6 (21%) | 24 (34%) |
| Participant characteristics | ||||
| Sex | ||||
| Male | 10 (42%) | 8 (42%) | 14 (50%) | 32 (45%) |
| Female | 13 (58%) | 11 (58%) | 14 (50%) | 39 (55%) |
| Age, years, mean (SD) | 58 (15) | 59 (11) | 53 (14) | 56 (14) |
| Relationship | ||||
| Spouse/partner | 15 (63%) | 5 (26%) | 14 (50%) | 34 (48%) |
| Parent | 2 (8%) | 5 (26%) | 0 (0%) | 7 (10%) |
| Child | 3 (13%) | 6 (32%) | 11 (39%) | 20 (28%) |
| Immediate family | 4 (17%) | 3 (16%) | 2 (7%) | 9 (13%) |
| Other | 0 (0%) | 0 (0%) | 1 (4%) | 1 (1%) |
| Family meeting count, mean (SD) | 3 (1.7) | 2.7 (1.0) | 2.9 (2.1) | 2.9 (1.7) |
| Present at death | 18 (75%) | 14 (74%) | 22 (79%) | 54 (76%) |
APACHE = Acute Physiology and Chronic Health Evaluation; ICU = intensive care unit; SD = standard deviation. Due to the Australian and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation (ANZICS CORE) Adult Patient Database (APD) data dictionary diagnosis hierarchy (https://www.anzics.com.au/wp-content/uploads/2021/03/ANZICS-APD-Dictionary.pdf), patients must be allocated an APACHE III-J diagnosis, with priority given to diagnoses selection in the following order: 1, cardiac arrest; 2, sepsis; and 3, trauma. Such clinical diagnoses may not correlate with allocation of APACHE III-J diagnoses, potentially obscuring some clinical diagnosis due to hierarchical preference of cardiac arrest, sepsis or trauma diagnostic codes.
Interventions
None of the participants allocated to the control group received any communication from the ICU. Within the 36 participants assigned to the letter group, 33 letters (92%) were sent out, of which one was subsequently returned unopened. The reason that the letter was not sent was clerical error for two participants and because no address was available for one participant. Of the 37 allocated to a condolence call, 35 participants (95%) received a phone call, which was judged to have been essentially completed in 33 cases. These successful telephone contacts were made after a mean of 1.5 attempts (SD, 0.80), with the mean completed condolence call duration being 4.7 minutes (SD, 2.7). The recipient’s response to the intervention, as perceived by the caller, was judged to be positive in 30 (91%), neutral in two (6%), uncertain in one (3%), and negative in none. Most participants, 22 (67%), made no enquiries during the call, and 26 (79%) did not request any further follow-up. A clinical question was posed by six (18%), two (6%) requested (and later received) a followup call from nursing staff, one (3%) requested (and later received) a follow-up call from medical staff, one (3%) was referred to another support service, one (3%) was referred to the hospital’s consumer liaison service, and three (9%) had some other form of referral or support organised.
Primary outcome
Within the cohort of 71 cases with complete data, the 6-month mean HADS-T score of relatives who received no contact was 12.4 (95% CI, 2.51–22.2). Receipt of a letter was associated with an estimated mean HADS-T score increase of 1.4 (4.0 decrease to 6.8 increase), and a condolence call was accompanied by a mean decrease of 1.6 (6.6 decrease to 3.4 increase; P > 0.5; Table 2, Table 3 and Figure 2, A). Although qualitatively similar effect estimates for the trial interventions were returned after adjustment within a multivariable linear model (Table 3), several other potential influences were also identified. Greater HADS-T scores were independently associated with survey answers from spouses or partners of the deceased patient and lower levels of satisfaction with ICU care (as represented by a lower CAESAR score), while greater age in both respondents and patients was somewhat associated with decreased HADS scores.
Table 2.
Unadjusted total Hospital Anxiety and Depression Scale (HADS-T) score within the primary analysis set of 71 respondents with full data
| Intervention | n | Mean (SD) | Median (IQR) | Range |
|---|---|---|---|---|
| No contact | 24 | 16.1 (8.9) | 17.5 (9.5-20) | 1-33 |
| Letter | 19 | 17.5 (8.5) | 16 (15-19) | 2-36 |
| Telephone | 28 | 14.5 (9.5) | 13.5 (7-20) | 2-37 |
| Overall | 71 | 15.8 (9.0) | 16 (9-20) | 1-37 |
IQR = interquartile range; SD = standard deviation. HADS-T scores ≥ 13 are suggestive of clinical anxiety and depression.22
Table 3.
Estimated incremental effects of covariates on the total Hospital Anxiety and Depression Scale (HADS-T) score within the primary analysis set of 71 respondents with full data
|
Unadjusted* |
Adjusted† |
|||
|---|---|---|---|---|
| Mean difference (95% CI) | P | Mean difference (95% CI) | P | |
| Intervention | ||||
| No contact | Reference* | Reference† | ||
| Letter | 1.4 (-4.0 to 6.8) | 0.60 | 4.9 (-1.1 to 11) | 0.11 |
| Telephone | -1.6 (-6.6 to 3.4) | 0.53 | -1.1 (-6.5 to 4.4) | 0.71 |
| Relationship (spouse or partner) | 4.9 (0.8-9.0) | 0.02 | 12 (6.8-18) | < 0.0005 |
| Patient age, years | ||||
| < 50 | Reference | Reference | ||
| 50-59 | -2.5 (-9.7 to 4.7) | 0.50 | -7.9 (-14 to -1.4) | 0.02 |
| 60-69 | -5.4 (-12 to 1.2) | 0.11 | -9.3 (-16 to -2.4) | 0.01 |
| ≥ 70 | -5.4 (-12 to 1.2) | 0.11 | -1.4 (-8.6 to 5.7) | 0.69 |
| Participant age, years | ||||
| < 50 | Reference | Reference | ||
| 50-59 | 2.6 (-3.0 to 8.2) | 0.37 | -0.6 (-7.5 to 6.4) | 0.87 |
| 60-69 | 0.0 (-5.4 to 5.3) | 1.0 | -1.7 (-6.7 to 3.4) | 0.52 |
| ≥ 70 | -2.5 (-8.8 to 3.7) | 0.43 | -10 (-19 to -1.3) | 0.03 |
| Patient female | -4.0 (-8.3 to 0.3) | 0.07 | -1.1 (-6.3 to 4.0) | 0.67 |
| Participant female | 5.0 (0.9-9.0) | 0.02 | -0.1 (-4.4 to 4.2) | 0.96 |
| Successful organ donor | 3.9 (-1.4 to 9.2) | 0.15 | 2.4 (-2.7 to 7.5) | 0.35 |
| Patient non-operative | 8.2 (1.7-15) | 0.01 | 0.9 (-7.7 to 9.4) | 0.84 |
| APACHE III score | ||||
| 0-74 | 1.4 (-4.3 to 7.0) | 0.64 | -3.0 (-9.0 to 3.1) | 0.34 |
| 75-89 | 0.0 (-5.7 to 5.8) | 1.0 | -3.1 (-10 to 3.8) | 0.38 |
| 90-109 | 5.7 (-0.6 to 12) | 0.08 | 2.2 (-3.4 to 7.8) | 0.44 |
| ≥ 110 | Reference | Reference | ||
| CAESAR total score | ||||
| 0-59 | 8.3 (1.9-15) | 0.01 | 7.5 (1.7-13) | 0.01 |
| 60-69 | 5.4 (1.4-9.4) | 0.01 | 5.0 (0.1-10) | 0.05 |
| 70-75 | Reference | Reference | ||
| Metastatic cancer, leukaemia, lymphoma | -0.2 (-4.9 to 4.5) | 0.93 | 0.8 (-4.0 to 5.6) | 0.74 |
APACHE = Acute Physiology and Chronic Health Evaluation; CAESAR = instrument designed specifically to assess the experience of relatives of patients who die in the intensive care unit.26 HADS-T scores ≥ 13 are suggestive of clinical anxiety and depression.22 Estimates from 71 cases with complete data within standard least squares linear models. The 95% CI were each derived from 1000 bootstrap replications.
Univariable estimates are unadjusted for the influence of any other variable in the table. The intercept for unadjusted univariable effect estimates for interventions = 16.1 (95% CI, 12.4–19.8), consistent with the results in Table 2.
Multivariable estimates are adjusted for the influence of all other variable in the table. The intercept for these adjusted multivariable effect estimates for interventions = 12.4 (95% CI, 12.5–22.2). The overall fit of the above multivariable adjusted linear model for HADS-T was supported as adequate using i) a regression specification-error test [Ramsey RESET indicating no omitted variables (F (3, 48) = 0.72; Prob > F = 0.54], and ii) a linktest for overall model misspecification (P > 0.20).
Figure 2.
Box and whisker plots demonstrating the median, mean (×), interquartile range, range and outliers of the primary and selected secondary endpoints — the different shaded groups represent the three types of contact with bereaved relatives
MTDM = medical treatment decision maker.
A sensitivity analysis of the primary outcome was performed using multiple imputation to generate 25 missing HADS-T and CAESAR outcomes for the participants with consent limited to existing demographic data and no anxiety and depression outcomes. Analysis of these pooled with the 71 cases with known outcomes returned estimates for the test interventions that were similar qualitatively to the primary unadjusted estimates (Table 4 and Online Appendix, tables S2–S5).
Table 4.
Estimated incremental effects of covariates on total hospital anxiety and depression scale (HADS-T) score after imputation of missing HADS-T and CAESAR satisfaction scores
|
Adjusted* |
||
|---|---|---|
| Mean difference (95% CI) | P | |
| Intervention | ||
| No contact | Reference** | |
| Letter | 2.7 (-3.3 to 8.6) | 0.37 |
| Telephone | -1.0 (-6.6 to 4.6) | 0.73 |
| Relationship (spouse or partner) | 7.7 (2.2-13) | 0.01 |
| Patient age, years | ||
| < 50 | Reference | |
| 50-59 | -5.7 (-14 to 2.8) | 0.18 |
| 60-69 | -5.1 (-12 to 2) | 0.16 |
| ≥ 70 | -2.9 (-11 to 4.8) | 0.45 |
| Participant age, years | ||
| < 50 | Reference | |
| 50-59 | -1.1 (-9.3 to 7.1) | 0.79 |
| 60-69 | -1.7 (-8.0 to 4.5) | 0.58 |
| ≥ 70 | -4.5 (-13 to 4.2) | 0.30 |
| Patient female | -1.9 (-7.3 to 3.5) | 0.49 |
| Participant female | 0.9 (-4.8 to 6.6) | 0.74 |
| Successful organ donor | 0.9 (-4.8 to 6.6) | 0.75 |
| Patient non-operative | 1.2 (-6.2 to 8.6) | 0.74 |
| APACHE III score | ||
| 0-74 | -2.9 (-10 to 4.2) | 0.41 |
| 75-89 | -3.3 (-10 to 3.4) | 0.32 |
| 90-109 | 1.1 (-5.9 to 8.2) | 0.75 |
| ≥ 110 | Reference | |
| CAESAR total score | ||
| 0-59 | 8.9 (1.7-16) | 0.02 |
| 60-69 | 5.3 (-0.9 to 12) | 0.09 |
| 70-75 | Reference | |
| Metastatic cancer, leukaemia, lymphoma | 0.6 (-5.2 to 6.5) | 0.83 |
APACHE = Acute Physiology and Chronic Health Evaluation; CAESAR = instrument designed specifically to assess the experience of relatives of patients who die in ICU.26 Estimates from 71 cases with complete data combined with 25 additional cases with imputed HADS-T and Caesar scores for participants who provided limited consent. Modelling using a standard least squares linear model, using the same terms as in Table 3. The 95% CI were each derived from 1000 bootstrap replications. The chained equations multivariable imputation process used truncated regression for HADS-T (a continuous variable with a restricted range from 0 to 42). Separately, the three levels of Caesar scores (0–59; 60–69; 70–75) were modelled within an ordered logistic regression.
Multivariable estimates are adjusted for the influence of all other variable in the table. The intercept for adjusted multivariable effect estimates for interventions = 13.9 (95% CI, 2.8–25).
Secondary outcomes
Secondary outcomes were also not markedly different between groups (Table 5 and Figure 2, B–F
Table 5.
Secondary outcomes according to randomised intervention
| No contact | Letter | Telephone | Total | P | |
|---|---|---|---|---|---|
| n | 24 | 19 | 28 | 71 | |
| HADS-A, mean (SD) | 8.3 (4.6) | 9.7 (4.5) | 7.9 (5.2) | 8.5 (4.8) | 0.43 |
| HADS-D, mean (SD) | 7.8 (5.0) | 7.8 (4.4) | 6.6 (4.9) | 7.4 (4.7) | 0.59 |
| IES-R total, mean (SD) | 36.3 (21.3) | 38.1 (19.5) | 33.9 (25.2) | 35.8 (22.3) | 0.81 |
| ICG total | 32 (16.9) | 31.9 (13.7) | 33 (23.0) | 32.4 (18.6) | 0.97 |
| ICG ≥ 25 | 15 (63%) | 13 (68%) | 17 (61%) | 45 (63%) | 0.91 |
| CAESAR total, mean (SD) | 66 (8.4) | 65 (8.9) | 64.5 (10) | 65.1 (9.1) | 0.83 |
HADS-A = Hospital Anxiety and Depression Scale, Anxiety Subscale; HADS-D = Hospital Anxiety and Depression Scale, Depression Subscale; IES-R = Impact of Events Scale – Revised; ICG = Inventory of Complicated Grief; SD = standard deviation. A binary threshold of ≥ 25 has been previously described.2 All P values derived from application of one-way analysis of variance, except for the use of the Fisher exact test for the binary variable ICG ≥ 25.
Discussion
This randomised trial evaluated two bereavement interventions32 at 4 weeks after deaths of patients in the ICU. Compared with no follow-up, early receipt of a condolence letter or telephone call from ICU staff was not associated with evidence of substantial improvement in the HADS-T score reported by relatives at 6 months after the death. The neutral effect persisted after adjustment for other possible influential factors within a multivariable linear model.
Studies using different methodologies have evaluated various bereavement interventions on persisting psychological distress in family members of patients who died in the ICU.8, 10, 11, 12 Despite widespread use, there are only three published randomised clinical trials that evaluate such interventions. Hitherto, only 366 relatives have been included in trials and there is no evidence of benefit.14, 20, 33 Furthermore, the largest randomised clinical trial with the greatest internal validity, which was conducted across 22 ICUs in France and included 208 relatives, found that a personalised condolence letter sent 2 weeks after the death was associated with greater rates of anxiety, depression and PTSD at 6 months.20 The present trial also does not support the utility of early condolence correspondence.
Similarly, the other intervention tested (ie, directly spoken communication) also did not offer benefit. Telephone condolences to relatives of ICU decedents has been assessed previously in two non-randomised single-centre studies. An observational study of 30 families has been conducted in which investigators offered printed bereavement material, a condolence card at one week, and additional psychological support via telephone 4–5 weeks later. The investigators reported a positive perception of the call but there was no measurable benefit on relative wellbeing.34 The other study was a cross-sectional pilot study that included 30 families who were offered printed bereavement material at the time of death, a condolence card at one week, a phone call at both 5 weeks and 6 months, and a written condolence letter at one year. The latter study reported a reduction in prolonged grief at 12 months associated with the telephone contact, but no effect on depression, anxiety, or PTSD.35 Of importance, the present trial, which was the first to randomly allocate participants to personalised telephone contact, did not identify substantial harm from the ICU bereavement intervention by telephone.
Our results are consistent with the existing literature from both ICU and non-ICU settings, showing a lack of efficacy for strategies implemented only after the patient’s death to prevent complicated grief.16, 36, 37, 38 However, interventions commenced by ICU staff before death, which may even be continued through the post-death phase, are of benefit.39
Limitations
This study has several limitations. There were missing data with an overall response rate of 65% at 6 months. However, this is comparable to previous work.40, 41 There were no obvious differences in baseline characteristics between participants who consented to fully participate in this trial, compared with those who consented only for use of demographic data. However, there were apparent differences in the proportions of consent across the control and two intervention groups, particularly in those allocated to receive a condolence letter, raising the possibility of systematic missing data within at least the primary analysis cohort. To address the limitation of missing data, further analysis was conducted with imputation. While imputation of missing outcome data did not alter meaningfully the observed effect estimates of condolence calls or letters, it remains possible that the results were biased by influences not included in the above analyses.
In addition, the size of the present single-centre trial was truncated by the local onset of the COVID-19 pandemic. Moreover, social restrictions enforced under COVID-19 public health regulations in Melbourne, Australia, may have increased anxiety in the population,42 and this may have influenced grieving in some of our participants.43
All measured markers of psychological distress were greater in our cohort compared with previous studies. This may be due to geo-social differences in grieving, extrinsic stressors from the pandemic, or other factors. Future studies should consider whether a prognostic enrichment strategy to include only individuals at high risk of complicated grief are studied.2
While 6 months is the interval required to diagnose complicated grief and is the outcome used in other trials,5, 25, 30 it should be recognised that other time periods and/or outcomes beyond those reported in the present study may well be important to families.
Implications for clinical practice
Family members of ICU decedents are reported to have expressed a desire for bereavement follow-up and many clinicians report interest in the provision of these interventions.44, 45 The current results, when viewed alongside existing data,20 do not indicate either strong efficacy or lack of harm from telephone bereavement contact from ICU clinical staff. Despite understandable enthusiasm, bereavement contact from ICU clinical staff — either by letter or telephone — requires further research to understand the impact of these efforts. Given the existing data, such efforts should focus on the pre-death phase of the patient’s illness.
Conclusion
The effect of bereavement contact from ICU staff — either a semi-structured, short telephone call or a condolence letter — remains uncertain with respect to symptoms of anxiety, depression, PTSD, or complicated grief at 6 months in relatives of patients who die in the ICU.
Acknowledgments
Acknowledgements:
We thank Lewis Hackenberger (Royal Melbourne Hospital, Intensive Care Unit, Data Report Lead) and Alison Wells (Royal Melbourne Hospital, Intensive Care Unit, Data Report Manager) for their investigation and assistance with patient data collection. We also thank Briannah Miles (Royal Melbourne Hospital, Intensive Care Unit, Research Scientist) for project administration and assistance with annual report submission and manuscript preparation. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Authors’ contributions:
LS: conceptualisation, methodology, formal analysis, investigation, data curation, writing (original draft), project administration.
LR, MC, MT and AG: investigation, data curation, writing, review and editing.
BT: conceptualisation, methodology, project administration.
JP: conceptualisation, methodology, formal analysis, writing (review and editing), supervision.
CM and YA: conceptualisation, methodology, writing (review and editing).
AD: conceptualisation, methodology, writing (original draft), supervision, project administration.
Competing interests
All authors declare that they do not have any potential conflict of interest in relation to this manuscript.
Supplementary Information
References
- 1.Kentish-Barnes N., McAdam J.L., Kouki S., et al. Research participation for bereaved family members: experience and insights from a qualitative study. Crit Care Med. 2015;43:1839–1845. doi: 10.1097/CCM.0000000000001092. [DOI] [PubMed] [Google Scholar]
- 2.Kentish-Barnes N., Chaize M., Seegers V., et al. Complicated grief after death of a relative in the intensive care unit. Eur Respir J. 2015;45:1341–1352. doi: 10.1183/09031936.00160014. [DOI] [PubMed] [Google Scholar]
- 3.Kross E.K., Engelberg R.A., Gries C.J., et al. ICU care associated with symptoms of depression and posttraumatic stress disorder among family members of patients who die in the ICU. Chest. 2011;139:795–801. doi: 10.1378/chest.10-0652. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Cook D., Swinton M., Toledo F., et al. Personalizing death in the intensive care unit: the 3 Wishes Project: a mixed-methods study. Ann Intern Med. 2015;163:271–279. doi: 10.7326/M15-0502. [DOI] [PubMed] [Google Scholar]
- 5.Mancini A.D., Griffin P., Bonanno G.A. Recent trends in the treatment of prolonged grief. Curr Opin Psychiatry. 2012;25:46–51. doi: 10.1097/YCO.0b013e32834de48a. [DOI] [PubMed] [Google Scholar]
- 6.Downar J., Koo E., des Ordons A.R., et al. Prevalence and predictors of severe grief reactions and desire for support following a death in the intensive care unit: a multicentre observational study. Intensive Care Med. 2018;44:521–522. doi: 10.1007/s00134-017-5027-2. [DOI] [PubMed] [Google Scholar]
- 7.Curtis J.R., Treece P.D., Nielsen E.L., et al. Randomized trial of communication facilitators to reduce family distress and intensity of end-of-life care. Am J Respir Crit Care Med. 2016;193:154–162. doi: 10.1164/rccm.201505-0900OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Efstathiou N., Walker W., Metcalfe A., et al. The state of bereavement support in adult intensive care: A systematic review and narrative synthesis. J Crit Care. 2019;50:177–187. doi: 10.1016/j.jcrc.2018.11.026. [DOI] [PubMed] [Google Scholar]
- 9.Downar J., Barua R., Sinuff T. The desirability of an Intensive Care Unit (ICU) clinician-led bereavement screening and support program for family members of ICU Decedents (ICU Bereave) J Crit Care. 2014;29(311):e9–16. doi: 10.1016/j.jcrc.2013.11.024. [DOI] [PubMed] [Google Scholar]
- 10.Zante B., Camenisch S.A., Schefold J.C. Interventions in postintensive care syndrome-family: a systematic literature review. Crit Care Med. 2020;48:e835–e840. doi: 10.1097/CCM.0000000000004450. [DOI] [PubMed] [Google Scholar]
- 11.Pattison N.A., White C., Lone N.I. Bereavement in critical care: A narrative review and practice exploration of current provision of support services and future challenges. J Intensive Care Soc. 2020;22:349–356. doi: 10.1177/1751143720928898. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Buckley T. End of life and bereavement care in the intensive care unit: A need for more quality empirical research. Aust Crit Care. 2017;30:137–138. doi: 10.1016/S1036-7314(17)30182-0. [DOI] [PubMed] [Google Scholar]
- 13.Beiermann M., Kalowes P., Dyo M., Mondor A. Family members’ and intensive care unit nurses’ response to the ECG Memento during the bereavement period. Dimens Crit Care Nurs. 2017;36:317–326. doi: 10.1097/DCC.0000000000000269. [DOI] [PubMed] [Google Scholar]
- 14.Barnato A.E., Schenker Y., Tiver G., et al. Storytelling in the early bereavement period to reduce emotional distress among surrogates involved in a decision to limit life support in the ICU: a pilot feasibility trial. Crit Care Med. 2017;45:35–46. doi: 10.1097/CCM.0000000000002009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Johansson M., Wåhlin I., Magnusson L., et al. Family members’ experiences with intensive care unit diaries when the patient does not survive. Scand J Caring Sci. 2018;32:233–240. doi: 10.1111/scs.12454. [DOI] [PubMed] [Google Scholar]
- 16.Rait L.I., Yeo N.Y., Ali Abdelhamid Y., et al. The impact of bereavement support on psychological distress in family members: A systematic review and meta-analysis. Crit Care Resusc. 2021;23:225–233. doi: 10.51893/2021.2.sr1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Department of Health and Human Services . Victoria State Government; Melbourne: 2016. Simplifying medical treatment decision making and advance care planning. Ensuring Victorians can make decisions about their future medical treatment.http://www.grpcc.com.au/wp-content/uploads/2016/07/Simplifying-medical-treatment-decision-making-and-advance-care-planning.pdf (viewed Apr 2022) [Google Scholar]
- 18.Harris P.A., Taylor R., Minor B.L., et al. The REDCap consortium: building an international community of software platform partners. J Biomed Inform. 2019;95 doi: 10.1016/j.jbi.2019.103208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Harris P.A., Taylor R., Thielke R., et al. Research electronic data capture (REDCap) — a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381. doi: 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Kentish-Barnes N., Chevret S., Champigneulle B., et al. Effect of a condolence letter on grief symptoms among relatives of patients who died in the ICU: a randomized clinical trial. Intensive Care Med. 2017;43:473–484. doi: 10.1007/s00134-016-4669-9. [DOI] [PubMed] [Google Scholar]
- 21.Pochard F., Darmon M., Fassier T., et al. Symptoms of anxiety and depression in family members of intensive care unit patients before discharge or death. A prospective multicenter study. J Crit Care. 2005;20:90–96. doi: 10.1016/j.jcrc.2004.11.004. [DOI] [PubMed] [Google Scholar]
- 22.Singer S., Kuhnt S., Götze H., et al. Hospital Anxiety and Depression Scale cutoff scores for cancer patients in acute care. Br J Cancer. 2009;100:908–912. doi: 10.1038/sj.bjc.6604952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Azoulay E., Pochard F., Kentish-Barnes N., et al. Risk of posttraumatic stress symptoms in family members of intensive care unit patients. Am J Respir Crit Care Med. 2005;171:987–994. doi: 10.1164/rccm.200409-1295OC. [DOI] [PubMed] [Google Scholar]
- 24.Creamer M., Bell R., Failla S. Psychometric properties of the Impact of Event Scale – Revised. Behav Res Ther. 2003;41:1489–1496. doi: 10.1016/j.brat.2003.07.010. [DOI] [PubMed] [Google Scholar]
- 25.Prigerson H.G., Maciejewski P.K., Reynolds C.F., et al. Inventory of Complicated Grief: a scale to measure maladaptive symptoms of loss. Psychiatry Res. 1995;59:65–79. doi: 10.1016/0165-1781(95)02757-2. [DOI] [PubMed] [Google Scholar]
- 26.Kentish-Barnes N., Seegers V., Legriel S., et al. CAESAR: a new tool to assess relatives’ experience of dying and death in the ICU. Intensive Care Med. 2016;42:995–1002. doi: 10.1007/s00134-016-4260-4. [DOI] [PubMed] [Google Scholar]
- 27.Grasselli G., Pesenti A., Cecconi M. Critical care utilization for the COVID-19 outbreak in Lombardy, Italy: early experience and forecast during an emergency response. JAMA. 2020;323:1545–1546. doi: 10.1001/jama.2020.4031. [DOI] [PubMed] [Google Scholar]
- 28.Tan E., Song J., Deane A.M., Plummer M.P. Global impact of coronavirus disease 2019 infection requiring admission to the ICU: a systematic review and meta-analysis. Chest. 2021;159:524–536. doi: 10.1016/j.chest.2020.10.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Johannesson K.B., Lundin T., Hultman C.M., et al. The effect of traumatic bereavement on tsunami-exposed survivors. J Trauma Stress. 2009;22:497–504. doi: 10.1002/jts.20467. [DOI] [PubMed] [Google Scholar]
- 30.Shear M.K., Simon N., Wall M., et al. Complicated grief and related bereavement issues for DSM-5. Depress Anxiety. 2011;28:103–117. doi: 10.1002/da.20780. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Burrell A.J., Pellegrini B., Salimi F., et al. Outcomes for patients with COVID-19 admitted to Australian intensive care units during the first four months of the pandemic. Med J Aust. 2021;214:23–30. doi: 10.5694/mja2.50883. [DOI] [PubMed] [Google Scholar]
- 32.Mitchell M., Coombs M., Wetzig K. The provision of familycentred intensive care bereavement support in Australia and New Zealand: Results of a cross sectional explorative descriptive survey. Aust Crit Care. 2017;30:139–144. doi: 10.1016/j.aucc.2016.07.005. [DOI] [PubMed] [Google Scholar]
- 33.Lautrette A., Darmon M., Megarbane B., et al. A communication strategy and brochure for relatives of patients dying in the ICU. N Engl J Med. 2007;356:469–478. doi: 10.1056/NEJMoa063446. [DOI] [PubMed] [Google Scholar]
- 34.Santiago C., Lee C., Piacentino R., et al. A pilot study of an interprofessional, multi-component bereavement follow-up program in the intensive care unit. Can J Crit Care Nurs. 2017;28:18–24. [Google Scholar]
- 35.McAdam J.L., Puntillo K. Pilot study assessing the impact of bereavement support on families of deceased intensive care unit patients. Am J Crit Care. 2018;27:372–380. doi: 10.4037/ajcc2018575. [DOI] [PubMed] [Google Scholar]
- 36.Moss S.J., Wollny K., Poulin T.G., et al. Bereavement interventions to support informal caregivers in the intensive care unit: a systematic review. BMC Palliative Care. 2021;20:66. doi: 10.1186/s12904-021-00763-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Wittouck C., Van Autreve S., De Jaegere E., et al. The prevention and treatment of complicated grief: a meta-analysis. Clin Psychol Rev. 2011;31:69–78. doi: 10.1016/j.cpr.2010.09.005. [DOI] [PubMed] [Google Scholar]
- 38.Mason T.M., Tofthagen C.S., Buck H.G. Complicated grief: risk factors, protective factors, and interventions. J Soc Work End Life Palliat Care. 2020;16:151–174. doi: 10.1080/15524256.2020.1745726. [DOI] [PubMed] [Google Scholar]
- 39.Kentish-Barnes N., Chevret S., Valade S., et al. A three-step support strategy for relatives of patients dying in the intensive care unit: a cluster randomised trial. Lancet. 2022;399:656–664. doi: 10.1016/S0140-6736(21)02176-0. [DOI] [PubMed] [Google Scholar]
- 40.Fincham J.E. Response rates and responsiveness for surveys, standards, and the Journal. Am J Pharm Educ. 2008;72:43. doi: 10.5688/aj720243. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Sitzia J., Wood N. Response rate in patient satisfaction research: an analysis of 210 published studies. Int J Qual Health Care. 1998;10:311–317. doi: 10.1093/intqhc/10.4.311. [DOI] [PubMed] [Google Scholar]
- 42.Serafini G., Parmigiani B., Amerio A., et al. The psychological impact of COVID-19 on the mental health in the general population. QJM. 2020;113:531–537. doi: 10.1093/qjmed/hcaa201. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Wallace C.L., Wladkowski S.P., Gibson A., White P. Grief during the COVID-19 pandemic: considerations for palliative care providers. J Pain Symptom Manage. 2020;60:e70–e76. doi: 10.1016/j.jpainsymman.2020.04.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Kalocsai C., des Ordons A.R., Sinuff T., et al. Critical care providers’ support of families in bereavement: a mixed-methods study. Can J Anaesth. 2020;67:857–865. doi: 10.1007/s12630-020-01645-0. [DOI] [PubMed] [Google Scholar]
- 45.Milberg A., Olsson E.C., Jakobsson M., et al. Family members’ perceived needs for bereavement follow-up. J Pain Symptom Manage. 2008;35:58–69. doi: 10.1016/j.jpainsymman.2007.02.039. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials