A three-step support strategy for relatives of patients dying
in the intensive care unit: a cluster randomised trial

Kentish-Barnes. Lancet 2022. Published Online;

Clinical Question

  • In relatives of patients dying in the ICU, does a 3 step support strategy compared with standard care reduce prolonged grief disorder?


  • Relatives of patients admitted to the intensive care are often affected by symptoms such as anxiety, depression and post-traumatic stress disorder
  • Relatives of patients who die in the ICU are at risk of suffering from prolonged grief that can be distressing and disabling and is associated with increased consumption of health-care resources
  • Studies have demonstrated that communication with ICU clinicians has a major effect on relatives’ experiences
  • This study aimed to investigate whether a communication strategy could reduce prolonged grief


  • Cluster randomised controlled trial
  • Participating ICU’s randomly assigned into an intervention cluster and a control cluster
  • Central randomisation using permuted blocks of non-released size
  • Stratified based on the recruitment period and whether the centre had previously participated in a study from the author’s research group
  • Intention to treat analysis
  • Registered with
  • Study was conducted prior to COVID-19 and was therefore unaffected by restricted visiting
  • Treating centres were not blinded
  • Outcomes assessed by trained psychologists during telephone follow up at 1, 3 and 6 months
  • Outcome assessors blinded to group assignment
  • Prolonged grief defined as score of ≥30 on the prolonged grief-13 questionnaire (PG-13) 6 months after the death
  • 454 relatives needed to obtain 90% power to decrease the primary outcome from 50% to 35% with a type 1 error of 5%. Between-cluster variation was accounted for using an inflation factor leading to a resulting sample size of 704. To allow for a 25% drop off, planned to enrol 874 relatives


  • 34 intensive care units France
  • Patients enrolled from 2017 – 2019


  • Inclusion:
    • Relatives of patients >18 years who had a treatment withdrawal or withholding decision
    • ICU length of stay ≥2 days
    • Relative who was most involved with the ICU team
  • Exclusion:
    • Relatives who did not speak enough French
    • Patient was an organ donor
  • 875 family members enrolled, 87% completed 1 month follow up, 79% completed 6 month follow up
  • Comparing baseline characteristics of intervention vs. control group
    • Number of centres enrolled: 17 vs 17
    • Male: 69% vs 65%
    • Age: 70 vs 71 years
    • Length of ICU: 8 vs 7 days
    • Patient type – Medical: 84% vs 81%
    • Cancer or haematological malignancy: 35% vs 34%
    • Psychiatric illness: 3% vs 5%
    • End of life characteristics
      • Decision to withhold treatment: 17% vs 24%
      • Withdrew treatment: 83% vs 76%
      • Patient sedated at time of death: 79% vs 85%
      • Family present at patient’s death: 69% vs 58%
      • Intervention of a chaplain: 25% vs 9%
      • End of life family conference: 93% vs 67%
      • Physician entered the patient’s room during the end of life: 93% vs 66%
      • Nurse entered the patient’s room during the end of life: 92% vs 74%
      • Physician and nurse met the relative after the patient’s death: 85% vs 29%
    • Relatives characteristics
      • Number of relatives included in study: 433 vs 352
      • Female: 70% vs 66%
      • Relationship
        • Spouse or partner: 36% vs 37%
        • Child: 45% vs 42%


  • Proactive communication and support intervention
    • Clinicians (all staff) attended interactive educational meetings focused on end of life communication
    • 3 meetings held with relatives (% that occurred)
      • Family conference with physician and nurse to prepare the relatives for the imminent death: 100%
        • Give opportunities to ask questions: 100%
        • Give opportunities for relative to express feeling and emotions: 83%
        • Encourage relative to talk to the patient and say goodbye: 86%
        • Discuss being present at time of death: 97%
        • Discuss possible involvement in the patient’s physical care: 65%
        • Discuss spiritual beliefs and needs: 85%
      • ICU room visit to provide active support
        • During the dying process the physician and nurse entered the patient’s room separately at least once (P = physician, N = nurse)
          • Meet relative’s emotional needs: 85% P, 84% N
          • Active listening: 97% P, 99% N
          • Offer material support (eg chair, drink): 86% P, 96% N
          • Discuss spiritual practices: 66% P, 64% N
          • Discuss relative’s implication in patient support and tenets of palliative care: 63% P, 62% N
          • Clarify and answer any questions: 95% P, 92% N
          • Highlight the relative’s commitment: 79% P, 76% N
      • Meeting after the patient’s death to offer condolences and closure
        • Physician and nurse met the relatives in a dedicated room within the hours following the patient’s death or when the relatives returned to hospital the next day for administrative formalities
        • If the relatives were unavailable the physician was asked to telephone the relative to address the same issues
        • Express condolences: 87%
        • Encourage questions about the patient’s ICU stay and death and address doubts: 89%
        • For administrative procedures, guide relatives towards specific professionals: 98%
        • Offer a visit with the ICU team later on: 86%
        • Show empathy and give the relative opportunities to express feelings: 92%
    • 1 month implementation phase to allow clinicians to practice


  • Standard care
    • Applied their best standard of care in terms of support and communication with relatives of dying patients


  • Primary outcome: Proportion of relatives with prolonged grief – (score on the prolonged grief-13 questionnaire [PG-13]15 ≥30) 6 months after the death – significantly reduced in intervention group
    • 15% vs 21%
    • Median difference -6.5% (95% CI -12 to -0.6), p=0.03
  • Median PG-13 score – significantly reduced in intervention group
    • 19 (IQR 14-26) vs 21 (IQR 15-29)
    • Mean difference 2.5 (95% CI 1.04-3.95), p=0.003
  • Secondary outcomes:
  • Favouring intervention group
    • Difficult end of life experience – CAESAR Score ≤59
      • 13% vs 22%
      • Mean difference  -8.5 (95% CI -12.9 to -3.1), p=0.002
    • Quality of death and dying 1 (QODD-1 scale)
      • Median 9 vs 8
      • Mean difference 0.4 (95% CI 0.1-0.7), p=0.02
    • Hospital Anxiety & Depression Scale
      • 1 month
        • Median 14 vs 14
        • Mean difference -1.4 (95% CI -2.7 to -0.1), p=0.036
      • 3 month
        • Median 11 vs 13
        • Mean difference -1.6 (95% CI -2.8 to -0.4), p=0.009
      • 6 month
        • Median 9 vs 11
        • Mean difference -2 (-3.3 to -0.7), p=0.003
    • Presence of PTSD related symptoms (Impact of event scale-revised score ≥33)
      • 3 months
        • 12 % vs 22%
        • Mean difference -10 (95% CI-15.6 to -4.3), p=0.005
      • 6 months
        • 8% vs 17%
        • Mean difference -8.1 (95% CI -13.1 to -3), p=0.014

Authors’ Conclusions

  • Among relatives of patients dying in the ICU, a 3 step support strategy significantly reduced prolonged grief symptoms


  • Multi-centre
  • Randomised
  • Blinding of outcome assessors
  • 6 month follow up
  • Intention to treat analysis
  • Registered on
  • Cluster randomised trials are well suited when an approach to patient care is being tested (as opposed to a physiological or pharmacological intervention)
    • It reduces the risk of contamination where aspects of the intervention may be adopted in caring for patients in the control group which would dilute the observed differences


  • Only conducted in France which limits external validity as communication styles and risk of prolonged grief may be different in different populations
  • In the intensive care unit I work in it is standard practice to have an end of life family conference where as this only occurred in 67% in the control group. It is unknown how much impact the end of life family conference had on prolonged grief compared with some of the other interventions utilised in this 3 step support strategy
  • Rates of prolonged grief in trial centres prior to study period was not known. We therefore do not know if there were any baseline differences between the intervention and the control centres
  • PG-13 score has not been validated in ICU, there is no established minimally important clinical difference in score
  • Only 79% completed follow up at 6 months. However sample size based on 25% drop out
  • The application of these results to linguistically diverse patients and relatives is not clear

The Bottom Line

  • Communication skills have a vital importance in reducing the risk of prolonged grief in relatives of patients who died in the ICU
  • In intensive care units in France, this 3 stage support strategy, compared with standard care, significantly reduced prolonged grief as well as symptoms of anxiety and depression, and symptoms of post-traumatic stress disorder
  • All intensive care units should review their communication styles and skills to determine what improvements can be made

External Links


Summary author: @davidslessor
Summary date: 28/02/2022
Peer-review editor: @celiabradford

Image by: Tomislav Jakupec from Pixabay 



Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.