A Randomised Trial of a Family-Support Intervention in Intensive Care Units

White et al. June 21 2018. NEJM 2018;378:2365-75. DOI:10:1056/NEJMoa1802637

Clinical Question

  • For surrogate decision makers of critically ill patients, does a multi-component family support intervention impact on the surrogates’ hospital anxiety and depression scale (HADS) at 6 months?


  • Intensive care admissions are a psychologically taxing time for patients and their families. Residual grief and stress can persist well after the ICU admission particularly if the surrogate’s loved one has died, has not returned to baseline function or is needing extra support from their family or community.
  • Communication between Intensive Care staff and families is often ad hoc, time-pressured and delivered by miscellaneous health care professionals.
  • A study published in JAMA in 2016 reported a randomised trial of meetings with families with standard communication (ICU physicians) compared with extra palliative care-led meetings (meeting at Day 7 + Day 17) did not show any benefit to the HADS at 3 months and worsened symptoms of PTSD.
  • A trial randomising families who had lost a loved one in ICU to receive a letter of condolence or standard care failed to reduce PTSD or improve satisfaction. Read the TBL summary of this Kentish-Barnes trial.
  • Multi-disciplinary support of families in crisis is a logical and inexpensive approach to communication but data is lacking as to how effective it is.


  • PARTNER = Pairing Re-engineered ICU Teams with Nurse-Driven Emotional Support and Relationship-Building
  • Multicentre, stepped-wedge, cluster-randomised trial
  • 5 different ICUs represented the unit of randomised
    • Each started in the control phase which lasted a randomly assigned time
    • Each unit then transitioned to the intervention phase
  • Telephone interviews to the surrogates were conducted 6 months after the patient’s hospital discharge by research staff who were unaware of the allocation
  • The surrogates provided written informed consent for their participation
  • The intervention was classified “Quality Improvement” whilst the surrogate follow-up was classified “Research”


  • 5 ICUs in 5 hospitals in the University of Pittsburgh Medical Centre Health System
    • 2 medical-surgical ICU, 1 medical ICU, 1 specialising in neurologic illness, 1 specialising in transplant surgery
  • Patient enrolment occurred between July 2012 and August 2015 with follow-up of surrogates ending in February 2016


  • Inclusion: Patients >18 years without decision-making capacity, with at least 1 of:
    • Ventilated (> 4 days),
    • An estimated chance of death of at least 40%, or
    • Estimated chance of long-term impairment of at least 40%
  • Exclusion: no surrogate decision maker or patient receiving palliative treatment
  • 1420 patients: 547 in intervention arm, 873 in control arm
  • There were some baseline imbalances; those in the intervention group were older, had higher severity of acute illness scores and greater number of chronic comorbidities


  • The intervention had 3 main components:
    1. The interprofessional team was overseen by 4-6 nurses (PARTNER nurses) who received advanced communication training focusing on supporting the families of the critically ill
    2. A PARTNER nurse would meet with the family/surrogate on a daily basis and arranged a clinician-family meeting within 48 hours of admission and every 5-7 days thereafter
    3. A quality-improvement specialist was utilised to incorporate the family-support pathway into the clinician’s workflow


  • Usual care (none of the ICUs had a protocolized approach to family communication)

Management common to both groups

  • All other aspects of care were determined by the treating clinicians


  • Primary outcome: The surrogates’ symptoms of anxiety and depression as assessed by a HADS scale at 6 months, with scores ranging 0–42 with a higher score indicating worse symptoms
    • Unadjusted and adjusted (detailed below) analysis: no difference
      • Intervention: 11.7 +/- 7.9
      • Control: 12.1 +/-8.5
      • Estimated effect of intervention: -0.34 (95% CI -1.67 to 0.99; P = 0.61)
  • Secondary outcome:
    • The IES score assessing PTSD; no significant difference intervention vs control  (P = 0.49)
    • QOC score assessing clinician-family communication was significantly better in the intervention group (P = 0.001)
    • PPPC score measuring the patient and family-centredness of care was significantly better in the intervention group (P = 0.006)
    • The ICU length of stay was shorter in the intervention group (P = 0.045), mediated by increased early mortality in the intervention group
  • Patients’ Clinical Outcomes: Intervention vs Control
    • In-hospital death: 38% vs 30.2% (P = 0.008)
    • Death at 6 months: 62% vs 54.1% (P = 0.17)
    • Living independently at home at 6 months: 1% vs 2.6% (P = 0.9)

Authors’ Conclusions

  • Among critically ill patients and their surrogates, a family-support intervention delivered by the existing interprofessional ICU team did not affect the surrogates’ symptoms of depression and anxiety at 6 months, but the surrogates’ ratings of the quality of communication and the patient- and family-centered-ness of care were better and the length of stay in ICU shorter with the intervention than with usual care.


  • The trial design (step-wedged, cluster-randomised design) is a clever way to assess the impact of a quality intervention where whole ICU care processes are changed
  • HADS is a well validated measure of anxiety and depression
  • The effect of the intervention was examined from the perspective of patients, health care providers, surrogates and use of health care resources.
  • Appropriate statistical adjustment given known baseline differences between intervention and control group


  • Out of 1420 patients included only 57% completed the 6 month follow-up
    • The drop-out rates were similar between the intervention and the control group
    • The supplementary appendix suggests that there were a higher proportion of blacks who were lost to follow-up than not (9.9% vs 5.9%; P = 0.01)
    • This high loss to follow-up severely compromises this study’s validity
  • There was significant baseline imbalance
    • This was corrected for in the adjusted analysis but may contribute to confounding bias
  • Analysis was by intention to treat, but this was only applied to those who agreed to be interviewed at 6 months, not all who were randomised
  • The study was conducted within the confines of the UPMC health system potentially limiting its external validity
  • There was no quantitative analysis of how the intervention changed the interactions with families, although there were regular site visits with the PARTNER nurses observed in their interactions with families
  • The reduced length of ICU stay (due to increased mortality in the intervention arm) is unlikely due to the communication bundle, but more likely due to the mismatch in the severity of the acute illness between the groups, as the difference is almost entirely due to more deaths in the intervention arm rather than live discharge

The Bottom Line

  • An interprofessional support package for families/surrogates of critically ill patients is a logical and cost-effective way to improve quality of communication and  patient and family centred support
  • The baseline imbalances and high loss to follow-up in this trial limit drawing conclusions from this intervention
  • Further trials with greater internal and external validity would add to this important area of ICU research

External Links


Summary author: Celia Bradford
Summary date: 8 August 2018
Peer-review editor: Duncan Chambler

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.