Effect of Systematic Intensive Care Unit Triage on Long-term
Mortality Among Critically Ill Elderly Patients in France
A Randomized Clinical Trial

Guidet. JAMA 2017; 318(15):1450-1459. doi:10.1001/jama.2017.13889

Clinical Question

  • In critically ill patients aged over 75 years, does a recommendation for systematic ICU admission compared with standard care reduce 6 month mortality?


  • The ageing population has led to an increased demand for both hospital admissions and intensive care admissions. Elderly patients are more commonly affected by multiple co-morbidities, frailty and have reduced physiological reserve. It is unclear if ICU admissions benefit this population as observational trials have reported mixed results and there have been no previous randomised controlled trials.


  • Cluster randomised controlled trial
  • Computer generated randomisation
  • Stratified by median number of emergency department visits, presence/absence of geriatric ward, and geographical area
  • 6 month follow up period. For patients discharged alive, outcomes assessed by telephone follow up with patients, relatives, general practioners or appropriate legal institutions
  • Statistical analysis
    • Sample size calculation:2802 patients required to have 74% power to detect a 6% difference in mortality rates from a baseline of 32% and an estimated intracluster correlation coefficient of 0.01, with a 2-sided alpha of 0.05. Number of hospital participating in study increased from 20 to 24, therefore planned sample size increased to 3000 patients
    • Bi-variable associations evaluated using t test for continuous variables and the χ2 or Fisher exact test for categorical variables
    • Binary outcomes analysed using logistic regression models
    • Crude overall survival at 6 months estimated by Kaplan-Meier method and compared with log-rank test
    • 6-month survival adjusted for baseline characteristics estimated using Cox model
    • Intention to treat analysis


  • 24 hospitals in France, both academic and non-academic with at least 1 emergency department (ED) and one ICU
  • Data collected: January 2012 – November 2015


  • Inclusion:
    • Age >=75 years presenting to the ED
    • Predefined critical condition that potentially require organ support
    • Had preserved functional status (Index of Independance in Activities of Daily Living score of >=4 or not evaluable)
    • Preserved nutritional status (absence of cachexia, subjectively assessed by physician at the bedside)
    • Free of active cancer (patients in whom cancer was diagnosed after inclusion remained in the statistical analysis)
  • Exclusion: ED stay of > 24 hours, secondary referral to the ED, refusal to participate
  • 3037 patients randomised
  • Comparing baseline characteristics of intervention vs. control groups – Patients in the intervention group had a higher recruiting rate and a higher severity of illness
    • Recruitment continued for 22.5 vs. 28.5 months
    • Hospital included: 11 vs. 13
    • Mean number of ED visits during study period: 12,746 vs. 16,580
    • Geriatric ward in the hospital: 10 vs. 11
    • Academic hospital: 7 vs. 10
    • Medical ICU: 5 vs. 7
    • Patients included: 1518 vs. 1518
    • Median age (IQR): 85 (81-89) vs. 85 (81-89)
    • Co-existing conditions
      • Ischaemic heart disease or hypertension: 41% vs. 42%
      • Respiratory disorder: 30% vs. 31%
      • Congestive heart failure: 15% vs. 11%, p=0.04
      • Cognitive impairment: 10% vs. 14%, p=0.06
    • Median SAPS 3 score at enrollement: 64 vs. 59, p<0.01
    • Seniority of ED physician >=2 years: 58% vs. 74%, p<0.001
    • Initial clinical diagnosis, p<0.01
      • Respiratory failure: 32% vs. 32%
      • Acute respiratory failure requiring intubation: 4% vs. 2%
      • Cardiac disorder: 12% vs. 15%
      • Shock: 21% vs. 16%
      • Coma: 12% vs. 9%
    • Living situation, p=0.01
      • Home without assistance: 71% vs. 65%
      • Nursing home/long-term care facility/hospital: 13% vs. 14%


  • Programme to promote ICU admission
    • ED & ICU physicians asked to systematically recommend an ICU admission for all included patients during trial period
    • ED physician required to systematically call the attending ICU physician; the ICU physician was required to systematically evaluate the patient at the bedside; and the ED & ICU physicians were required to jointly decide whether to admit the patient to the ICU
    • If no ICU bed was available the patient had to be transferred to another hospital ICU
    • 62% of patients admitted to ICU
    • 16% of patients admitted to intermediate care or specialised unit


  • Standard care
    • No recommendations regarding the ICU triage
    • Decision for admission made by physician at bedside
    • 34% of patients admitted to ICU
    • 21% of patients admitted to intermediate care of specialised unit

Differences in Management Between Groups

  • In intervention group, physicians inquired significantly more often about patients’ or surrogate decision makers’ opinions about ICU admission: 49% vs. 24%, p<0.001
  • ICU physciands more often involved in triage process 97% vs. 62%, p<0.001
  • ICU physians were more favourable to ICU admission 75% vs. 66%, p<0.001
  • Patients were more favourable to ICU admission: 88% vs. 66%, p<0.001


  • Primary outcome: Overall mortality at 6 months
    • Adjusted for baseline characteristics – no significant difference
      • RR1.05, 95% C.I. 0.96-1.14
    • Crude mortality – significantly higher in intervention group
      • 45% vs. 39% (difference 6%, 95% C.I. 3-10%, p<0.001
  • Secondary outcomes:
    • ICU admission – significantly higher in intervention group
      • 61% vs. 34%
        • Unadjusted RR 1.8, 95% C.I. 1.66-1.95
        • Adjusted RR 1.68, 95% C.I. 1.54-1.82
    • In-hospital mortality – significantly higher in intervention group
      • 30% vs. 21%
        • Absolute risk difference -9% (95% C.I. -12 to -5), p<0.001
        • Adjusted absolute risk difference -4 (95% C.I. -8 to -1), p=0.03
    • Decrease from baseline Index of Independence in Activities of Daily Living at 6 months – significantly greater in intervention group
      • Median difference between follow up and baseline score
        difference in means, -0.2 (95% C.I. -0.37 to -0.03)
  • Post-hoc analysis
    • Comparing treatments given to patients admitted in the intervention vs. control groups
      • Underwent mechanical ventilation: 374 vs. 147, p<0.001
      • Underwent non-invasive ventilation: 251 vs. 170, p=0.004
    • ICU and hospital stay – no significant difference
    • Patients discharged alive had increased length of hospital stay: 16.8 vs. 13.6 days, p<0.001
    • Self-reported physical quality of life at 6 months – no significant difference
      • Mean score 36.7 vs. 36.2
    • Self-reported mental quality of life at 6 months – significantly higher in the intervention group
      • Mean score 44.6 vs. 43.7, difference in means 0.9 (95% C.I. 0.1-1.6)

Authors’ Conclusions

  • In critically ill elderly patients a programme to promote systemic ICU admission increased ICU use but did not reduce mortality.


  • Randomised controlled trial
  • Multi-centre
  • Large number of patients randomised
  • Significant difference in ICU admission rate achieved between intervention and control group


  • The recruiting rate was lower in the control group suggesting a selection bias.
  • There were significant differences in baseline characteristics
  • The trial was set in only one country which may limit external validity as ICU admission criteria vary from country to country
  • The use of cluster randomisation means that systemic differences between hospitals may have accounted for some of the results reported.

The Bottom Line

  • In elderly critically ill patients in France a policy to promote systemic ICU admission increased ICU utilisation but had no affect on 6 month mortality. In-hospital mortality was significantly increased in the intervention group. Potential reasons for this include differences in baseline characteristics, methodological weaknesses, or harm resulting from ICU care.
  • Due to the differences in baseline characteristics and methodological weaknesses this paper should not change current practice but highlights the importance of thinking about whether an ICU admission will actually benefit or potentially harm the patient in front of you.

External Links


Summary author: @davidslessor
Summary date: 30th October 2017
Peer-review editor: Adrian Wong

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.