Association Between Day and Time of Admission to Critical Care and Acute Hospital Outcome for Unplanned Admissions to Adult General Critical Care Units: Cohort Study Exploring the ‘Weekend Effect’

Arulkumaran N, Harrison DA and Brett SJ

British Journal of Anaesthesia 2017;118 (1): 112-22.


Aim of Study

  • Does the day and time of admissions to ICU influence overall mortality?
  • ICU must operate as a 24/7 speciality so answering the question of the existence of the weekend effect is an important clinical investigation.
  • It is imperative to ascertain whether out-of-hours services are adequate to cover emergency care without any adverse impact on outcome for patient


Design and Location

  • Prospectively collated cohort study
  • Data from UK Intensive Care National Audit and Research Centre (ICNARC) Case Mix Programme (CMP) database
  • Not including specialist units (neuro, cardiac) nor standalone HDU
  • Day split into “routine hours” (0800-1759) and “out of hours” (1800-0759)
  • Missing physiological data presumed normal
  • Potential confounders identified in advance and adjusted for in analysis:
    • Age
    • Severe conditions in past medical history
    • Prior functional dependency
    • Number of days from hospital admission to critical care admission
    • location before admission
    • CPR in 24 hours prior to critical care admission
    • Primary reason of admission to critical care
    • Acute severity of illness score (ICNARC and APACHE scores used)
  • Stats:
  • 3 multilevel logistic regression models used:
    • Unadjusted with single covariant of day/time of admission
    • Adjusted model adjusting for key potential confounders:
    • Adjusted model including delay/no delay covariate
  • 216 General/mixed Intensive Care Units in the UK that submitted data for the ICNARC CMP
  • April 1 2013-March 31st 2015



  • Inclusions:
    • Age >16
    • Unplanned admission
  • Exclusions:
    • Planned ICU admission (including post theatre, transfers and repatriation)
    • Readmissions – only the first episode was counted
    • Organ donation
    • Missing data (primary outcome or key confounders)
  • 300,469 admission screened, 195,428 included in final analysis
  • Baseline characteristics:
    • mean age of patients admitted was 60 yr
    • mean APACHE II = 17
    • ~1/4 had some degree of prior dependency
    • Overall, 4% of patients received in-hospital CPR during the 24 h before admission and a further 4% had out-of-hospital CPR
    • ~1/4 admitted with respiratory pathology
    • No significant differences between the two cohorts seen in any of the major criteria, including age, severity of illness, and co-morbidities

Weekend days cohort

  • Day of admission analysis: Patients admitted on Saturday or Sunday
  • Time of admission analysis: During routine working hours (0800-1759) and out of (1800 0759)

Week days cohort

  • Day of admission analysis: Patients admitted on Monday, Tuesday, Wednesday, Thursday or Friday
  • Time of admission analysis: During routine working hours (0800-1759) and out of (1800 0759)


  • Primary outcome:
    • Total ICU mortality: 18.8%
    • Total hospital mortality: 26.6%
    • Using Monday as reference day, no significant difference seen between odds of admission out of hours or at a weekend in adjusted analysis (P value =0.61)
  • Secondary outcome:
    • 8,295 (4.2%) patients experienced a delay (documented decision to admit to time of admission) of at least 1 hour prior to admission to ICU. 2,097 (1.1%) had delay >4hrs
      • Both groups associated with an increased risk of death (OR 1.08 for < 4hrs, 1.17 for >4 hrs, P=0.04)
      • Adjusting to account for delay did not change lack of effect of day/time of admission


multilevel logistic regression models.

The output of each model is presented as odds ratios with 95% confidence intervals (CIs) and P-values.

Hypothesis testing.

Results of the hypothesis tests are reported as the P-value (Wald test) and the absolute risk difference with 95% CI (based on linear combinations of the marginal predicted mortalities).


  • After risk adjustment with detailed clinical data, there was no difference in acute hospital mortality for unplanned admissions to ICU between weekdays and weekends, or daytime and nighttime
  • A small proportion of patients experienced delays in admission to ICU and this was associated with an increased risk of death

Stated Limitations from the Study


  • Important question
  • Large database
  • Multi centre
  • Detailed risk adjusted analysis
  • Secondary analysis an important question


  • Without delving into the inner workings of each ICU, shows association not causation
  • Difference exists between units in the ICNARC CMP
  • Delayed admission is a complex entity; may represent increased complexity of patient and further investigations or treatments.
  • UK based study


Discussion from Journal Club Meeting (?Change of Practice)

Ensure no delay to admissions to ICU

Ensure access to ICU specialist consultant at all times

Comments are closed.

Blog at

Up ↑

%d bloggers like this: