Does prompt admission to the ICU have an impact on mortality?
EJRC ARTICLE REVIEW
Guidelines recommend that critical care admission should be delivered within 4 hours. However, the evidence behind the recommendation is limited. Randomised trials are deemed unethical and non-randomised studies are often confounded by allocation bias. An alternative to experimental randomisation is to seek an instrument that naturally randomises patients to prompt admission or not, as in this study focused on critical care unit strain.
The aims of the study were first to describe the effect of critical care strain on decision-making, to explore whether delays to admission engendered by high strain allowed for estimation of the effect of delay on patient outcome and to focus on the subgroup of patients recommended for critical care.
The study was a prospective cohort study of consecutive ward patients assessed for critical care. Prompt admissions (within 4 hours of assessment) were compared to a “watchful waiting” cohort. Critical care strain (bed occupancy) was classified as low, medium or high.
The analyses included 12,380 patients from 48 UK hospitals assessed for critical care and 4,560 (37%) hereof assessed and recommended for critical care. Each hospital registered data for a median of 8 months between September 2010 and December 2011. For the whole cohort, 20% had prompt admissions, and of the “watchful waiting” group, 20% had delayed critical care and 80% no critical care. For the subgroup, there were 51% prompt admissions and among the “watchful waiting” group, 45% had delayed critical care. A total of 81% of assessments were made when there were two or more empty critical care beds, 11% with one empty bed and 8% when the ICU was fully occupied. As strain increased, the proportion of prompt admission decreased (21%, 15% and 9%, p<0,001) for the whole cohort and for the subgroup (53%, 38%, and 23%, p<0,001).
Overall seven-day mortality was 14% and 90-day mortality was 30%. For those admitted promptly, unadjusted 90-day mortality was 37% and for the watchful waiting group 29%, with the prompt admission group having higher severity scores. After risk adjustments survival was equal.
With instrumental variable analysis based on critical care strain, prompt admission reduced 90-day mortality by 7% (p=0.12) for the whole cohort and by 16% (p=0.04) for the subgroup recommended for critical care.
The results indicate that prompt admission reduces 90-day mortality, although only statistical significant for those who are recommended for critical care.
Strengths of the study include the large cohort, the inclusion of multiple centres, and the prospective data.
The study results rely on the assumption that critical care strain acts as a natural randomisation event and as such minimises treatment biases (confounding by indication).
This article review was submitted by Associate Professor Hanne Irene Jensen, member of the ESICM Ethics Section and the Journal Review Club.
Steve Harris et al. Impact on mortality of prompt admission to critical care for deteriorating ward patients: an instrumental variable analysis using critical care bed strain. Intensive Care Med 2018. doi.org/10.1007/s00134-018-5148-2 [Epub ahead of print]