Boarding time and mortality: the association points both ways

Design

post hoc subgroup analysis of a retrospective multicentre cohort (2009-2020), logistic regression by transfer-time quintile, adjusted for hospital and APACHE III, stratified by hospital type

Population

18,798 adults admitted directly from the emergency department to intensive care across 12 Dutch hospitals; median transfer time 1.9 hours

Primary outcome

hospital mortality by emergency-department-to-ICU transfer time

Effect

out-of-hospital cardiac arrest in academic hospitals OR 2.94 (95% CI 1.80-4.78) beyond 3.4 hours, but odds ratios below 1.0 in non-academic teaching hospitals; no association in overdose, sepsis, pneumonia, respiratory failure or intracranial haemorrhage

A secondary analysis of a Dutch cohort across twelve hospitals, 2009 to 2020, examined 18,798 adults admitted directly from the emergency department to intensive care, across seven diagnostic groups each with over 1,500 patients. Median emergency-department-to-ICU time was 1.9 hours. Analyses adjusted for hospital and APACHE III score and were stratified by hospital type.

For out-of-hospital cardiac arrest the associations were in opposite directions. In academic hospitals, longer transfer time went with higher mortality (OR 1.48, 95% CI 1.08-2.02 at 1.1-1.6 hours; OR 2.94, 1.80-4.78 beyond 3.4 hours). In non-academic teaching hospitals, longer transfer time went with lower mortality, odds ratios below 1.0 across all quintiles. For non-operative trauma an association appeared in the overall cohort (OR 1.90, 1.12-3.21 beyond 3.4 hours) but disappeared on stratification by hospital. For overdose, sepsis, pneumonia, respiratory failure and intracranial haemorrhage there was nothing.

The opposing directions are the point. The most likely explanation is that transfer time is measuring different things in different systems — in one, a delay; in another, a patient stable enough to wait, or one in whom further escalation had been reconsidered. Boarding time is widely used as a quality metric, and this says it cannot carry that weight without knowing what it encodes locally.

• Do not use emergency-department-to-ICU time as a standalone quality metric without local process data
• Where you audit it, stratify by diagnosis; the seven groups here behaved completely differently
• Treat a subgroup association that reverses between hospital types as a warning about confounding, not a discovery about either
• Cardiac arrest and trauma are where transfer time plausibly matters; sepsis and pneumonia showed nothing here
• Prospective data linking specific delays to specific care processes is what would answer this

Why it matters

It undercuts a metric that emergency departments are increasingly measured against.

Don't overread it

Retrospective, post hoc and subgroup-based — none of these associations should be read as an effect of the delay itself.

The statistics, in plain English

When the same exposure is associated with higher mortality in one hospital type and lower in another, the honest reading is that the exposure is not causing either — it is standing in for something unmeasured that differs between them, most likely how sick the patients were and how the decision to admit was made. Adjusting for APACHE III reduces that problem without removing it.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.