Unexpected intensive care transfer of admitted patients with severe sepsis

Patients with severe sepsis generally respond well to initial therapy administered in the emergency department (ED), but a subset later decompensate and require unexpected transfer to the intensive care unit (ICU). This study aimed to identify clinical factors that can predict patients at increased risk for delayed transfer to the ICU and the association of delayed ICU transfer with mortality.

This is a nested case-control study in a prospectively collected registry of patients with severe sepsis and septic shock at two EDs. Cases had severe sepsis and unexpected ICU transfer within 48 h of admission from the ED; controls had severe sepsis but remained in a non-ICU level of care. Univariate and multivariate regression analyses were used to identify predictors of unexpected transfer to the ICU, which was the primary outcome. Differences in mortality between these two groups as well as a cohort of patients directly admitted to the ICU were also calculated.

Of the 914 patients in our registry, 358 patients with severe sepsis were admitted from the ED to non-ICU level of care; 84 (23.5%) had unexpected ICU transfer within 48 h. Demographics and baseline co-morbidity burden were similar for patients requiring versus not requiring delayed ICU transfer. In unadjusted analysis, lactate ≥4 mmol/L and infection site were significantly associated with unexpected ICU upgrade. In forward selection multivariate logistic regression analysis, lactate ≥4 mmol/L (OR 2.0, 95% CI 1.03, 3.73; p = 0.041) and night (5 PM to 7 AM) admission (OR 1.9, 95% CI 1.07, 3.33; p = 0.029) were independent predictors of unexpected ICU transfer. Mortality of patients who were not upgraded to the ICU was 8.0%. Patients with unexpected ICU upgrade had similar mortality (25.0%) to those patients with severe sepsis/septic shock (24.6%) who were initially admitted to the ICU, despite less severe indices of illness at presentation.

Serum lactate ≥4 mmol/L and nighttime admissions are associated with unexpected ICU transfer in patients with severe sepsis. Mortality among patients with delayed ICU upgrade was similar to that for patients initially admitted directly to the ICU.