Septic shock: An analysis of outcomes for patients with onset on hospital wards versus intensive care units

Objective: To determine if early interventions for septic shock were associated with reduced mortality. Design: Retrospective cohort study. Setting: University hospital intensive care unit (ICU) and general wards. Patients: Forty one consecutive patients prospectively identified with positive blood cultures and septic shock. Although all patients were eventually treated in an ICU, ten (24%) patients were on a general ward at the onset of septic shock, and 31 (76%) were in an ICU setting. Interventions: None. Measurements and Main Results: Over a period of 9 mos, a cohort of 41 patients who had positive blood cultures and septic shock was prospectively identified, The 28 day crude mortality was 46% (19 deaths). We compared the management of septic shock and outcome for patients on a general ward vs. those patients in an ICU setting. Of the ten patients on the ward at time of shock onset (median age 55.5 yrs; median Acute Physiology and Chronic Health Evaluation [APACHE] II score of 18.5), seven (70%) died. In contrast, the 31 patients receiving intensive care when shock developed were older and more ill (median age 66 yrs; median APACHE II 24), yet had a mortality of 39% (12 deaths). The odds ratio (OR) for death for ward patients compared with ICU patients was 3.57 (p =.17). In a multivariate logistic regression analysis, two risk factors for mortality were important: APACHE II score (p =.015) and ward status (p =.08). Candida species in the bloodstream is known to have a high attributable mortality. When type of bloodstream pathogen (Candida species vs, bacteria) was added to the model, APACHE II (OR 2.64 for IO unit increase) remained significant (p =.014), but ward status (OR 3.97) became statistically nonsignificant (p =.222). The patients who were on a general ward when their shock developed had a median delay of 67 mins before transfer to an ICU setting. Ward patients received an intravenous fluid bolus after a median delay of 27 mins, whereas those in the ICU who received a fluid bolus did so after a median of 15 mins (p =.48). Ward patients also had a median delay of 310 mins to receive inotropic support compared with a median 22.5 mins (p =.037) for the patients in an ICU setting when shock started. Conclusions: The data suggest that for patients with septic shock on wards, there were clinically important delays in transfer of patients to the ICU, receipt of intravenous fluid boluses, and receipt of inotropic agents. However, the most powerful predictors of mortality were APACHE II scores and bloodstream infection with Candida species.