Sepsis Presenting in Hospitals versus Emergency Departments: Demographic, Resuscitation, and Outcome Patterns in a Multicenter Retrospective Cohort

Abstract

Background: Differences between hospital-presenting sepsis (HPS) and emergency department-presenting sepsis (EDPS) are not well described.

Objectives: We aimed to (1) quantify the prevalence of HPS versus EDPS cases and outcomes; (2) compare HPS versus EDPS characteristics at presentation; (3) compare HPS versus EDPS in process and patient outcomes; and (4) estimate risk differences in patient outcomes attributable to initial resuscitation disparities.

Design: Retrospective consecutive-sample cohort.

Setting: Nine hospitals from October 1, 2014, to March 31, 2016.

Patients: All hospitalized patients with sepsis or septic shock, as defined by simultaneous (1) infection, (2) ≥2 Systemic Inflammatory Response Syndrome (SIRS) criteria, and (3) ≥1 acute organ dysfunction criterion. EDPS met inclusion criteria while physically in the emergency department (ED). HPS met the criteria after leaving the ED.

Measurements: We assessed overall HPS versus EDPS contributions to case prevalence and outcomes, and then compared group differences. Process outcomes included 3-hour bundle compliance and discrete bundle elements (eg, time to antibiotics). The primary patient outcome was hospital mortality.

Results: Of 11,182 sepsis hospitalizations, 2,509 (22.4%) were hospital-presenting. HPS contributed 785 (35%) sepsis mortalities. HPS had more frequent heart failure (OR: 1.31, CI: 1.18-1.47), renal failure (OR: 1.62, CI: 1.38-1.91), gastrointestinal source of infection (OR: 1.84, CI: 1.48-2.29), euthermia (OR: 1.45, CI: 1.10-1.92), hypotension (OR: 1.85, CI: 1.65-2.08), or impaired gas exchange (OR: 2.46, CI: 1.43-4.24). HPS were admitted less often from skilled nursing facilities (OR: 0.44, CI: 0.32-0.60), had chronic obstructive pulmonary disease (OR: 0.53, CI: 0.36-0.78), tachypnea (OR: 0.76, CI: 0.58-0.98), or acute kidney injury (OR: 0.82, CI: 0.68-0.97). In a propensity-matched cohort (n = 3,844), HPS patients had less than half the odds of 3-hour bundle compliant care (17.0% vs 30.3%, OR: 0.47, CI: 0.40-0.57) or antibiotics within three hours (66.2% vs 83.8%, OR: 0.38, CI: 0.32-0.44) vs EDPS. HPS was associated with higher mortality (31.2% vs 19.3%, OR: 1.90, CI: 1.64-2.20); 23.3% of this association was attributable to differences in initial resuscitation (resuscitation-adjusted OR: 1.69, CI: 1.43-2.00).

Conclusions: HPS differed from EDPS by admission source, comorbidities, and clinical presentation. These patients received markedly less timely initial resuscitation; this disparity explained a moderate proportion of mortality differences.

FIG 1

Displays contributions of Hospital vs ED-presenting sepsis to overall case prevalence and patient outcomes. Hospital presentations are also stratified into sepsis presentations that occurred during an ICU admission versus those that occurred outside the ICU on the hospital wards. Brackets indicate 95% CIs. Abbreviations: EDPS, ED presenting sepsis; HPS, hospital presenting sepsis; ICU, intensive care unit.

FIG 2

(A) Displays the adjusted likelihood of HPS vs EDPS patients to receive specific initial resuscitation interventions. (B) Displays the adjusted likelihood of all HPS patients to experience the specified outcome, with and without correction for differences in resuscitation delivery. The ARD is the percentage of the outcome difference between HPS and EDPS that was explainable by differences in initial resuscitation delivery. (C) Displays the likelihood of outcome as in (b) but excluding HPS patients whose sepsis presentation occurred in the ICU after admission for another reason. Brackets indicate 95% CIs. Abbreviations: ARD, adjusted risk-difference; EDPS, ED presenting sepsis; HPS, hospital presenting sepsis; ICU, intensive care unit.