118 Delayed Emergency Department Fluid Resuscitation May Lead to Increased Mortality in Sepsis: A Call for an Optimal Fluid Resuscitation Interval

Abstract

Many hospitals struggle to maintain compliance with sepsis guidelines and to achieve a targeted fluid resuscitation of 30mL/kg of crystalloid within 3 hours. Recent data suggest that even earlier administration of these recommended fluids reduces mortality from septic shock, but the optimal timing of fluid delivery in the emergency department (ED) is still unknown. The objective of this study was to analyze the timing of fluid resuscitation as it relates to patient outcomes for severe sepsis/septic shock in the ED. A retrospective cohort study in an urban/suburban (pop. 1.1 million) community health system from January 2017 through December 2019. Included were all adults (18 years and up) admitted with an ICD10 diagnosis coded for sepsis at discharge or a sepsis DRG code with associated bacteremia. The sepsis diagnosis had to be present at admission (POA). Excluded were patients with a sepsis discharge in the 90 days prior to the study case sepsis event. Primary outcome was hospital mortality. Other outcomes included ED and hospital length of stay (LOS) and any ICU admission. Detailed fluids data with time stamps was available from our institution database allowing us to compute multiple time-based fluid variables. Multiple tables of data were extracted from Epic using SQL query language, and then validated and analyzed using IBM SPSS Statistics v26. Analysis included student t-test, Mann-Whitney U and Chi-Square tests as appropriate. Adjusted odds ratios (OR) with 95% confidence intervals (CI) for mortality were determined by multivariate logistic regression using those factors found to be significant at the univariate level. This minimal risk study was approved by the hospital’s Institutional Review Board. Of 2,696 patients, 2375 had sepsis POA: septic shock (44.9%), severe sepsis (42.4%), sepsis without organ dysfunction (5.4%), and sepsis with no classification (7.2%). Mean age was 66.2 years (SD=17.3) and 48.8% were female. ESI levels 1, 2 and 3 were 16.5%, 62.1%, and 21.3% respectively. Any ED hypotension (SBP<90mmHg or MAP<65mmHg) occurred in 51.9%. ‘Surviving Sepsis’ targeted 30mL/kg fluids was met for 75.1% at 3 hours, 63.7% at 2 hours, 55.1% at 1 hour and 53.6% at 30 minutes. Median (IQR) time to 30mL/kg was 7.0 hours (3.1-19.1), ED LOS was 5.5 hours (4.3-7.1) and hospital LOS 6.2 days (3.6-11.0). Admission to ICU at any point of hospitalization was 60.5%. Overall mortality was 13.7%. Variables associated with mortality at the univariate level (median;IQR) any ED hypotension (1.79;1.41-2.28), any ED MAP<65mmHg (1.63;1.28-2.21), discharge diagnosis of septic shock (3.60;2.78-4.63), and ICU admission (5.40;3.85-7.59). Younger age and longer ED LOS were associated with lower mortality. Multivariate analysis (OR;95% CI) identified increasing age (1.018;1.010-1.027), ESI level 1 (2.598;1.933-3.492), septic shock (3.176;2.383-4.231) and increasing time to 30mL/kg fluids (1.006;1.003-1.010) as independent factors associated with mortality. Theoretical application of the logistic equation to the recommended 30mL/kg fluids in 3 hours would result in 32 fewer deaths compared to the study period. Delays to fluid resuscitation may result in increased mortality in sepsis. Emphasis on measures to improve the timing and delivery of resuscitation fluids in the ED are warranted. Leveraging large health system data and predictive analytics will help us to optimize and customize fluid delivery to improve sepsis outcomes.