SEPSIS RESEARCH WSHFT: THE IMPACT OF PREHOSPITAL SEPSIS SCREENING

Transcription

1 SEPSIS RESEARCH WSHFT: THE IMPACT OF PREHOSPITAL SEPSIS SCREENING Dr. Duncan Hargreaves QI Fellow Worthing Hospital Allied Health Sciences Network 2017

2 SEPSIS IMPROVEMENT AT WSHFT QUESTcollaboration -> Sepsis 6 BUFALO -> Track and trigger Sepsis stickers in ED significant improvement in screening locally > Kaizen projects in ED/EF Identification of specific process problems e.g. : Booking of GP patients on AMU delaying Abx Flow in ED when crowded Education of Junior Staff across the trust Creation of Nurse pulled rather than Dr. pushed system using Patientrack

3 SEPSIS DATABASE QUEST Project Additional data gathering as part of MEDS score and Bacteraemia research Previous projects including validation of qsofa vs NEWS in septic patients from an acute medical intake cohort External validation of Shapiro Score to predict bacteraemia -> development of modified Worthing-Shapiro model Now ongoing data collection to allow further process improvement and analysis Multi-disciplinary process -> hard work & persistance Deteriorating patient work: previous experience of automated decision tools effecting mortality reduction and process improvement with AKI

4 AIMS OF CODE YELLOW STUDY Primary: investigate the differences in patient outcomes and managementrelated processes between code-yellow patients flagged by SECAMB and those identified in the ED Secondary: To describe a cohort of patients from two ED s who were identified as possibly septic. To analyse the utility of the qsofa and NEWS scores as predictors of those ED patients with suspected infection at risk of significant outcomes (death at 30 days, admission to ICU, LoS, Confirmed bacteraemia) To identify independent predictors of outcome (e.g. mortality) which could be combined into a more discriminating prediction model

5 WHAT WAS MEASURED? Outcomes: Length of Stay Death at 30 days Escalation to ICU Process Time to antibiotics Time to delivery of Sepsis 6 BUFALO Development of true bacteraemia (positive culture within 48 hours) Final coding of Sepsis

6 WHAT IS CODE YELLOW? Derived by SECAMB themselves Rooted in national guidance (e.g. Sepsis Trust) Still using SIRS criteria (predates SEPSIS -3) Initial audit promising but low numbers No formal external validation of efficacy hitherto

7 SELECTION OF PATIENTS Ambulance code yellow vs walk-in Triage nurse decision priority patient Code yellow expedites this process Sticker inserted in notes Doctor assessment for sepsis

8 COHORT CHARACTERISTICS M > F (53% vs 47%) 31% Hypertensive 23% Diabetic 23% Vascular disease 86% Triaged as?sepsis (sticker) 4.7% AKI on admission Lactate 2.14 (±1.68) 1 st NEWS ward : median 4 IQR [2-6] 25% already shocked (SBP < 90) 3% Altered mentation (AVPU < A) 16% RR >22

9 CODE YELLOW COHORT WORTHING M > F (55% vs 45%) No statistical difference in comorbidities (incidences similar to cohort overall) Slightly less code yellow sepsis identified by triage sticker (76% vs 81%) Similar rates of shock, mental status and RR > 22 & similar to cohort overall (qsofa) Similar rates of AKI and lactate & similar to cohort overall No difference in NEWS on discharge to ward

10 CODE YELLOW OUTCOMES NEWS on leaving ED not different between cohorts. Marker of mortality so already suggestive that outcome may not be affected. No difference in 30 day mortality No difference in LoS / ICU Admission No difference in True Bacteraemia (positive BC within 48 hours) DIFFERENCE in final coded diagnosis of Sepsis or infection.

11 WHAT ABOUT PROCESS MEASURES? Time to antibiotics- significant difference in the first hour. Time to delivery of Sepsis 6 - Resuscitation Bundle Sepsis 6 at WSHFT: Blood cultures Urine output monitoring / catheter Fluid challenges to 30ml/kg Antibiotics Lactate and Hb measurement Oxygen and titrate to maintain SaO2 > 96%

12 CODE YELLOW & TIME TO ANTIBIOTICS 72% 53%

13 CODE YELLOW & COMPLETION OF SEPSIS 6 32% 23%

14 LUCA 2 SCORE DERIVATION Univariate analysis identified 15 variables associated with 30-day mortality (p < 0.01). Multiple imputation to adjust for missing values. Multivariate logistic regression in a stepwise backwards elimination method in 11 steps was performed until 6 variables remained: LUCA 2 Lactate > 4 mmol/l Urea > 10 mmol/l CCF in PMHx Cold < 36C Age > 65 AVPU < Alert

15 AUROCS FOR LUCA 2 Area under receiver operator curves (AUROC) for the new predictive model, qsofa and NEWS at the point of leaving the ED (first available): LUCA 2 : 0.79 ( ). NEWS : 0.67 ( ) qsofa : 0.63 ( )

16 LUCA 2 CUT OFFS Cut-off of 4 points had the best sensitivity of 97% but poor specificity of 23%. Cut-off of 5 points, sensitivity 80%, Specificity 68%, PPV 30% and NPV 95%. Favourable comparison to NEWS and qsofa Operationally LUCA 2 4 likely to be low risk Further temporal / external validation studies underway.

17 SO WHY SHOULD WE CARE? Performance improvement efforts for sepsis are associated with improved patient outcomes A recent meta-analysis of 50 observational studies: Performance improvement programs associated with a significant increase in compliance with the SSC bundles and a reduction in mortality (OR 0.66; 95% CI ). Mandated public reporting: NYS, CMS, UK

18 SKEPTICISM NOT CYNICISM Although a significant number of aspects of care have relatively weak support, evidence-based recommendations regarding the acute management of sepsis and septic shock are the foundation of improved outcomes for this important group of critically ill patients. (Dellinger 2013, CCM)

19 DETERIORATING PATIENTS: WHAT NEXT? NEWS and qsofa observations from point of admission: data sharing project with SECAMB Identifying Code Yellow in SRH cohort PRESEP validation of a pre-hospital score in this cohort Validation studies for LUCA 2 score Modified Worthing-Shapiro Scores for Bacteraemia validation Crowding study validating crowding measures (ICMED and NEDOCS) by correlation with clinician opinion in our EDs

20 THANK ANY YOU QUESTIONS? FOR LISTENING