embedded research quality improvement initiative · –collects all sirs, investigative and...
TRANSCRIPT
Embedded
Research
Quality
Improvement
Initiative
Amith Shetty
Objectives
Embedded research
Shoe stringing
Background
Background
Research to Evidence based practice – delays
Capturing the effects of practice change – are we really doing better
Quality initiatives –
Usually very focused
System targets
Intended and unintended consequences
SEPSIS KILLS program:reduce preventable harm to patients with sepsis
RECOGNISE:
Risk factors, signs and symptoms of sepsis and inform senior clinician
RESUSCITATE:
With rapid antibiotics and IV fluids within one hour
REFER:
To specialist care and initiate retrieval if needed
Sepsis Bundle
– Oxygen
– Lactate
– Monitor
– Empirical Antibiotics
– Blood Cultures
– Intravenous Fluids
Sepsis Kills
Embedded research
Shoe stringing
Sepsis Pathways
Pathways guide clinicians to THINK about sepsis
NOT prescriptive ……clinical judgement is KEY
SEPSIS KILLS results
NSW hospital sepsis mortality
10
12
14
16
18
20
22
2009-2011 2012 2013 2014 2015
De
ath
s w
ith
an
d w
ith
ou
t A
uto
psy
(%
)
Principal only P+4 Comor P+5 Comor P+25 Comor P+50 Comor MJA - Comor 1-5
SMEDSA
– Sydney Multicentre Emergency Department Sepsis Archive
– Retrospective chart review populated sepsis registry approved at 5
Western Sydney EDs patients placed on the sepsis pathway
– Patients identified through clinician reported EMR alert for sepsis
based on CEC SIRS criteria or senior clinician suspicion
– Collects all SIRS, investigative and in-hospital outcome data for
identified patients
What we can already do!
Track and trigger
Self reported Time to antibiotics
Data reports
Research outcomes
At state level – CEC sepsis register – Broad coarse system level data
At district level – Multicentre data-rich Sepsis archive
Lactate in Suspected sepsis –
CEC sepsis register
ED Lactate levels risk
stratificationLactate group (mmol/L) Age median (IQR) Total, n (Died n/%)
[p]*
AE
n (%) [p]*
0 to <1 66.7 (48.1-79.4) 847 (37/4.37)
[NA]
54 (6.38)
[NA]
1 to <2 72.1 (57-82.1) 3531 (181/5.13)
[0.36]
244 (6.91)
[0.58]
2 to <3 73.1 (60.3-83) 1922 (145/7.54)
[0.0003]
198 (10.3)
[<0.0001]
3 to <4 74.3 (61.9-83.5) 897 (105/11.71)
[0.0003]
135 (15.05)
[0.0003]
≥ 4 74.1 (60.9-84) 1113 (283/25.43)
[<0.0002]
352 (31.63)
[<0.0002]
Total 72.6 (58.1-82.6) 8310 (751/9.04) 983 (11.83)
*p-values calculated for proportion difference against group below lactate group
Data learning to guideline
translation
State Level
– Time to antibiotics target extended to 120 minutes
– Lactate trigger for high degree of adverse outcome risk ≥ 2 mmol/L included
Registry data
– SIRS algorithms performance
– Broad spectrum antibiotic usage and AMS initiatives
– Multicentre data validation of qSOFA and SOFA sepsis definitions
What more have we done?
– Large dataset evaluation of sepsis algorithms in state-wide datasets
– Over 4 million events in NSW
– Cerner alert, Severe sepsis alert and qSOFA – sensitivity and specificity compared
– Multicentre ED data-sharing for validation and improvement of qSOFA
– 12555 events across multiple EDs in Australia and the Netherlands
– qSOFA – sensitivity 47.6% Specificity 89.1%
– LqSOFA(2) – sensitivity 65.5% Specificity 81.5%
Why do QI Research
Lessons learnt
– Clinician leadership locally critical
– Engagement carrots!
– Sustainability crucial
– Reproducibility
– DATA DATA DATA
– Implementation science – guidelines, knowledge generation, reflection,
adaptation and reimplementation + monitoring
Future challenges
– Clinician decision support versus Clinician Automation
– How do we track clinicians’ behaviour and suspected infection cohorts?
– Triaging patients in ED
– Tracking clinician test ordering
– What is acceptable test performance statistics?
Acknowledgements
– Dr Harvey Lander, Malcolm Green, Mary Fullick and CEC Sepsis kills team
– All NSW ED QI and staff – data in data out
– ED clinician leads at various sites and many others