University of DundeeSchool of Medicine Improving the management of sepsis in general hospital wards Dr Charis Marwick CSO Clinical Academic Fellow & SpR.

University of Dundee School of Medicine

Improving the management of sepsis in general hospital wards

Dr Charis MarwickCSO Clinical Academic Fellow & SpR Infectious Diseases

Prof. Peter DaveyProfessor and Consultant in Infectious Diseases

In comparison with severe sepsis on arrival at hospital, less is known about...

• Hospital inpatients who develop sepsis • The potential to improve care for these patients

in general hospital wards• Management in earlier stages of sepsis

– Logical to intervene before deterioration

• Patients without proven bloodstream infection– Previous studies focus on positive blood cultures– Only includes 7-17% of septic patients1

– Mortality and morbidity similar whether +/– ve1,2

1.Jones and Lowe 1996, 2.Kumar et al 2006

Defining the problem• Prospective case-note reviews hospital inpatients

– Develop case identification method: blood cultures taken

– Quantify deficiencies in patient management– Baseline Sept 2008 – Feb 2009 – Post-intervention Oct 2009 – Mar 2010

• Mortality among septic inpatients

Baseline clinical data

Demographic characteristics (n=339)

Mean age : 67 years (range 18-95)

Male gender: 193 (57%)

Ward type:General medicineGeneral surgeryOrthopaedicOther

140 (41%)120 (35%)

31 (9%)48 (14%)

Suspected site of infection:Respiratory tractSkin or soft tissueUrinary tractIntra-abdominalLine infectionOther More than one site

145 (43%)46 (14%)79 (23%)79 (23%)35 (10%)30 (9%)

68 (20%)

Intervention target

• 1144 patients screened, 339 (30%, 95%CI 27-32%) valid cases

Sepsis patients per ward

0

1

2

3

4

5

6

7

Medicine:26 patients per month;

11 Wards

Surgery:21 patients per month;

6 Wards

Orthopaedics:5 patients per month;

4 Wards

Seps

is p

atien

ts p

er w

ard

per m

onth

Mean 2.3

Mean 3.6

Mean 0.7

Baseline study outcomes

Timing antibiotic therapy after sepsis onset (n=279)

Within four hours 107 (38%, 95%CI 33-44%)

Mean 10.9 hours (95%CI 9.3-12.4)

Median 6.0 hours (IQR 2.4-13.0)

Within eight hours 169 (61%, 95%CI 55-66%)

Timely medical review (n=291) 139 (48%, 95%CI 42-54%

Blood cultures before antibiotics (n=268) 212 (79%, 95%CI 74-84%)

Severity assessment (n=339) 80 (24%, 95%CI 19-28%)Main component of delay = time between medical review and antibiotic prescription (mean 7.2 hours, median 2.5 hours)

Where do delays occur???

?? 1.0

0.0

3.2

7.1

0.9

Mean time in hours

Median time in hours

Main delay is from review to prescription

2.4

Improvement strategy• Implement intervention in Medical, Surgical and

Orthopaedic wards– 86% patients, feasible

• Sepsis “tools” = clinical care pathways– Recognition, risk stratifying and management

• Education and raising awareness – Presented to >300 clinical staff in Ninewells

• Monthly performance feedback to clinicians– Displayed as posters on intervention wards – Emailed to clinical staff

Outcome measure Pre-intervention cohort

Post-intervention cohort

Difference and significance test result

Antibiotics within four hours

91/241 (38%)(95%CI 32-44%)

139/297 (47%)(95%CI 41-52%)

9%X2=4.44, df=1, p=0.04

Antibiotics within eight hours

145/241 (60%)(95%CI 54-66%)

198/297 (67%)(95%CI 61-72%)

7%X2=2.43, df=1, p=0.12

Mean time to antibiotics

11.0hrs(95%CI 9.3-12.7hrs)

9.5hrs(95%CI 8.1-11.0hrs)

1.5hrst=1.30, df=536, p=0.19

Median time to antibiotics

6.0hrs(IQR 2.5-13.3hrs)

4.5hrs(IQR 2.0-12.0hrs)

1.5hrsU=32460, p=0.06

Timely medical review 118/251 (47%)(95%CI 41-53%)

126/250 (50%)(95%CI 44-57%)

3%X2=0.58, df=1, p=0.49

Blood cultures taken before antibiotics

183/230 (80%)(95%CI 74-85%)

246/290 (85%)(95%CI 81-89%)

5%X2=2.46, df=1, p=0.12

Blood lactate level measured (severity)

31/291 (11%)(95%CI 7-14%)

87/346 (25%)(95%CI 21-30%)

14%X2=21.99, df=1, p<0.01

Summary• Sepsis is common (>40 cases per month)

in Medical and Surgical Specialties • BUT, each Ward only has 1-6 patients per

month• Main delay in Time to First Antibiotic

Dose occurs AFTER medical review• Guidelines, education, audit &feedback

at Specialty level had little impact

Conclusions• Collection and reflection on measures for

improvement should be at Ward level– Weekly identification of case(s)

• EWS charts• Antibiotic prescriptions• Blood cultures• HDU transfers

– Weekly run chart of individual patient Time to First Antibiotic Dose

– Monthly report on Sepsis Six

Run Chart, Medical Ward, Sepsis & EWS 4+

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Tim

e to

Firs

t Anti

bioti

c D

ose

(hou

rs)

Cases in date order

RESULTS: MORTALITY

Sepsis at Ninewells Hospital• 12 months data

Total Per Month

Blood cultures taken 2603 217

Patients screened for sepsis 2157 180

Patients with sepsis 1342 (62% BCs) 111

Hospital onset sepsis 641 (48% sepsis) 53

0

1

2

3

4

5

6

Any blood culture versus comparators

BC with Sepsis versus comparators

BC without sepsis versus comparators

Odd

s Rati

o 30

Day

Mor

talit

y

Odds ratio for mortality in comparison with patients hospitalised on the same wards with the same length of stay (+ 1 day), adjusted for age, gender and co-morbidity

13% definite +ve

2% definite +ve

Mortality, multivariable analysis• 30 day: 124/640 (19%, 95%CI 16-22%)• 90 day: 180/640 (28%, 95%CI 25-32%)• Age (not comorbidity, gender or SIMD) associated• Severity scores risk-stratify, CURB65 performed best• Admission type, days to onset, and ward associated

Proposal• Mortality (30 day) in any patient who has

had a blood culture taken is likely to be a more specific outcome measure for sepsis than total hospital mortality

• Further work with SPSP hospitals & ISD– Prevalence of sepsis in BC patients– Identification of BC patients by Ward– Record linkage to standardise mortality