EGDT – the final chapter…?Kathy Rowan

Disclosures

• None

• Single centre trial• N = 263

• Protocolised EGDT versus usual care

• 16% absolute hospital mortality reduction

• Cited > 4600 times

Residual questions

• Single centre: are the results generalisable?

• High control arm mortality (46%)?• Subsequent trial ~20%1

• Low initial ScvO2 (49%)?• Subsequent studies > 65-70%1,2

• Are the catheter, blood, inotropes necessary?• Subsequent studies 3-11% transfusion1,2

1 – Jones et al. JAMA 20102 – Nguyen et al. CCM 2007

ProCESS/ARISE/ProMISe

✓ 1341 patients

✓ 31 US hospitals

✓ 1600 patients

✓ 51 hospitals in 5 countries

✓ 1260 patients

✓ 56 hospitals in England

www.icnarc.org

0.39

0.64

0.53

0.72

0.41

P value

Favors EGDT Favors usual resuscitation

3.5 or more2.5 - 3.51.8 - 2.50.2 - 1.8

Time to randomisation (hr)

6 - 145

3 - 40 - 2

SOFA Score

10 - 207 - 95 - 60 - 4

MEDS Score

78 - 9568 - 7757 - 6718 - 56

Age

HighLow

Degree of protocolised care

0.5 1.0 2.0 3.0Adjusted odds ratio (95% CI)

ProMISe subgroups

Residual questions

“…. EGDT may still be beneficial in the most severely ill patients…”

PRISMProtocolised Resuscitation In Sepsis Meta-Analysis

An IPDMA of the EGDT trials

On behalf of the PRISM Investigators

March 12, 2007

Delaney A, ANZICS CTG meeting

• Determine effect of EGDT versus usual care on 90-day mortality and other outcomes

• Compare effect of EGDT across pre-specified patient and care-delivery subgroups

• Each trial• Randomised ED patients w/n 2 hours of meeting inclusion criteria c/w Rivers trial

• Powered to detect 6-8% absolute mortality reduction

• Trained teams to deliver EGDT

• Variation in country healthcare context

Huang et al. Intensive Care Med 2013

• Prospective• Planned before enrollment of first patient

• Harmonised trial design and data collection

• Published SAP and a priori hypotheses (before unblinding of parent trials)• ClinicalTrials.gov NCT02030158

• Patient-level meta-analysis > traditional meta-analysis• Greater power to explore subgroups

Reade et al. Intensive Care Med 2010

Methods

• PRISM database preparation• Data sharing agreement put in place

• Compared trial grants, protocols, CRFs, data dictionaries to identifyany re-coding needs

• Each trial team prepared their trial’s data for pooling using a prepared, detailed dataset specification

• Data checked for missing-ness, duplication, inconsistencies, etc.

Overall effect

• EGDT versus usual care

Addressing heterogeneity of treatment effect

• Pre-randomisation, baseline patient characteristics• Age, sex, comorbidities, site of infection

• Severity of illness• shock criteria (hypotension, high lactate or both)

• lactate

• APACHE II Acute Physiology Score

• APACHE II

• SOFA

• mechanical ventilation (Y/N)

• vasopressors (Y/N)

• predicted risk of death

Addressing role of underlying treatment patterns (“usual care”)

• Pre-randomisation, care delivery characteristics• Time from ED presentation to randomisation and to first abx

• Time of randomisation

• Underlying usual care intensity• based on propensity models for use of vasopressors and fluids

Statistical analysis

• Intention-to-treat

• Hierarchical regression modeling• Patients nested in sites nested in trials• Adjusted for age, sex, SBP, APII, mechanical ventilation

• Subgroups• Categorised a priori subgroup variables into tertiles• Multivariable risk modeling to best isolate potential subgroup interaction• Examined for interaction between treatment and subgroup

• Cost-effectiveness analysis• From health services perspective to 90-days

Results

Baseline characteristics

• Cohort• 3723 pts

• 138 hospitals

• 7 countries

• Demographics• 65yo

• 58% male

• Lungs #1 site of infection (34%)

• Physiology• MAP 67

• lactate 4.3

• APACHEII 16

• Pre-randomization Rx• 93% abx

• 2L IVF (28cc/kg)

Overall results - secondary outcomes

• Increase in• ICU LOS (0.5 days)

• Vasoactive agent duration (0.3 days)

• Cost ($1-2K, to 90 days)

• No change in all other outcomes, including• Mechanical ventilation

• Renal replacement therapy

• QOL

No evidence of benefit in subgroups with highest severity of illness

• Highest APACHE II tertile• 1217 pts

• Mean score 25

• Mortality 42%

• Highest predicted risk of death tertile• 1227 pts

• Mortality 46%

No evidence of benefit in subgroups with highest severity of illness

• Both hypotension + hyperlactatemia• 628 pts

• Mean SBP 89, mean lactate 6.7

• Mortality 38%

• Highest lactate (> 4.1 mmol/L)• 1796 pts

• Mean lactate 6.7 mmol/L

• Mortality 34%

What about ScvO2?

• Unavailable pre-randomization

• Unavailable post-randomization in usual care

• No benefit in highest lactate group

Nguyen et al, Crit Care 2016

Considerable variation in usual care intensity

• We divided participating hospitals into tertiles based on propensity for• Vasopressor use

• Mean propensity = 23%, 44%, 65%

• Post-randomization fluid• Mean volume = 1.3L, 2.0L, 3.4L

No evidence of benefit at sites providing less aggressive resuscitation in usual care

Conclusion

• Harmonisation across large, international multicentre trials is feasible and valuable

• EGDT did not result in better outcomes than usual care,including in• Patients with the highest severity of illness

• Hospitals with lowest propensity to use vasopressors or fluids in usual care

Caveats and limitations

• All patients had early recognition and treatment• Lactate screening part of usual care • Prompt delivery of IVF and antibiotics• “possible that general advances in the provision of care for sepsis…explain

part or all of the difference in findings”

• PRISM did not test whether…• Early recognition was better than late recognition• Prompt care was better than delayed care

• Therefore, PRISM does not undermine efforts to…• Promote sepsis awareness• Diagnosis early• Treat early

EGDT – the final chapter…?

Implications

• Early Goal Directed Therapy• All agree on early recognition and directed care

• Only debate now is which goals and therapies…?!

• Future research• Novel goals/monitoring technologies

• Novel therapies

Acknowledgements

• Patients and their loved ones

• Site investigators, coordinators and clinical staff

• Staff and faculty of the trial coordinating centres

• TSC and DSMB members

• Funding agencies

• Dr. Rivers for his pioneering work in this area