do ppis reduce bleeding in icu? revisiting stress ulcer ... · special considerations as of...
TRANSCRIPT
ICU-Acquired Upper GI Bleeding
Case series of 300
ICU patients describing ‘stress-
related erosive syndrome’
Frequent
Fatal
Endoscopically proven
Lucas et al, Arch Surg 1971
All About Risk
• Etiologic risk - pathophysiology
• Temporal risk - trends over time
• Baseline risk – incidence of upper GI bleeding
• Attributable risk – associated morbidity & mortality
• Risk factors - clinical
• Risk minimization – stress ulcer prophylaxis
• Risk : benefit ratio of stress ulcer prophylaxis
• Risk of withdrawing ‘standard practice’
Past Present Future
Clinically Important
Bleeding
common uncommon rare?
Prophylaxis universal targeted highly
selected?
# Prophylactic
Drugs
2 1 none?
Which Prophylactic
Drugs
Antacids,
Sucralfate
H2RAs
PPIs
H2RAs
PPIs, if any?
Impression of Upper GI Bleeding Rates
> 3.5%
< 1 %
1999 2017
Inci
den
ce
Years
…..Is SUP still needed?
Bleeding Definitions Are Key
• Endoscopic: erosions or ulcerations only
• Microscopic: occult, non-visualized
bleeding
• Macroscopic: nasogastric blood,
hematemesis, melena, hematochezia
• Clinically important bleeding: overt
bleeding with hypotension, tachycardia,
need for blood transfusions in the
absence of other causes
• >75%
• >50%
• >20%
• ~2-4%
• Objective: to describe the prevalence, risk factors, and attributable mortality of upper GI bleeding
• Design: 7 day observational study• Population: 1034 patients (without GI bleeding on
admission) • Setting: 97 ICUs in 11 countries (UK, Denmark, Sweden,
Finland, Canada, New Zealand, Australia, Norway, Iceland, Netherlands, Italy)
• Overt GI bleeding: 4.7% (3.4-6.0%)• Clinically important GI bleeding: 2.6% (1.6%-3.6%)
ICM 2015
Attributable Mortality & Length of ICU Stay
of Clinically Important Upper GI Bleeding
Mortality
Relative Risk Increase
LOS
Mean Difference
Crude comparison 2.2 (1.6 - 2.9) 17.2 (13.2 - 21.3)
Matched cohort 2.9 (1.6 - 5.5) 3.8 (-0.01 - 7.6)
Model-based
matched cohort
1.8 (1.1 - 2.9) 6.7 (2.7 - 10.7)
Regression 4.1 (2.6 - 6.5) 7.9 (1.4 - 14.4)
Adjusted regression 1.0 (0.6 - 1.7) 6.2 (1.0 - 11.4)
Cook et al & the CCCTG, Crit Care 2001
• Multicenter observational study of withholding stress ulcer prophylaxis from 2,252 patients
• Multiple logistic regression revealed only 2 independent risk factors with 4% bleeding risk:
– mechanical ventilation >48 h (OR 16)
– coagulopathy (OR 4)
• Other patients have a very low risk <0.01%CCCTG, NEJM 1994
Risk Factors for Clinically Important GI bleeding
• Independent risk factors
– >3 co-existing diseases (OR 8.9, 2.7-28.8)
– chronic liver disease (OR 7.6, 3.3-17.6)
– renal replacement therapy (OR 6.9, 2.7-17.5)
– acute coagulopathy (OR 5.2, 2.3-11.8)
– chronic coagulopathy (OR 4.2, 1.7-10.2)
– acid suppressing agents (OR 3.6, 1.3-10.2)
– higher organ failure score (OR 1.4, 1.2-1.5)
ICM 2015
57 RCTs enrolling 7293 patients
High risk of bias in 30 trials
Low risk of bias in 16 trials
Unclear risk of bias in 11 trials
ICM 2018
Design: Blinded pilot RCT
Inclusion: 4 months – 18 years
>48 h expected MV
Intervention: Pantoprazole 1
mg/kg or placebo q24h
The PIC-UP Pilot RCT
85 of 120 children randomized
7 centers
Outcomes: 1. Effective screening
2. Satisfactory enrolment
3. Protocol timeliness
4. Protocol adherence
Funding: CIHR, Hamilton Health
Sciences, IWK Health Centre
Meta-analysis of PPIs vs Placebo: 6 RCTs enrolling 713 Patients as of October 2018
Outcomes All outcomes: low quality evidence; I2=0%
Upper GI Bleeding
OR 0.96; 95% CI 0.24, 3.82; p=0.95
Clostridium Difficile
OR 2.10; 95% CI 0.31, 14.07; p=0.44
VAP OR 1.45; 95% CI 0.84, 2.50; p=0.18
Mortality OR 1.11; 95% CI 0.76, 1.61; p=0.58
Why Is the Intervention Placebo?
• Article 6
“Even the ‘best proven interventions’
must be evaluated continually
through research for their safety,
effectiveness, efficiency, accessibility
and quality”
Declaration of Helsinki
SUP-ICU Trial
Aim
• To evaluate the effects of pantoprazole vs placebo
Hypotheses• Pantoprazole will decrease bleeding
• Pantoprazole will increase infections
Sample Size • 3350 patients would give 90% power for PPIs to detect a
decrease in the 90d mortality rate of 25% by 20%, to 20%
Design • Randomized concealed blinded superiority trial
Setting • 33 ICUs in Denmark, Finland, Netherlands, Norway,
Switzerland, UK
ResultsOutcomes PPI Placebo Relative Risk
(95%CI)
Primary:
90d mortality
510
(31.1%)
499
(30.4%)
1.02 (0.91-1.13)
Secondary:
>1 Clin impt event
(Clinically important bleed,
Pneumonia, Clostridium
difficile, Myocardial ischemia)
Days alive without
advanced life support
360
(21.9%)
92
(60,97)
372
(22.6%)
92
(65,97)
0.96 (0.83-1.11)
--
Do PPIs Increase Risk of Death?
Subgroup PPI Placebo Relative Risk
(95%CI)
SAPS II >53
(37.5% sample)
272/579
(47.0%)
229/558
(41.0%)
1.13 (0.99-1.30)
SAPS<53
(p=0.05)
205/929
(22.1%)
231/967
(23.9%)
0.92 (0.78-1.09)
Pantoprazole
N~1642
Placebo
N~1640
Relative Risk
(95%CI)
Clinically
important
bleeding
41
(2.5%)
69
(4.2%)
0.58 (0.40-0.86)
Pneumonia 266
(16.2%)
266
(16.2%)
1.00 (0.84-1.19)
Clostridium
Difficile
19
(1.2%)
25
(1.5%)
0.76 (0.42-1.39)
Myocardial
ischemia
77
(4.7%)
66
(4.0%)
1.17 (0.84-1.65)
Secondary Composite
Bleeding Characteristics
PPI Placebo
Decrease in SBP, DBP or MAP by > 20mmHg
25/41 46/69
Vasopressor started or increased by > 20%
22/41 35/69
> 2g/dl hemoglobin drop 23/41 41/69
> 2U PRBC transfusion 29/41 39/69
Pts transfused overall 535/1644 (32.5%) 488/1647 (29.6%)
Transfused RBCs/patient 0 (0,1) 0 (0,1)
Bleeding Characteristics
PPI
N=41
Placebo
N=69
Endoscopy 16 28
Surgery 3 5
Coiling 2 4
Ulcer 10 17
Gastritis 4 4
Other 6 14
Bye Bye,
PPI!
This raises concern about high risk
patients!
This will
change my
practice!
In the twittersphere
Editorialists’ Take Home Messages:
“Though no mortality difference, bleeding reduction may still support PPIs, based on admittedly small 1.7% ARR”
“Additional data needed to determine effects of PPIs in the ICU, especially in patients at very high risk for this complication & to quantify any protective or harmful effects attributable to coadministration of enteral nutrition”
NEJM 2018
Decrease in
Clinically Important
Bleeding
No effect on:
Mortality
LOS
Life support
Clinically important events
Pneumonia
Clostridium difficile
Myocardial ischemia
Transfusions
Concern about harm in most severely ill
Questionable cost-effectiveness
Uncertain utility with enteral feeding
Mechanisms for Enteral Nutrition Prophylaxis
Buffering acid
InducingProstaglandins
Enhancing blood flow
Sparse RCT data For this Daily Intervention
Mortality
6 RCTs, 1124 events
3884 patients
Pneumonia
5 RCTs, 565 events
3863 patients
Clinically Important
GI Bleeding
6 RCTs, 118 events
3893 patients
Clostridium Difficile
Infection
3 RCTs, 48 events
3596 patients
Special ConsiderationsAs of November 2018
• 6 RCTs of 3800 patients
• Frailty for some outcomes
• Variable practice
• Unclear approach to pre-ICU PPI use
• Prophylactic potential of enteral nutrition
• Remaining concern about harm
• Not enough data to continue adoption
• Not enough data to continue de-adoption
• Not enough data to inform cost-effectivenes
• Unclear policy implications
INCLUSION
• > 18 years old in ICU• Invasive mechanical ventilation • Enteral nutrition
Pantoprazole indicated or contraindicated Active or high risk for GI bleed Mechanical ventilation > 72hrs Received > 24 h PPI or H2RA in ICU Dual anti-platelet therapy Limitation of life support Pregnancy
EXCLUSION
Placebo Pantoprazole
Clinically impt upper GI bleed, pneumonia, C Difficile,
AKI, mortality
Randomize