Goals of presentation • beware trials stopped early for benefit • beware prestigious journal publication bias • stay skeptical of expert recommendations that are not rigorously evidence-based • beware composite endpoints • note the unbearable lightness of physiological reasoning • beware the limitations of sub-group analysis • beware limitations observational studies – comparative effectiveness
• challenges of recommendation that appropriately balance benefits and risk
Background •
100 Million adults worldwide undergo noncardiac surgery annually 1 million suffer major perioperative CV event
•
Noncardiac surgery increases catecholamines –
leads to increase in HR, BP, free fatty acid • • •
increases myocardial O2 demand increases arrhythmias beta-blockers attenuate both effects
Poldermans, NEJM, 1999 • elective vascular surgery
– positive dobutamine stress echo
• compared bisoprolol to placebo – unblinded
• primary endpoint death or nonfatal MI
Poldermans NEJM 1999 • planned sample size 266 • prior planned single look at 100 pts – stop if exceeded O’Brien-Fleming boundary • p < 0.001
• stopped after 112 patients • primary endpoint – 2 of 59 (3.4%) in bisoprolol group – 18 of 53 (34%) in placebo
• RR 0.09, 95% CI 0.02 to 0.37, P< 0.001
Composite – Random Effects Study ID
Year
RR (95% CI)
Jakobsen
1997
3 (0.13 to 69.09)
Wallace
1998
0.65 (0.17 to 2.41)
Bayliff
1999
0.73 (0.15 to 3.52)
Poldermans
1999
0.12 (0.03 to 0.43)
Raby
1999
0.25 (0.01 to 5.62)
Zaugg
1999
0.07 (0 to 1.26)
Urban
2000
0.43 (0.07 to 2.81)
MaVS
2006
0.88 (0.49 to 1.57)
Combined Estimate
0.48 (0.24 to 0.97)
p=0.1 for heterogeneity, I-squared=39% 0.001
0.01
0.1
1
10
RR (log scale)
Total events 83/1152
100
POISE - Methods • does metoprolol CR reduce 30 day risk of major CV events after noncardiac surgery • eligibility – > 45 yrs, undergoing noncardiac surgery and – CAD, PVD, stroke, CHF admission, major vascular surgery, or – 3 of 7 risk factors • high risk surgery, CHF, DM, renal insufficiency, age > 70 yrs, TIA, or urgent/emergent surgery
• intervention – metoprolol CR or placebo started 2-4 hrs preoperatively, continued for 30 days • primary outcome measure – 30 day composite of
Validity • randomization concealed • blinded – – – –
patients clinicians date collectors adjudicators
• follow-up 99.8%
Outcomes Outcome
Metoprolol (N=4174)
no.
Placebo (N=4177)
HR (95% CI)
P
0.04
Primary outcome (CV death, nonfatal MI, nonfatal CA)
243 (5.8%)
290 (6.9%)
0.83 (0.70-0.99)
nonfatal MI
151 (3.6%)
215 (5.1%)
0.70 0.0007 (0.56-0.86)
CV death
75 (1.8%)
58 (1.4%)
1.30 (0.92-1.83)
0.14
21 (0.5%)
19 (0.5%)
1.11 (0.60-2.06)
0.74
nonfatal CA
Primary Outcome
Non-fatal MI
Subgroups primary outcome Primary Subgroup
Cardiac Risk Index Risk 0 Risk 1 Risk 2 Risk 3 Risk 4+
INTERACTION P-VALUE
//2.74 0.126
//3.69
Secondary Subgroups Gender
Males Females
0.965
Surgery
Vascular Orthopedic Intraperitoneal
0.271
Secondary Outcomes Outcome
total mortality
Metoprolol (N=4174)
129 (3.1%)
Placebo (N=4177)
HR (95% CI)
P
97 (2.3%)
1.33 (1.02-1.74)
0.03
Causes of death as reported by centres Cause of death
CV death cardiogenic shock or CHF Non-CV death sepsis or infection
Metoprolol (N=129)
Placebo (N=97)
P
75 (1.8%) 7 (5.4%)
58 (1.4%) 9 (9.3%)
0.14
54 (1.3%) 36 (27.9%)
39 (0.9%) 18 (18.6%)
0.62 0.12 0.02
Total Mortality
0.02 0.01
Risk
0.03
HR(95%CI)=1.33(1.02-1.74), p=0.032
Metoprolol
0.0
Placebo
0
No. at Risk 4177 P 4174 M
10
20
30 Days
4116 4113
4091 4066
4069 4038
Total Mortality
0.02 0.01
Risk
0.03
HR(95%CI)=1.33(1.02-1.74), p=0.032
Metoprolol
0.0
Placebo
0
No. at Risk 4177 P 4174 M
10
20
30 Days
4116 4113
4091 4066
4069 4038
Secondary Outcomes Outcome
stroke
Metoprolol (N=4174)
41 (1.0%)
Placebo (N=4177)
19 (0.5%)
HR (95% CI)
P
2.17 0.005 (1.26-3.73)
Stroke
Nonfatal MI – Fixed Effects
Summary of Findings Quality Assessment Illustrative comparative risks
Outcome
Number of participants (studies)
Limitations
Consistency
Directness
Precision
Reporting Bias
Myocardial infarction
10,125 (9)
No
OK
OK
OK
No
Quality
Relative Effect Relative (95% CI) or WMD
High
0.71 (0.57 to 0.86)
5.1%
3.6% (2.9% to 4.4%)
Mortality – Fixed Effects
Summary of Findings Quality Assessment Illustrative comparative risks Quality
Relative Effect Relative (95% CI) or WMD
No
High
0.71 (0.57 to 0.86)
5.1%
3.6% (2.9% to 4.4%)
No
Moderate or low
1.23 (0.98 – 1.55)
2.3%
2.8% (2.2% to 3.6%)
Outcome
Number of participants (studies)
Limitations
Consistency
Directness
Precision
Reporting Bias
Myocardial infarction
10,125 (9)
No
OK
OK
OK
Mortality
10,205 (7)
No
Possible ↓
OK
Imprecise
Stroke – Fixed Effects
Summary of Findings Quality Assessment Illustrative comparative risks Quality
Relative Effect Relative (95% CI) or WMD
No
High
0.71 (0.57 to 0.86)
5.1%
Imprecise
No
Moderate or low
1.23 (0.98 – 1.55)
2.3%
OK
No
High
2.21 (1.37 – 3.55)
0.5%
Outcome
Number of participants (studies)
Limitations
Consistency
Directness
Precision
Reporting Bias
Myocardial infarction
10,125 (9)
No
OK
OK
OK
Mortality
10,205 (7)
No
Possible ↓
OK
Stroke
10,889 (5)
No
OK
OK
3.6% (2.9% to 4.4%) 2.8% (2.2% to 3.6%) 1.1 (0.69% to 1.8%)
Recommendations • beta blockers – for – against
• Strength of recommendation – strong – weak
Goals of presentation • beware trials stopped early for benefit • beware prestigious journal publication bias • stay skeptical of expert recommendations that are not rigorously evidence-based • beware composite endpoints • note the unbearable lightness of physiological reasoning • beware the limitations of sub-group analysis • beware the limitations of observational studies – comparative effectiveness
• challenges of recommendation that appropriately balance benefits and risk