Table A15:

GRADE Evidence Profile for Comparison of TAVI Versus SAVR (Effectiveness Outcomes at 1 to 2 Years)

No. of Studies (Design)	Risk of Bias	Inconsistency	Indirectness	Imprecision	Publication Bias	Upgrade Considerations	Quality
Composite End Point of All-Cause Mortality, Stroke, or Rehospitalization at 1 Year—Noninferiority Hypothesis
1 RCT18	No serious limitations	Cannot be evaluated	No serious limitations	No serious limitations	Undetected	Not applicable	⊕⊕⊕⊕ High
Composite End Point of All-Cause Mortality or disabling stroke at 2 years—Noninferiority hypothesis
1 RCT17	Serious limitations (−1)a	Cannot be evaluated	No serious limitations	No serious limitationsb	Undetected	Not applicable	⊕⊕⊕ Moderate
All-cause mortality at 1 year
2 RCTs17,18	No serious limitationsc	No serious limitations	No serious limitations	Very serious limitations (−2)d	Undetected	Not applicable	⊕⊕ Low
All-cause mortality at 2 years
1 RCT17	Serious limitations (−1)a	Cannot be evaluated	No serious limitations	Very serious limitations (−2)d	Undetected	Not applicable	⊕ Very Low
Any Stroke at 1 Year
2 RCTs17,18	No serious limitationsc	No serious limitations	No serious limitations	Very serious limitations (−2)d	Undetected	Not applicable	⊕⊕ Low
Disabling Stroke at 1 Year
2 RCTs17,18	No serious limitationsc	No serious limitations	No serious limitations	Serious limitations (−1)e	Undetected	Not applicable	⊕⊕⊕ Moderate
Disabling Stroke at 2 Years
1 RCT17	Serious limitations (−1)a	Cannot be evaluated	No serious limitations	Serious limitations (−1)f	Undetected	Not applicable	⊕⊕ Low
Nondisabling Stroke at 1 Year
2 RCTs17,18	No serious limitationsc	No serious limitations	No serious limitations	Very serious limitations (−2)d	Undetected	Not applicable	⊕⊕ Low
Transient Ischemic Attack at 1 Year
2 RCTs17,18	No serious limitationsc	No serious limitations	No serious limitations	Very serious limitations (−2)d	Undetected	Not applicable	⊕⊕ Low
NYHA Symptoms at 1 Year
2 RCTs17,18	No serious limitations	No serious limitations	No serious limitations	Serious limitations (−1)g	Undetected	Not applicable	⊕⊕⊕ Moderate
Quality of Life (KCCQ Overall Score) at 1 Year
2 RCTs17,18	No serious limitations	No serious limitations	No serious limitations	Serious limitations (−1)g	Undetected	Not applicable	⊕⊕⊕ Moderate
6-Minute Walk Test at 1 Year
1 RCT18	No serious limitations	Cannot be evaluated	No serious limitations	Serious limitations (−1)g	Undetected	Not applicable	⊕⊕⊕ Moderate
Aortic Valve Reintervention at 1 year
2 RCTs17,18	No serious limitationsc	No serious limitations	No serious limitations	Very serious limitations (−2)d	Undetected	Not applicable	⊕⊕ Low
Rehospitalizations at 1 year
2 RCTs17,18	No serious limitationsc	No serious limitationsb	No serious limitations	Serious limitations (−1)e	Undetected	Not applicable	⊕⊕⊕ Moderate
Valve hemodynamics at 1 year
2 RCTs17,18	No serious limitations	Serious limitations (−1)h	No serious limitations	Serious limitations (−1)i	Undetected	Not applicable	⊕⊕ Low

Abbreviations: GRADE, Grading of Recommendations Assessment, Development, and Evaluation; KCCQ, Kansas City Cardiomyopathy Questionnaire; NYHA, New York Heart Association; RCT, randomized controlled trial; SAVR, surgical aortic valve replacement; TAVI, transaortic valve implantation.

^aAlthough appropriate imputation methods were used in this study, large extent of imputation used (i.e., approximately 90% of patients had not reached 2 years of follow-up at time of the analysis) could potentially increase risk of bias for this outcome.

^bDifference between TAVI and SAVR was not statistically significant; however, noninferiority criterion (main end point of this study) was satisfied (P < .0001 for noninferiority).

^cOne study had a large degree of data imputation (i.e., approximately 45% of patients had not reached 1 year of follow-up at time of analysis. Although appropriate methods for imputation were used, this could potentially affect estimates obtained. Because we consequently cannot determine extent to which the risk of bias might be increased, we decided not to downgrade for risk of bias.

^dStudy had very low statistical power to detect difference between groups. Evidence was downgraded by two levels given very few events in intervention and control groups of both studies and given fragility of estimates (an increase of a few events in either group could change direction of point estimate or render results not statistically significant).

^eOne study reported a statistically significant result, but the other study did not.

^fStudy result was statistically significant. However, we decided to downgrade owing to fragility given the very small number of events in each group (fragility of estimate).

^gStudies were not adequately powered to detect difference between groups.

^hResults were inconsistent between studies.

ⁱIt is unclear whether small differences observed between groups are clinically important.