Study Limitations
The study end point was reoperation, which limited the conclusions drawn
about bioprosthesis failure rates. First, the valve failure rates may be
lower than actual with this approach, as patients may be lost to
valve-related mortality,26 inoperable
status,19 out-of-province reoperations, and
non-operative management of failing aortic valves rather than referral
for reoperation. Indeed, studies that focus only on reoperation rates
systematically underestimate the real incidence of
SVD.27 Thus our approach underestimates valve failure
rates, resulting in lost statistical power in comparing SVD by valve
type. Overall event rates were low, limiting statistical power as well
as the potential generalizability of the results.
This has also caused the Cox proportional hazards model to fail to show
significance of smoking on prosthesis reoperation, which is otherwise a
known predictor of SVD27. This precluded us from
examining other known predictors of SVD, or to make a meaningful
distinction between SVD and non-SVD mechanisms.
Echocardiogram data can reveal specific cause of valve failure, such as
SVD, non-structural VD, or infective endocarditis.28Such data would demonstrate more accurate rates of prosthesis failure
requiring reoperation.