Materials and Methods
Data Source Patients for this study were drawn from the Northern New
England Cardiovascular Disease Study Group (NNECDSG) Cardiac Surgery
Registry. Data in the NNECDSG registry are validated against billing
data from each hospital every 2 years to ensure complete capture of
cases and accurate vital status at discharge. The institutional review
boards at all but one of the seven hospitals have designated the NNECDSG
a quality improvement registry, and, for this reason, patient consent is
waived. The last hospital obtains patient consent. For information about
patient survival beyond hospital discharge, the NNECDSG data were linked
to the National Death Index (through 2001) and the Social Security
Administration Master Death File with complete data through the end of
2010. In 2012, the NNECDSG became a certified user of the Social
Security Administration data and, as such, receives monthly updates of
death data as well as death data from the departments of vital
statistics of Maine, New Hampshire, and Vermont.
Patient Groups and Operative Details
We compared 262 patients who underwent sutureless aortic valve (SLV)
implantation with 394 patients who underwent standard sutured aortic
valves (SAVR). These operations occurred between August 2015 and
December 2018 and were performed by three surgeons who use both valve
types at a single institution. Patients with mechanical valves or
homografts and those with endocarditis, aortic dissection or emergency
presentation were excluded. Hospital cost data (including the valve
cost) were reviewed.
Study End Points
The primary end point of this study was all-cause mortality among
patients undergoing aortic valve replacement. In addition, we examined
whether mortality outcomes differed based on type of operation and type
of valve used for the operation. Other outcomes studied included
operative times, postoperative morbidity (reoperation for bleeding,
perioperative stroke, acute kidney injury, mediastinitis or sternal
dehiscence, pneumonia, and prolonged ventilation), and length of
hospitalization.
Statistical Analysis
Baseline characteristics between groups were compared by using the c2
test for categorical variables and the Wilcoxon rank-sum test for
continuous variables. To account for differences between groups, a
non-parsimonious, multivariable propensity model was used. The model
included surgical procedure, age, sex, body surface area, preoperative
white blood cell count, prior percutaneous coronary intervention,
vascular disease, diabetes, chronic obstructive pulmonary disease,
congestive heart failure, preoperative dialysis or creatinine of 2 mg/dL
or more, New York Heart Association class 4, prior stroke, ejection
fraction, left main stenosis 50% or greater, three-vessel coronary
disease, recent myocardial infarction (within 7 days), acuity at time of
operation, and hospital where operation was performed. Standardized
differences (std dif) of means are reported for the comparison between
the two groups in the unadjusted and inverse probability weighting
adjusted data. Hazard ratios (HRs) and 95% confidence limits were
generated by using Cox proportional hazard regression models. All data
were analyzed with Stata statistical software, version 14.1 (Stata Corp,
College Station, TX).