Patients and methods

Study design

Between June 2005 and June 2015, a total of 61 patients underwent TVS at the Prince of Wales Hospital, Hong Kong. Local paper and electronic health records were retrospectively reviewed to collect perioperative data. Database lock date was 1st March 2021. Paper and electronic health records were reviewed to determine the occurrence of postoperative complications and cardiac events. The Hong Kong Patient Master Index provided data on vital status. The study was authorized by the Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee and consent was waived in view of its retrospective nature.

Operative technique

Median sternotomy and aorto-bicaval cannulation were the standard surgical approach. Cardioplegic arrest was achieved by administering cold crystalloid cardioplegia into the aortic root in antegrade fashion. Coronary ostia were selectively cannulated and perfused with cardioplegia in patients with aortic regurgitation. Tricuspid valve repair was routinely performed with the heart arrested using a remodeling annuloplasty ring and bicaval snares.

Definition of endpoints

The primary endpoint was operative mortality, which was defined as death within any time interval after the operation if the patient had not been discharged from the hospital or within 30 days of the operation. The secondary endpoint was late cardiac events, which were defined as the occurrence of recurrent congestive heart failure or cardiac death after discharge.

Calculation of EuroSCORE and simplified MELD-Na

Standard EuroSCORE definitions of preoperative risk factors and postoperative adverse events were used. EuroSCORE was calculated by the first author using the EuroSCORE web site (www.euroscore.org).
Most patients had blood sampling multiple times in the months leading up to surgery, usually due to prior hospitalizations for heart failure. In addition, our unit routinely requests an organ function biochemistry panel for all patients on the day prior to surgery. To ensure fair comparison and account for day-to-day fluctuations in creatinine and bilirubin measurements, it was the measurement taken the day prior to surgery that was used in the final calculation.
The following formulas were used for calculation of MELD and MELD-Na:
\begin{equation} Simplified\ MELD=\ 9.57\ x\ log\left(Creatinine\ in\ mg/dL\right)+\ 3.78\ x\ log(Bilirubin\ in\ mg/dL)\ +\ 6.43\nonumber \\ \end{equation}\begin{equation} Simplified\ MELD\ including\ Na=\ MELD-Na\ in\ mEq/L-\left[\ 0.025\ x\ MELD\ x\ \left(140-Na\right)\ \right]+140\nonumber \\ \end{equation}
International Normalized Ratio (INR) was omitted from the formula because 85.2% of the cohort had atrial fibrillation and 82.0% were on oral anticoagulants preoperatively.

Assessment of right ventricular systolic function

Significant heterogeneity exists when comparing right ventricular systolic function based on echocardiographic data alone. The most common modalities used by referring institutions were tricuspid annular plane systolic excursion (TAPSE) and fractional area change (FAC). Reporting standards vary from one institution to another. Some institutions report both while other institutions report just one parameter. For the purpose of comparison, right ventricular systolic function was classified as normal and abnormal using a TAPSE cut-off of 16mm and FAC cut-off of 35%, based on American Society of Echocardiography guidelines from 2016.8

Statistical analysis

Statistical analysis was performed with SPSS version 20.0. Descriptive statistics were reported as mean with standard deviation for continuous variables and as frequencies and percentages for categorical variables.
Univariate analysis was performed to determine risk factors for early and late mortality. For categorical data, the χ2 test was used to evaluate univariate categorical data when the minimum number of observations in a category was over 5; otherwise, likelihood ratios G-tests were used. For continuous data, simple logistic regression was used. Variables with a p-value less than 0.05 on univariate analysis were considered statistically significant.
Using logistic EuroSCORE as the base model, a hierarchical logistic regression model incorporating MELD-Na was used to ascertain the association between MELD-Na and operative mortality and major morbidities.
Survival was estimated using the Kaplan–Meier method. Survival curves were compared using Cox proportional-hazards model. The discriminatory power of MELD-Na and EuroSCORE for in-hospital mortality evaluated using area under the receiver operating characteristic (ROC) curve. The ROC analysis results were interpreted as follows: AUC <0.70, low diagnostic accuracy; AUC in the range of 0.70–0.90, moderate diagnostic accuracy; and AUC ≥0.90, high diagnostic accuracy. Survival curves were generated with MedCalc version 19.4.