Observed versus predicted mortality after isolated tricuspid valve surgery

Abstract Background Aim of this study is to analyse the performances of Clinical Risk Score (CRS) and European System for Cardiac Operative Risk Evaluation (EuroSCORE)‐II in isolated tricuspid surgery. Methods Three hundred and eighty‐three patients (54 ± 16 year; 54% female) were enrolled. Receiver operating characteristic analysis was performed to evaluate the relationship between the true positive fraction of test results and the false‐positive fraction for a procedure. Results Considering the 30‐day mortality the area under the curve was 0.6 (95% confidence interval [CI] 0.50–0.72) for EuroSCORE II and 0.7 (95% CI 0.56–0.84) for CRS‐score. The ratio of expected/observed mortality showed underestimation when considering EuroSCORE‐II (min. 0.46–max. 0.6). At multivariate analysis, the CRS score (p = .005) was predictor of late cardiac death. Conclusion We suggest using both scores to obtain a range of expected mortality. CRS to speculate on late survival.

The overall operative mortality reported was 8.8% with a significant advantage of repair over replacement (p = .009). 1 The preoperative risk assessment represents a key step in patient management. The European System for Cardiac Operative Risk Evaluation (EuroSCORE) II 2 and The Society of Thoracic Surgeons (STS) score 3 are currently used in clinical practice for preoperative risk estimation, both in surgical and transcatheter valve patients. 4 These scoring systems are based on several different preoperative comorbidities. Despite they are not validated in the field of isolated tricuspid valve procedures. In this setting, several factors not contemplated by conventional scoring systems such as right ventricular function, etiology of disease, or liver function could play a special role in the patient's outcome. With the increasing number of transcatheter procedures for the treatment tricuspid insufficiency, a dedicated method to preoperatively address the risk profile of a patient is of increasing importance. 5,6 Indeed, just recently, based on the results of a multivariable model on more than 2000 patients, an easily clinical calculable score (Clinical Risk Score [CRS] score) was established by LaPar and coauthors to estimate the probability of perioperative mortality and major morbidity after isolated TV surgery. 7 Using CRS values derived from rounded adjusted odds ratios for each factor entered into the final regression models, probability event rates were calculated for categories of clinical risk scores with a CRS score range from 1 to 10+. Depending on the calculated total CRS, predicted probability of death ranged from 2% to 34%. However, the CRS score is neither externally validated nor largely used in the current clinical practice.
The aim of the current study is the comparison the performances of two different scoring systems, EuroSCORE II and CRS score, in the setting of isolated tricuspid valve surgery. Thirty-day mortality is defined as the rate of death that occurred up to the 30th postoperative day after the surgical procedure. Inhospital mortality is defined as any death which occurred before discharge from the hospital at any time interval while operative mortality is defined as any death which occurred before discharge or up to the 30th postoperative day when the patient could have left alive the hospital before (30-day mortality + in hospital mortality). In our study population, in-hospital and operative mortality were equal.

| Registry design and data collection
All deaths for unknown reasons were considered cardiac death for statistical purposes. The observed mortalities are described as a linear rate (%).
The expected-to-observed mortality ratio were obtained by dividing the expected number of events according to the risk score by the observed one, with a value of 1.0 indicating optimal prediction, value < 1.0 an underestimation of the risk score while value >1.0 an overestimating effect.
Receiver operating characteristic (ROC) curves were generated for the risk scoring systems analyzed (EuroSCORE II and CRS score).
The ROC curves are used in clinical research to determine a special cut-off value for a special test or, in our study, a special calculator.
This algorithm analyses the relationship between the true positive fraction of test results and the false-positive fraction for a procedure.
To measure the accuracy of the risk calculator, the area under the curve (AUC) is reported. It could vary between 0.5 (lowest accuracy) and 1.0 (highest accuracy). Results are presented as AUC and 95% confidence intervals (CI). Kaplan-Meier analysis was performed to assess freedom from cardiac death. This study represents an "as-treated analysis"; given the retrospective design of the study, data regarding the "intention-to-treat population," and the relative cross-over from one treated to the other one were not collected. Statistical analysis has been executed with the IBM SPSS Statistics 27.0 (IBM Corp).

| Risk score calculation
Risk score were recorded from the analysis of the medical documentation when specifically stated in the operation report of the patient or retrospectively recalculated. The online calculator (http://www.euroscore.org/calc.html) was used for EuroSCORE II calculations. The CRS score was calculated as published by LaPar et al. 7 All variables were calculated and classified according to the exact definitions set out in each of the scoring systems.

| Patient demographic
A total of 383 consecutive patients (mean age 54 ± 16 years old; 54% female) were enrolled in the International SUR-TRI Registry.
Indication for surgery was endocarditis in 23%, functional regurgitation in 45%, rheumatic disease in 10%, and other etiologies (degenerative, pacemaker-related, and carcinoid syndrome) in the remaining 22%. In 20% of cases, surgery was executed in an urgency/ emergency setting. 157 (40%) patients had already undergone left side cardiac surgery and severe symptoms (NYHA III/IV) were present in 47% of the population. Preoperative left ventricular ejection fraction was 56 ± 10%. Moderato-severe tricuspid regurgitation (TR) was present in 96% of the population while in the rest indication was associated with valve stenosis or severe mixed disease. An isolated repair procedure (TVr) was performed in 48% of case (n = 185) and a beating heart strategy was applied in 38% of the whole population (n = 149). In the TVr group (n = 185), 68% underwent tricuspid ring implantation, 15% suture annuloplasty, 10% tricuspid valve bicuspidalization while remaing cases underwent other techniques as clover technique, vegetation removal, pericardial patch augmentation. In the TVR group (n = 198) 76% received a biological valve, 12% a mechanical valve while in 12% this data was unknown. A right mini-thoracotomy approach was performed in 21% of patients.

| In-hospital mortality
Twenty-four patients experienced in-hospital death (6.26%). Twelve The expected/observed in-hospital mortality ratio was 0.46 for EuroSCORE II and 0.95 for CRS score.

| Repair versus replacement
A subgroup analysis was performed to compare results of repair (TVr; n = 187) versus replacement (TVR; n = 196) approaches.
The expected/observed 30-day mortality ratio for the Euro-SCORE II in the TVr and TVR groups was 0.67 and 0.57, respectively (Table 2). Further, the expected/observed 30-day mortality ratio for CRS score was 1.5 (TVr) and 0.90 (TVR).

| Late results
During the follow-up period a total of 54 patients did not survive due to cardiac reasons. Freedom from cardiac death was 85 ± 2%, 81 ± 2%, and 76 ± 3% at −3, −5, and −7 years, respectively. Univariate analysis showed that age, NYHA Class III/IV, TV replacement, LVEF < 55%, COPD, preoperative EuroSCORE II as well as CRS score, and urgent surgery were predictive factors for cardiac death during follow-up. No association with type of valve implanted, biological versus mechanical, has been recorded.

| DISCUSSION
The clinical and prognostic importance of tricuspid valve disease has been increasingly studied in the past few years and the optimal treatment strategy, as well as the proper timing for invasive approaches, is still a matter of debate. 8,9 Although the tricuspid valve has been for a long time termed the "forgotten valve," recently In the present series we report the results of a multicenter experience of 383 consecutive patients enrolled in 12 different international centers with the aim to compare the performance of EuroSCORE II and CRS scores in predicting in-hospital, 30-day mortality, and late cardiac survival.
Herein we describe an AUC of 0.63 (EuroSCORE II) and 0.74 (CRS score), respectively, for the two scores considering in hospital The results of our study show that both the EuroSCORE II and the CRS score have an acceptable, but still not optimal, predictive value in the setting of isolated tricuspid surgery. More the CRS score seems to better predict mortality when compared with its counterpart.
T A B L E 2 ROC analysis for 30-day mortality. To the best of our knowledge, this is the first study that aimed to validate risk score for isolated tricuspid procedures in a multicenter design. The stratification of the patient profile and risk estimation are still lacking in the field of tricuspid disease. Despite this, the current era of clinical practice sees tricuspid disease as a hot topic both in international conferences and every day's heart team discussions. According to our data we believe that the combination of these two scores considering the EuroSCORE II and the CRS score as the lower and higher limit of a range of expected mortality rates could be an interesting approach to define the preoperative risk-profile of a patient planned for isolated tricuspid procedures. Accordingly, we recently presented a practical and integrated decisional flow chart to guide heart team decision, surgical or interventional approach as well as patient transfer to dedicated centers. 22 Moreover, according to the data here presented, the CRS score maybe an interesting tool to speculate late survival and might help to define the best treatment option when life expectancy is limited.
T A B L E 3 ROC analysis for in-hospital mortality Recently, the new TRI-SCORE has been introduced by Prof.
Dreyfus. It represents a modern and interesting tool for the evaluation of TR patients. 23 The score aimed to deal with all specific factors of tricuspid disease that were not previously considered neither in the EuroSCORE nor in the CRS Score, as right ventricular function, sing of right decompensation, dose of furosemide needed, bilirubin values. The score provides the expected in hospital mortality and should be further externally validated.

| STUDY LIMITATIONS
Data collection was retrospective and the risk calculation was strictly dependent on the information present in the medical documentation.
The study involved several etiologies determining TR and several patient's profile included endocarditis. Therefore differences in a patient's profile may alter the results. Several aspects may have played a special role as the decision-making process to perform repair or replacement and the center's expertise. Distribution of cases in the different centers have been previously reported and may represent a selection bias 24 A prospective assessment of this rare cohort of patients is the main objective of the SUR-TRI registry data collection and is ongoing. More, the number of patients enrolled, despite already considerable for the pathology itself, is still limited and larger sample size would probably lead to statistically relevant differences in the observed trends.