TRI-SCORE: a single-centre validation study

Abstract

Objectives: The TRI-SCORE is a recently published risk score for predicting in-hospital mortality in patients undergoing isolated tricuspid valve surgery (ITVS). The aim of this study is to externally validate the ability of the TRI-SCORE in predicting in-hospital and long-term mortality following ITVS.

Methods: A retrospective review of our institutional database was carried out to identify all patients undergoing isolated tricuspid valve repair or replacement from March 1997 to March 2021. The TRI-SCORE was calculated for all patients. Discrimination of the TRI-SCORE was assessed using receiver operating characteristic curves. Accuracy of the models was tested calculating the Brier score. Finally, a COX regression was employed to evaluate the relationship between the TRI-SCORE value and long-term mortality.

Results: A total of 176 patients were identified and the median TRI-SCORE was 3 (1-5). The cut-off value identified for increased risk of isolated ITVS was 5. Regarding in-hospital outcomes, the TRI-SCORE showed high discrimination (area under the curve 0.82), and high accuracy (Brier score 0.054). This score showed also very good performance in predicting long-term mortality (at 10 years, hazard ratio: 1.47, 95% confidence interval [1.31-1.66], P < 0.001), with high discrimination (area under the curve >0.80 at 1-5 and 10 years) and high accuracy values (Brier score 0.179).

Conclusions: This external validation confirms the good performance of the TRI-SCORE in predicting in-hospital mortality. Moreover, the score showed also very good performance in predicting the long-term mortality.

Keywords: Isolated tricuspid valve surgery; Risk scores; TRI-SCORE; Tricuspid regurgitation; Tricuspid valve disease.

Figure 1:

Predicted probability of in-hospital mortality. This figure highlights the exponential growth of the risk of in-hospital mortality as the TRI-SCORE increases above the value 5.

Figure 2:

Receiver operating characteristic curves for patients undergoing isolated tricuspid valve surgery. The value of the area under the curve was 0.82 indicating a high discrimination of the TRI-SCORE in these patients. AUC: area under the curve.

Figure 3:

Kaplan–Meier curve for all-cause death at follow-up in patients undergoing isolated tricuspid valve surgery.

Figure 4:

Receiver operating characteristic curves regarding mortality at follow-up in patients undergoing isolated tricuspid valve surgery. The area under the curve was >0.8 at all time-frames indicating high discrimination of the TRI-SCORE also at mid- and long-term follow-up.

Figure 5:

Comparison between TRI-SCORE, EuroSCORE II and Society of Thoracic Surgeons scores. Receiver operating characteristic curve regarding in-hospital mortality.

Figure 6:

Comparison between TRI-SCORE, EuroSCORE II and Society of Thoracic Surgeons scores. Receiver operating characteristic curve regarding long-term outcome. The performance of the TRI-SCORE was excellent, with an AUC constantly above 0.80, particularly the AUC was 0.84, 0.86, 0.85, 0.86 and 0.83 at 2, 4, 6, 8 and 10 years, respectively. The performance of the EuroSCORE II was worse with an AUC constantly below 0.8, particularly the AUC was 0.74, 0.79, 0.79, 0.75 and 0.75 at 2, 4, 6, 8 and10 years, respectively. Finally, the performance of the STS was better compared to the EuroSCORE II but worse compared to the TRI-SCORE, with an AUC of 0.79, 0.79, 0.80, 0.84 and 0.85 at 2, 4, 6, 8 and 10 years, respectively. The 6th panel describes the AUC over time. AUC: area under the curve.