Thoracoscore: Does it predict mortality in the Indian scenario? - A retrospective study

Abstract

Background and Aims: Preoperative risk stratification helps in better prognostication and allocation of resources. However, risk scoring models are less often used in thoracic surgery. Thoracoscore, a risk score model for thoracic surgery was originally developed on a French population and was later validated in many countries. As there is no literature on its ability to predict mortality in the Indian population, we aimed to validate Thoracoscore in Indian thoracic surgical patients. Methods: This retrospective study was carried out in a tertiary care centre after obtaining institutional ethics committee clearance. Patients who were operated for lung pathologies via a posterolateral thoracotomy incision between January 2014 and December 2018 were included in the study. Data on Thoracoscore variables and few additional factors (pulmonary arterial hypertension (PAH), redo surgery, blood loss, blood transfusion, duration of anaesthesia, one lung ventilation and surgery) was collected along with observed mortality statistics. Mortality was predicted using online calculator from the site https://sfar.org/scores2/thoracoscore2.php.Significant continuous and categorical variables in causation of mortality were identified using unpaired t-test and Chi-square tests, respectively. These variables were subjected to multivariate logistic regression to find independent risk factors for mortality. The calibration and discrimination of the Thoracoscore model was analysed by using Hosmer-Lemeshow test and area under the curve of receiver operating characteristic curves. Results: Overall observed mortality in the study was 3.2% while predicted mortality was 0.44%. The Thoracoscore had poor calibration and fair discrimination ability. PAH and re-operative surgery along with Thoracoscore were found to be independent risk factors of mortality in thoracic surgery. Conclusion: Thoracoscore fails to predict mortality in the Indian population.