Authors : Dheeraj Kapoor, Meghana Srivastava, Jasveer Singh, Komal Aggarwal, Manpreet Singh
DOI : 10.18231/2394-4994.2018.0019
Volume : 5
Issue : 1
Year : 0
Page No : 114-119
Introduction and Aims: Prognostic indices are infrequently practiced in surgical intensive care units (SICU) of Indian tertiary care hospitals to assess the overall outcomes. Simplified Acute Physiology Score 3 (SAPS 3) prognostic system is now practiced globally and comprises easily measurable parameters on admission of patient in intensive care unit (ICU).The objective of the present study was to evaluate SAPS 3 index as a predictor of mortality in postoperative critically ill surgical patients admitted to ICU in a northern Indian setup.
Materials and Methods: This prospective observational study was performed in the ICU of a tertiary hospital in northern India. SAPS 3 global model was applied to measure the predicted ICU mortality. Standardized mortality ratio (SMR) was computed by comparing the observed and predicted mortality rates. To predict ICU mortality the discrimination and calibration properties of the SAPS 3 index were analysed. Data were prospectively collected at the time of admission of surgical patients in ICU. Estimated mortality rates were measured by SAPS 3 scores. Discrimination was estimated by area under receiver operating characteristic (AUROC) curves. Calibration was interpreted by Hosmer-Lemeshow goodness-of-fit C-statistic test to appraise the agreement between observed and expected number of survivors and non- survivors in alliance to the probability of death. In this analysis, P > 0.05 denotes good test adjustment.
Results: A total of 55 postoperative patients were included over 3 months period. The observed ICU mortality was 17.1%. Standardised mortality ratio (SMR) was 1.07. The SAPS 3 score of 42 showed sensitivity and specificity of 83.3% and 45.8% respectively. SAPS 3 global index had fair discrimination with an area under the receiver operating characteristic curve (AUROC) of 0.743 (CI 0.55-0.93). Patient calibration by Hosmer-Lemeshow test displayed good adjustment (P -0.388 and X2 -6.32).
Conclusion: This group of postoperative SICU patients, the performance of SAPS3 prediction model showed reasonable discrimination and good calibration in predicting mortality risk in northern Indian tertiary hospital.
Keywords: Calibration, Discrimination, Intensive care unit, Mortality, Validation.