External validation of SAPS II score reported to the Norwegian Intensive Care and Pandemic Registry (NIPaR)

Abstract

Background: Simplified Acute Physiology Score II (SAPS II) is a mortality prediction model widely used to compensate for differences between intensive care units (ICU) in benchmarking and research. Accuracy of SAPS II is sparsely documented. We investigate accuracy by comparing patient journal SAPS II values with registry SAPS II values in the Norwegian Intensive Care and Pandemic Registry (NIPaR). Method: NIPaR personnel collected data from the patient journal during visitations to ICUs in ten different hospitals between 2017 and 2022 while blinded for registry SAPS II data. The patient journal SAPS II values were subsequently compared with the registry SAPS II values. Results: Difference of means for SAPS II score between patient journal and registry data was 5.2 points (95% CI 2.8–7.6; p < 0.001). SAPS II score depended significantly on ICU (p < 0.001) and data origin (p = 0.006), whereas the interaction term for these two variables was not significant. Conclusion: We find low accuracy of SAPS II score in a registry setting.