What is already known about this topic:
The Simplified Acute Physiology (SAPS) 3 was published in 2005, and it is the most commonly used severity of illness score in Brazil. The performance of a score is expected to deteriorate over time, particularly in terms of calibration. Therefore, it should be reassessed periodically to evaluate whether it remains appropriate. In addition, the parameters and metrics used to estimate ICU resources were reported in 2007 and, to the best of our knowledge, have not been revalidated. The present study evaluated SAPS 3 performance in a contemporary cohort of 1,306,811 patients admitted to 1,239 ICUs participating in the Brazilian ICU Registry (UTIs Brasileiras) and investigated the need for its customization.
What this study adds:
We found that the estimated hospital mortality using the standard equation of SAPS 3 (SAPS 3-SE) and ICU resources using the original metrics to estimate the length of stay (LOS) per survivor were significantly overestimated, which resulted in low standardized mortality and resource use rates, respectively. Therefore, we performed a first-level customization (recalibration) of SAPS 3 (SAPS 3-Custom) and derived a new set of metrics to estimate the expected number of ICU days per survivor. The customization procedures corrected these overestimations and resulted in more accurate predictions of the two outcomes of interest.
How this study may affect practice or policy:
The study results support the use of SAPS 3-Custom over SAPS 3-SE to evaluate ICU performance and efficiency and for benchmarking in Brazilian ICUs. However, reevaluations of the performance of the SAPS 3-Custom and its customized LOS per survivor should be regularly performed to assess whether they remain accurate in the near future.
