Mortality prediction using SAPS II: an update for French intensive care units
2005

Improving Mortality Prediction in French ICUs

Sample size: 77490 publication Evidence: high

Author Information

Author(s): Le Gall Jean Roger, Neumann Anke, Hemery François, Bleriot Jean Pierre, Fulgencio Jean Pierre, Garrigues Bernard, Gouzes Christian, Lepage Eric, Moine Pierre, Villers Daniel

Primary Institution: Hôpital Saint Louis, Paris, France

Hypothesis

Expanding the SAPS II by adding routinely collected variables would improve mortality prediction without increasing the burden of data collection.

Conclusion

The expanded SAPS II model performed significantly better than the original and customized models, making it suitable for benchmarking in ICUs.

Supporting Evidence

  • The expanded SAPS II model showed good calibration and discrimination.
  • The original SAPS II model was found to overestimate mortality.
  • Customization of the SAPS II improved calibration but not uniformity of fit.
  • The expanded model's SMR values for patient subgroups were not significantly different from 1.

Takeaway

Doctors wanted to make a better tool to predict how many patients might die in the ICU. They added some easy-to-get information to an old tool, and it worked much better.

Methodology

A retrospective database study was conducted in 106 French ICUs, analyzing 77,490 patient admissions to develop and validate three mortality prediction models.

Potential Biases

Some ICUs failed to provide the SAPS II score for more than 20% of admissions, potentially affecting the results.

Limitations

Data were not collected specifically for the study and were taken from standardized reports, which may have completeness issues.

Participant Demographics

Mean age was 56.7 years, with 59% male and 73% medical patients.

Statistical Information

P-Value

0.78

Confidence Interval

0.985–1.028

Statistical Significance

p<0.0001

Digital Object Identifier (DOI)

10.1186/cc3821

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication