Skip to main content

Comparative study of intensive care unit (ICU) perfomance

Background and objective

It is possible to evaluate ICU performance using of severity of illness score, but is this methodology objective enough to draw comparison between different units?


Data were prospectivelly collected at two hospitals during 42 months. To provide homogeneity, all data were collected following the same protocol, verified by one author and calculated with the same software, developed at Health Informatics Center-Universidade Federal de São Paulo-UNIFESP.

At hospital B patients were younger (55.8% were 50 years old or less) than those of hospital A, 27.8% were admitted due to acute trauma. At hospital A, 59.4% of patients were 60 years old or older, only 6.7% were admitted due to trauma, 37.2 % were admitted after elective surgery. The area under ROC curve showed good degree of calibration but discrimination was not adequate for both hospitals: hospital B had higher number of observed than expected deaths at all ranges of risk and hospital A had less observed than expected deaths only at low-risk admissions.

Discussion and conclusion

The comparisons between institutions must be corrected for several factors: variations in casemix, patients co-morbidites, status of previous disease, delays in referral, social factors and access to current technologies. Furthermore it is questionable the appropriateness of an American index to Brazilian hospitals. Therefore we suggest that hospitals should build up their own databases and adjust theses scores accordingly, so that they can make more relevant comparisons.

Table Results:

Author information

Authors and Affiliations


Rights and permissions

Reprints and permissions

About this article

Cite this article

Livianu, J., Orlando, J., Maciel, F. et al. Comparative study of intensive care unit (ICU) perfomance. Crit Care 2 (Suppl 1), P164 (1998).

Download citation

  • Published:

  • DOI: