OASIS is a novel severity score which aims to simplify risk adjustment in critical care. While calculation of traditional severity scores can be laborious, in particular as some scores have over 20 variables and 100 diagnostic categories, OASIS only requires 10 variables. Furthermore, these variables are routinely collected for patients in the intensive care unit (ICU) and can be used in a timely fashion (e.g. no laboratory measurements are used in OASIS).

Score card

Performance comparison

Data source	Score	AUROC	SMR
Hospitals across the USA (Cerner)*, 2010-2011	OASIS	0.837	0.91 [0.88, 0.94]
	APS III	0.822	0.95 [0.92, 0.98]
Oxford, Oxon, UK (John Radcliffe)	OASIS	0.776 [0.758, 0.795]	1.122 [1.057, 1.189]
	SAPS II	0.767 [0.748, 0.786]	0.704 [0.663, 0.745]
	APS III	0.751 [0.731, 0.771]	1.107 [1.042, 1.174]
Boston, MA, USA (MIMIC-II)	OASIS	0.790 [0.781, 0.799]	0.921 [0.890, 0.952]
	SAPS II	0.802 [0.794, 0.811]	0.559 [0.540, 0.578]
	APS III	0.785 [0.775, 0.794]	1.017 [0.984, 1.050]
	SAPS	0.764 [0.755, 0.774]	-
	SOFA	0.748 [0.738, 0.757]	-
Original publication of score	OASIS	0.837	0.91 [0.88, 0.94]
	SAPS II	0.860	-
	APS III	-	-
	SAPS	0.770	-
	SOFA	-	-

* AUROC confidence intervals were within 0.0003 of the reported value.

Details regarding these performance comparisons can be found in my thesis, available through the Oxford Research Archive, titled Mortality prediction and acuity assessment in critical care.

Code

This code is provided to ease the calculation of OASIS. If you have code for calculating OASIS in a different programming language, I would be happy to post it here and provide credit.

Clicking the below links will download the files to your computer.

• Excel calculator
• MATLAB code
• Python code - thanks Tom Pollard!

Acknowledgement

OASIS is described in the following publication: A new severity of illness scale using a subset of Acute Physiology And Chronic Health Evaluation data elements shows comparable predictive accuracy.

@article{johnson2013new,
  title={A new severity of illness scale using a subset of acute physiology and chronic health evaluation data elements shows comparable predictive accuracy},
  author={Johnson, Alistair E. W. and Kramer, Andrew A and Clifford, Gari D},
  journal={Critical care medicine},
  volume={41},
  number={7},
  pages={1711--1718},
  year={2013},
  publisher={LWW}
}

If you find OASIS useful in your research, I would be grateful if you would cite the above.