Introduction
Various scoring systems exist to predict morbidity and mortality following liver transplantation. There is lack of evidence regarding their accuracy with regard to Sri Lankan population. This study was designed in view of applying multiple risk stratification scores to the same group of patients and compare their ability to accurately predict morbidity and mortality.
Methodology
We retrospectively reviewed 11 patients who underwent deceased donor liver transplantations over 15 months. The scoring systems Model for End staged Liver Disease model including sodium (MELD Na)1, Acute Physiology and Chronic Health Evaluation (APACHE ii)2, Sequential Organ Failure Assessment Score (SOFA)3 and Child- Turcotte-Pugh score(CTP)2 was applied to every patient on admission to intensive care unit. Descriptive statistics and non parametric tests were used to determine associations since sample size was small.
Results
The incidence of 30 day survival was 81%. Statistically significant cutoff values for prediction were taken as MELDNa 29, APACHE ii 28, SOFA 14 and CTP 9.
	Maximum score	Minimum score	Mean score	Sensitivity	Specificity	Positive predictive value	Negative predictive value
MELD Na	35	19	287	100%	88%	66%	100%
APACHE ii	33	4	19	100%	88%	66%	100%
SOFA	15	5	10.6	100%	100%	100%	100%
CTP score	12	8	9	0%	77%	0%	77%

Conclusion
Various scoring systems showed different predictive accuracies. Commonly used MELD Na score seems to be a useful marker as a predictor of mortality as compared to other complex scores which require large number of parameters. A unique model is needed in order to achieve further improvement of prognostic accuracy.

[P149] A comparison of morbidity and mortality prediction tools following deceased donor liver transplantation in the intensive care unit in a resource poor environment – a single center experience