Objective:
Validity of PRISM-III score in predicting mortality in PICU.

Design:

Prospective study over one year.

Setting:

Four bedded PICU of a tertiary hospital.

Methods:

109 prospective sick admissions in PICU (1 month – 15 years) analyzed and PRISM-III (12 and 24) score calculated and outcome recorded as survived, death, left against medical advice (LAMA).

Results:

Majority of the patients (n = 51, 46.8%) were infants (mean age was 2.89 years, male: female ratio was 3:1). Duration of stay was 9.09 – 6.38 days. Total 27 patients died. 16 LAMA were excluded from the analysis. A score of 1-9 was found in 53 patients at PRISM 12 and 20.75% of these expired. Among children who had a score of 10-19, 30.8% (n = 8) died. There were 7 children each with a score between 20-29 and 30 and the mortality was 42.8% (n = 3) and 71.4% (n = 5), respectively. The risk of mortality was significantly high with higher scores (p <0.05). The area under the ROC curve was 70%, which validates the PRISM-III score in predicting morality. No correlation was found between age/sex to the outcome of PICU. Overall mortality of 24.77% can be attributed to late presentation and delay in PICU admission.

Conclusion:

PRISM-III score can effectively predict mortality in PICU in Indian setting.

NCPCC 2005 - Conference Abstracts.Pediatric Oncall [serial online] 2006 [cited 15 May 2006(Supplement 5)];3. Available from:
http://www.pediatriconcall.com/fordoctor/Conference_abstracts/
PRISM_score.asp