PRISM III in a Pediatric Intensive Care Unit with Multiple Disease Entities

Choi, Seung Jun; Moon, Cheong Jun; Chun, Yoon Hong; Yoon, Jong Seo; Kim, Hyun Hee; Kim, Jin Tack; Lee, Joon Sung

Abstract

BACKGROUND
We applied the pediatric risk of mortality (PRISM) III score to study patients in a pediatric intensivecare unit (PICU), where children with various kinds of diseases were hospitalized. We analyzed whether this scoring system was useful to predict patient mortality in the PICU.
METHODS
We retrospectively analyzed the medical records of patients hospitalized in a 5-bed PICU at a tertiary general hospital. Children who were transferred to other hospitals and remained under pediatric intensive care were excluded from this study.
RESULTS
We studied a total of 105 children, which included 63 boys (60%) and 42 girls (40%). The mean age was 4.2 years (range 0-17 years). The children were admitted to the PICU for various conditions, including respiratory disease (31 children), neurological disease (30 children), congenital anomaly or neonatal disease (11 children), hemato-oncological disease (10 children), accident or poisoning (7 children), cardiovascular disease (5 children), sepsis (2 children), and the other miscellaneous diseases (9 children). The mean period of PICU stay was 9 days (range 2-66 days). Out of the 105 patients, 94 survived and 11 died. Thus, the mortality rate was calculated as 10.5%. PRISM III scores of the patients were between 0 and 38, with a mean +/- SD of 5.0 +/- 6.7. In comparison with previous studies on PICU patients with similar PRISM scores, the patients included in our study exhibited a higher mortality. The area under the curve for the prediction of mortality by PRISM III was 0.107. Among the variables included in PRISM III, Glasgow coma scale, pupillary light reflex, and platelet counts were associated with patient mortality.
CONCLUSIONS
In a PICU with a wide spectrum of diseases, PRISM III was not a useful predictor of patient mortality.