Use of prism scores to predict hospital stay in diabetic ketoacidosis

Academic Article


  • Background. Pediatric Risk of Mortality (PRISM) scores are used to assess mortality risk. An emergency department score(PRISMed) was calculated to test its use in triage (ICU or floor). In patients>10yo, a pulse <80 is not a predictor of increasing severity so a PRISM3 was calculated omitting heart rate. This study evaluates the use of these scores in predicting ICU and total stay. Methods. A medical record review was conducted for patients admitted for diabetic ketoacidosis (DKA) (ICD9=250.11) between 1/92 and 6/94 to The Childrens Hospital of Alabama (n=72). Inclusion criteria were age<17 years, pH<7.35 or serum HCO3<18. 3 scores were calculated: PRISM-using the worst recorded values over the first 24 hours, PRISMed-using arrival values and PRISM3 (calculated on patients whose PRISM score was elevated due to heart rate <80) without including heart rate unless >150 or <50, Median PRISM score, median tests, and Spearman Rank Correlation were determined using True Epistat Results. Median scores were as follows: PRJSM=1 l;PRISMed=12;and PRISM3=11. Median scores of PRISM, PRISMed and PRISM3 among patients admitted to ICU were less than median scores among floor admitted patients: SCORE PICU median (mean) floor median (mean) p value PRISM 9.5 (8.6) 11.0 (10.3) p=0.04 PRISMed 11.0 (8.5) 15.0 (13.4) p<0.001 PRISM3 8.0 (7.1) 12.0 (11.7) p0.001 Spearman Rank Correlations were significant for all scores versus total stay: PRISM, r= 0.29, p=0.009; PRISMed , rs 0.60, p<0.001, PRISM3, rs= 0.48, p<0.001. Correlations were significant for all scores versus ICU stay. PRISM, rs= 0.22, p= 0.05; PRISMed, rs = 0.41, p<0.001, PRISM3 rs=0.41, p<0.001. Conclusion Triage guidelines for ICU versus floor admission for DKA patients could have significant economic impact (mean DKA-ICU charge= $11,417 versus mean DKA-floor charge- $4,447). PRISM scores may be an important variable to include in a multiple regression model used to predict the need for ICU monitoring. We question the validity of heart rate in calculating PRISM scores in patients > 10yo.
  • Published In

    Author List

  • Monroe K; King W; Atchison J
  • Volume

  • 44
  • Issue

  • 1