Facebook Twitter YouTube


Pediatric Mortality Risk Scores and Initiation of End-of-Life Discussions In a Tertiary Pediatric Intensive Care Unit

Sunday, October 21, 2012
Room 210 (Morial Convention Center)
Alicia DeMarco, MD1, Katherine Kruse, MD1, Surender Rajasekaran2 and Michael Stoiko, MD3, (1)Pediatrics, Grand Rapids Medical Education Partners, Helen DeVos Children's Hospital, Michigan State University-College of Human Medicine, Grand Rapids, MI, (2)Pediatric Critical Care, Helen DeVos Children's Hospital, Grand Rapids, MI, (3)Pediatric Critcal Care, Helen DeVos Children's Hospital, Grand Rapids, MI


The relationship between physiologic status and mortality risk has been investigated using tools such as the Pediatric Risk of Mortality III (PRISM III) score, Pediatric Logistic Organ Dysfunction (PELOD) score, and Pediatric Index of Mortality 2 (PIM 2) score.  To date, the utility of these scores in predicting time to withdrawal or limitation of life sustaining therapies has not been investigated.  We evaluated the PRISM III, PELOD, and PIM 2 scores as tools to predict time to critical conversations with families regarding end of life care in the pediatric critical care unit (PCCU).


Data were abstracted retrospectively from Virtual PICU Systems (VPS) and chart review. The study included 62 patients aged 0-18 years treated at a 30-bed PCCU between January 2009 and July 2011. Descriptive statistics were generated and expressed as mean +/- SD. T-test was used to compare the means of all continuous variables between groups. Pearson Chi-squared test was used to determine any relationship between categorical variables. Data analysis was done using SPSS software (V.17, 2008).

The mean age of the cohort was 7.6 +/- 5.9 years.  Patients had an average ICU stay of 6.1 days (median 2).  End of life discussions were initiated at 3.9 days (median 1), and withdrawal occurred 3.6 days (median 1) after discussions were initiated.  Death occurred in 5.1 hours (range 0 – 3 days) after withdrawal.  The mean PRISM III score of the cohort was 24.0 compared to 3.1 for all PCCU admissions over the same period; mean PELOD score was 28.9 versus 8.8. 


The sickest patients (PRISM III score > 10) had shorter PCCU stays (5.9 versus 14.9 days, P=0.002) and were less likely to have documentation of end of life discussions (25.2% versus 14.2%, P=0.043). Sicker children were significantly more likely to die within 48 hours of admission to the PCCU (58.3% versus 28.2%, P=0.03).  Patients whose charts had no documentation of end of life discussions spent less time in the PCCU (1.7 versus 7.3 days, P=0.003), but were not necessarily sicker (PRISM III 23.8 versus 24.1, P=0.66).


End of life discussions are vital to helping families accept limitation or withdrawal of aggressive support, yet these discussions are often not documented in the medical record in our PCCU. Patients with the highest severity of illness scores die sooner and are less likely to have end of life discussions documented. The decision to withdraw support is extremely difficult for caregivers and families, on average requiring 3.6 days of discussion before this decision is made. Most children in our cohort passed away quickly after the parents’ decision to withdraw support. Further studies are needed to better define a best practice for end of life discussions with families and improve documentation of those discussions.