Automating the Pediatric Index of Mortality 3 in the Electronic Health Record: Automated Risk Scoring Using Structured Clinical Documentation
In brief
Automated PIM3 scoring showed no meaningful mismatch in 20 tested cases
A structured electronic health record workflow captured Pediatric Index of Mortality 3 data in 95% of 923 pediatric intensive care encounters; 84% of completed entries were error-free. In a separate sample of 20 encounters, automated scores closely matched manual calculations, with no clinically meaningful differences. The small validation sample leaves broader accuracy to be confirmed.
- Journal
- Applied clinical informatics (Q2)
- Published
- 18 November 2026
- Study design
- Unclassified
- Evidence level
- Level 5, Expert Opinion (CEBM 5)
- Authors
- Katherine Heise, Robert Kahoud, Travis Kirkpatrick, Ruxin Dai, Arthur Gallardojohnson, Yu Kawai
- PMID
- 43051270
- DOI
- 10.1055/a-3004-6222
Why clinicians should know about it
- Picked for Health Informatics (top studies of the week, 22 November 2026): EHR automation evaluated for adoption, data quality, and score concordance
Abstract
Background The Pediatric Index of Mortality 3 (PIM3) is a validated severity-of-illness score widely used for risk adjustment, benchmarking, quality improvement initiatives, and research in pediatric intensive care units (PICUs). Its use in electronic workflows is limited by required variables that rely on clinical judgement and are traditionally abstracted manually from narrative documentation. Objectives To design, implement, and evaluate an automated PIM3 calculation within the electronic health record (EHR) using structured clinical documentation to capture non-discrete variables, and to assess clinician adoption, documentation accuracy, and concordance with manual abstraction. Methods A structured documentation template with standardized response options was developed to capture non-discrete PIM3 variables, including ventilation status, elective admission status, procedural recovery, and diagnosis category. These clinician-entered data were mapped to discrete structured data elements and combined with physiologic and laboratory inputs for automated score calculation. Documentation completion and accuracy were evaluated over a 12-month period. Automated scores were validated against manually abstracted gold standard PIM3 scores in a convenience sample of 20 targeted encounters. Results Among 923 PICU encounters, clinician compliance with structured PIM3 documentation was 95% (n=877). Of the completed entries, 84% (n=773) contained no documentation errors. Identified errors were primarily related to binary clinical judgements rather than diagnostic clarification. Of the 20 encounter validation sample, automated PIM3 scores and associated probabilities of death demonstrated near-complete concordance with manually calculated gold standard values, with no clinically meaningful discrepancies. Conclusion Automated PIM3 calculation using structured clinical documentation within the EHR is feasible and reliable for supporting standardized data capture and timely availability of risk-adjustment information. This approach reduces reliance on manual abstraction and provides a scalable informatics framework for quality improvement, benchmarking, and research in pediatric critical care.
Abstract as published, via PubMed.
For healthcare professionals. The summary is generated by AI from the published abstract, and the evidence level is assigned automatically from the study design on the Oxford CEBM hierarchy. Neither is medical advice. Read the full paper before changing practice.