Predictive Analytics In Healthcare: Medications as a Predictor of Medical Complexity.

Area of Science:

• Pediatric Healthcare
• Health Services Research
• Medical Informatics

Background:

• Children with special healthcare needs (CSHCN) require extensive services.
• Medically complex children (MCCs) are a growing subset of CSHCN with comorbidities, consuming significant healthcare resources.
• Rapid identification of MCCs in hospitals is needed for effective planning.

Purpose of the Study:

• To assess if medication data can predict medically complex children (MCC) status.
• To develop a simple screening tool for identifying MCCs in a hospital setting.

Main Methods:

• Retrospective analysis of inpatient medication data at Seattle Children's Hospital.
• Correlating medication counts (inpatient/outpatient) and drug classes with MCC status.
• Developing a predictive model based on medication variables.

Main Results:

• The number of inpatient and outpatient medications significantly correlated with MCC status.
• A model using outpatient and inpatient drug classes and individual inpatient drug names effectively ranked patient complexity.
• The model provided a simple, objective method for screening potential MCCs.

Conclusions:

• Medication data, specifically counts and classes, can effectively identify medically complex children (MCCs).
• This approach enables the development of simple, objective admission screens for predicting pediatric patient complexity.
• Implementation of these screens facilitates strategic planning for MCCs' healthcare needs.