Performance Characteristics of a Rule-Based Electronic Health Record Algorithm to Identify Patients with Gross and...

Area of Science:

• Urology
• Medical Informatics
• Health Services Research

Background:

• Hematuria affects 2 million patients annually, necessitating efficient evaluation methods.
• The American Urological Association guidelines recommend risk-stratification for hematuria evaluation.
• Limiting advanced imaging like CT scans to high-risk patients is a key recommendation.

Purpose of the Study:

• To develop and validate an algorithm for identifying hematuria cases within electronic health records (EHRs).
• To differentiate between gross and microscopic hematuria using EHR data.
• To enable population-level analysis of hematuria evaluation patterns.

Main Methods:

• Utilized International Classification of Diseases (ICD)-9/ICD-10 codes, urine color, and microscopy data.
• Developed and refined algorithms through iterative processes and chart reviews.
• Applied the validated algorithm to a large cohort of adult patients (n=539,516).

Main Results:

• Identified 51,500 hematuria cases and 488,016 controls with high accuracy (PPV 100%, NPV 99%).
• Successfully categorized hematuria types: 11,435 gross, 26,658 microscopic, 12,562 indeterminate.
• The gross hematuria algorithm demonstrated 100% PPV and 99% NPV; microscopic hematuria algorithm showed 78% PPV and 100% NPV.

Conclusions:

• An EHR-based algorithm accurately identifies and categorizes hematuria (blood in urine).
• This tool facilitates research into patterns of care for hematuria.
• The algorithm supports understanding and potentially improving the evaluation of this common urological condition.