Diagnostic Code-Based Screening for Identifying Children with Primary Hyperoxaluria

Gregory Tasian1,2, Kimberley Dickinson3, John Karafilidis4

  • 1Department of Surgery, Division of Urology, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania.

The Journal of Urology
|August 5, 2022
PubMed

Insights

Diagnostic codes for primary hyperoxaluria (PH) have low positive predictive value (PPV) for identifying children with this rare disease in large electronic health record databases. Careful validation is crucial for research accuracy.

Area of Science:

  • Pediatric Nephrology
  • Medical Informatics
  • Rare Diseases

Background:

  • Primary hyperoxaluria (PH) is a rare genetic disorder.
  • Accurate identification of PH patients is crucial for timely diagnosis and management.
  • Electronic health records (EHRs) offer potential for large-scale patient identification, but code accuracy can be a challenge.

Purpose of the Study:

  • To evaluate the utility of diagnostic codes for screening patients with primary hyperoxaluria (PH).
  • To determine the positive predictive value (PPV) of these codes in identifying children with PH within the PEDSnet network.
  • To assess factors influencing the PPV of diagnostic codes for PH.

Main Methods:

  • A cross-sectional study was conducted using PEDSnet data from January 2009 to January 2021.
  • Screening criteria using diagnostic codes were developed into three tiers based on hypothesized PH probability.
  • Electronic health records of potential PH cases were reviewed for diagnosis confirmation and code accuracy assessment.

Main Results:

  • Out of 341 screened patients, 33 (9.7%) had confirmed PH.
  • The overall PPV of diagnostic codes was low (20% for Tier 1).
  • PPV varied significantly by PH type (PH3: 100%, PH1: 22.8%) and was influenced by institutional data extraction accuracy.

Conclusions:

  • Diagnostic codes for PH demonstrate poor positive predictive value in large EHR databases.
  • Caution is advised when utilizing diagnostic codes for PH research without source data validation.
  • The accuracy of EHR data extraction impacts the reliability of identifying rare diseases like PH.
Abstract