Incidence and prevalence of IPF in an insurance claims database: Assessing accuracy using medical records

Abstract

Background: Incidence and prevalence of idiopathic pulmonary fibrosis (IPF) from electronic databases without case confirmation may be inaccurate.

Objectives: Assess the positive predictive value (PPV) of claims-based algorithms to identify IPF and estimate its incidence and prevalence in the US.

Methods: We developed two algorithms to identify IPF cases 50 years of age or older in the HealthCore Integrated Research Database^SM from 2006 to 2012. One algorithm was developed to be sensitive and the other specific. Medical records were reviewed to determine the PPV of each algorithm and measure the corrected incidence and period prevalence of IPF as identified by the sensitive algorithm.

Results: We identified 4,598 patients using the sensitive algorithm and 2,052 (44.6%) patients using the specific algorithm. After medical record review, the PPV of the sensitive and specific algorithms were 40.0% (95% CI 26.4-54.8) and 56.1% (95% CI 43.3-68.3). PPV was higher in patients over the age of 65 years. The incidence and prevalence of IPF identified by the sensitive algorithm and confirmed by chart review were 12.7 per 100,000 person-years and 50.1 per 100,000 persons respectively.

Conclusions: The low PPV of the sensitive algorithm confirmed that non-validated case-finding approaches overestimate the incidence and prevalence of IPF. A revised specific algorithm with improved PPV is needed.