An imputation method for calculating and comparing autoimmune disease incidence using partial case review.

Area of Science:

• Immunology
• Vaccinology
• Epidemiology

Background:

• Assessing autoimmune condition incidence after HPV4 vaccination is crucial for vaccine safety monitoring.
• Electronic health record (EHR) data presents challenges due to potential inaccuracies and incomplete coding.
• Manual chart review for all potential cases is often infeasible in large populations.

Purpose of the Study:

• To develop and validate a method for estimating autoimmune condition incidence in vaccinated and unvaccinated populations using EHR data and targeted chart review.
• To address limitations of relying solely on EHR data for vaccine safety surveillance.
• To identify potential safety signals that might be missed with standard EHR data analysis.

Main Methods:

• Incident cases were identified using ICD-9 codes, laboratory results, and medication data.
• A random sample of charts was manually reviewed to confirm diagnoses and determine onset dates.
• Multiple imputation using a Monte Carlo model estimated the case status of non-reviewed charts, calculating incidence rates across imputed datasets.

Main Results:

• The proposed method successfully estimated incidence rates, providing a distribution of results.
• The model accounted for varying confirmation rates based on age and case identification methods.
• A potential safety signal was identified that would have been missed using only EHR data.

Conclusions:

• This imputation-based method offers a feasible approach for calculating disease incidence when comprehensive chart review is not practical.
• The technique enhances the reliability of vaccine safety surveillance by integrating EHR data with sampled manual review.
• Accurate incidence estimation is vital for understanding vaccine safety profiles and informing public health decisions.