Estimation of vaccination coverage from electronic healthcare records; methods performance evaluation - A contribution of the ADVANCE-project

Abstract

INTRODUCTION: The Accelerated Development of VAccine beNefit-risk Collaboration in Europe (ADVANCE) is a public private collaboration aiming to develop and test a system for rapid benefit-risk (B/R) monitoring of vaccines, using existing electronic healthcare record (eHR) databases in Europe. Part of the data in such sources is missing due to incomplete follow-up hampering the accurate estimation of vaccination coverage. We compared different methods for coverage estimation from eHR databases; naïve period prevalence, complete case period prevalence, period prevalence adjusted for follow-up time, Kaplan-Meier (KM) analysis and (adjusted) inverse probability weighing (IPW). METHODS: We created simulation scenarios with different proportions of completeness of follow-up. Both completeness independent and dependent from vaccination date and status were considered. The root mean squared error (RMSE) and relative difference between the estimated and true coverage were used to assess the performance of the different methods for each of the scenarios. We included data examples on the vaccination coverage of human papilloma virus and pertussis component containing vaccines from the Spanish BIFAP database. RESULTS: Under completeness independent from vaccination date or status, several methods provided estimates with bias close to zero. However, when dependence between completeness of follow-up and vaccination date or status was present, all methods generated biased estimates. The IPW/CDF methods were generally the least biased. Preference for a specific method should be based on the type of censoring and type of dependence between completeness of follow-up and vaccination. Additional insights into these aspects, might be gained by applying several methods.