Report on confounding adjustment for comparative effectiveness research in electronic medical records and claims DBs

WP2 aims to build a global platform for pediatric pharmacoepidemiology studies. This deliverable is based on the need to develop methods for paediatric comparative effectiveness research. In children, drugs are often used off label because of lack of appropriate efficacy evidence. Data regarding drug prescriptions and health outcomes, captured as part of routine healthcare can be analysed to provide evidence of drug effectiveness but several methodological challenges should be overcome. Children are peculiar in terms of anatomy, physiology, drug pharmacokinetics and pharmacodynamics as well as drug utilization. It is not clear if the drugs that are commonly evaluated in pediatric comparative effectiveness research reflect routine drug use. Little is known about the types of outcomes that are commonly evaluated in children. Confounding caused by selective prescribing based on indication, severity and prognosis may threaten the validity of observational database studies because of limited details about clinical information. In order to adjust for such confounding, proxy variables that are derived from the real-life experiences of treated children may be utilized to construct high-dimensional propensity scores. In this deliverable we describe a systematic review of the literature focusing on comparative studies that assessed alternative drug treatments in children. We show the amount and type of studies that were conducted in the past and reflect on the features as well as need to improve the methods. In the next steps, we will develop and test propensity scores for confounding adjustment.