Description of Data Sources. A report on the identified healthcare databases and their characteristics plus literature on their experience with respect to paediatric studies

Introduction: The available healthcare databases on infants, children, and adolescents are not adequately utilized to conduct post-authorization drug utilization and safety studies. The lack of a federation of healthcare databases restricts the capacity for meaningful investigations in these vulnerable populations. Moreover, the lack of shared methodologies to specifically retrieve paediatric information hinders access to valuable information. Objectives: One of the aims of the Global Research in Paediatric (GRiP) network (http://www.grip-network.org) is to identify and describe automated population-based healthcare databases that can provide medication and clinical information for paediatric pharmacoepidemiological researches on a global scale. Methods: We performed a web-based survey among all global databases that were identified through manual revision of the pharmacoepidemiology/pharmacovigilance conference abstracts, Bridge.to.Data software and/or by databases directly identified by members of GRiP network. The survey included questions concerning: (i) contact information for database and responsible person; (ii) nature of database (possible linkage of drugs prescriptions and/or clinical data with population); (iii) demographic, clinical and drug/vaccine related data provided, (iv) accessibility of the database for future collaboration in paediatric studies, and (v) validity of the data. Results: Ninty-nine databases were identified globally (in Europe, North- and South-America, in Asian-Pacific area, and Africa) and were invited to participate to the survey. At the time this deliverable was written, only 16 answers were received, corresponding to a response rate of 15%. In total, 75% of the respondents (N=12) accepted to collaborate with the GRiP network for future pharmacoepidemiology studies. The collaborating databases are located in 5 different European countries: Germany, United Kingdom, Denmark, Netherlands, and Italy, except for the MediGuard database that is available in more than 1 country. The data sources were set up between 1986 and 2007 providing around 16 million of total cumulative number of paediatric population (0-18 years). Nine databases capture outpatient records and 3 both, outpatient and inpatient data from primary care physicians and/or insurance claims. Both medication and clinical information are described in 11 databases. Patient-level linkage between drug prescription and clinical data is feasible for all 12 databases. Conclusions: Those databases that replied to the survey and agreed to participate provide good potential for paediatric pharmacoepidemiological studies. Thos databases that did not yet reply will be contacted in the coming months which hopefully results in participation from automated population-based healthcare databases in North- and South-America, in Asian-Pacific area, and Africa. Creating an inventory of existing health care databases and their willingness to participate in future projects is important as large databases are needed for paediatric pharmacoepidemiology research in terms of power and long term follow-up.