Other Databases in Europe for the Analytic Evaluation of Drug Effects

Other Databases in Europe for the Analytic Evaluation of Drug Effects

INTRODUCTION

A systematic review of the abstracts presented at the 16th and 21st International Conference on Phar-macoepidemiology in 2000 and 2005, respectively, showed that the majority (53% and 51%) of submit-ted European pharmacoepidemiological studies were conducted by means of automated general practitioner (GP), pharmacy or insurance data (Table 29.1). Little has changed between 2000 and 2005. The United Kingdom ranked highest in number in 2000, basi-cally because of the wide use of the General Prac-tice Research Database (GPRD) within the United Kingdom itself. In 2005, the majority of GPRD-based abstracts comes from outside the United King-dom. Twenty-five abstracts were based on data from the GPRD, eight from the United States, nine from the United Kingdom, five from the Netherlands and three from Spain. The UK databases will not be further discussed in this chapter as they are covered elsewhere in this book.

In the ranking of abstracts that were based on auto-mated databases, the first positions were taken by Denmark, The Netherlands and Italy, which were consistent in 2000 and 2005. Spain ranked high in 2005 because of the use of the GPRD by the Spanish Center for Pharmacoepidemiology (CEIFE). A Span-ish general practice database is being established in collaboration with CEIFE (www.bifap.es), but valida-tion processes are on its way.

The Netherlands is well known for the PHARMO Record Linkage System (www.pharmo.nl) and the Integrated Primary Care Information (IPCI) (www.ipci.nl) GPRD. Other dispensing databases (Interaction) are occurring, but little information is available on them so far. In addition, there are fixed cohort studies that are linked to pharmacy data (e.g. Rotterdam study) and have proven useful for phar-macoepidemiological research, but ad hoc studies fall outside the scope of this chapter. Pharmacoepidemi-ology in the Netherlands is strong because of the availability of various academic training and doctoral programmes, the organization of health care and the availability of high-quality data (Leufkens and Urquhart, 2005).

Denmark is well known for its regional and national dispensing databases that can be linked to other national registries such as the cancer, mortality and hospitalization registry. Initially only Jutland and Funen county had prescription databases, but since 2003 the national prescription database can be used and linked to other registries at Statistics Denmark. This creates the unique possibility to study an entire country and provides a strong backbone for pharma-coepidemiologic research in Denmark.

The Italian story is quite different. In 2000, the Ital-ian studies were mostly based on claims databases that are compiled for National Health Service (NHS) payments (regional or local). Nowadays access to regional databases is complicated because of privacy legislations. Local databases are sometimes used for pharmacoepidemiological studies, but not all have the same structure and quality, combination of these databases at a local level may provide opportuni-ties in the future. Since 2000, a general practice database (former name Health Search) and paedia-tricians’ medical record database (PEDIANET) have gained importance in the field of pharmacoepidemi-ology. Both of these databases will be discussed.

The results from the abstract review underline that national differences in the organization and reim-bursement of health care have a major impact on the possibility to conduct pharmacoepidemiologic research (Leufkens and Urquhart, 2005). The coun-tries with the highest numbers of abstracts are flour-ishing because of specific features of the systems for health care delivery and the presence of (phar-maco)epidemiologists. Italy has many databases that are suitable for pharmacoepidemiological research, but the output is limited so far because of the scarcity of academic (pharmaco)epidemiology programmes.

This chapter describes general practice and record linkage databases from the Netherlands, Italy and Denmark, which have yielded peer-reviewed phar-macoepidemiological papers during the period 1990 and 2005. Since the quality of databases depends on the health care systems they are embedded in, a summary of the major health care characteristics will be provided for each of these countries. Table 29.2 provides a systematic overview of the characteristics of the databases.