Bollettino epidemiologico nazionale

Editorial

Roberto Raschetti Laboratorio di Epidemiologia e Biostatistica, ISS

“Each person in the world creates a book of life. This book starts with birth and ends with death. Its pages are made up of the records of the principal events in the life. Record linkage is the name given to the process of assembling the pages of this book into a volume” [H. L. Dunn, 1946]

As can be seen from this quotation, the concept of “record linkage” is not new and was known even before the widespread diffusion of information technology. The primary interest in using linked records rests with the possibility of using data that is routinely collected on large populations and and amplifying its informational content through linkage with other data sources. Although such linkage offers great potential, it is nonetheless important to underline some of the problems that can be encountered in using linked administrative data bases for epidemiologic purposes:

• Level of coverage: what are the levels of coverage of the target population, and, more precisely, what are the characteristics of those NOT included?

• Data quality: in this case it is appropriate to assume that only those data that have a direct impact on the management process have a level of quality that is high and that also can be evaluated. With other data sources, it is important to evaluate the possibility of missing or erroneous data.

The experience in developing information systems that are highly articulated and complex in some Italian regions, for example Friuli Venezia Giulia and Umbria, in which it is possible to correlate hospital discharge records, prescriptions, and death records, have offered and continue to offer interesting opportunities to conduct, for example, large observational studies. These, however, need to take into account the limits, not only of a technical and scientific nature, but also of the ethical and privacy considerations. The navigation of these records should not be analogous to the serendipitous nature of internet browsing but should involve defined study protocols.

In this number of the BEN, the issue of the use of large data sets and linkage is raised. The first article shows how the linkage of data sets depends on the appropriate identification of identifiers on which the linkage is based, while the second presents an interesting use of the hospital discharge records to study a specific health problem.