The purpose of this study is to present an experience of using virtual databases on health, Adsaúde, Lilacs, Bireme_MS (Ministry of Health), Bireme_ENSP (National School of Public Health)), FSP(USP-University of São Paulo) , Bireme_Lilacssp (Public Health), Scielo and Capes Teses e Dissertações (CAPES Theses and Dissertations)in order to quantify health publications  dealing with economics. The study focuses on the problems encountered to handle, standardize and flexibilize works made available by these bases, when processing information.

The data were downloaded during the second week of December 2004, searching content by means of key-words, without descriptors and specifically working with economic information that was used, discarding the other references. The record was standardized, because a few fields did not belong to all bases and their linear disposition resulted in difficulties in filtering and classifying. Once the programming was eliminated, the data were reallocated manually , absorbing 80% of the time available to execute the ongoing work, seeking solutions to count the works without duplicating the information and maintaining full importation.

As a result, several difficulties were encountered regarding data manipulation. For instance, it is there are some difficulties to define the number of publications existing in Brazil by topic, since the same publication appears in more than one database, or else through different key-words.

A problem of another order was the impossibility of exporting data in a same format. The  Adsaúde and Lilacs bases use xml, html, ISO2709(Microisis0 and RIS FORMAT. Bireme (MS, ENSP and LILACSSP) uses xml, html, ISO 2709. Bireme FSP(USP ) and Scielo use it only in html and ISO 2709. And finally, the CAPES Teses e Dissertações base does not allow any kind of export.

It showed that Library Science addresses the content user, concentrating on the use of classification and thesaurus and their coexistence in an informatio n system, an extremely pertinent fact, since a great step forward occurred when virtual information was made available, but the statistical user who intends to quantify data encounters the difficulties already mentioned.

It concluded that the great availability of information and the means to disseminate it created several needs, the reliability of information, easy access, portability, standardization and, finally, flexible use, a weighty requirement in a society permeated by technological advances and continuous creation of tools and processes, but experience has shown that what is planned does not always fulfill all needs.