Chemical Abstracts Service (CAS; http://www.cas.org) and its online partner, STN International (http://www.cas.org/stn.html or http://www.stn-international.de), have launched a new toxicology database called Toxcenter. The database combines input from several biological information services, including BIOSIS, the American Society of Health-System Pharmacists (ASHP), the National Library of Medicine (NLM), and CAS itself. Future plans include adding input from other data providers. According to CAS representatives, the creation of the database was driven, in part, by the need to replace the NLM TOXLINE database already on STN. NLM had stopped delivering new TOXLINE data in the old format in January, requiring STN to post the TOXLINE database as a static one. NLM had then, according to CAS staff, ordered the removal of the file completely by July, but extended the order to this month. NLM continues to run a newly designed version of TOXLINE, available free on the TOXNET service (http://toxnet.nlm.nih.gov).
Toxcenter, produced by CAS, provides bibliographic information on the toxicological, pharmacological, biomedical, and biochemical effects of drugs, chemicals, and foods, as well as methodology, industrial hygiene, and legal issues and standards. Currently it contains over 5 million records, covering not only published literature but patents as well. Four major sources feed into the database: CAplus (1947 to present), MEDLINE (1958 to present), BIOSIS (1969 to present), and International Pharmaceutical Abstracts (IPA; 1970 to present). All four file segments carry CAS Registry Number indexing for efficient substance retrieval. The new file is now available on STN International's online network as well as via STN on the Web and STN Easy. The file updates weekly.
The relationship between Toxcenter and TOXLINE offers a number of interesting insights on changes in the online industry over the last decade. TOXLINE was created in the early days of online, before CAS or other leading secondary publishers had their own direct outlets through proprietary services like STN International—much less the near-universal outlet the Web now provides. NLM had its own online network in place. In those early days, the federal government had a much more visible role in providing online services than the private sector.
The TOXLINE database incorporated data licensed from CAS. That stopped several years ago when CAS, now a competitor with NLM for life sciences and biomedical online users, raised its prices beyond the reach of NLM budgets. These days the NLM has committed itself fully to a policy of putting its data onto the Web for open access at no cost. Jeanne Goshorn, chief of the Biomedical Information Services Branch of NLM's Specialized Information Services, indicated that the licensing costs for CAS data, now that they had gone to the open Web, would be completely prohibitive.
With NLM's TOXLINE files long since reduced substantially in content, the Toxcenter database approaches replacement status for the old file. A CAS representative pointed out that CAplus records supply some 40 percent of the records in Toxcenter, which, by inference, shows the scope of the data lost on TOXLINE when CAS input was withdrawn. STN charges for Toxcenter are $49/connect hour and $2 for a full record display, 82 cents for a bibliographic citation, 91 cents for an abstract, and 27 cents for indexing. Off-line prints cost 17 cents each.
Meanwhile, back at NLM, a TOXLINE file continues to survive online. But instead of one massive database, NLM has taken a modular approach with an array of databases in its TOXNET service. According to Goshorn, users going to the TOXLINE Core file, essentially a replacement of TOXBIB, reach a filtered version of input from PubMed (MEDLINE), while the TOXLINE Special file leads to historical material and other current sources. Special journal and other research literature comes from Developmental and Reproductive Toxicology (DART), the International Labour Office's CIS, and Riskline from the Swedish National Chemicals Inspectorate, along with federal technical reports and research project information, including toxicology documents and data from NTIS. The archival material (no longer updated) includes that from BIOSIS and IPA, two of the sources supplying the new Toxcenter database, as well as eight more old collections. (For more details on the restructuring, go to http://sis.nlm.nih.gov/whatsnew.cfm or to http://www.nlm.nih.gov/pubs/techbull/jf01/jf01_toxline.html) Among others, the Toxcenter database covers journal articles, conference proceedings, books and book chapters, monographs, patents, presentations, reviews, technical reports, translations, and theses.
Clearly the Toxcenter database provides more information, and in one unified database. Searching on the TOXNET service, for example, offers no de-duplication of records, while—according to Karen Sanderson of CAS—the Toxcenter database supports the Duplicate Remove feature usually tied to multiple database searching. (To be fair, de-duplication matters more to searchers paying by the hit than it does to open Web searchers.) Users can even choose which supplier's records they prefer for their de-duped results.
However, the unification of sources in the Toxcenter database apparently does not include a user-proof, integrated design. Each major segment comes with its own thesaurus. One thesaurus—NLM's MeSH—has special searchable status in the Controlled Term (CT) field, but using that field automatically limits all search output to MEDLINE-only records. To avoid this problem, users can use the Basic Index where they can make phrases by surrounding words with quotation marks, a Web-standard technique. However, that approach will stretch searching well outside just index terms, regardless of the thesaurus used.
Nor does the STN International service currently offer anything like Questel•Orbit's recently announced "super-record," through which users can pick and choose fields for display from multiple coverage of the same item. If you want to see MEDLINE headings and CAS indexing for a record, you must buy it twice. You could probably save some money by buying a full record ($2) from one source and then adding only an indexing display for another matching record (27 cents) and patching the results together off-line.
In other words, for efficient and effective searching in Toxcenter, searchers should expect to immerse themselves in details of database construction and search techniques. (How would you like to master simultaneous left- and right-handed truncation of a search term?) When expecting a substantial amount of material to come from MEDLINE records, budget-minded searchers with more time than money will probably choose to start with TOXNET searches. Besides TOXLINE Core and TOXLINE Special, the TOXNET service includes other databases, such as the Hazardous Substances Data Bank, the EPA's Integrated Risk Information System, Gene-Tox, and the National Cancer Institute's Chemical Carcinogenesis Research Information System.
CAS plans to continue negotiation for new suppliers to Toxcenter. Goshorn told me that, starting in January, NLM will open licensing to its special files at a very attractive price: free.