Mosby-Year Book, Inc., acquired GenRx (The Complete Reference for Generic and Brand Drugs) from another publisher in early 1995. At that time this 3,000-page drug reference was issued annually in print. From the beginning, Mosby's goal was to continuously update the content and sell it in print, on CD-ROM, on the Web, and in custom multititle editions. Texterity was chosen in June 1995 to design and build the editorial and production systems.

A year later, the print and CD editions were being produced directly from a single set of SGML (Standard Generalized Markup Language) master files. In January 1997, Mosby released its very successful intranet and Internet product, GenRx. [Editor's note: By 2001 Mosby had discontinued its Internet version of GenRx.]

GenRx has become a model for how Mosby wants to manage and sell its reference content. This is the story of what was involved in achieving that success.

Condition of Acquired Content

The content acquired by Mosby was in a proprietary database format. All coding was strictly for typesetting purposes. The genius of GenRx is its compilation of data from multiple sources: A genericized package insert is introduced by brand, cost of therapy and classification information, and followed by dosages, forms and suppliers. The original data-management approach, however, had mixed all those content types to make it easy to compose pages. From the typesetter's point of view, elements that looked the same were coded the same, losing all the distinctions about the source. As a result, updating was laborious, accuracy was difficult to check, and creating effective electronic products was impossible. Mosby realized it had a lot of work to do to achieve the potential of GenRx.

Texterity created DTDs (Document Type Definitions) for each component of GenRx content. For example, one DTD was created for drug inserts, another was created for suppliers (pharmaceutical manufacturer or distributor), and another for "how-supplied" listings (brand, package size, etc.). Separate DTDs for each type of content was the cornerstone that enabled each type of content to be updated independently.

Once the DTDs were complete, the typesetting files were programmatically transformed into the targets; each DTD was a target for each class of content. That was a two-step process in which Texterity derived the structure and then Mosby staff added missing semantics. Examples of semantic coding that was added include unit doses, HCFA (Health Care Financing Administration) classifications, and section-to-section and drug-to-drug cross references.

Editorial System

The editorial system had to meet several challenges:

  • Each class of content needs to be updated on different schedules. For example, drug inserts can change at any time during the year but HCFA classifications are issued every quarter.
  • Editors, for historic and economic reasons, are in different locations, so remote access and a foolproof means of check-out/check-in was required.
  • Some content types, such as data feeds from HCFA and other sources, have to be completely reloaded on a regular basis.

To meet those requirements, an SGML-based editorial system was developed by Texterity. Editors can select drugs based upon drug ID or generic name. The editors check out the SGML files for those drugs, edit them, and check them back in. The database is updated automatically from the SGML. Reports on the status and contents of the SGML files can be generated at any time, and that reporting function has become an important tool for enhancing quality.

The system is browser-based and uses Java applets for each component of the editorial and production workflow. Freelancers and work-at-home editors are very productive with modem connections.

Production System

The primary function of the production system is to integrate the multiple SGML-content types into a composite record for each generic drug. Once complete, the composite records are sorted alphabetically by generic name because that is needed for print (the electronic media are indifferent to order). The system supports three media — print, CD-ROM, and the Web. The composite SGML files are provided to selected distributors.

"A visionary champion is critical"

Each medium required a one-time setup to transform the SGML into the target media. Those setups were done by the compositor for print, by the CD packager, and by Texterity for the Web products. The composed pages are quite dense, containing almost 20,000 characters each. The book is produced annually and an interim printed supplement is issued twice a year. CDs are updated quarterly and are targeted primarily to the single-user market. The Web products are updated every 60 days and are targeted to institutions such as hospitals, managed-care organizations, and drug companies.

Lessons

We learned three major lessons in creating GenRx's success.

  1. A visionary champion is critical. Key people within Mosby shared a vision for GenRx and secured the funding to make the project possible.
  2. SGML is very powerful for reference products. GenRx is an example of the substantial benefits that come from changing the center of the universe from the typesetting system to SGML.
  3. A winner is the best advocate. Mosby's success in publishing GenRx in multiple media is informing all reference publishing within the company.

Comments from the Front Line

I asked the people we worked with at Mosby to comment on the process now that it's been running for over a year, and the following quotes are from those interviews/e-mail responses:

  • "Most of the difficulties were not really SGML-related, except for our compositor's initial conversion to SGML. He's off the Valium, now, and he's doing just fine. In getting the editorial system up and running, our biggest problem was eXceed X-Windows software for PCs; switching to a higher version mostly solved this problem. Our editors have very little trouble learning to use the SGML editor." [Note: That difficulty with eXceed is why all later versions of Texterity's editorial systems are browser-based.]

  • "Once the initial wrinkles were ironed out, our biggest problems are not technical but typical editorial ones, e.g., acquiring source data in a timely fashion. SGML actually helps us editorially, because the tags allow us to query the database using criteria like, 'Which drugs don't have a monograph element?' We can easily identify gaps in our database."

Mosby has achieved all its goals for GenRx. It continuously updates the content and sells it in print, on CD, and in Web editions. Unanticipated benefits have been lower composition costs and a later editorial close for the annual print edition. The Web product is licensed for local installation on intranets and its great success has encouraged Mosby to launch other content collections via the same channel [formerly http://www.texterity.com:8081/books/mosby/Web/welcome-carelink.htm]. (Subscriber ID is jep and password is jep.)



Martin Hensel has worked in electronic publishing for more than 20 years. As a founder of LaserData he was an early pioneer in videodisk and CD-ROM publishing, and was chairman of the National Information Standards Organization (NISO) committee that developed ISO 9660, the worldwide CD-ROM logical format standard. As general manager of Thomson's Investext division, he was instrumental in its success. His experience creating print and electronic products for many publishers gave him insights into the power and flexibility that can come from detailed markup. In 1991, he founded the company that became Texterity to provide publishers the means to easily re-purpose their content in all media. Martin received an MBA from Harvard University in 1975. He may be reached at martin.hensel@texterity.com.