When the National Library of Medicine announced that its database would be provided free on the web through the web site PUBMED, Vice President Al Gore was on hand to encourage Americans to visit "their neighbourhood medical library", "the home page that makes house calls." Major U.S. television news programs, the Wall Street Journal, and Ann Landers devoted copy to the announcement. If not quite achieving cult status, it was clear that 's rich and unique content would now reach consumers directly.'
Librarians, who prior to 1996 had been the principle mediators between and its users, were involved with the testing of the PubMed user interface. They continue to have reservations about its value in all but limited information-seeking scenarios. For the next few minutes I want to give you information about the evolution of free on the Net, and identify some of the major issues surrounding consumer use of the database, and address issues about its value.'
What is ?
provides in-depth indexing and abstracting of articles in 3800 biomedical journals. It is produced by the National Center for Biotechnology Information (NCBI), at the National Library of Medicine, which is part of the National Institutes of Health. Its coverage includes the U.S. and 70 foreign countries, with a total of 8.6 million records dating back to 1966. Very few of the journals have any consumer focus, nor are they even intended to be read or understood by consumers.'

MEDLINE is but one of 40 databases in the MEDLARS family of databases, which includes AIDSLINE, TOXLINE, and HEALTHSTAR. There is overlap among many of the MEDLARS databases, but some unique coverage in each as well.'

The Journey from Fee to Free

How did we get to free MEDLINE on the Internet? Until the mid-70's the National Library of Medicine provided access to both the MEDLINE database and its modes of access. There were two: the printed INDEX MEDICUS, and a complex computer search interface. This search interface permitted very precise searching of medical topics. It was intended for use by medical librarians who received in-depth training in searching methodologies, and it was expected that they would use this interface regularly. That search interface is still available and is still the choice of many medical librarians for speed and effectiveness.'

The purchase of MEDLINE data by other American organizations began in early 1970's. A few third-party vendors, such as SILVERPLATTER, OVID, and DIALOG, saw several untapped markets. These vendors took the data and developed their own user interfaces, then repackaged everything and sold it to libraries, business information services, and directly to health professionals, mainly physicians.'

Grateful Med is software created by NLM for accessing its MEDLARS databases by health care professionals who are not librarians. Its introduction coincided with the increasing availability and affordability of personal computers. Using a template to guide searching and a sophisticated thesaurus called the Unified Medical Language System , the Grateful Med software enabled medically sophisticated users to conduct searches without having to know special search language. They still had to dial into the database, pay connect time charges and citation charges, but the software would mediate their encounter and guide them through the database's complex structure.'

Grateful Med was great software. A 1988 study conducted by the NLM, one of the very few available on actual success rates of MEDLINE use, showed user satisfaction with search results at 86%, and well over half of the users surveyed were Grateful Med end-users. The Grateful Med software was modified for the web and introduced on the NLM web site in April 1996. Partly experimental, it was hoped that the emerging HTML standard could solve the pesky problem of having to create variant versions of Grateful Med for all computing platforms. It was also hoped that now that there was no requirement to buy software, the web would make Grateful Med more accessible to searchers. It did not become free until June 1997. And limitations of HTML design features at the time made it appear substantially less elegant and less effective than the desktop versions of Grateful Med software.' Around mid-1996, web entrepreneurs such as Medscape, Healthgate and others began purchasing MEDLINE data and mounting the MEDLINE database on their web sites. They hoped to attract and increase audience share in order to sell advertising space or attract sponsorship. And compared to other databases, the MEDLINE database wasn't very expensive. So like the third party vendors twenty years earlier, the web entrepreneurs developed their own web-based interface to MEDLINE, with varying results. IT IS THIS DEVELOPMENT THAT SIGNALS THE BEGINNING OF CONSUMER AWARENESS OF AND INTEREST IN MEDLINE.Until its appearance on the web, hardly anyone outside the health professions community was even aware of MEDLINE's existence.'

PreMEDLINE was a new database of quickly marked up records for material that would eventually be added and fully indexed in MEDLINE. Although not directly related to this discussion, it addressed the continuing problem of information currency in MEDLINE. It is not generally known that until PreMEDLINE was introduced, some journals were added to the MEDLINE database up to a year after they were published, according to a priority indexing schedule in use at the National Library of Medicine. Internet Grateful Med became free in June 1997, and at the same time PUBMED was introduced, with great fanfare. The databases are different: PUBMED includes PreMEDLINE and MEDLINE, while Internet Grateful Med includes MEDLINE, plus AIDSLINE, TOXLINE, HEALTHSTAR and others. PUBMED uses a very simplified interface which permits complex Boolean search strategies IF you know how to construct them. Internet Grateful Med offers the user more choices at first but in a manner that I find preferable to PUBMED. Neither one is intuitively understandable to either the professional or the end user.'
'
What makes a database valuable, and how valuable is free MEDLINE on the net? If we create a list of generally accepted criteria for database value, such as those in the slide above, then MEDLINE as a database definitely meets the ones in the left hand column. The main problems are in the interface and not the database. They are important problems because an inadequate interface can dramatically undermine the quality of search results, and in certain conditions, this can have important research or clinical implications.

No consumer health focus. Regardless of the interface selected, MEDLINE is a poor choice for many topics that are of great interest to consumers, such as common conditions, lay descriptions of diseases and conditions, and alternative and complementary therapies.'

The contents of MEDLINE was never intended for consumers. The journals covered by MEDLINE almost without exception contain no titles with a consumer health focus, and if they do, they are intended for health professionals not patients. (Note: at this writing Medline has plans to include a number of consumer-focused journals in its database later in 1998) A valuable index for health professionals. MEDLINE has been demonstrated to be exceedingly valuable for health care professionals - again I refer to the 1988 study at NLM which showed a wide variety of uses of MEDLINE by medical professionals, over half in support of medical decision making, but also for research, physician education, and medical administration, and with high levels of success in search ...