Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 101
6 Accessing and Navigating the TEHIP Databases Although the audience served by the TEHIP program encompasses a number of different user communities with a variety of information needs and a range of sophistication in searching online databases, there are common issues involved in accessing and navigating the TEHIP databases. These issues encompass the entire search process from establishing the necessary computer connections, to the development of a search strategy, to the evaluation of the search results. Recommendations to NLM involving the refinements in the technology associated with the TEHIP databases will facilitate use by all interested searchers. This chapter focuses on searching the TEHIP databases directly through NLM and does not address the diverse interfaces and search mechanisms available through commercial vendors or academic institutions. Several search options for NLM users are discussed throughout the chapter and are highlighted in Box 6.1. Additionally, this chapter discusses a range of issues that affect access and navigation and presents the committee's short- and long-term recommendations for refining the TEHIP databases. Access: Getting Connected To The Databases Although this is a time of increased computer use by health professionals, input to the committee from focus group participants indicates that access to the TEHIP database system is not readily available (Appendix C). Access to the databases was selected as the primary factor that limits use of the TEHIP databases by those individuals responding to a committee questionnaire (Appendix B). Although access to MEDLINE is a standard feature of end-user
OCR for page 102
BOX 6.1 Methods of Searching the TEHIP Databases Grateful Med: Grateful Med was developed to simplify MEDLINE searching. However, input screens have gradually been added for other databases, and users can now access 20 NLM databases through Grateful Med. Grateful Med currently offers input screens for all of the TEHIP databases except DART/ETICBACK, EMIC/EMICBACK, GENE-TOX, and TRIFACTS. In 1996 Internet Grateful Med was introduced and is currently available for searching MEDLINE, HealthSTAR, AIDSLINE, and PREMEDLINE. Direct Searching: The only fully operational method available for accessing all of the TEHIP databases is direct searching using NLM's search command language. Direct searching involves the use of commands, mnemonics (abbreviations indicating database fields), and Boolean logic for combining terms. This interface requires a high level of search proficiency and expertise. Menu Searching: Two of the TEHIP databases, TRI and HSDB, have menu search interfaces with step-by-step options for assisting users in developing a search strategy. Experimental World Wide Web Interface: SIS staff are in the process of developing a new approach to searching the TEHIP databases that uses the hypertext and graphics capabilities available through the World Wide Web. workstations in health science libraries, fewer library workstations provide end-user access to the TEHIP databases. NLM is in the somewhat unique position of playing a number of different roles in the database process. For a few of the TEHIP databases, such as HSDB and ChemID, NLM is the creator of the database and is responsible for the content. In other cases, such as CCRIS and IRIS, other federal agencies are responsible for content and NLM is responsible for maintaining the database on the NLM computer servers (see Chapter 2). These agencies may provide additional access points for their databases such as leasing them in online or CD-ROM formats. For all of the databases in the MEDLARS system, NLM acts as a database provider and offers online search access for registered users. Additionally, NLM as a database producer leases its databases to commercial vendors, academic institutions, libraries, and other organizations. The result is a variety of access routes available to users interested in searching the NLM databases. However, the number of access routes varies depending on the specific database (Figure 6.1). MEDLINE is offered by many commercial vendors in numerous CD-ROM and online formats, and most health science libraries and biomedical organizations provide their users with access to MEDLINE. On the other hand, the more specialized databases have fewer points of access, and there is only one access point—direct searching—that allows the user to connect with the entire TEHIP complement of all 16
OCR for page 103
databases. This is particularly problematic because direct searching's complex command-line interface requires the most expertise in online searching (see discussion below). There are advantages and disadvantages to the various access modes. Commercial vendors may add value to the databases by providing easy-to-use search interfaces and may integrate the database into the vendor's network of databases on similar topics. Academic institutions and hospitals may offer access to the NLM databases as a component of the campus-wide network with connections to a wide range of applications including library services and email. Accessing the databases directly through NLM is lower in cost than accessing them through commercial databases because the search costs (user fees) are charged only to cover access costs (e.g., telecommunications; see Chapter 2). NLM's federal budget appropriation is used to build and maintain the databases. As with many databases, users are required to register to access the NLM databases and are assigned an NLM User Identification (ID) code and password.1 In June 1996 a new feature was added to Internet Grateful Med (Box 6.1) allowing users to establish a new NLM account online and to use it for immediate searching. In most cases, however, registration is handled through mail or fax. There is no fee to register for a user ID code, and the code allows access to all of the NLM databases. However, the time and effort required to apply for the ID and the inconvenience of having another set of numbers and codes to remember may present barriers to some users. Simplifying the registration process and promoting the availability of online registration will encourage health professionals and other interested individuals2 to become connected to the TEHIP databases. NLM offers technical support to individuals having difficulties in connecting to any of its databases. The MEDLARS Service Desk is available through a toll-free telephone number during weekday business hours as well as 1 There are exceptions to the requirement for user IDs. Institutions organizationscontracting with NLM to provide networked access to the NLM databases can developtheir own policies on user IDs and billing procedures for their users. 2 Access to the NLM da
OCR for page 104
FIGURE 6.1 Primary access points to the TEHIP databases.
OCR for page 105
through email. In addition, the Regional Medical Libraries of the National Network of Libraries of Medicine provide technical support and assistance in searching. The committee commends NLM for these efforts and believes that increased publicity regarding the availability of this assistance may encourage more health professionals, especially those without networked access to attempt to search the NLM databases. Cost The NLM databases are available under a number of different pricing structures. NLM makes its databases available online by connecting directly to the NLM servers, which is one of the least expensive search options. In addition, NLM leases its databases to commercial businesses, which may repackage the databases in online or CD-ROM format. NLM also leases its database tapes to academic institutions and professional organizations, which incorporate the NLM databases into their internal computer networks and determine their own pricing structures. In an effort to expand its accessibility, NLM is offering organizations a new option: a fixed-fee access rate for Internet searching of the NLM databases (Wallingford et al., 1996). A cost issue specifically affecting access to the TEHIP databases is the wide cost variations between the royalty and nonroyalty databases. TOXLIT, TOXLIT65, and CHEMLINE incorporate information from the Chemical Abstracts Service which charges royalties for usage. All other TEHIP databases and other MEDLARS databases are nonroyalty, and fees are assessed at the same rates for these databases.3 Examples of cost variations include the citation charge (i.e., the cost per citation viewed, printed, or downloaded). The nonroyalty databases charge $.01 per citation, whereas the charges for TOXLIT and TOXLIT65 are $1.15 per citation, and the charges for CHEMLINE are $.69 per citation (NLM, 1996a).4 The charges for each search statement submitted show similar variations: $.06 for nonroyalty databases, $1.78 for TOXLIT and TOXLIT65, and $1.07 for CHEMLINE (NLM, 1996a). It is important that cost issues be considered in examining the barriers to use of the TEHIP databases. NAVIGATING THE TEHIP DATABASES The complexities involved in navigating a database system are not unique to NLM. There is no standardization of search and retrieval methods, and each 3 Note that there are no charges for the use of NLM’s AIDS-related databases (see Chapter 2). 4 Online charges listed are for regular billed domestic (U.S.) access.
OCR for page 106
BOX 6.2 TOXNET Selection Menu National Library of Medicine's Database Selection Menu Which TOXNET database would you like to access: Hazardous Substances Data Bank - [HSDB] Registry of Toxic Effects of Chemical Substances - [RTECS] Chemical Carcinogenesis Research Info Sys - [CCRIS] Integrated Risk Information System - [IRIS] GENE-TOX - [GENETOX] Environmental Mutagen Information Center Backfile - [EMICBACK] Environmental Mutagen Information Center - [EMIC] Environmental Teratology Information Center Backfile - [ETICBACK] Developmental and Reproductive Toxicology - [DART] TRIFACTS - [TRIFACTS] Toxic Chemical Release Inventory Files - [TRI] database developer produces a search system with unique features and its own degree of user-friendliness. Selecting a Database TEHIP users may have difficulty accessing the desired information because the content of each database is not immediately apparent from the initial TOXNET screens. NLM's present configuration of the TEHIP databases displays a list of databases from which to choose when logging onto the TOXNET system via the direct search mode (Box 6.2) or Grateful Med. Yet, the 16 TEHIP databases vary widely in the scope of their subject content and in the level of detail of the information. The name of the database provides only limited clues to its relevance to the user's information query. Additionally, because some of the TEHIP databases (TOXLINE, TOXLIT, DIRLINE, ChemID, and CHEMLINE) are available through the ELHILL server they are not listed on the TOXNET menu, and users may be unaware of the gateway between the ELHILL and TOXNET servers. Although the committee realizes that frequent users know which database they need to access, numerous other potential users, including many health professionals, reach the list of databases and do not know in which direction to proceed. Efforts to assist users in determining which database to search by providing information on the scope of the subject matter and the type of information available (e.g., summaries, experimental data, and bibliographic citations) would be beneficial. This could be accomplished by the development of a decision-tree structure that would walk users through selection criteria,
OCR for page 107
thereby simplifying this process for novice as well as experienced users (this issue is further discussed in Chapter 7). Scope of the Databases As described in Chapter 2, a wide range of information that may be useful to health professionals is available in the TEHIP databases. The program evolved from a number of separate legislative mandates, interagency cooperative agreements, and NLM initiatives without the benefit from the beginning of a long-range plan. As a result, the content of the information in the databases overlaps and is somewhat uneven. The NLM Long Range Planning Panel on Toxicology and Environmental Health (NLM, 1993) noted areas for expanded subject coverage in the TEHIP program. These included morbidity and mortality statistics related to environmental health, community emergency preparedness and response, radiation, and molecular medicine (NLM, 1993). Similarly, the IOM committee solicited input from health professionals regarding directions in which to expand the subject coverage of the TEHIP databases. The results indicate the need for additional information on occupational exposures and risk characterization; requests for expanded coverage of international research, epidemiologic and demographic research, and health outcomes data; and the exploration of the possibility of integrating geographic information system tools as part of the data systems. The IOM committee did not explore the economic and interagency implications of supplementing the TEHIP databases with additional information or complementary databases. However, with the refinement of new technologies including the Internet, NLM may not need to directly provide the databases on its server but could provide pointers and linkages to other relevant databases (see Chapter 7). User Interface The balancing act in designing a user interface is to make it intuitive and easy to understand while still offering the user the power and depth of all of the features of the database program. Computer users are increasingly demanding interfaces that include simplicity while performing complex tasks; designs that work equally well for novice searchers as well as experts; graphics and text; customization options; and the ability to use the keyboard, mouse, pens, touch screens, and voice recognition (Alberico and Micco, 1990). Focus group participants and respondents to the committee's questionnaire indicated that a primary barrier to searching the TEHIP databases was the user interface.
OCR for page 108
For many of the TEHIP databases the user has several options in choosing an interface. The following section notes the strengths and weaknesses of each of the NLM search interfaces. Direct Searching As described in Box 6.1, direct searching is the only fully operational method of searching all of the TEHIP databases. Direct searching uses a command-line interface in which the user enters a search statement and receives input on the results. Command-line searching was developed for expert searchers who used early online bibliographic systems (Hersh, 1996). Consequently, this interface presents only skeletal directions, and the user must know the commands, the correct syntax, and Boolean logic to develop a targeted and precise search strategy (Box 6.3). The major disadvantage for novice and infrequent users of direct searching is its complexity and the steep learning curve needed to learn and master the commands and logic. Furthermore, extensive training or documentation is required to use the commands effectively and to determine the precise ways of keying in the search strategy. Menu Searching In an effort to simplify searching and thereby assist infrequent searchers, a menu interface has been developed for two of the TEHIP databases, TRI and HSDB. Menu searching provides the user with step-by-step screens that prompt the user in developing a search strategy and then modifying the strategy as needed. The initial TRI menu gives the user six choices (Box 6.4) and then continues to provide menus to assist the user in selecting the parameters for the search. For many health professionals and other user communities who are interested in searching the TEHIP databases but who do not have the time to learn the command language, menu searching offers a preferred search interface. The trade-off with the menu system is the loss of the intricacies of direct searching, including the ability to modify the search quickly and to narrow or expand the search strategy easily. Grateful Med The user interface for Grateful Med was designed by NLM for use by health professionals and other end users who do not have the time or inclination to learn the syntax and commands needed for direct searching (Hersh, 1996).
OCR for page 109
BOX 6.3 Sample Search on HSDB [HSDB] SS1/cf? Indicates that the user should enter the first search statement User: (USE) solvent First search statement requesting records for chemicals used as solvents Search in progress SS(1) PSTG (468) System message indicating 468 records match the first request [HSDB] SS2/cf? User: (htox) spinal and cord Second search statement requesting records with both the words ''spinal" and "cord" in the human toxicity field Search in progress SS(2)PSTG (32) System message indicating 32 records match the second search statement [HSDB] SS3/cf? USER: 1 and 2 Requests that records that meet the search requirements of both searches 1 and 2 be retrieved Search in progress SS(3) PSTG (2) Indicates that there are two records with information on solvents affectingthe spinal cord [HSDB] SS4/cf? USER: prt htox Print command; indicates that the human toxicity (htox) field should be printed NAME –Diethyl ether RN –60-29-7 HTOX –The principal physiological effect is anesthesia. Repeated exposures in excess of 400 ppm may cause nasal irritation, loss of appetite, headache, dizziness, and excitation, followed by sleepiness. Mental disorders have been reported, as has kidney damage. [Encyc Occupat Health & Safety 1983, p. 786] **Peer-reviewed** HTOX –Repeated contact with the skin may cause it to become dry and cracked. [Encyc Occupat Health & Safety 1983, p. 786] ** Peer-reviewed** HTOX –Cases of human death in industry due to acute inhalation are rare. One such case, subject developed acute mania and died in uremic convulsions. [Patty. Indus Hyg & Tox, 3rd ed. Vol. 2A, 2B, 2C. 1981-82. p. 2508] ** Peer-reviewed** HTOX –Clinically, albumin may appear in the urine, and polycythemia in blood. Nephritis may develop in rare cases. [Patty. Indus Hyg & Tox, 3rd ed. Vol. 2A, 2B, 2C. 1981-82. p. 2509] ** Peer-reviewed** NOTE: This is only a partial printout of the search results. Boldface type indicates input by the searcher.
OCR for page 110
BOX 6.4 Initial TRI Menu Do you wish to search the Toxic Chemical Release Inventory by: Facility Location? Facility Identification? Chemical Identification? Publicly Owned Treatment Works Identification? Other Off-Site Locations Identification? Environmental Releases? This program simplifies the search process for novice users while also offering the option of direct searching for experienced searchers. The user can input terms or may access the MeSH thesaurus and select the appropriate MeSH terms. The Grateful Med program automatically inserts the Boolean logical operators into the search request and handles other technicalities (e.g., formatting the author's name correctly. As with menu searching, expert searchers may find Grateful Med limiting, whereas novice searchers get useful results with minimal effort or training. Internet Grateful Med debuted in April 1996 and is available for searching MEDLINE, HealthSTAR, AIDSLINE, and PREMEDLINE. The interface is user friendly (buttons and pull-down selection boxes are used) and offers a number of ways to limit a search (e.g., by type of publication, date, or language). One valuable addition to this interface is the incorporation of the Unified Medical Language System Metathesaurus (discussed below), which analyzes the search strategy and suggests additional terms that could be used to expand or narrow the search strategy. Experimental World Wide Web Search Interface An experimental approach to searching the TEHIP databases is available to users on the World Wide Web (WWW) through the Special Information Services (SIS) homepage (http://sis.nlm.nih.gov). There are two search interfaces in prototype form: Webline, which is designed to search the bibliographic files MEDLINE, TOXLINE, and AIDSLINE; and the TOXNET Experimental WWW Search Interface, which searches the TEHIP factual databases. Both interfaces use hypertext linkages, a point-and-click interface, and pull-down selection boxes. The initial screen for the TOXNET Experimental WWW Search Interface provides choices of subject areas that can be searched: health effects, emergency medical treatment, or scientific literature (Box 6.5). A search on all of the fac
OCR for page 111
BOX 6.5 Experimental World Wide Web Search Interface: Criteria for Narrowing the Search Strategy Concepts for Query: ___Health Effects ___Human Toxicity ___Carcinogenicity ___Mutagenicity ___Reproductive Toxicity ___Others ___Emergency Medical Treatment ___Clinical Effects ___Treatment Overview ___Range of Toxicity ___Risk Assessment ___Noncarcinogenic Assessment—Lifetime Exposure ___Carcinogenicity Assessment—Lifetime Exposure ___Animal Toxicity Studies tual TEHIP databases is conducted, and it is only when the search results are presented that the searcher becomes aware of the multiple databases that were searched. This interface presents choices that are of interest to health professionals (e.g., emergency medical treatment) and does not require special training or extensive documentation. The disadvantages of this search method are not a result of the user interface but, rather, are a question of access to the Internet. Although a recent survey of 2,500 searchers of the NLM databases found that 75 percent have Internet access, 36 percent of rural users and almost half of users at hospitals indicated that they do not yet have access to the Internet (NLM, 1996b). Because access to the Internet is growing rapidly, this may not be a concern in the near future. However, it will be important to coordinate the development of this interface with Internet Grateful Med to produce a single interface (or similar interfaces) for searching all of the NLM databases on the Internet. Conducting a Search Searchers begin a database search with a specific query or information need. Focus group participants stated that many of their queries for toxicology and environmental health information begin with a specific chemical name. However, they also discussed the need for other starting points for their queries that fall into three general categories:
OCR for page 112
exposure-related queries, particularly those based on a job description or a geographic location, symptom-related queries, and regulation-related queries. Additionally, primary care professionals, emergency medical personnel, and poison control center personnel need to be able to conduct searches on the basis of the brand names of household or other products to retrieve treatment information. Commercially-available databases, such as POISINDEX®, provide brand name searching and interfaces that are developed to provide rapid access to the acute care information needed by poison control centers and emergency departments. One of the hallmarks of online databases has been the use of controlled vocabulary indexing to provide users with a tool to increase the precision of their search strategies (Siegel et al., 1990). NLM has been a leader in this field with the development and continual refinement of the MeSH thesaurus. Many health professionals are familiar with MeSH through searching the MEDLINE database and may assume that all of the NLM databases are indexed with MeSH terms. However, that is not the case, due to the diversity of the originating federal agencies responsible for the TEHIP databases. The TEHIP databases are designed to facilitate chemical-related searches, but they are less well-equipped to handle searches from other entry points. The field that is common to all of the TEHIP factual databases except DIRLINE is the CAS Registry Number field. NLM has recently added the capacity to perform chemical-based cross-file searches across six of the databases (HSDB, CCRIS, GENE-TOX, IRIS, RTECS, and TRIFACTS) by linking the files through the Registry Number field. Additionally, free-text searching allows the searcher access to words in the abstract and other text fields; however, indexing across the databases is not consistent. Although this issue may present a barrier for health professionals using the TEHIP databases, a simple cost-effective solution is not available. Indexing is a labor-intensive process, and the cost of adding MeSH terms to the TEHIP databases would be prohibitive (Siegel et al., 1990). The new technologies discussed at the end of this chapter, including natural language searching, should ease current constraints and should, in the near future, facilitate effective searching from a variety of entry points. Evaluating and Refining the Search: Quality Indicators The last step in the search process is an evaluation of the search results and a refinement of the search strategy, if needed. One of the factors involved is an evaluation of the validity and reliability (i.e., quality) of the data. Health professionals, through input provided by focus groups and questionnaires,
OCR for page 113
expressed the need for clear indicators of the level of scientific review on all data retrieved from the TEHIP factual databases. This is particularly important for summarized information when the searcher is not presented with the original data. Focus groups brought up the additional concern that there are widely differing scientific opinions on the toxicities of certain chemicals, in part because of the paucity of data from studies with humans and difficulties in extrapolations from animal studies. These different opinions point to the need for peer review and for clear indications of the source and quality of the data. The issue of quality indicators was discussed in the NLM Long Range Planning Panel report (NLM, 1993), and NLM has made efforts to address this concern. As described in Chapter 2 and summarized in Table 6.1, the levels of review differ between databases. To indicate the extent of review, NLM has added statements that are displayed after the user has logged onto the NLM system and has selected a database. For example, after selecting the GENE-TOX database, the user receives the message, "The mutagenicity data in this file have been peer-reviewed." Each data element in HSDB is marked with one of three tags indicating the level of review: "No review," "QC reviewed," meaning that the data have been reviewed for quality errors such as missing or contradictory data, or ''Peer-reviewed," meaning that HSDB's Scientific Review Panel has reviewed this information. In many of the databases the data are not reviewed for content but have been internally validated to confirm that there were no input errors. Indicators of the extent of peer review and other quality control measures are of utmost importance to the medical and scientific communities. It would be advantageous for searchers to have as an option the ability to limit their searches to only peer-reviewed information. This option could be considered a component of new interfaces such as those being developed on the World Wide Web. TABLE 6.1 Review Process for TEHIP Factual Databases Peer-Reviewed by Reviewers External to the Originating Agency Scientifically Reviewed by Reviewers in the Originating Agency Data Validation Checks HSDB CCRISa ChemID GENE-TOX IRIS CHEMLINE DIRLINE RTECS TRI TRIFACTS a Experts in carcinogenicity and mutagenicity select sources and provide scientific evaluation.
OCR for page 114
FUTURE DIRECTIONS Effective Interfaces Effective user interface design incorporates the psychological, behavioral, and technological factors that are involved in the human-machine interaction (human factors engineering) (Kohoutek, 1992). The committee received input from focus group participants, questionnaire respondents, invited speakers, and discussions with colleagues on the issues involved in simplifying the current user interface to the TEHIP databases with the goal of achieving an interface that assumes no a priori knowledge on the part of the user with respect to the data sources, their format, or content. The committee realizes that the state-of-the-science in interface design is rapidly changing and chooses to highlight a few of the innovations that could enhance the TEHIP interface: transparent or seamless interface between databases (an interface in which the user is unaware of the complexity of the database system underlying the search); graphics-based interface (an interface that uses buttons, icons, pull-down selection boxes, and other methods of inputting searches [e.g., Internet Grateful Med and the TEHIP Experimental World Wide Web interface]); relevance searching (the information retrieved is ranked in order of relevance to the search strategy); query by example (the searcher selects one reference that is on target, and the system then searches for all references similar to the example); and search analysis (the search program analyzes the search strategy and suggests terms that could narrow or expand the search [e.g., Internet Grateful Med]). Access As previously discussed, users can access the entire TEHIP complement of 16 databases in only one way: direct searching. However, direct searching is the most difficult interface for novice users due to its complex command-line interface. SIS is developing a more intuitive search interface using the graphical and hypertext capabilities of the Internet World Wide Web. The committee believes that this effort would provide an attractive search option for health professionals. However, the committee hopes that this effort will be coordinated with the development of Internet Grateful Med and other future Internet developments at NLM, to provide health professionals with similar interfaces whether they are searching MEDLINE or one of the TEHIP databases.
OCR for page 115
The rapid expansion of the World Wide Web has also created possibilities for links from other sites to the TEHIP Web site and has led to the development of new information dissemination approaches for health professionals. One example is Physicians' Online, a Web-based resource sponsored by pharmaceutical companies that provides free access to several NLM databases (MEDLINE, AIDSLINE, CANCERLIT, and HEALTH) for all physicians and other health care professionals who write prescriptions. Adding free access to the more clinically-relevant TEHIP databases (e.g., HSDB and TOXLINE) would promote use by a number of health professionals who are not familiar with these databases. Although these decisions are not in NLM's purview, NLM could explore these as possible access points. Additionally, NLM could explore the possibilities of providing hypertext linkages between the World Wide Web homepages of health professional organizations and the TEHIP Web site and the possibility of using Web search engines to provide pointers to TEHIP's Web site. Search Vocabularies One of the stumbling points for database users has been the fact that many databases or database systems use their own unique thesaurus of keywords or descriptors. If the user wants to perform a comprehensive search, he or she must become familiar with the search vocabulary, or if the database does not use a controlled vocabulary, the searcher must determine all relevant synonyms and related terms on which to search. To assist users in overcoming these quandaries, NLM has initiated an extensive research and development project, the Unified Medical Language System (UMLS). The basic assumptions of the UMLS project are that databases will continue to have diverse vocabularies and that users will continue to use a variety of terms in searching for biomedical information (Humphreys and Lindberg, 1993). The UMLS project has developed four knowledge sources, three of which (the Metathesaurus, Semantic Network, and SPECIALIST lexicon) integrate diverse vocabularies by linking terms on the basis of conceptual, semantic, and lexical connections, and the fourth, the Information Sources Map, is used to link the user with the appropriate information resource (McCray et al., 1993, 1994; Schuyler and Hole, 1993; see also Chapter 7). The UMLS knowledge sources are available to systems developers, who can incorporate them into expert systems that provide cross-database searching, natural language queries, and a variety of other applications. The UMLS knowledge sources have been incorporated into the Internet Grateful Med program and provide users with additional terms that can be used to narrow or expand the search strategy. The UMLS project is especially applicable to the TEHIP databases because these databases are not indexed with
OCR for page 116
MeSH terms or any other single controlled vocabulary. Consequently, efforts to facilitate free text and natural language searching will result in increased search effectiveness. This will be particularly useful for nonchemical-based search strategies such as searching by symptoms or queries on exposure regulations. Applying this technology to the TEHIP databases will assist users in constructing effective search strategies and will result in increasingly precise search retrieval results. CONCLUSIONS AND RECOMMENDATIONS The issues involved in accessing and navigating the TEHIP databases are in a state of flux because of rapidly changing computer technology and the expansion of the Internet. However, the committee wishes to stress that the TEHIP databases should be included in the technological changes of the future. The pioneering efforts applied to MEDLINE should, when applicable, be incorporated into the TEHIP program. As discussed in Chapter 1, the demands for toxicology and environmental health information by health professionals and other interested user communities, including the general public, are increasing, and it is critically important for information resources to be easily accessible with intuitive search interfaces. The committee recommends that a two-step approach be implemented in making refinements to the TEHIP databases and emphasizes the need for evaluation components to be incorporated and then monitored throughout this process. The first step would entail improvements that could be made in the short term while the user analysis (recommended in Chapter 4) is being conducted. The second step would be implemented over the long term and would be based on the results of the user analysis. Once the results of the user analysis are examined and it has been determined which of the TEHIP databases are most useful to health professionals, then a prioritization of activities (including outreach and training, access, and navigation) should be undertaken around those most useful databases. The committee's short- and long-term recommendations with regard to access and navigation are detailed below. The committee recommends that in the short term (during the time that the user analysis is being conducted) NLM continue its efforts to increase access to the TEHIP databases and to simplify navigation of the databases by: coordinating the development of the TEHIP Experimental World Wide Web Interface with Internet Grateful Med, promoting online registration for database access, and
OCR for page 117
exploring the possibilities of linking TEHIP's World Wide Web site with the Web sites of other health professional organizations and establishing pointers to the TEHIP databases from World Wide Web search engines. The committee recommends that in the long term and on the basis of priorities set as a result of the user analysis, NLM expand its efforts to facilitate access and navigation of the TEHIP databases by making full use of the navigational tools being developed within NLM and beyond. This includes: implementation of more efficient and intuitive user interfaces, incorporation of UMLS knowledge sources and other expert systems that would enhance symptom-related and other natural language searches, incorporation of navigational tools and interfaces that would create a seamless interface between databases, and implementation of indicators of peer review into new search interfaces. REFERENCES Alberico R, Micco M. 1990. Expert Systems for Reference and Information Retrieval. Westport, CT: Meckler. Hersh WR. 1996. Information Retrieval: A Health Care Perspective. New York: Springer-Verlag. Humphreys BL, Lindberg DAB. 1993. The UMLS project: Making the conceptual connection between users and the information they need. Bulletin of the Medical Library Association 81(2):170–177. Kohoutek HJ. 1992. Assuring quality of the human-computer interface. Quality and Reliability Engineering International 8(5):427–440. McCray AT, Aronson AR, Browne AC, Rindflesch TC. 1993. UMLS knowledge for biomedical language processing. Bulletin of the Medical Library Association 81(2):184–194. McCray AT, Srinivasan S, Browne AC. 1994. Lexical methods for managing variations in biomedical terminologies. Proceedings of the 18th Annual Symposium on Computer Applications in Medical Care. Pp. 235–239. NLM (National Library of Medicine). 1993. Improving Toxicology and Environmental Health Information Services. Report of the Board of Regents Long Range Planning Panel on Toxicology and Environmental Health. NIH Publication No. 94-3486. Bethesda, MD: NLM. NLM. 1996a. NLM Online Charges [http://www.nlm.nih.gov/databases/us_price.html]. December.
OCR for page 118
NLM. 1996b. Survey of Online Customers: Usage Patterns and Internet Readiness. NIH Publication No. 96-4181. Bethesda, MD: NLM. Schuyler PL, Hole WT. 1993. The UMLS Metathesaurus: Representing different views of biomedical concepts. Bulletin of the Medical Library Association 81(2):217–222. Siegel ER, Cummings MM, Woodsmall RM. 1990. Bibliographic retrieval systems. In: Shortliffe EH, Perreault LE, eds. Medical Informatics: Computer Application sin Health Care. Reading, MA: Addison-Wesley. Wallingford KT, Ruffin AB, Ginter KA, Spann ML, Johnson FE, Dutcher GA, Mehnert R, Nash DL, Bridgers JW, Lyon BJ, Siegel ER, Roderer NK. 1996. Outreach activities of the National Library of Medicine: A five-year review. Bulletin of the Medical Library Association 84(2 Suppl).
Representative terms from entire chapter: