Selecting CD*ROM databases for nursing students:

Transcription

1 Selecting CD*ROM databases for nursing students: a comparison of MEDLINE and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) By Ellen Okuma Instruction and Outreach Librarian Hawai'i Community College Hilo, Hawaii With the introduction of the Cumulative Index to Nursing and Allied Health Literature (CINAHL) on CD-ROM, research was initiated to compare coverage of nursing journals by CINAHL and MEDLINE in this format, expanding on previous comparison of these databases in print and online. The study assessed search results for eight topics in 1989 and 1990 citations in both databases, each produced by SilverPlatter. Results were tallied and analyzed for number of records retrieved, unique and overlapping records, relevance, and appropriateness. An overall precision score was developed. The goal of the research was to develop quantifiable tools to help determine which database to purchase for an academic library serving an undergraduate nursing program. INTRODUCTION Libraries that support nursing education programs may choose from several options for providing access to journal literature. The two primary indexes for nursing journals are International Nursing Index (INI) and the Cumulative Index to Nursing and Allied Health Literature (CINAHL). INI includes all English-language nursing and pertinent non-english titles [1]; CINAHL indexes only English-language journals [2]. In 1989, SilverPlatter was the first vendor to make CINAHL accessible on CD-ROM. Since then, Compact Cambridge and CD-PLUS have added CINAHL to their list of CD-ROM products. The purpose of this research was to compare coverage of nursing journals in MEDLINE and CINAHL, focusing on their usefulness for an academic library serving baccalaureate nursing students. Evaluation criteria were number of retrieved records, overlapping and unique records, relevance, and appropriateness of the retrieved records to the user. An overall precision score also was developed for each of the databases. LITERATURE REVIEW After CINAHL's online debut in 1983, Fishel found that the controlled vocabulary in CINAHL produced efficient online search results [3]. Although 70% of the subject terms in CINAHL are based on Medical Subject Headings (MeSH), many other terms unique to CINAHL were found to be specific to the nursing field. In Fishel's comparison, CINAHL produced more documents than did MEDLINE and also was easier to search. Lansing and Edmondson also compared the two online databases and their print counterparts for subject indexing of the American Journal of Occupational Therapy (AJOT). MEDLINE was found to have more than twice the number of subject headings for each of the 176 articles from a three-year period than did CINAHL. CINAHL had more terms specific to rehabilitation, but MEDLINE produced better results when the search term was not found in the CINAHL thesaurus [4]. Research by Brenner compared indexing of fifty randomly selected articles from nursing journals by the two databases. Differences in the use of major descriptors were slight, but the differences were greater in common and unique indexing access points. Brenner concluded that further research is needed to look at quality or content of indexing in both databases [5]. Research by Fried, Killion, and Schick on "novice searchers" found major differences between CINAHL and MEDLINE [6]. Their findings were especially relevant to this study, because the advent of CD-ROM, particularly for MEDLINE and CINAHL, encourages more end-user searching. Two topics were searched in online versions of CINAHL and MEDLINE using BRS/Colleague; results were compared for ease of access and retrieval. CINAHL produced more results Bull Med Libr Assoc 82(1) January

2 Okuma Table 1 Free-text search strategies for eight topics compared in MEDLINE and CINAHL (SilverPlatter) 1. Assessing cultural differences among patients strategy-cultural WITH assess* (MEDLINE) strategy-cultural WITH assess*; set # AND py > 1988 (CINAHL) 2. Developing an educational program for teaching breast self-examination strategy-breast WITH self WITH exam* WITH educat* (MEDLINE) strategy-breast WITH self WITH exam* WITH educat*; set # AND py > 1988 (CINAHL) 3. Massage therapy strategy-massag* WITH therap* (MEDLINE) strategy-massag* WITH therap*; set # AND py > 1988 (CINAHL) 4. Cultural differences in nutrition strategy-nutrition WITH cultural (MEDLINE) strategy-nutrition WITH cultural; set # AND py > 1988 (CINAHL) 5. Rape trauma strategy-rape WITH traum* (MEDLINE) strategy-rape WITH traum*; set # AND py > 1988 (CINAHL) 6. Assessing religious differences among patients strategy-(spirit* or religio*) WITH assess* (MEDLINE) strategy-(spirit* or religio*) WITH assess*; set # AND py > 1988 (CINAHL) 7. Self-abusive behavior strategy-self-abus* OR self-inflict* OR self-injur* (MEDLINE) strategy-self-abus* OR self-inflict* OR self-injur*; set # AND py > 1988 (CINAHL) 8. Developing an educational program for premenstrual syndrome (PMS) strategy-pms OR (premenstrual WITH syndrome); set # WITH educat* (MEDLINE) strategy-pms OR (premenstrual WITH syndrome); set # WITH educat*; set # AND py > 1988 (CINAHL) because appropriate terminology was found in the thesaurus. MEDLINE required more complex search strategies than did CINAHL because of the lack of specific terms. Words were free-text searched and then narrowed to the field of nursing. The authors concluded that both databases covered the "mainstream" nursing literature but that CINAHL is the database of choice for novice searchers because of the ease of quickly and effectively determining relevant publications. They recommended that an extensive search include both databases to ensure comprehensiveness. Kilby, Fishel and Gupta also reviewed the differences and similarities between the online and print versions of MEDLINE and CINAHL. They stated that because the National Library of Medicine serves a wide range of professionals, including physicians, dentists, nurses, and veterinarians, its thesaurus "cannot meet the specific needs of any one audience" and that searching MEDLINE and CINAHL comes down to a difference of "depth versus breadth" [7]. The research addressed two questions: which database provides the best starting point for nursing literature, and which produces the best results for a nursing search? MEDLINE's list of 3,470 indexed journals included 8% in the field of nursing, while 65% of CI- NAHL's 300-plus journals were specific to nursing. When they compared overlap, 91% of MEDLINE's journals also were indexed in CINAHL, with only sixteen titles unique to MEDLINE. Eighty-six percent of CINAHL's nursing journals also were indexed by MEDLINE, with twenty-five titles unique to CI- NAHL. MEDLINE also proved to be more restrictive in terms of article inclusion and indexing than was CINAHL. For a five-year period of Maryland Nurse, MEDLINE included twenty-three citations and CI- NAHL had thirty-one. As the preceding review indicates, comparisons of the two indexes have concentrated on print and online versions. Little has been done to compare the indexes in the newer format, CD-ROM, other than in reviews of the products. Articles in Database Searcher, Laserdisk Professional, and Health Libraries Review [8-10] reviewed CD-ROM availability of CINAHL, while MEDLINE on CD-ROM continues to be reviewed regularly (for example, in Database Searcher [11]). Of interest to reference librarians training end-users to do CD-ROM searching is the recent study by Potter, who found that such training is labor intensive [12]. And collection development librarians may want to read the bibliometric research by Burnham et al. who utilized MEDLINE and CINAHL CD-ROM versions to develop a methodology to evaluate journal collections [13]. METHODOLOGY As the literature indicates, relevance, precision, and appropriateness of search results are perhaps the primary criteria to be used for comparing the two databases. Therefore, the present comparison of MED- LINE and CINAHL on CD-ROM quantified and analyzed relevance and appropriateness of the retrieved documents to undergraduate nursing students. To develop data for comparison, a question array technique was used. Eight research topics typical of upper-division nursing students at Hawai'i Loa College were chosen. The eight topics covered a wide range of subjects. SilverPlatter was chosen as the database vendor for the study products because it was the first to produce CINAHL on CD-ROM. This version was installed at Hawai'i Loa College's library and thus was easily accessible to the author. The library of the School of Public Health at the University of Hawai'i at Manoa had MEDLINE on SilverPlatter, which was made available for this project. Because MEDLINE does not always use nursingspecific terminology, a generic rather than a controlled vocabulary search strategy was employed to search for each topic in both databases. Truncation is available in SilverPlatter (by using an asterisk) and was incorporated as appropriate. Because the Boolean operator "AND" produced too many irrelevant results in a trial search, SilverPlatter's connector "WITH" was incorporated as necessary. Identical search strategies were used for each topic in both databases, ex- 26 Bull Med Libr Assoc 82(1) January 1994

3 A comparison of MEDLINE and CINAHL Table 2 Retrieval of citations for individual questions by database MED- MED- Ad- LINE LINE justed Question '89 '90 Total total* CINAHL 1. Culture Breast self-examination Massage therapy Nutrition Rape Religion Self-abuse Premenstrual syndrome * Eliminates 1989/90 duplications. Table 3 Overlap of records in CINAHL and MEDLINE Records In CINAHL in MEDLINE Questfon both (%)* 1. Culture 12 12/20 (60) 2. Breast self-examination 6 6/28 (21) 3. Massage therapy 2 2/9 (22) 4. Nutrition 2 2/5 (40) 5. Rape 0-6. Religion 8 8/14 (57) 7. Self-abuse 3 3/4 (75) 8. Premenstrual syndrome 0 - * X% = 34.4% (average). cept that searches in CINAHL included publication years. All years of CINAHL are on one disk, while MEDLINE has a different disk for each year. The two years chosen for the study were 1989 and Table 1 shows the search strategies and topics used for the comparison. DATA ANALYSIS Results of each of the eight searches in each database were tallied. Then uniqueness and overlap of citations were compared. To judge relevance and appropriateness, a panel of three judges was used because of their expertise in nursing education and their familiarity with nursing literature. Harter alludes to the necessity of using judges for defining the concept of relevance because it is "a semantic relation between a document and a question. A document may be said to be relevant to a question (that is, a formalized or compromised information need) if it is objectively judged to be so by a consensus of those practicing in a discipline" [14]. Each of the judges received a packet that included a questionnaire with a four-point scale asking for judgments of relevance for each record and the full records retrieved for each of the eight topics from MEDLINE and CINAHL. A section also was provided following each question for comments, particularly if records were judged to be irrelevant. FINDINGS As expected, based on results and observations in the literature and the publishers' descriptions of coverage for each database, MEDLINE had the highest number of citations retrieved for all questions except number 2 (breast self-examination). Retrieval results for each question are presented in Table 2. In MEDLINE, there was some overlap or duplication of records in the two disks covering 1989 and In those cases, retrieval totals were recalculated by subtracting the duplicates. The final number used in the study appears in the "adjusted total" column. Many of the records in CINAHL were expected to appear in MEDLINE because of MEDLINE's inclusion of the International Nursing Index. However, this expectation was not borne out by results for all of the questions. Table 3 shows the overlap of records in the two databases. The last column represents the percentage of CINAHL records found in MEDLINE. An average percentage (X%) of overlap for all questions was determined, computing an arithmetic mean for the sum of the percentages in the last column of Table 3, resulting in a figure of 34.4%. Numbers of unique records for each database also were calculated. These figures represent the converse of overlap. Results showed that CINAHL has, on the average, 65.6% records that are unique to that database, while 91.5% of MEDLINE's records are unique. It is worth noting that almost two thirds of the records found in CINAHL were found to be unique; overlap had been expected to be higher because of MED- LINE's coverage of nursing literature. MEDLINE was found to have a much higher percentage of records with abstracts (88.6%) than did CINAHL (43%). As described earlier, relevance of records to the user was determined by a panel of three judges. Individual judges were asked to evaluate each record according to a four-point scale of "very irrelevant," "irrelevant," "relevant," and "very relevant." Responses then were collapsed into two categories- "irrelevant" and "relevant." Comments were solicited specifically for "irrelevant" responses. Raw data tables were used to determine the number of "relevant" and "irrelevant" responses for each search question for each judge. A relevance percentage for each judge (identified as A, B, and C) was computed for each question within each database. Then all three judges' relevance percentages for each question were averaged (arithmetic mean, identified as X%) to create a relevance percentage for each ques- Bull Med Libr Assoc 82(1) January

4 Okuma Table 4 Judges' precision scores and mean precision scores (X%) for each question in MEDLINE and CINAHL Judges-MEDLINE (%) Judges-CINAHL (%) Question A B C X% A B C X% 1. Culture Breast self-examination Massage therapy Nutrition Rape Religion Self-abuse Premenstrual syndrome Overall average tion within each database. This arithmetic mean is called a precision score. Numbers were rounded to a whole number (.5 was rounded higher when the preceding number was even and lower when the preceding number was odd). The usefulness of having a panel of judges determine relevance is evident in Table 4. For CINAHL, there is no consistent pattern of differences. For MED- LINE, however, judge C generally rated more records as being relevant than did judges A and B. Judge C stated that "upper division students should have broad-based literature searches and I encourage [their] use of allied sciences." This perception will be addressed in more depth in the following section. Once the average for each question was computed (X% in Table 4), all of the averages for the questions in each database were added together and divided by eight (total number of questions) to achieve a final, average precision score. MEDLINE had a precision score of 57.89%, while CINAHL had a much higher score, 79.36%. An appropriateness factor was developed by examining the "irrelevant" responses and comments and corresponding records for a subset of the eight questions. Typical comments for "irrelevant" records in MEDLINE were "too technical," "too medical," and "non-english." A closer look at these comments formed the basis for an analysis of appropriateness of the records for undergraduate nursing students. The records for topics ranked the lowest in precision by the judges were chosen for in-depth analysis of appropriateness. In MEDLINE, the eighty-eight records for "massage therapy" received the index's lowest precision score of 22.7%. In CINAHL, the records for the two topics, "nutrition" and "rape," tied with the lowest precision score of 66.7%. However, because these two topics produced so few records (five and one, respectively), the topic with the nextlowest precision score was added-"massage therapy," with nine records. Comments from the judges were divided into two 28 categories. The responses "too technical," "too medical," and "too specific" became "non-nursing," while the "non-english" comment was left as one category. Only the responses that included comments to explain the "irrelevant" rating were used. In some cases, a record was judged to be irrelevant for both "nonnursing" and "non-english" reasons; each reason was counted. The nonrelevant percentages for the three topics then were calculated. For three of the eight topics, more than 50% of the nonrelevant records in MEDLINE were judged to be so because of the nonnursing focus. Non-English records also appeared to influence the judges' responses for 41% of the nonrelevant records related to massage therapy. CONCLUSION Although CINAHL has been available in print since 1956, its availability on CD-ROM has contributed to a continuing interest in comparing it to MEDLINE. At the same time, as institutions respond to nursing shortages by developing and offering nursing education, academic librarians are faced with major purchasing decisions-such as whether to buy MEDLINE or CINAHL on CD-ROM, to cancel paper index subscriptions, or to do online searches on demand. CD- ROM has expanded user access to journal literature as no other technology has. With CD-ROM, the enduser can shape personal search strategies. An undergraduate library must decide which CD-ROM index provides undergraduate nursing students with the best access to the nursing literature. This project was done to develop a framework to help academic librarians choose between CINAHL and MEDLINE. CD-ROM purchases are expensive compared to their print equivalents; quantifiable tools are required if librarians are to make informed decisions about acquisitions. Tenopir summarizes the options librarians face when databases become available on CD-ROM after the library has print subscriptions. One option, obviously, is to cancel print sub- Bull Med Libr Assoc 82(1) January 1994

5 A comparison of MEDLINE and CINAHL scriptions and use resulting funds to buy the CD- ROM versions [15]. However, a library's collection may become noticeably thin in certain areas as endusers perform more and more searches on CD-ROM and discover obscure journal titles. Future research might compare how many times each database indexes the most frequently used nursing journals in the college library. One approach to choosing a database is to lease CINAHL on CD-ROM for nursing students and conduct more comprehensive MEDLINE searching online as needed. MEDLINE also is available for enduser searching at a nominal online rate with a GRATEFUL MED software subscription. This approach was taken at Hawai'i Loa College, where much of the research reported here was done. GRATEFUL MED has several features that facilitate its use for undergraduate nursing students. The menu-driven searching program is self-explanatory, and options for English language, nursing subheadings, and review articles can refine search results to meet the needs of the students. The results of this study are consistent with the findings of Kilby et al. and Fried et al. regarding relevance of records in CINAHL to nursing. The current study also analyzed other factors, including precision and appropriateness of search results to the undergraduate nursing student. Precision will affect the success and satisfaction of the end-user, whichever CD-ROM database is chosen. CINAHL was found to provide more precision in its coverage of nursing than did MEDLINE. The final criterion, appropriateness, was found to be more common for the CINAHL database than for MEDLINE. Many of the records in MEDLINE that were judged irrelevant were described as non-english or were not related to the field of nursing. Although CINAHL is the preferred database for nursing students, MEDLINE also should be consulted for comprehensive searches because of its high percentage of unique citations. Effective searching of MEDLINE for the field of nursing requires the use of specific parameters. ACKNOWLEDGMENTS The author wishes to express her gratitude to Carol Tenopir, professor of library and information studies, and Virginia Tanji, librarian for the School of Public Health, both at the University of Hawai'i at Manoa; and to the three Hawai'i Loa College nursing faculty judges: Carol Winters, Ellen Akaka, and Celia Garrison. REFERENCES 1. FRIED A, KILLION V, SCHICK L. Computerized databases in nursing: information at your fingertips. Comput Nurs 1988;6(6): SilverPlatter CD-ROM pricing guide. Newton Lower Falls, MA: SilverPlatter Information, 1991:1. 3. FISHEL C. The nursing & allied health (CINAHL) database: a guide to effective searching. Med Ref Serv Q 1985; 4(3): LANSING P, EDMONDSON M. Subject indexing of the American Journal of Occupational Therapy in MEDLINE and NAHL. Med Ref Serv Q 1987;6(2): BRENNER S. CINAHL and MEDLINE: a comparison of indexing practices. Bull Med Libr Assoc 1989;77(4): FRIED, op. cit. 7. KILBY S, FISHEL C, GUPTA A. Access to nursing information resources. IMAGE: J Nurs Scholarship 1989;21(1): New data, improved service features for CINAHL medical data. Database Searcher 1992;8(4): PRINGLE RM, JR. Nursing & allied health (CINAHL) on CD. Laserdisk Prof 1990;3(1): WILLLAMSON J, STEvENs D. CINAHL on CD-ROM. Health Libr Rev 1990;7(1): SilverPlatter revises PC SPIRS CD-ROM software. Database Searcher 1992;8(4): PoTTmR LA. Measuring library staff time spent training patrons to use health-related CD-ROM databases. Bull Med Libr Assoc 1992;80(3): BURNHAM JF, SHEARER BS, WALL JC. Combining new technologies for effective collection development: a bibliometric study using CD-ROM and a database management program. Bull Med Libr Assoc 1992;80(2): HARTER S. Online information retrieval: concepts, principles, and techniques. San Diego, CA: Academic Press, 1986: TENopiR C. Costs and benefits of CD-ROM. Libr J 1987; 112(14): Received January 1993; accepted March 1993 Bull Med Libr Assoc 82(1) January