More than ever, information is critical for the benefit of effective practice, research, educa-tion, and management in medicine, as in many other fields. However, as beneficial as it may be, information that cannot be accessed cannot be used -- having access precedes use. Information rich and information poor are two modern information crises, each in its own way impeding access to information.

2. BACKGROUND

Since its inception in 1913, The Rockefeller Foundation has recognized the critical importance of biomedical information to the progress in medical and public health practice, research and education. As a result, The Rockefeller Foundation has supported throughout its history projects for development of biomedical libraries, dissemination of health information, and study and demonstration of information services in many parts of the world. The tradition continues with two demonstration projects:

1. Comprehensive information System in Medicine (CISmed) started in 1985 and also called from 1985 until 1989 the Selective Medical Library on Microfiche (SMLM), and

2. Comprehensive Information System in Public Health (CIShealth) started in 1989, with installations established in 1990.

Most positive experiences with CISmed (or SMLM), including a high degree of user accep-tance and use provided impetus for evolution and establishment of CIShealth.

Both projects are aimed toward improving access to biomedical and health literature to sup-port indigenous talent in developing countries. The specific objectives are to demonstrate and evaluate a novel and comprehensive approach in alleviating information problems and providing access to information services to workers in biomedicine and public health, particularly including researchers, educators, clinicians, administrators and students. The stress is on access to and use of information.

The Rockefeller Foundation has supported operation of CISmeds as demonstrations in a medical school in four countries: Columbia, Egypt, Indonesia, and Mexico for five years, 1985-89; after that period the sites became self-supporting. General information about these schools is given in Table 1. Similarly the Foundation started in 1989 support for CIShealth as demonstra-tions in the Ministries of Health also in four countries: Brazil, China, Mexico, and Zimbabwe.
 
 

While the Foundation will not support any further demonstration sites, it is hoped that the results will encourage other institutions and donors to purchase commercially similar systems on their own. University Microfilm International (UMI) (address: 300 North Zeeb Road, Ann Arbor, Michigan, 48109, USA) is the company that was the prime contractor with the Foundation to assemble and install CISmeds and CIShealths. As a result, UMI is now providing a commercial package for the two systems for purchase or donation by anybody. The basic approach involving user and use orientation, high quality in contents, relatively low economic costs and possibility of many information services is particularly suitable for environments that have difficulties in accessing contemporary world information, or that have no access at all.

The paper reports the use of the journal collection for CISmed at the four demonstration sites for the three year period 1987-89, with an in-depth analysis of individual journals used for two years, 1987 and 1988, for which complete data is available. CIShealth use cannot be reported because the operations just started in 1990, however, use data is being collected.
 
 

3. STRUCTURE

CISmed is a user and use oriented modular information system with high quality in contents, and relatively low economic costs, particularly suitable for medical schools, research institutions, teaching hospitals, and clinics.

The system, together with details on its philosophy, design, and implementation, is described in great detail in the articles listed in the Bibliography, thus only a brief description is provided here. CISmed consists of three modules, each with a specific function and technology:

3.1. Core Collection Module

To provide access to a set of biomedical journals found to be of high quality and high use worldwide. Judiciously selected, the collection consists of 99 journals, which can be classified into 35 subject specialties; these journals are found to be of high quality and high use worldwide. Journals are delivered by UMI on microfiche. The associated technology includes two fiche readers and two reader-printers, the latter serving as photocopying machines for users to obtain their own copies of articles.

3.2. Information Workstation Module

To provide for searching of indexes to identify relevant articles within and outside the col-lection. For CISmed the indexes provided were Index Medicus on microfiche, and a special printed Index Medicus with abstracts taken of Medline and containing only the journals in CISmed collection. For CIShealth the indexes are MEDLINE ON CD-ROM. (CISmed indexes were provided in print and CIShealth on CD-ROM because CD-ROMs were not widely available when CISmed started, while it became standard when CIShealth started. Obviously, a present CISmed could be installed with CD-ROMs instead printed or microfiche indexes. In 1990, CISmeds were equipped with CD-ROMS.) The technology where CD-ROM is involved includes: a PC microcomputer, CD-ROM player and printer. The information workstation can also function for providing a number of user services, such as alerting services.

3.3. Document Delivery Module

To provide a "window to the world" -- a contact with a document delivery organization to obtain for users articles (or other documents) not in the collection. The technology includes a facsimile machine (fax). Requests can be made by mail, fax or online; delivery can be by fax, mail, or courier service.

4. JOURNAL SELECTIONS

The selection of CISmed journals was based on a novel method of general applicability developed by William Goffman (see bibliography). The procedure is based on the often repeated observation that a small number of journals, generally considered of high quality, has a very large impact on the literature as measured by the number of citations received or the number of times requested from a large library collection. About 10% of journals in a field such as biomedicine account for some 80% of citation; about 20% (or even 10%) of journals in a larger library account for 80% (or even 90%) of use, providing for the famous 20-80 rule [Trueswell (1969), Burrell (1985)]. Citation data from Journal Citation Reports (JCR) were used to generate a list of 105 of the highest cited journals in biomedicine.

Next these lists were merged; it was found that 38 journals were common to all lists -- these 38 journals became the nucleus. Eventually, UMI obtained copyright permission for reproduction for 99 journals out of the 105, the rest of the publishers refused to grant such permission. These 99 journals, classified into 35 specialties, are listed in an alphabetic and classified order in Saracevic (1988b).

It is estimated that the CISmed collection contains about 80% of the most relevant material for any medical specialty in a given year. However, the selected journals have to be considered a core collection of world biomedical literature, not a whole collection; each site can add to this core according to its own needs, particularly in respect to local, national and regional journals.
 
 

5. OVERALL USE OF THE COLLECTION

As seen in Table 2, the total number of uses for the 99 journals in the CISmed collection over the three years (1987, 1988 and 1989) at all four sites was 59,026. Overall use grew steadily over the three years: from 16,377 uses in 1987, to 19,960 (+22%) in 1988, to 22,689 (+ 14% over 1988) in 1989; in two years, from 1987 to 1989, the overall use grew 39%.

The use across the four sites was not evenly distributed: Colombia and Mexico had approxi-mately the same amount of use for the three years (between 22,000 and 23,500), Indonesia had about 11,500 uses, and Egypt about 2,000. The relatively low amount of use in Egypt can be explained, among others, by the fact that this a brand new medical school; the first graduates started emerging in 1987. Moreover, the orientation of the school is toward community medicine, where students from the early years on spend a lot of time in different communities.

In general, the overall use should be considered relatively high, especially since the whole collection is in English, which is not a native language in any of the CISmed sites. The medical knowledge ("the medical language") does not have any linguistic or national barriers, it is truly international. Librarians at all four sites made the observation that this set of journals was used much more than any other journals of local, national or regional origin.
 
 

6. DIVERSITY OF JOURNALS USED

The data presented from now on represents use for two years, 1987 and 1988, which, added together, amounts to 36,337 uses for all four sites. There would be no significant difference in general conclusions if each year was looked at separately.

As can be seen from Table 3, the top journals used are also the most popular and best known biomedical journals in the world. While the ranking in use of top journals differs from site to site, the differences are not pronounced. In other words, pretty much the same journals received the highest amount of use.

However, ranking alone does not give a complete picture. The pattern of CISmed use differs significantly from the 20-80 observed rule of library use, mentioned previously and documented in the bibliography (i.e., the rule states that 20% of the journals in the collection account for 80% of use; in very large academic and research libraries the rule is even 10-90). In CISmed the top 20% of journals account only for about 50% of use. It takes about 50% of journals to account for 80% of use, thus, for CISmed collection the observed use rule is 50-80 instead of the usual 20-80. (The percentage of the collection accounting for 80% of use at each of the sites was as follows: Colombia 49%, Egypt 40%, Indonesia 50%, and Mexico 35%; all four sites 48%.)

In general, the dispersion of use across the whole CISmed collection is significantly higher than observed in regular, larger library collections. The high degree of selectivity, by which a small collection of high quality is selected for CISmed may account for higher diversity in use. Selectivity seems indeed to be working in that a higher percentage of the collection is being used.
 
 

7. PHOTOCOPYING: A MAJOR ASSET

CISmed included reader-printers; these are machines that make photocopies from fiche on plain paper on the same principles and at the same costs as other photocopiers, such as Xerox. The users could obtain copies of articles and/or any pages on request. Tables of contents were copied and sent to potential users as alerting services; selective dissemination of information (SDI) services were also provided using the reader-printers. The machines were used considerably. In 1987/88 the monthly average of photocopied pages were as follows:

Each site instituted a per page charge for copies which was roughly the same or slightly lower than other photocopying charges. The proceeds were used for supplies and maintenance.

It turned out that the ability to obtain a copy easily and inexpensively became a major factor in

the success and popularity of the system at all sites. Users could get or make copies of articles for their own use -- and this they valued very much. For technological and administrative reasons it was much easier for a user to get a copy from a fiche using a reader-printer, than a copy of an article from a printed journal (or a book, report, etc.) using the regular library photocopiers. Fiche reader-printers could be used only with a fiche thus they were not heavily guarded and controlled by the library as other photocopiers that can be used for copying anything.

The usual and anticipated resistance to fiche did not materialize, primarily because of reader-printers. To the contrary, many users expressed a preference of fiche over printed journals, again because of easy photocopying for their own use. Furthermore, fiche was not stolen, borrowed indefinitely, or had pages ripped as is the case with many print journals, assuring the users that what is claimed to be there is indeed there. Thus the technological combination of fiche with reader-printers was a major factor in the success, acceptance, and high rate of user and user satis-faction. While we anticipated this, the actual high impact by which this technological combination became a major asset was a surprise.
 
 

8. IMPACT

CISmed had considerable impact on the library and the schools of medicine where it was located. While evidence of the impact cannot be documented by numbers, excerpts from a letter by Arturo Morillo, M.D., Academic Dean, Faculty of Medicine, Pontifical University Javeriana, Bogota, Colombia, tell the story:

Over the decades after the Second World War, a number of approaches were tried with vary-ing degrees of success to alleviate the problem of access to world literature for users in developing countries. Quite evidently, the approach taken here is a success at the four demonstration sites. Despite great differences in the context and environment at different sites, the CISmed collection received a relatively high degree of use at all sites, suggesting that the approach may be generaliza-ble, with a high probability of success anyplace. The approach was demonstrated to work through an extensive evaluation over several years.

The CISmed approach involves four key elements:

1. A very high degree of selectivity, concentrating only on journals of proven high quality and high degree of use worldwide.

2. A concentration on economics, to keep costs as low as possible. High degree of selectivity contributes to low subscription costs. Use of fiche results in space saving.

3. Application of useful technology to facilitate a number of user services and other aspects, such as portability, accommodations, etc. The technology as build in can be upgraded with technological changes over time.

4. A pronounced orientation toward users and use, facilitated by a number of features, among the most important being the ability for users to obtain copies of articles for their own use. Numerous information services can be launched, based on the technology involved.

The result of selectivity was a relatively small collection of about 100 biomedical journals that received a relatively high degree of use -- about 60,000 uses in three years at the four sites. More-over, use was dispersed across the collection, rather than as usual in larger libraries where the mi-nority of collection accounts for a lot of use, and the majority is used very little, if at all. Selectivi-ty works in the sense that diversity of the collection use is significantly higher. The usual library use rule changed from 20-80 (20% of the collection accounts for 80% of use) to 50-80 for CIS-med. This represents an increase of 2.5 times in the diversity of collection use -- a most significant increase by any standards! Data show not only that the use was heavy, but that the diversity was high and that subject or specialty categories that were covered turned out to be essentially relevant.

Of course, adjustments can be made based on data about use, i.e., study of use can lead to readjustments. Least used journals can be dropped. Journals can be added to specialty categories recording the highest use.

The technology that was selected at the outset was relatively simple and it did not scare people off. To the contrary, it was a major contributing factor in the use and popularity of the system. Librarians liked the technology, in some ways it increased their status and standing. Although experience with subsequent CD-ROM technology and information workstations is still limited, it seems that this newer technology will also be eagerly adopted. The often promoted notion of a different or "appropriate" technology necessary for conditions in developing countries does not hold at all for these experiences. The same information technology is accepted as eagerly in developing environments, as in developed ones.

A similar general conclusion can be made about biomedical knowledge and literature. What is relevant, sought, accepted, and used in the developed world is also relevant, sought, accepted and used in developing countries. Quality is quality, no matter where. Biomedical knowledge is universal and so is the biomedical literature. When provided, the best literature is used -- given that using is easy and the service is user, rather than collection, oriented. The English language seems not to be a barrier, but a lingua franca.

The matter of economics is critical for any approach to information problems. Selectivity keeps not only the use up, but the costs down, particularly if considered on the amount of use per cost of subscription for the collection.

In 1990, annual subscription costs for the CISmed collection on fiche is close to U.S. $6,000; annual subscriptions to MEDLINE on CD-ROM is about $1,000-$1,500 (depending on how many back years are desired); a reader-printer can be obtained at about $4,100; and an information workstation at about $3,200. The system can be accommodated in a room 8 x 8 meters. While these costs are not insignificant, they are lower than establishing and running a regular library or information center.

The approach described here is not a replacement for a library. However, it is an approach by which the growing problems of information access can be reduced, as convincingly demons-trated at the four CISmed sites. We believe that the approach is generalizable.
 
 

BIBLIOGRAPHY

Bruer, J. T., W. Goffman and K. S.Warren, "Selective health libraries and library networks for developing countries," American Journal for Tropical Medicine and Hygiene, 30 (6): 1133-40 (1981).

Bruer, J. T., & T. Saracevic, "Selective medical library on microfiche," International Journal of Micrographics & Video Technology, 5 (1): 1-6 (1986).

Burrell, Q. L., "The 80/20 rule: Library lore or statistical law?" Journal of Documentation, 41 (1): 24-39 (1985).

Goffman, W. & K. S. Warren, K. S. Scientific Information Systems and the Principle of Selectivity. New York, NY: Praeger, 1980.

Goffman, W., "A pragmatic approach to literature selection," In: Selectivity in Information Systems: Survival of the Fittest, ed. by K. S. Warren. New York, NY: Praeger, 1985. pp. 117-143.

Huth, E. J., "First-rate, low-cost medical libraries for the Third World," CBE Views 10 (2): 22 (1987).

Saracevic, T. Selective libraries for medical schools in developing countries. New York, NY: The Rockefeller Foundation, 1980.

Saracevic, T., "Evaluation procedures for the Selective Medical Library on Microfiche," Interna-tional Journal of Micrographics & Video Technology, 5 (1): 7- 17(1986).

Saracevic, T., "An experiment in the delivery and use of a high-quality, low-cost medical journal collection for developing countries.," Proceedings of the American Society for Information Science. 25: 10-16 (1988).

Saracevic, T., "Selective Medical Library on Microfiche: An International Experiment Supported by The Rockefeller Foundation," Bulletin of the Medical Library Association, 76 (1): 44-53 (1988).

Trueswell, R.W., "Some behavioral patterns of library users: The 80/20 rule,"Wilson Library Bulletin, 43: 458-461 (1969).

Warren, K. S., "Selective aspects of the biomedical literature," In: Warren, K. S. (ed.) Coping with the biomedical literature: A primer for scientists and clinicians, ed. by K.S. Warren. New York, NY: Praeger, 1981

Warren, K. S., "The evolution of selective biomedical libraries and their use in the developing world," JAMA- Journal of the American Medical Association, 257 (19): 2628-9 (1987).

Wolff, A., "A cold eye on mediocrity -- filtering out the best of the biomedical literature," RF Illustrated, March 1986, pp. 9-11.