Managing bibliographic citations using microcomputers

Managing bibliographic citations using microcomputers

SPECIAL ARTICLES Managing Bibliographic Citations Using Microcomputers CALVIN M. KUNIN, M.D. cokJnlbus.Ohio From the University partment of Medici...

859KB Sizes 0 Downloads 93 Views

SPECIAL ARTICLES

Managing Bibliographic Citations Using Microcomputers

CALVIN M. KUNIN, M.D. cokJnlbus.Ohio

From the University

partment of Medicine, Ohio State liege of Medicine, Columbus, Ohio. eequests P(1 reprintsshould be addressed to Dr. Calvin M. Khin, Department of Medicine, Ohio state unlverslty, Room 110, starling Loving Hall, Columbus, Ohio 43210. Manuscript accepted September 13, 1984.

Programs are now available to construct and retrieve lists of blbllographlc cltatlons by microcomputers. Although conslderabfe effort must be expended to learn to use them, once mastered, they can be useful to anyone who must manage large collections of references. There are several ways to use personal computers to store and retrieve blbllographlc citations. Word processors can be used to manage relatively small lists of cltatlons. Preprogrammed blbllographlc systems are available that are designed speclflcally for this purpose. A general data base management program can also be adapted for bibliographic purposes and used for other functions as well. This essay has been prepared to provide gukfance to those who have a working knowledge of microcomputers and wish to expand this to use a data management system for blbllographlc purposes. The dBASE II program Is used to Illustrate how to set up a bibliographic system. Methods are described on how to prepare cltatlons for storage and retrieval using comblnatlons of key words and Boolean operators, how to prepare selected lists of references and arrange them In alphabetic order or by subject heading, and how to print tailored lists of cltatlons. The system was found to be hlghiy responslve to commands and able to provlde rapid retrieval of lnformation. The biomedical literature has become so large and is growing so rapidly that it is difficult even for an expert in a limited field to keep up with advances. To keep current, a great deal of effort must be expended to collect bibliographic material. A system is needed to store the information and retrieve it rapidly. One method is to maintain large collections of reprints of important articles. Another is to record the citations on cards. References can be also obtained from articles and reviews or by use of computer-based bibliographic services. Regardless of the method used, a great deal of effort is often required to find a key citation and to prepare a comprehensive list of references for a conference, an article, or a research proposal. Collections of references can grow so large that they may become difficult to manage. It would be helpful, therefore, to be able to add new references conveniently to a collection and to be able to store and retrieve them rapidly. It would be particularly useful to be able to call up lists of references on a given topic and to have them arranged alphabetically or in a desired sequence and format suitable for publlcation. This essay was prepared to prqvide guidance to those who have a working knowledge of microcomputers and would like to use a computer data management system for bibliographic purposes. It is

April 1995

The American Journal of Medklne

Volume 79

627

COMPUTERS

FOR

BIBLIOGRAPHIC

CITATIONS-KUNIN

based on the lessons I have learned, often by trial and error, in adapting a dBASE II, a popular computer data base program, to manage a large collection of references in the field of urinary tract infections, USE OF THE WORD

BOOKENDS: The reference management system (Ashwell, J. Microware Division of Churchill Livingston, Inc. 1983). This has a companion program, BOOKENDS MED-TRANS, which allows the user to obtain references from MEDLINE and incorporate selected references into BOOKENDS. Other programs available for managing citations are Bibliofile (Power Tools, Amherst, Massachusetts) and Personal Bibliographic System (Ann Arbor, Michigan). The characteristics of several of these systems have been described in helpful reviews

PROCESSOR

A word processor may be all that is needed to prepare lists of commonly used references. Since citations need not be retyped after they are entered, errors in transcription is reduced. The citation can be modified by the word processor to meet the style of a particular journal. References can be entered as they become available, and the “find” or equivalent command of a good word processor program can be used to search for the name of an author, a citation number, or an identifying word in the title of the paper. Once the reference is found, it can be transferred as a block to a blank space, repositioned, and edited as needed to meet the requirements of a specific journal. A typical 5 l/4 inch double-density, double-sided floppy disk holds about 280,000 bytes or typewriter spaces. Some of the larger or newer disks have even greater capacity. Since most citations take up less than 300 bytes of space, very large collections can be stored on a single disk. Fewer citations could be stored if additional space were occupied with abstracts. This method is useful to manage relatively short lists of citations for preparation of a manuscript. lt avoids the need to retype each citation. The only skill required is knowledge of word processing and familiarity with some of the special commands of the system. With careful planning, citations can be organized in sections according to topic headings. The disadvantages are that most word processing programs are unable to sort citations in alphabetic or numeric order. Also each citation must be managed individually and cannot be manipulated in groups without considerable effort. This method is therefore cumbersome for managing very large collections of references. Nevertheless, the relative simplicity and immediate utility of the word processor may be all that is needed for most purposes. PROGRAMMED

BIBLIOGRAPHIC

[WI. A system called PaperChase is available at the Beth Israel Hospital in Boston [3]. This enables a user to search MEDLINE’s bibliographic base and to retrieve references without the need for the services of a specially trained librarian. The needed references can be selected and printed. The system does not however, have a citation management program to arrange or edit the references. APPLICATION OF A GENERAL PROGRAM FOR BIBLIOGRAPHIC

A general data management system has the advantage over other programs in that it can be used also for a wide variety of data management tasks. When one system has been thoroughly learned, the same principles can be applied to other functions. These include manipulation of patient and laboratory data, performance of mathematical calculations, construction of forms for computer analysis, inventory control, and patient management. It is not necessary to learn to program the computer, but rather to understand how to use the procedures and commands that are already built into the program. Several sophisticated data base programs can be adapted for management of bibliographic citations. The dBASE II program (Ashton-Tate, Culver City, California) will be described in this report. It was selected because it is used widely and is compatible with the operating systems of most microcomputers. These include the CP/M and MS-DOS (IBM-compatible) systems. Other data systems are the PFS Filing System (originally designed for the Apple computers) and DB Master (Stoneware, Inc., San Rafael, California), which can also be used with the Apple computer. Regardless of which data management system is selected, the general concepts used to construct the programs are sufficiently similar that the method for organization and retrieval of citations to be described in this paper can be applied to most of them. How to Begln. If you are not familiar with dBASE II or other data management systems, scan the material that follows to gain some idea of how the data are organized and retrieved to prepare tailored lists of references. If you wish to adapt this method for your own use, or

SYSTEMS

Specialized programs can be purchased to retrieve citations from bibliographic data bases such as MEDLARS, Chemical Abstracts, or other large collections. Some of these are coupled with another program that allows the citations to be inserted into the user’s own computer and saved (down-loaded) for storage, managment, and retrieval. Examples of these systems are Sci-Mate (Institute for Scientific Information, Philadelphia, Pennsylvania). It can be purchased as the Universal Online Searcher to tap a large data base (MEDLINE). Personal Data Manager is a companion program for management of the citations. A similar program is

628

April

1885

lhe

Amerkan

Journal

of Madklne

Volume

DATA MANAGEMENT PURPOSES

78

COMPUTERS

consider how dBASE II might be used for other tasks, I suggest that you read the first few chapters of the primer [4]. Chapter 3 contains a brief but lucid account of how computer hardware and software systems function. Even if you do not plan to work with computers yourself, the primer is a helpful introduction and can be read for general interest. If you are already familiar with another data base management system, you may wish to compare the concepts provided here for assigning fields and establishing key words, Creatlng the Structure for a Reference File. Let us assume that you have some appreciation for how dBASE II works and how to create a file, enter and edit data, and prepare reports. It is now appropriate to prepare the structure needed to enter individual citations into a file. The file will be called BOOK (to avoid confusion all dBASE commands will be capitalized). Each citation is set up as an individual record within the file BOOK. Each record has an identical descriptive format made up of fields. Each field contains information about the citation. The first field is used to accommodate the citation, the next two are for key words, and the last is to provide an accession number; dBASE II can accept up to 32 fields for each record. We will only require four for our purposes. It can hold up to 1,000 bytes (typewriter spaces) for each record. We will use fewer than 400. It can manage 65,535 records. This is well beyond any single user’s needs. The fields will contain all the information needed to describe each citation or record. The records can be retrieved and manipulated within and among fields to arrange the citations in a wide variety of lists. Establlshlng the Format for Each Cltation or Record. To set up the format that will be used for each record, type CREATE. This will bring up a series of prompts to enable you to form the structure of the record to be used to characterize each citation. The responses to be typed in are shown in Flgure 1. Once established, the same template will be saved by the program for use to enter or edit each record in the file BOOK. By typing DISPLAY STRUCTURE, you will call up the format and be told the number of citations that have been entered. Use of Fields to Characterize Each Citation. The first field is called REF, for reference. This is the largest field and occupies all of the 254 bytes that can be used in any one field. The letter C indicates that this is designated as a character field. Character fields accepts alphanumeric information as opposed to logical or numeric fields, which are used for other purposes. Most references, typed according to the format of the Index Medicus and limited to the first three authors (with “et al” if needed), will fit into this space. It is preferable to include the entire citation in a single field rather than to use separate fields for authors or titles and references. This avoids having to recombine

FOR

BIBLIOGRAPHIC

CITATIONS-KUNIN

CREATE

iNTEE FILENAME: BOOK 'R'JTER RECORDSTRUCTURE AS FOLLOWS: FIELD NAME,TYPE,WIDl'X,DBCIMAL PLACES 001 RE3r,C,254 002 CAT,C,lS 003 suB,c,100 004 NBC,4 005 INPUT DATANOW? N Figure

dBASE

Structure 1. II program

as

for setting it WOUM

up file appear

records on the

using the computer

screen.

segments of the citation when printing a list of references. Although this slows the search to find a particular string of letters or numbers within the citation (since the entire field of each record must be scanned by the computer), it avoids further editing later on. The only modification needed in the citation is to place a comma after the year of publication rather than a semicolon. The semicolon has special meaning to the program. The dBASE II program can search the entire citation to find a particular word, phrase, or number (such as the year or volume). The second field is called CAT for category. A field size of 15 bytes is adequate, since only one key word will be used. CAT is used to identify broad topics such as chapter headings for a book or a title of a reprint file. For example, in the field of cardiology, a CAT could be electrocardiography, echocardiology, endocarditis, valve, etc. The field CAT is set up for several reasons. First, it can correspond to hard files of reprints of articles that can be pulled for preparation of a manuscript. Second, it can be used to search the data base more rapidly. Third, it can be combined with key words in SUB to increase specificity and allow identification of given citations. The third field is SUB for subcategory of key words. This field is purposely made large, employing 100 bytes. It can contain as many key words as will fit into 100 bytes, as long as each is separated by a comma to aid the computer search. The position of the key word in the field does not matter. For the example of cardiology, key words could inciude tricuspid, mitral, prolapse, the species of a microorganism, or any other appropriate term. You can think of CAT as the heading of a chapter and SUB as the index of a book, or they can be considered the major and minor headings in a book index. The fourth field is called N for number. It could have been called NUMBER, but then six letters would have to be used when only one is needed to search the files.

April

1985

The

American

Journal

ol Medlclne

Volume

78

829

COMPUTERS

FOR BISLIOGRAPHIC

CITATIONS-KUNIN

The program provides an entry number for each record, but this is used to define the position of the citation in the computerized list and is not a permanent part of the individual record. It will change automatically as the file is rearranged-for example, when alphabetized or when records are inserted or deleted. I recommend you use, in the beginning, the same number for N as appears in the record file. This will keep the list in order in the master file. By giving a permanent number to each entry, you have the opportunity to pull up a series of citations for preparation of reports, regardless of how the citations are arranged in the computerized list. List of citations may be organized by entry number or category or alphabetically according to first author, depending on the format preferred. Preparing a Citation for Entry. Before the information is entered into the computer, write on the paper copy of each citation the key words to be assigned in CAT and SUB and the citation number to indicate that the data have.been entered. I recommend you enter the citation in the form used by Index Medicus, since this style is now used by most journals. The proper abbreviation of the name of the journal can be found in the National Library of Medicine List of Journals Indexed. It should take only a few minutes to scan a paper and to assign the most descriptive key words. Selecting Key Words. Key words are helpful to retrieve complete lists of relevant citations. The titles can be searched for a topic as well. This may allow identification of a single citation when the author or title is known, but it is inadequate for assembly of specific reference lists, since the titles may not contain the critical identifying word that is needed. Combining key words makes it possible to prepare tailored reference lists from a large data base. This is accomplished by selecting a single key word or combination of key words that best identifies a given topic. In preparing a book on urlnary tract infections with about 2,500 citations, I found that I needed 60 items in the field CAT for major chapter cr section headings, and about 300 different key words in SUB to characterize the citations further. Key words need to be chosen very carefully to suit a future bibliographic task. Once selected, they should not be changed if at all possible. The National Library of Medicine publishes a list of key words used in its bibliographic citations, This can be helpful to maintain uniformity when many persons use the same data base, but you may wish to construct your own list. Enough room is available in the fields of CAT and SUB to enable use of full descriptive terms or well-accepted abbreviations. A record listing each key word selected should be maintained. If this is lost, key word can be retrieved from the computer by typing LIST CAT or LIST SUB. The key words in each record will then appear on the screen or can be printed.

630

April

1985

The American

Journal

of Madklno

Volume

Useful Commands. The program allows you to search, manipulate, and prepare specific lists of the citations by use of the commands DISPLAY, LIST, INDEX, SORT, COUNT, and REPORT. If you type DISPLAY ALL, all the records will appear and flow down the screen in sequence. If you type COUNT and specify what you want counted, you will be inform of the number of records with this characteristic. REPORT will provide an organized form containing the data you wish to see according to a format you have previously established. If you add the words TO PRINT, the report will be printed for you as well as displayed on the terminal. Each of these commands can be conditioned by specifying what should be included in the search. A specific list of citations is called up by typing a dBASE command followed by a string of words or letters that characterize a portion of one or more fields. These are combined with the field name in which they are located. Delimiters (’ ‘) surround the word or sequence of letters or numbers that are selected from a field. Parts of a word or groups of numbers can be used to increase the sensitivity of the search. The full name or number or a longer sequence increases the specificity. The first few letters of a name or word may be used to find all related names when you do not know the exact spelling. However, I recommend that whenever possible the entire name or word in the citation or key word in the fields CAT and SUB be used to allow the search to be as specific as possible. Combining Key Words with Boolean Operators to Increase Specificity. With cardiology as an example, key words can be developed to characterize each major topic. If only the field CAT were used, you would need hundreds of specific key words. By combining it with the field SUB, a major topic can be placed in CAT and several conditional characteristic topics can be placed in SUB. For example, if ‘valve’ is placed in the field CAT, SUB can be used to define the characteristics of the valve, Identifying words in SUB could include ‘prolapse’, ‘ruptured’, ‘stenosis’, ’ insufficiency’, ‘mitral’, ‘tricuspid’, and ‘aortic’. The Boolean operators .AND., .OR., .NOT. can then be used to link ‘valve’ with any series of conditions such as ‘prolapse’ and ‘mitral’. (Boolean operators are functions that allow information within a field and among fields to be combined in order to construct specialized lists. They may be defined as follows: The operator .AND. means “only with,” .OR. means “and with,” and .NOT. means “not with.“) If the list of citations were limited to the field of cardiology, it would not be necessary to use ‘valve’ since both ‘prolapse’ and ‘mitral’ imply valves. If the file also contained gynecologic citations, it would be necessary to distinguish a prolapsed valve from a prolapsed uterus by linking ‘uterus’ with ‘prolapse’ by interposing .AND. between them. If you typed CAT=‘valve’.AND. (‘pro-

78

COMPUTERS

FOR BIBLIOGRAPHIC

CITATIONS-KUNIN

USE BOOK : REPORT ENTERREPORTFORMNAME: BOOK PNTea OPTIONS,MaLEFTMARGIN,IrLINESIPAGE, W=PAGEWIDTE M=lO,L=66,W40 PAGEHEADING?(YIN) Y PNTEBPAGEHEADING:Reference List DOUBLESPACEREPGET?(YIN) Y ARE TOTALSREQUIRED?(YIN) N COL wIDTR,coNTENTs 001 70,REF Wl'EIt HEADING: 002 Flgwe

2.

Standard form to be used for preparing reports of citation lists using dBASE II as it would appear on the computer

screen. lapse’$SUB.AND.‘mitral’$SUB), you would obtain all the references that exist in the file concerned with prolapsed mitral valve. Even if there were gynecologic as well as cardiologic citations, you could achieve the same result by typing ‘prolapse’$SUB.AND.‘mitral’ $SUB, since mitral excludes any gynecologic term. The use of the $ or string operator together with delimiters will be described later on. Preparation of a Report. In order to prepare a specific list of references, a form must be constructed to describe how the material is to appear when printed. Once this report form is prepared, it can be used for all reference lists. Type REPORT. This will bring up a series of prompts used to develop the report form. Fill in the queries according to the form shown in Figure 2. This form is called BOOK.FRM and will be retained in the computer files. The form specifies its name, BOOK, the length, width, and margins of the page to be printed, and a title for the report to appear as a heading. Reference List was chosen, but the title can be left blank and filled in at a later time. The form specifies that the report will list the citations (REF) only (it will not include CAT, SUB, or N) and will occupy 70 typewriter spaces per line. If you type Y after the prompt DOUBLE-SPACE REPORT? (Y/N), each citation will be separated by one blank line. Once the report form is set up, selected reference lists may be prepared using the same form over again. This can be accomplished by several approaches using specific commands to the computer. Example 7. Let us suppose that your name is David Smith and you wish to prepare a list of papers in which you were cited as an author. You also wish to have a title for the list called Smith’s List of Publications. Type (after the dot prompt) USE BOOK SET HEADING TO Smith’s List Of Publications REPORT FORM BOOK FOR ‘Smith’$REF TO PRINT The report will now appear as:

Smith’s List of Publications Reference List Smith, DH, Jones, MS. Observations on computers in medicine. Ann Comput Sci 1983, 110: 22-44. Black, JS, Jones, WS, Smith, DH. Further observations on computers. Ann Comput Sci 1984, 111: 332-336. etc. Use of String Operators. Two citations with the name ‘Smith’ are shown in this example. In the first citation, Smith was listed as the first author and, in the second, as the third author. This is because $, called a string operator, was used to tell the computer to find the name ‘Smith’ regardless of where it was listed in the citation (in the fieM REF). Think of $ as signifying “included in.” The alternate method to search for ‘Smith’ is to use the expression REF=‘Smith’. Since the $ is left out, only those papers will be listed in which Smith was first author. Use of $ and the name of the field (REF) following ‘Smith’ Instructs the computer to pick out all Smiths who happen to be listed in the citations. If there were a ‘Smith phenomenon’ or ‘Smithson’ in the titles, these would also be detected. This problem can be avoided by asking for ‘Smith, D’ to specify a particular Smith. Similarly, a year of publication can be distinguished from an identical page number by typing ‘1984,‘. The comma is unique in following the year of publication and does not follow a page number. Use of Other Commands. The same approach is used with other commands such as DISPLAY or COUNT. For example, type DISPLAY FOR REF=‘Smith’ This will provide a list on the screen containing all the records in which Smith was cited as the first author. The record includes all the fields used to define the citation (REF, CAT, SUB, and N), as opposed to the report form, which will display only the citation. If you type COUNT FOR ‘Smith’$REF, you will be told the number of times your name was cited regardless of its position in the citation. If you type COUNT FOR REF=‘Smith’, you will

Apdll999

The

Anwkan

Journal

of Medklne

Volume

79

631

COMPUIERS

FOR BISLIOGRAPHIC

CITATIONS-KUNIN

be told the number of times you were cited as a first author. Listing Reference by Number. Let us suppose that you wish to have a list of citations that are characterized by their number. They should have been listed in field N by at least four complete digits, such as 0005 or 0095 or 2001. Even more digits will be needed for very large lists, The computer is very literal. If you ask for ‘I’, it will find all citations in which ‘1’ is included, such as 10 or 121 or 1000. To make the number unique, all four (or more) spaces must be used. Example 2. Let us assume that you wish the series of citations to appear in the following sequence 0015, 0095,000 1,0045, and 0010. Type REPORT FORM BOOK FOR ‘0015’$N.OR.‘0095’$N.OR.‘OOOl’$N.OR.’OO45’$N. OR.‘00 1O’$N The report will then appear with the title Reference List (already placed in the form), with each reference listed in the sequence it appeared in the file. Unfortunately, dBASE will not list them in the sequence stated in the command. The Boolean operator .OR. tells the computer to include all of these citations in series. An alternative method would be to type REPORT FORM BOOK FOR N=‘0015’.OR.N=‘0095’, etc. TO PRINT In the first statement, you are asking the program for the number included in the field N; in the second statement, you are asking it to search the field N for the number listed. In this example, both mean the same, but seemingly minor changes in the construction of these commands often have very different meanings. The program is more strict in demanding proper use of syntax than a tough third-grade teacher of English (as opposed to a third-grade English teacher-which may have a very different meaning). Preparation of Specific Lists of References. Example 3. You may want to develop a very special list of references for electrocardiographic findings in patients with diabetes. You have already placed electrocardiography in CAT and diabetes in SUB. You want to restrict your list to those references that include just electrocardiography and diabetes. Type SET HEADING TO EKG Findings in Diabetes REPORT FORM BOOK FOR CAT=‘EKG’.AND.‘diabetes’$SUB TO PRINT The report will appear as: EKG Findings in Diabetes Reference List followed by a list of the appropriate citations. Arranging the Citations in Alphabetical or Numeric Order or by Subject Headings. The entire list of citations can be alphabetized by setting up an index file. The index file will be called REFNDX to denote that the field

832

April 1982

The American Journal d Medicine

Volume 78

REF (references) is inixed. To index the file, type INDEX ON REF TO REFNDX In order to prevent the program from alphabetizing each letter in the list of citations to achieve perfect order (an enormous task that is not needed), this command is modified to INDEX ON $(REF, 1,lO) TO REFNDX. This limits alphabetizing to the first 10 letters of each citation. The computer will respond with the number of citations indexed. To use the indexed file, type USE BOOK USE BOOK INDEX REFNDX BOOK is the name of the file we are using for all the citations; it needs to be identified in the command. Once the file is indexed, a very responsive command called FIND can be used to call up the name of the first author in the citations. Type FIND. Then type the first 10 letters (or fewer if you wish) of the last name and initials of the first author. The citation will appear immediately when you type DISPLAY. If there are other citations with the same name they can be brought up by typing CONTINUE and DISPLAY once again. This can be repeated until all citations with the same first authors are displayed. Indexing may be considered to be a method to conveniently arrange a file to search rapidly for a reference. Files can be indexed by CAT and SUB or by N. All new records added to the file will also be automatically placed in the index file in alphabetic or numeric order, depending on which is used. index files are not, however, permanent records-but simply a method by which the program can rapidly find an entry. In order to produce a more permanent alphabetized file for preparing reports, you should use the command SORT. Type SORT ON $(REF, 1,lO) TO (new filename). To call up the sorted file, type USE (new filename). Once the records containing citations have been entered together with their categories, subcategories and numbers (REF, CAT, SUB, and N), and indexed, you can review the entire collection, record by record, to make corrections and changes. The contents of records can be modified, added, or deleted at any time. By indexing the file, you will be able to note multiple entries of the same citation, since they will be grouped together alphabetically. They can then be marked for deletion to avoid duplication. Other Uses of Index and Sort. If you wish to arrange the list of citations by major categories (included in the field CAT), type INDEX ON CAT The citations will then be arranged according to each category in the alphabetic order of the categories. SORT can be used in the same manner to develop a permanent file or series of files subdivided by categories.

COMPUTERS

TABLE I

FOR

BIBLIOGRAPHIC

Response Time of the Bibllographlc Program In Search for Cftatlons’ Time Required

Fislds searohed Number BYtee 1 1 1 1 1 1 1 1 2 2 3 3

CITATIONS-KUNIN

250 250 250 250 250 15 15 100 265 265 365 365

Numberol Key Words Ussd

Numberol Records FOUnd

1 1 2 3 4 1 1 1 2 3 3 4

39 50 4 3 1 25 250 21 3 4 1 2

Time Required

IO co&

lo Dieplay Records wfseoond

139 187 202 240 275 132 132 139 141 171 166 176

15.7 13.1 10.8 9.1 7.9 18.5 16.5 15.7 15.5 12.8 13.2 12.4

RMOrdS Defseoond 186 228 207 241 278 162 442 166 141 169 168 177

11.7 9.8 10.5 9.1 7.9 13.5 4.9 13.2 15.5 12.9 12.9 12.3

The program was adapted from dBASE II (Ashton-Tate, Culver City, California). It was run on a Vector 4-S computer using a IO-megabyte hard disk. The data base consisted of 2,183 records of 408 bytes each. Note that these times were measured for the COUNT and DISPLAY commands. The FIND command used with an indexed file gave an almost instantaneous response. l

seconds to examine the entire data base to find two citations identified by four key words.

construCtIon of Special Lists of Citatkms. Regardless of the number of citations entered into the computer, the index file using the command FIND will retrieve a reference almost instantaneously. It will take longer, however, to DISPLAY, COUNT, or REPORT, since every letter in the field must be scanned. Thus, it can be very useful to develop smaller and more specific subfiles that can be more readily rearranged by the SORT command and edited further. To accomplish this, for example, for all references that deal with a prolapsed mitral valve, type COPY FIELD REF,CAT,SUB,N TO B:PROLAPSE FOR CAT=‘prolapse’OR.‘mitral’$SUB The computer will respond with the number of records copied. To inspect the new file, type USE BPROLAPSE, and then DISPLAY STRUCTURE to examine the number of records, followed by DISPLAY ALL to examine all the citations. This new file can then be managed as described earlier. Response Time of the Program. In order to determine how rapidly the program could count citations and display them on the screen, I tested the response time to these commands using several strings of key words located in various fields (Table I). A rather large data base of 2,183 records containing 408 bytes each was searched. The COUNT command was generally more rapid than DISPLAY, depending on how many records were brought to the screen. For both commands, the more records found by a command the longer it took to search the entire data base. For example, the longest time it took to examine the entire file was for a key word that detected and displayed 250 records. This took 7 minutes and 22 seconds. It took only 2 minutes and 57

COMMENTS This essay was written to help guide those who wish to consider the use of computers for management of bibliographic citations. It is possible to do this without the use of computers, but they can be very powerful time-saving tools. In order to acquire skills in the use of computers, considerable time and effort must be expended. The need, therefore, must not be trivial. File cards and reprint collections, coupled with the resources of a good medical library, are adequate for most purposes. Use of computers becomes worthwhile only when there is a need to manage very large collections of citations for research or scholarly work. Despite their limitations, word processors can serve as useful devices for storing lists of citations. By arranging the citations according to topics, a user may retrieve and edii the collection for further use. Members of a very active research group who must keep up with the current literature and prepare reports frequently may find their needs met best by systems such as Sci-Mate and BOOKENDS, which offer the opportunity to retrieve citations from MEDLINE and to manage them effectively. For those who have already developed a large collection of citations in their field of interest and continuously scan the literature in a selected manner, a computerized bibliographic data management system may be most suitable. A special program can be purchased for the purpose, or one of the more generalized data management systems can be adapted. In this essay, I have described how dBASE II can be

April

1985

The

American

Journal

of Medlclne

Volume

78

033

COMPUTERS

FOR BIBLICGRAPHIC

CITATIONS-KUNIN

ACKNOWLEDGMENT

adapted for management of bibliographic citations. Similar programs can be used for the same purpose. The advantage of acquiring the skill to use the more general data management systems should be apparent to those who have the need to manage large amounts of clinical or laboratory data-it allows you to take advantage of the revolution in microcomputers without requiring the assistance of programmers. Management of bibliographic citations, as described herein, illustrates a relatively simple task that can be performed by these programs. For example, only four of the 32 fields in dBASE II were needed for this purpose. Many data base management systems can perform much more complex procedures and prepare special reports. The most important limitation is your own imagination in defining a useful task.

634

rrpH

1985

llm

Amrican

Journal

d Modklno

Volume

I thank Betsy M&ath for her collaboration in adapting dBASE II for management of lists of citations and John Buck and Kieth Lamb of Microcenter, Inc., Columbus, Ohio, for their review of the manuscript and helpful suggestions. REFERENCES 1. 2.

Taylor SI: Review of BOOKENDS: the reference management system. N Engl J Med 1984; 300: 134. Wi9ton RS: Review of S&Mate universal on-line searcher, S&Mate persoml data marqsr, and Bibliophile. Ann Intern Med 1984;

3. 4.

78

100: 821422.

Horowitz GL, Jackson JD, Bleich K: Paperchase: self-service bibliographic retrieval. JAMA 1983; 250: 2494-2499. Byers R/k Everyman’s dat&ase primer. Culver City, Celifomia: Ashton-Tate, 1983.