Indexing the British Medical Journal Richard Jones As a general medical journal, The BMJ contains a wide range of subject matter, and many types of information need to be incorporated in its semi-annual index. Index Medicus vocabulary can be used for clinical articles, but non-clinical matter presents problems of 'soft' language. A weekly publication, the BMJ runs to about 1600 pages a volume, so succinct indexing is important, as is keeping to schedule. The number of authors and the vagueness of journal users present particular problems that can be ameliorated by the design of the index. Medline is a useful adjunct for subject access. Both the journal and the index have changed during a decade in which social and political aspects of medicine have assumed greater importance.

I have indexed the BMJ—as the British Medical Journal is commonly known—since the start of 1992, taking over from Ann Shannon who indexed it for over ten years. This change coincided with a reorgani zation of information services in the British Medical Association. When Ann left I took over as indexer as part of my responsibilities for technical services in the library. The library also provides other information and enquiry services. I have always catalogued and indexed books but I was surprised to find how much more difficult it was to index a journal! The weekly frequency means that I cannot allow myself to fall behind. Each issue needs about 300 entries, and 25,000 copies of the index are distributed, so errors would be very public. After developing the index for over ten years, Ann Shannon was prepared to invest time in patiently teaching the indexing style of the BMJ. We started four months before the end of her final volume and graduated to my test-indexing an issue.

Working to schedule Each Friday morning, after scanning through the BMJ, I start to assign subject headings to the articles, consulting previous indexes to assign headings consist ently. I number the title words and use these numbers as shorthand for words when noting the headings and subheadings in the margins. Chemical names or other phrases occurring in the text that will need cross-refer ences are circled and lettered. Rules for author entries are consistent and given in the indexing notes, now collated in a manual. I try to be as neat as possible, particularly in marking letters about journal articles, which are added on to the original entries. I use Macrex software to produce the index and index with MeSH (Medical Subject Headings) head

ings. When all of the issue has been assigned subject headings, 1 enter it into Macrex. If the headings for each article are entered consecutively, then one can take advantage of Macrex's powerful add entry and editing facilities. The Indexer Vol. 19 No. 1 April 1994

Since I am training an assistant, the index is entered in stages. He started entering author cross-references, and now enters author entries (the style of which varies according to section). I then enter subject head ings and cross-references. By the end of the year, he will be able to index the journal. This is particularly important as the weekly schedule does not allow any let-up because of illness, holidays or resignation. The indexing notes give full information on style which, 1 hope, would allow a trained indexer to carry on the index. Each week's entries are saved as a WordPerfect file, which is used for checking against the text. I have found it difficult to find the time to do this as the index is being constructed, and have done it over a concentrated period of four or six weeks. In the last couple of weeks I check the consistency of the entries and match up cross-references. I use Macrex's search facility to look for style or spelling mistakes. Since we use the index in a WordPerfect file, we have used its spellchecker to check the index. This is laborious as proper names are picked up too, but it probably takes less time, and is more accurate, than checking for errors manually. The index is quickly disseminated throughout the BMA. 1 send a computer file to the BMJ office, to be put on the editorial office network. Members of the library staff are able to access the index through the library's network. Member libraries and BMA members are able to dial up our fileserver number and access the index file. Computer files are also sent to the BMA press office and economic research unit.

A general journal Like any other indexing problems. journals in Britain, Smith, the current

journal, the BMJ has its peculiar It is one of the best known medical with a wide readership. Dr Richard editor, recently described the BMJ as the most general of the world's general medical journals and its readership as including academic 13

INDEXING THE BRITISH MEDICAL JOURNAL

researchers, hospital consultants, general practitioners and public health doctors all over the world—and, given the political and social importance of health reforms, it is read by managers, politicians, health ser vice researchers, patient groups, and people interested in health policy. General medical journals are not particularly easy to index, however. The range of subject matter is wide. Most subject entries only have one page reference. The subject language is softer in scope than that of other major general scientific and medical journals such as Nature or The Lancet, with a bias towards social medi cine. Many other types of information are included: editorials, news stories, news notes, original papers,

to the chemical name, but sometimes that chemical name is not in the Index Medicus vocabulary so the drug has to be cross-referred to a more general form: GLYCERYL TRIN1TRATE See under NITROGLYCERIN. Authors don't use vocabulary consistently. Consecutive articles might refer to aged, elderly, older people and so on. The preferred heading is aged but I am unsure how far I should try to standardize this vocabulary for subheadings. I tend to use the subject

regular

(the title term) or aged (the subject heading), and whether the subheading for aged should be colonic

series,

occasional

series,

obituaries,

letters,

medicopolitical news, fillers, columnists, book reviews, exhibition/play/tv reviews, personal views, and the final Minerva page (generally the first page to be read). My greatest problems are: those associated with this general language of a general medical journal; adapting to the style demanded by the different sec tions; and indexing succinctly.

The language of medicine The wide range of subject information means that I have to adapt the hard scientific vocabulary of MeSH. Subject headings are chosen from the Mesh (medical subject heading) vocabulary used by the US National Library of Medicine in Medline and Index Medicus, which are changed to British English. The controlled vocabulary of Index Medicus is a massive thesaural resource (with keyword access) which is internationally available: important for an international journal. It creates problems, however. I have to make many cross-references from English to American preferred vocabulary, as in cars See auto mobiles. I do the same for title phrases such as granny battering, which is referred to the Index Medicus term elder abuse. This last isn't the term used by the author and may not be the one used by the reader, but it is the term used in the international vocabulary of medicine and so should be internationally understood. Often, cross-reference phrases have to be inverted,

as in lighteners: skin. In such cases I try to make cross-references for title phrases or words the reader might remember. The phrase difficult patients is cross-referred to doctor-patient relations. This has the advantage of grouping together related articles. This applies also to the phrase health checks, which is grouped with other entries at mass screening. MeSH does not cover NHS administrative terms and I use some of these freely if they seem useful, such as Family Health Service Authorities or Health Authorities. Drug names sometimes give problems. Obviously, a cross-reference needs to be made from the trade name

14

heading or the shortest term. This problem might arise with a title: 'Drug therapy for colonic cancer in elderly people'. I would use the MeSH terms colonic neo plasms and aged, but am not sure whether the sub headings for COLONIC NEOPLASMS should be ELDERLY

NEOPLASMS or COLONIC CANCER.

The more general news items or columns often have eye-catching but descriptively useless titles, such as 'Killers in the basement' about US product tamperers, or 'Smug's game' about religion, or 'A long way to go yet' about doctors' communication of bad news. I index material on doctors' communication at doc tor-patient relations under the subheading commu nication—with the sub-subheading bad news rather than long way to go.

Different sections, different styles The style for different entries also took some get ting used to. Series are indexed not only to aid recog nition but also to indicate the depth of information. Thus, the user could gauge that a Regular Review on asthma treatment would contain more clinical infor mation than, say, a Personal View or a letter. The style I inherited contained rules regarding the position of the series indicator as an additional indication of its importance. Indications of irregular continuing series such as Lesson of the Week or Regular Review are placed after the author, whereas weekly series (which contain a set number of articles) are placed before the author. Other style matters concern preceding punctu ation, italics and parentheses. All of these matters have been written up into a manual. A journal index cannot look as neat as a book index because pieces on the same subject generally have different authors. The series needs to be included as an aid to recognition and the same applies to the authors. The series notations indicate depth of infor mation, but because this does not apply to author notations, I would like to stop recording the author with subject entries. Presumably a user who wanted to locate David Brown's article on stress asthma in adolescents would look under the author rather than try to guess which subject heading the indexer chose. A user who chose to look a known item up by a sub ject heading might, in a general index such as the BMJ, find only one or two page references there. The

The Indexer Vol. 19 No. 1 April 1994

INDEXING THE BRITISH MEDICAL JOURNAL

subheadings would provide additional clues to the known item the user is trying to find. Most users who look up subject headings in journal indexes are trying to find information by any author. I have tried to make the journal more consistent in its subject indexing and, if several articles discuss the same subject, to add the page numbers to the same heading, and to collocate two slightly different articles at the same heading.

How short the title? Shortening the title of the article to be indexed is a bugbear. Authors probably choose their titles with a

view to maximizing their retrievability from keyword

based retrieval aids such as Current Contents, but this may make them unnecessarily long. I have learnt that age, location and methodology seldom need to be included in an index entry to indicate its general sub ject content. Is a user who wants information on leukaemia in infants aged 0-4 unlikely to consult an article which refers generally to leukaemia in children? Long titles are irritating, and, while I might choose four or five access points, I try to simplify the subheading(s) if there is no other entry to distinguish it from. This is often the case in a general medical jour nal covering many different subjects. Often the subject headings with the most entries are the news stories, such as EC specialist accreditation, or coma patients. 'Double blind clinical and laboratory study of hypoglycaemia with human and porcine insulin in dia betic patients reporting hypoglycaemia unawareness after transferring to human insulin'—this is too long as a title, let alone as a possible index entry. It is an accurate summary of the content of the article but the reader does not need to know all this to decide if the article is likely to be of interest. I indexed this as 'Hypoglycaemia, Human and porcine insulin and unawareness'. A researcher hypoglycaemia unawareness not neglect to consult this index entry did not mention

who was interested in and insulin type would article just because the the context in which they

were used. Is it necessary to mention the research design in the title? I recognize that a 'double blind clinical and labo ratory study' is a thorough and systematic study, but these details don't need to be mentioned in the index entry.

Succinct indexing Indexing depth is another problem area. I try to index with a view to the amount of information there, but this is often difficult. Short pieces may often con

I prefer to make an additional entry, rather than to make one subject entry with a second cross-reference: if only because the second entry need not take any more space than a cross-reference and it would save the user a further look-up. I am more likely to do this if there are already entries at this second subject. In the most recent index, I have made some changes to the structure. As well as see alsos between related headings such as medical audit and quality assurance health care, I have made them between narrower terms and broader terms for diseases and countries. There is a see also cross-reference from anus neoplasms and from cervix neoplasms to neo

plasms, but not between equivalent terms. A reader who wanted to locate everything on neoplasms would be referred from the narrower heading to neoplasms, where all of the narrower headings are listed alphabet ically. Similarly, there are cross-references from coun tries to their continent where other countries in, say, Africa are listed. 1 like to limit the number of indentations to about three or four because I think the index looks messy with more, national health service has a subhead ing staffing and then seven further subheadings, one of which is whistleblowers. There could be further subheadings, new dh guidance or zeitlin: helen but this would make the entry look untidy. Even though this repeats headings, related information is adjacent. Generous use of subheadings in a general index often expands the total size. Since the indexing vocabulary is general, and hence large, most headings will only have one or two page references. The layout I inherited shunts the second line of text, for second and third subheadings, back to the left margin of the first subheading. This makes the text difficult to understand; I would prefer to align sub headings in blocks.

More language problems The style I inherited deletes definite and indefinite articles from the index entry. Of course, these have no descriptive value, but some entries make odd reading without them. An article entitled 'Waiting for the tide' drew an analogy between a sailor attempting to sail against a tide and medical staff learning not to strug gle against the inevitable. But does 'Waiting for tide' have a useful meaning in an index? These issues of language and succinctness were neatly encapsulated in a recent personal view1 by Steve Chaplin, a medical journalist, who expressed his con cern about labelling people by their illness in the way

bered by readers. A series called 'Drug Points' includes

we might talk about 'asthmatics' or 'hypotensives'. Most BMJ articles refer to the patient with the disease rather than the disease itself. The author concludes

case reports on adverse drug reactions. Although each entry is less than one column, it is often impossible to choose fewer then three or four headings.

that 'there comes a point when depersonalizing some one is unacceptable ... If you can't use the term to a person's face then don't use it at all.' What respect

tain several discrete aspects which could be remem

The Indexer Vol. 19 No. 1 April 1994

15

INDEXING THE BRITISH MEDICAL JOURNAL

should indexes pay to political correctness? Is it legiti mate indexing practice to refer to 'asthmatics' rather than 'people with asthma'?

Too many authors? The issue of authorship has been raised recently. Richard J. Epstein wrote of 'Six authors in search of a citation: villains or victims of the Vancouver system?'2 Multiple authorship is

the bane of many indexers'

work. Epstein's survey of eight biomedical journals showed that most journals showed a trend towards an increasing number of authors from 1982 to 1992, but general medical journals {Lancet, NEJM) published far fewer seven-author than six-author articles. His premise

is

that

this

may

be

influenced

by

the

Vancouver convention, which precludes citation of more than six authors. His survey suggests that papers

with more than six potential authors are organized so that only six are chosen (who all earn a Medline

entry) and other potential researchers are acknowl edged at the end of the paper. Principal authors, who

are traditionally listed last, would not be included in citations if there were more than six authors. His solu tion to this increase in the median number of authors is to cite the first and last authors and replace et al with et int (and intervening). The citation would still be unambiguous, indexing would be quicker and less laborious, the index would be less unwieldy, and—his main concern—only genuine academic contributions would be acknowledged. The convention for the BMJ index is to make cross-references for all authors, although if letters have more than six authors only the first is entered and the number of collaborators is given after et al, as in et al (10) if there were 11 in all. The BMTs editorial staff

query the number of authors (particularly for short reports), but papers with seven or eight authors are not uncommon. How useful are these author crossResources, using efficiently: CCHMS asks consultants

N Nabarro proposal!). See under Medical manpower NAEVUS (R H Champion), 265, 1633 (AQ ?) Laser treatment of port wine stains, 285, 215 (C) Nagana. See Trypanotoma brucei

NAHA. See National Association of Health Authorities NAIDOO, A. See BURN AND, K G, 285, 1071 NAILS White flecks (P D Samman), 285, 1484 (AQ ?)

NAILS, INGROWING

Treatment in child (P Gornall), 285, 716 (AQ ?) Nails. Kiintscher's. See under Femoral fractures NAKANO, J H. See HEYMANN, D L, 285, 531 NALIDIXICACID Haemolytic anaemia, fatal acute immune, caused by (O Tafani and others), 285, 936 (SR) NAMBU, S. See MURAKAMI, K, 285, 543 NANSON, E M. See HETZEL, M R, 285, 815 NAQVI, N: Cardiology in district hospital, 285,1358 (C) NARIMAN, S. See NEVILLE, E, 285, 796 NATHAN, A W, and HAVARD. C W H: Paralytic ileus and urinary retention due to hypothyroidism, 285, 477 (SR), 896 (C) NATIONAL ASSOCIATION OF HEALTH AUTH ORITIES University cuts: NAHA's dismay, 285, 905 (S), 1589 (S) NATIONAL COAL BOARD Coalworker's pneumonconiosis in Britain today and tomorrow, 285, 64 (C) NATIONAL CONGENITAL RUBELLA SUR VEILLANCE PROGRAMME 1971-81 (R W Smithells and others), 285, 1363 (E) NATIONAL HEALTH SERVICE Administrative staff, senior, appointment, 285, 1209 (C) Administrators: data for management: Komer Report (D Black), 285, 1227 (L) Administrators: let he who is without sin . . ., 285, 817 (C) American GP practising in: making visit (J J Frey), 285, 614