House style

Medical staff who often work overtime are likely to suffer from stress.
Medical staff, who often work overtime, are likely to suffer from stress.

• Use commas before "and," "or," "but" in two-sentence sentences (when the coordinate conjunction joins two main clauses):
Half received drug treatment, but their symptoms did not resolve more quickly.
We could make an omelette, or you could go and get a takeaway.

• Note that when a comma is used, both main clauses must have a subject:
The patients stopped smoking, and they felt better for it.
The patients stopped smoking and felt better for it.

• Minimal hyphenation - use hyphens only for words with non-, -like, -type, and for adjectival phrases that include a preposition (one-off event, run-in trial). Not using hyphens will help you to avoid noun clusters (see Grammar below).

• Quotation marks - please use double, not single, inverted commas for reported speech. Full stops and commas go inside quotation marks:
She said, "We will."
• No exclamation marks, except in quotes from other sources.
• Reference numbers go after commas and full stops, before semicolons and colons.
• Minimal capitalisation. Use capitals only for names and proper nouns.

• Organisations and groups of people take singular verbs:
The government is; The team has researched...

• Avoid noun clusters:
"Patient in coronary care unit" rather than "coronary care unit patient."

• Watch out for "danglers" (unattached participles and misrelated clauses):
Joining the service in 1933, his first post was... (the post didn't join the service)
Joining the service in 1933, he was first posted to... (this is correct)

The BMJ allows a mixture of English and American spelling, depending on the provenance and main target audience of the article

aetiology/etiology
oestradiol/estradiol
anaemia/anemia
haemorrhage/hemorrhage
practice (noun)
practise (verb)
foetus and fetus are both acceptable in English: the BMJ uses fetus.
minimise/minimize
organisation/organization
capitalisation/capitalization

Use Whitakers Almanac and Times Gazetteer as sources for geographical names. Use European spelling in city names that sound the same as in English — eg, Hannover, Lyon

We allow minimum use of abbreviations because they're hard to read and often the same abbreviation means different things in different specialities and contexts.

Drugs should be referred to by their approved non-proprietary names, and the source of any new or experimental preparations should be given.

Scientific measurements may be expressed using either conventional or SI units, with the conversion factor expressed in parentheses only at first mention. Articles that contain numerous conversion factors may list them together in a box. In tables and figures, conversion factors may be presented in the footnote or legend. Conversion factors can be obtained here. The metric system is preferred for the expression of length, area, mass, and volume.
Blood pressure, which should be expressed in mm Hg.

Numbers under 10 are spelt out, except for measurements with a unit (8 mmol/l) or age (6 weeks old), or when in a list with other numbers (14 dogs, 12 cats, 9 gerbils).

Raw numbers should be given alongside percentages, and as supporting data for P values.

Submitted articles should not include Trademark or Registered trademark symbols in the main text, tables or figures.

These should be kept to a minimum and should be clear and consistent. If you need to justify corrections to the proofs, please do so in a covering letter, not on the proof.

Authors must verify references against the original documents before submitting the article.

References should be numbered in the order in which they appear in the text. At the end of the article the full list of references should follow the Vancouver style.

Please give the names and initials of all authors (unless there are more than six, when only the first three should be given followed by et al).

The authors' names are followed by the title of the article; the title of the journal abbreviated according to the style of Index Medicus; the year of publication; the volume number; and the first and last page numbers.

References to books should give the names of any editors, editor, and year. Examples:

21 Soter A, Wasserman SI, Austen KF. Cold urticaria: release into the circulation of histamine and eosinophil chemotactic factor of anaphylaxis during cold challenge. N Engl J Med 1976;294:687-90.

22 Osler AG. Complement: mechanisms and functions. Prentice-Hall, 1976.

For material published online, give the authors, title, date or year of publication as given on the web page, and URL.

Please add the URL if material (such as official reports) is available online as well as in print.

Information from manuscripts not yet in press or not yet published online, papers reported at meetings, or personal communications should be cited only in the text, not as a formal reference.

Authors should get permission from the source to cite personal communications.

You may know of other websites that will interest people reading your article. If you know the web addresses (URLs) of those sites, please include them in the relevant places in the text of your article. If we accept your article we will insert hotlinks in the electronic version so that people using bmj.com can jump directly from your article to those related sites.

Please try to provide informative and relevant photographs, figures, or other illustrations when you’re submitting articles to the BMJ. If you cannot provide pictures with your article, perhaps you can suggest some for our picture editor to find. Please see our detailed advice on submitting images to the BMJ.

You must seek the patient’s written consent to publication in the BMJ if there is any chance that he or she may be identified from a picture, from its legend or other accompanying text. Patients are almost always willing to give such consent. We no longer publish pictures with black bands across the eyes because bands fail to mask someone’s identity effectively.

If you are using a figure that has already been published elsewhere, get the publisher’s permission to reuse it.

Illustrations should be used only when data cannot be expressed clearly in any other way. When graphs, scattergrams, or histograms are submitted the numerical data on which they are based should be supplied; in general, data given in histograms will be converted into tabular form.

Tables should be simple and should fit across a maximum of two landscape pages, and they should not duplicate information in the text of the paper. Any caption text should be concise and relevant to the table. Ideally, columns should not change headings midway down a table.

Ethnicity and culture are socially determined variables of limited use in biological research, though they are useful in health services research. All the variables are confounded by socioeconomic status. Try to use accurate descriptions of race, ethnicity, and culture rather than catch all terms in common use. In the methods section of research articles describe the logic behind any ethnic groupings used. It is best to present a range of information including:

• genetic differences
• self assigned ethnicity, using nationally agreed guidelines
• observer assigned ethnicity
• country or area of birth (participant's own, or parents' or grandparents' if applicable)
• years in country of residence
• religion

There are now many evidence based clinical guidelines and good systematic reviews, some of which may relate directly to your article. We would like you, therefore, to provide suitable evidence for key statements - the Cochrane Library or a medical librarian should be able to help you.

Guidelines

Try to indicate probabilities and levels of evidence:

• Frequency information: when giving a differential diagnosis for a particular presentation please provide population estimates of the frequency of each separate diagnosis
• Diagnostic tests: please give their false positive and false negative rates (or sensitivities and specificities) and, if possible, the causes of false positives and false negatives
• Prognosis: information on prognosis or natural history is often missing from medical articles but is vital for rational decision making. Wherever possible, give information about remission, progression and risks of disease
• Treatments: please mention the level of evidence on which main treatment recommendations are made. This can be kept relatively simple, using three levels:
• No clear evidence: opinions based on clinical experience, anecdotal case studies, or descriptive articles; conflicting evidence from studies or poorly designed studies, even if randomised controlled trials
• Suggestive evidence: evidence from cohort, case control, before-and-after studies; evidence from non-randomised experimental studies
• Firm evidence: evidence from at least one properly designed randomised, controlled trial with adequate sample selection, sample size, and appropriate controls; with double or single blinding; and with clear outcome(s)

Please try to quantify the benefit of treatment, giving the relative risk reduction or the typical number needed to treat.