Write the introduction that leads to the question
An introduction moves from the broad problem to the gap the study addressed, and it ends on the question and on what was done. It narrows like a funnel within a limited length, it can waste its first sentence on a banal opening, it works through its steps in order from background to what was done, and it sets out the problem and background in the present tense. Whether it should also announce the findings is disputed.
From the problem to the question
The main purpose of the introduction is to set the stage. A description of the gaps the manuscript fills should lead directly into the research question, which should be written at the end of the introduction.
The introduction should cover three areas: the general nature of the problem, the most pertinent, timely published studies that relate to it, and the specific, focused question the study set out to answer, with any formally stated hypothesis. It classically opens with a broad statement that puts the problem in context, and good introductions are often written as a three-level inverted pyramid, from the broadest statement through a narrower one to a statement of the study's purpose. Prior work should be selected only for its direct bearing on the research question, not presented as an exhaustive review.
The introduction works like a funnel, wide at the top and narrow at the bottom. It starts from who the reader is, what they really know and what they absolutely need to be told, and the writer decides whether to start from prevalence figures or from the fact that nobody even knows how to treat the disease. It then turns to the research under way and to what is missing from it, such as side-effect data in an otherwise well-studied area, or a specific problem that needs solving. The final paragraph states the objective and the hypothesis. Anything not directly related to the study objective should be removed, however interesting it is, and every statement takes a reference unless it is established science, with a reference given when in doubt.
A short example of the funnel narrows from a first point to a more restrictive one, that too little first aid reaches patients who suddenly have a stroke, then gives an illustrative figure for those patients, offers a solution, adds research showing that the solution has been useful, and finishes with the hypothesis.
An outline for an introduction on prostate cancer also shows the funnel. It runs from prevalence, to incidence rising over twenty years, to 417,000 new cases in Europe in 2012, to the range of available treatments from hormone therapy to surgery, radiotherapy and chemotherapy, to androgen deprivation therapy used in a third of patients, to a new generation of that therapy shown to extend survival, with its mode of action and supporting studies, and finally to what this study was designed to do.
The introduction should guide the reader's thoughts gradually to the study aims, which its last paragraph describes. Its ideas should be organised so that, immediately before reading the aim, the reader understands the relevance of the topic and anticipates which gap in knowledge has to be filled.
Length
An introduction should have no more than five paragraphs, each carrying one main idea, and three to five points can work. The ideas can be planned as five words, built into five sentences that each become the first sentence of a small paragraph. The whole section should be approximately 500 to 600 words, because more than that bores readers and leads them astray.
The first sentence
The first sentence should grab the reader and draw them immediately to the crucial issue the paper addresses. Too many papers instead open with banal information, the written equivalent of clearing the throat. A chapter on renal blood flow in a book on specialised cardiovascular physiology opens by stating that the kidneys are bilateral, bean-shaped organs lying in a retroperitoneal position on either side of the vertebral column beneath the diaphragm. That is unnecessary even for a second-year medical student, let alone the book's actual reader.
The same pattern appears in a leading respiratory journal. A paper on prognostic indicators in idiopathic pulmonary fibrosis opened by telling that journal's readers that the disease is a relatively common interstitial lung disease. Of its 34 references, 29 could be read in full, and twelve of those opened with a variation on the same kind of sentence. Only two had focused openings that told the reader what was coming: that there is an unmet need for an accurate non-invasive measure of disease severity in the condition, and that the condition has undergone important redefinition based largely on revised histopathologic classification criteria.
Sometimes deleting the first sentence and starting with the second improves a paper instantly. Sometimes a banal first sentence signals that the whole introduction needs rewriting, often because the authors have treated it as a general review of the topic. A general review is appropriate in a thesis but unnecessary and boring in a research paper that asks and answers a circumscribed question. An orthopaedic paper on dislocations after total hip replacement shows the alternative: it opens by stating that between January 1972 and June 1975, 300 procedures were performed by five surgeons on one hospital's orthopaedic service, and so puts the reader straight in with the surgeons and their results.
A common error in an introduction or background section is to begin too broadly or too far off topic. The introduction can begin with a startling statistic that illustrates the seriousness of the problem the study addresses, and it should get to the point as soon as possible. Dictionary definitions of terms that readers already know should not be given.
The steps from background to what was done
The first sentence should put the main topic up front, so the reader knows at once where the article is headed. In a clinical paper, Herpes zoster, also known as shingles, is a common and often debilitating disease that occurs primarily in older or immunocompromised individuals tells the reader that the subject is herpes zoster. The same move works for basic research: Activation of the epidermal growth factor receptor stimulates the growth of many cell types and is implicated in some forms of breast and other cancers announces the receptor as the subject.
The rest of the first paragraph gives the general background of the area. For a clinical study, that is the incidence and prevalence of the disease, its effects on the patient and its economic impact; for basic research, it is the essential characteristics of the system. What to include depends on the journal's audience. An expert audience needs no basic information, while a more general one may need the fundamentals of the research area or the disease, so an article for an infectious-diseases journal may not need to explain what streptococcal bacteria are, while one for a molecular biology journal might. This background should generally be limited to a single paragraph.
Three steps follow the background. The introduction first describes where things stand now: how the disease is currently treated or what the current level of understanding is, what options or solutions exist, what improvements or new developments there have been, and where the field is heading. If the study used a special technique or method, this is a good place to establish its validity. Next it states what is missing or what problem remains, explains what the study or set of experiments tried to solve, and says why that question matters. Finally it says briefly what was done: for a clinical study, what kind of study it was and what the principal objectives were; for basic research, what the work set out to determine, with the hypothesis stated here if there was one.
Tense
The general problem and the background work should be described in the present tense, and a research question presented as a statement is usually in the past tense.
Announcing the findings
Whether the introduction should announce the findings is disputed. The ICMJE recommendations advise against including data or conclusions from the work being reported, while others advocate stating the conclusions early in the article rather than keeping the reader in suspense, and the introduction may be the best place for what the findings imply for clinical practice. Where this controversy applies, the writer should use their best judgement, based on the data being presented.
The introduction should not end with a summary of the findings. Some articles used to end that way, and a senior author may insist on it, but the abstract already summarises the results, and the introduction's goal is to explain why the study was done rather than what was found.
A summary of the key findings straight after the study question also gives the story away. The AMA Manual of Style, in section 19.1.2, states that results or conclusions do not belong in the introduction.