How to write a medical paper

She says a scientific paper should be the first disclosure of original research findings. The manuscript should be sufficiently detailed to allow other scientists or clinicians to assess the observations, repeat the experiments, and evaluate the reasoning and conclusions drawn by the authors.

"An experiment or clinical trial is not complete until the results are published," Dr. Robinson says. "This is recognized in part by the authorization of publication costs from federal research grant and contract funds."

Some journals, such as the Archives of Dermatology, only publish primary work and will not consider derivative manuscripts. Other journals publish summaries, reviews, conference reports, meeting abstracts and commentaries on primary work.

The paper should be written in a straightforward, unadorned manner.

"Literary tricks, metaphors and flowery language divert attention from the message to the style," Dr. Robinson says. "Scientific writing is more about substance than style."

Begin early Dr. Robinson recommends beginning the manuscript while the research is still under way. Not only is the writer actively engaged in thinking about the research, but the process of writing may help clarify the research.

"The writing process may point out inconsistencies in the results or suggest other avenues that may be followed," Dr. Robinson advises.

To get over writer's block, she recommends beginning with the materials and methods section. Be careful not to include results in this section, and be sure to include a paragraph describing the statistical methods used. Consider presenting detailed information, such as inclusion criteria, in tabular form for easier comprehension.

Although the results section is certainly the most important section in the manuscript, it often has the least amount of narrative text of any of the sections. Data from large, multifactorial trials are best presented as tables and graphs. Tables and figures should be cited in the text, but not duplicated.

The introduction should give the reader sufficient background to understand the scope of the problem being investigated. Relevant scientific literature should be cited in a concise literature review. The approach taken to study the problem under investigation should be summarized, and previously published abstracts describing the present study should be cited, along with any closely related manuscripts submitted for publication or in press.

The discussion section is where the author describes the relationships and generalizations deduced from the results. Here, the author should point out limitations of the study, inability to extend the findings to larger populations, or unexpected findings. The author should use the discussion to place the study's findings in the context of previously published studies, and to explain how these findings contribute to the field.

Postpone composing a title and writing the abstract until the manuscript is otherwise complete, Dr. Robinson advises. Choose the title with care, and make it as brief as possible, using short, specific, familiar words. Remember that the title is meant to serve as a label, not a sentence.

The last step is to write the abstract.

"Make the abstract awesome. Sometimes it is the only part of the paper that is read by a reviewer," she says. "If it is a great abstract, the reviewer reads the whole paper. If the abstract is awful, the reject box gets marked and no reviewer comments are made to the author."

A structured abstract is concise: usually 250 words or less. References should never be cited in the abstract, and all information in the abstract should be included in the main text of the paper.