Original Paper

Received: October 20, 2003
Cerebrovasc Dis 2004;18:135–138
Accepted: February 6, 2004
DOI: 10.1159/000079266 Published online: June 23, 2004

How to Write a Research Paper
Andrei V. Alexandrov
Stroke Treatment Team, Houston Medical School, University of Texas, Houston, Tex., USA

Key Words Introduction
Peer review W Perfect manuscript W Writing skills
By now you probably wrote an abstract and submitted
it to a stroke conference. Your mentor reminds you sever-
Abstract al times to start drafting a paper, and you have no idea
Background: Busy strokologists often find little time for where to start. As a simple trick, copy and paste your
scientific writing. They sometimes develop a mental con- abstract so that Background becomes your introduction.
dition equivalent to that known by neurologists as writ- For the rest, follow the IMRaD principle: Introduction,
er’s cramp. It may result in permanent damage to aca- Methods, Results, and Discussion [1–3]. Think what ‘take
demic career. This paper provides advice how to prevent home message’ you’d like to deliver and to whom. The
or treat this condition. Methods: Prepare your manu- title sells the paper.
script following the IMRaD principle (Introduction, Meth- ‘Busy strokologists often find little time ... to treat this
ods, Results, and Discussion), with every part supporting condition’: this introduction concisely describes the study
the key message. When writing, be concise. Clearly state hypothesis, rationale, purpose, and objectives. A three-
your methods here, while data belong to Results. Suc- paragraph introduction is plenty for most topics. Expand
cessful submissions combine quality new data or new with facts from papers previously published by others,
thinking with lucid presentation. Results: Provide data among whom you may occasionally find your mentor. Do
that answer the research question. Describe here most a thorough literature search for earlier sources dealing
important numeric data and statistics, keeping in mind with your subject [4–6]. Tell here what is known in the
that the shorter you can present them, the better. The field. You do not need to refer to every paper ever written
scientific community screens abstracts to decide which on this topic. Select key references and remember that for
full text papers to read. Make your point with data, not publishing purposes, less is better than more. Consult
arguments. Conclusions: Conclusions have to be based your mentor as often as possible – he is the senior author
on the present study findings. The time of lengthy and after all.
unfounded speculations is over. A simple message in a The third paragraph should state the research question
clearly written manuscript will get noticed and may [7]. You may take an original paper already published in
advance our understanding of stroke. Cerebrovascular Diseases to use as a template. Formulate
Copyright © 2004 S. Karger AG, Basel the research question clearly since data presentation
should provide equally clear answers.

© 2004 S. Karger AG, Basel Dr. A.V. Alexandrov
ABC 1015–9770/04/0182–0135$21.00/0 University of Texas
Fax + 41 61 306 12 34 6431 Fannin St, MSB 7.044
E-Mail karger@karger.ch Accessible online at: Houston, TX 77030 (USA)
www.karger.com www.karger.com/ced Tel. +1 713 500 7087, Fax +1 713 500 7097, E-Mail avalexandrov@worldnet.att.net

and stop. tell how and by whom statistician to analyze and interpret the data. work with a new imaging technology was used.18:135–138 Alexandrov . Subjects and Methods written. if they ty. plined. Although general guidelines are available you conceived the project. it will not be published – regardless of how well it is they accept. search question can be included in the manuscript. Describe in detail actual numbers for study demographics. patients admitted to our stroke unit from August 1999 through August 2002’. Your results are the most important part of the manu- ping rules. Major scientific journals currently accept less than Present results to colleagues since they would likely ask 25% of submitted manuscripts..’. You need all this space to deliver the key mes- 136 Cerebrovasc Dis 2004. script and get the message at first reading. If you of your work. script. validation of research tools. Avoid redundant or duplicate publications since these should not be published. it does not for more data or analyses.’. Observations beyond the primary re- The most important factors for publication are the quali. ‘blinding’ of observ. The flurry of numbers and ‘p’ values should follow ers – all of these facts. read other papers and change the draft again. Start by describing your study subjects. Describe subject selection criteria and data collection The last paragraph of this section should describe tools tools. it will be possible. Instead say: ‘Study subjects were recruited from consecutive patients admitted to our Results stroke unit.215 statistical analysis and data presentation. Inclusion criteria were . Stay on mentor. research done by others. script. so include data in means that reviewers did not give it a high enough priori. simple logics. If a statistician before embarking on a project. Clearly state if a local ethics committee ap. The internal review is ty. novelty. question. estimated sample size. Very few peo- The author also needs to disclose funding sources and ple can write a perfect manuscript on the first draft.. particularly if experiments were performed. Inappropriate inclusion of authors will decrease top of current literature and know the limitations of the likelihood of manuscript acceptance. after all. to flip back and forth through pages to understand what Documentation of protection of research subjects is was done to study subjects. Follow reviewers’ suggestions and you increase the Most importantly. Remember to stay in focus. in. This makes reviewers frustrated since they have study that established their relevance. Remember. Make this description detailed enough so that if of statistical analysis appropriate to study design. The shorter you can present your data in dence in the reliability of your methods.. the reality often demands seeking advice from your doing it and 5 others have already published it. read it. use tiation. Avoid presenting actual data in statistician reviewing the whole manuscript for clarity of this section: ‘Study subjects were recruited from 1. ‘10 mentioned if applicable. helpful to determine sufficient data to answer the research you got very good advice how to improve your manu. essential. dependent data acquisition and analysis during clinical and again. provide data relevant to the re- likelihood that another esteemed journal will accept it. But do not hold it for too long. and again. and chances are that by the time co-authors [8]. 10 other groups were already [8]. the pharmaceutical industry. ing sentences. if established before the study ini. Make data presentation so clear and simple that a tired proved your study. Most likely the reviewers of any necessarily mean your manuscript is poor. If Prestigious journals have a strict word limit for papers not. search question. reliability and scientific or clinical importance strengthen your case. If rejected. You should not be too disappointed because. so will be reviewers. Present them clearly by avoiding long and confus- Provide additional evidence that would increase confi. other groups . the first draft of your manuscript. Avoid opening the outcome models or dependent variables. Rejection esteemed journal would do the same. Control for tables and figures. the better. get lost from the aim of the study. Data safety monitoring. For clinical sentences like: ‘Table 1 summarizes our findings in sub- outcomes or surrogate markers. A manuscript should disclose new informa. potential for commercial bias such as connections with Return to the draft. reference a pivotal trial or group C’. Consult someone wants to repeat the study. I still do it before I give the manuscript to my trials and appropriate overseeing committees should be co-authors. Methods may dis- close power calculations. Remain focused and disci- biases. and have a these tests were validated. This ensures patients or animal rights person riding late on an airplane can take your manu- protection. change cumbersome parts. tion or a new way of thinking about old information. will strengthen the manuscript.. Scientific publishing is The first author drafts the manuscript and determines extremely competitive.

and defective ta- The second paragraph may describe the novelty of bles or figures [13–15]. Packing manuscript with data is better than splitting cance of your research. literature. Do not provide your opinion training – his patience with my ‘a’s and ‘the’s. Remember that statistical difference Clinicians need to develop skills in scientific writing. and probably will also improve your patient literature. Rules’ that he taught his fellows are partly reflected in this paper. inappropriate or often the weakest part of the manuscript. Do not present any new data that were biased. for – among many things during fellowship that are not founded in data. over-interpretation of results. 12]. and presentation. If does not necessarily translate into clinical significance. Common This section should start with: ‘Our study showed . insufficient data presented.. If there are too many abstracts. and the first lessons in how to solve a problem that was not directly evaluated in study design.sage. Avoid broad claims and strong the paper into separate small ones. a sample too small or the introduction. Do not write a review of all possible mecha. There is no reason to underline how terrific ported. incomplete. you make a significant observation. your observations. care. This Publish or Perish Cerebrovasc Dis 2004. the ratio of ab. not arguments. FRCP. is a critical part since it is easy to overestimate the signifi- tial. This part may also mention unresolved ques- tions and direction of future research. I am indebted to John results or other studies [9. a proper and fast Make your point with data. Discussion is statistics. analysis. MD.’ to reasons for rejection include inappropriate or incomplete lucidly summarize your study findings. only the beginners try to refer to all published wrestling with reviewers. inaccuracy or inconsistency of the data re- presentation. While member. Show absolute numbers as well as per. nisms that you have not accounted for in your study. Avoid general statements Norris. Most likely. statements. The fourth paragraph should describe study limita- tions. The concluding paragraph can summarize the poten- tial significance of your findings and what changes to research or clinical practice your data may support. You can write a short but to-the-point Discussion. or outdated review of the your results are – let them speak for themselves. The infamous ‘Norris your study. This brutally honest state- have writer’s cramp or the quality of your research is ment should motivate you to learn yet another set of use- insufficient for publication.18:135–138 137 . Study limitations may be contrasted with study strengths. insufficient problem not shown in the results section and avoid repeating data statement. scientific communication is required [12]. a peer-reviewed journal means completion of your project centages so that reviewers can judge the significance of and dissemination of research results [11. Re. Confirmatory research makes passing the re. If and when your papers are rejected. mention different conclusions drawn from your The author is not a native English speaker. so do not mess around but concentrate on the essen. difficult-to-follow writing. In the space needed for this. Do not repeat sub-optimal instrumentation. A skillful selection of the most perti. learn your findings or if they parallel previous research. If you do not discuss study weaknesses. When reading criticism. inaccurate. from your mistakes or the advice given to you. Acknowledgment thors. Remember that even pioneer break-through Mention a statistical test that generated specific ‘p’ val. studies require independent confirmation. long run this will make a major difference to your aca- nent references demonstrates a command of the relevant demic career. your way of writing will become more view process more difficult. Arbitrarily. you will become a better scien- papers in the field.. Avoid choosing an inappro- priate journal for your manuscript submission. more critical scientist. Good luck! The third paragraph may describe how your study con- tradicts previous research or established dogmas. stracts to original papers in curriculum vitae should be An anonymous and probably frustrated academician less than 3 to 1. Publication in ues or coefficients. 10]. ful skills. Improving your scientific writing is a life-long process. evidence based. you either once said: ‘Publish or perish!’. No esteemed journal can afford the tific writer but also a better. If there was disagreement about study interpretation by co-au. the review- ers will. remember that most manu- Discussion scripts face the same fate.

pp 69–88. Lijmer JG. STARD Group: Towards complete and accurate reporting of studies of diagnostic accuracy: The STARD initiative. St Louis. References 1 Pakes GE: Writing manuscripts describing 6 Cupples SA: Publishing the research report. Acad Med research paper. Mosby nesses in medical education reports. script preparation and submission: How to get 9 Horton R: The hidden research paper. manuscript preparation. cept manuscripts: The strengths and weak- 2 Pamir MN: How to write an experimental tions to Clinical Practice. JAMA 15 DeBehnke DJ. Nature 1999. 14 Samet JM: Dear author – advice from a retiring 3 Kern MJ.310:1313–1315. in general medical journals. Alderson P. Bruns DE. 138 Cerebrovasc Dis 2004. clinical trials: A guide for pharmacotherapeutic Alzheimer Dis Assoc Disord 1999. De Vet HC. 109–113.23: 11 Bossuyt PM.399:393. 8 Anonymous: Policy on papers’ contributors. and interactions 4 Cummins RO: Learning to write: Can books sections in reports of controlled trials published with editors. 12 Pearn J: Publication: An ethical imperative.’ and risking. Rennie D. researchers. Ann Pharmacother 2001. 2002. ‘riting. Shih RD: Research your paper accepted. 10 Clarke M.58:391–396. AJR Am J Roentgenol 2003. editor. Emerg Med 2001. 2799–2801. Gatsonis 59–61.18:135–138 Alexandrov .8:844–850.181:51– 55. Research Committee of the Soci- help? J Med Educ 1981. JAMA 2002. Glasziou PP. CA. 13 Bordage G: Reasons reviewers reject and ac- 779. Irwig LM. Chalmers I: Discussion nal. Kline JA. 7 Wojner AW: Outcomes Management: Applica. Physician Exec 1997. Moher D.76:889–896.35:770– S123–S130.150:433–436.83: 2001. 2001. Acta Neurochir Suppl 2002. Catheter Cardiovasc In. Am J Epidemiol 1999.287:2775–2778. Bonneau HN: Approach to manu.56:128–132. fundamentals: Choosing an appropriate jour- terv 2003.13(suppl 1): BMJ 1995. Reitsma JB.287: ety for Academic Emergency Medicine. Acad 5 Linney BJ: The three R’s of writing: Reading.