THE ARCHIVES editorial staff is dedicated to publishing outstanding
research articles leading to the improved health and health care of children
and adolescents. As part of doing so, we will publish a series of articles
that we hope will improve the readability and quality of submitted papers
and improve the peer-review process. This first article provides suggestions
to authors on writing for the ARCHIVES.
GENERAL APPROACH TO MEDICAL WRITING
We are firm believers that most writers are made rather than born. Our
own first attempts at medical writing were terrible; hopefully they have improved
over the years. Writing skill comes from experience; it may never get easy
for some, but it will get better the more you do it.
The first draft of a paper is often the hardest to write, but the great
advantage of computers is the ease of revision. Don't aim for perfection on
the first draft. Just get it down! Write the first draft early in the study,
perhaps even before you know the results. When your ideas are on paper, it
is easier for you and your colleagues to see what further work is needed.
In writing, as in conducting research, keep the focus narrow. The best
studies and the best papers are those that ask a single focused question and
answer it well. We do not advocate splitting a study into many publications;
rather, we believe that a study and the resulting publication will be more
readable and informative if the focus is on one question.
One way to make papers more readable is to use the active voice and
the first person. Many of us were taught that scientific writing sounds more
objective if we use the passive voice and the third person; however, articles
are more interesting and more easily read if the writer says, "I conducted
a randomized trial" or "We carried out a case-control study," rather than,
"A study was done." Science is done by human investigators, not robots; authors
should say what they did and take credit or blame for their work.
EDITING
Plan on putting your paper through many drafts. We never submit our
first or second draft for publication, and sometimes the number of drafts
gets into double digits. Multiple drafts are important to get a high-quality
product.
When revising, seek out critical reviewers. Only your parents should
think your work is perfect. Potential editors include coauthors as well as
mentors and colleagues. You want editors who will improve your work, not merely
correct your typos. Don't take their comments (or those of journal editors
or peer reviewers) personally. The comments are meant to make the paper better,
and usually they will. Comments are like clinical advice from a consultant;
you take them into consideration, weigh them, and then make a decision about
what is best. Blind acceptance of each comment from multiple coauthors may
lead to a manuscript that does not make sense. The first author's job is to
make sure that the final paper reads well.
You should be your own best editor. Print out your work in double-spaced
format and go over it with a pen. To see your writing with a fresh eye, set
it aside for a week or 2, then read it again. Try to be a severe critic of
your own writing.
Strive for clarity above all else. Avoid unnecessary jargon. If a $1
word will do the job, choose it over a $10 word. Readers will find a clear
manuscript more persuasive and enjoyable than one that attempts to makes its
authors sound scholarly.
Brevity is a virtue. Omit unnecessary phrases, sentences, and paragraphs.
Carefully examine your work for empty phrases. For example, "a majority of"
can be replaced with "most." "It is of interest that" can just be deleted.
Search for useless sentences at the start and end of paragraphs; they are
often lurking there. Editors try to publish as many articles as they can given
a fixed number of pages allotted each month by the publisher. Most editors
will publish 2 manuscripts of 2500 words each rather than one manuscript of
5000 words. In 272 words, Lincoln at Gettysburg changed the way America thought
about its founding principles.1 Watson and
Crick described the structure of DNA and won the Nobel Prize with a paper
that was 1 page long.2 The most complicated
studies are now routinely published in fewer than 3500 words.
Don't use abbreviations unless you have good reason to believe that
the abbreviation is well known to most readers, and then only if you are going
to use the abbreviation many times in your paper. If you use an abbreviation,
define it when first used.
FORMAT
Abstract
Don't attempt to write this until you are nearing your last draft. Be
sure that it contains nothing that is not also in the manuscript. Include
your main results using numbers, not just words. Be sure that the numeric
results you give are the same as those in the manuscript; when the results
in the text and the abstract differ, reviewers are not favorably impressed.
Introduction
Make this succinct. The purpose is not to give a comprehensive review
of the literature. It should state the importance of the problem (without
being trite), point out the area in which important knowledge is lacking,
and state the research question that your study will answer.
Methods
The ARCHIVES typesets the "Methods" section into a special box not to
suggest that it should be optional reading, but rather to emphasize its importance.
State the research design; don't make the reader guess. Define the study population,
the study setting, and the time interval for the research. How were the data
collected and what instruments were used? Clearly state the main outcome measures
that will be estimated. Tell us about the statistical analysis in clear language
and with appropriate citations to new or unusual methods. Don't tell the reader
the name of a statistical package unless the analysis used unusual methods
only available in a few software packages.
Results
Put all the results in the "Results" section; results should not appear
in the "Methods" and no new results should be introduced in the "Comment"
section. Some description of the study population is usually needed. Then
describe those results that are directly related to the main study question.
Readers do not need to see every number that was generated by your computer.
Use tables and figures to display the results, and use the text to succinctly
point out key findings in the tables, without repeating all of the information
in the tables. Don't waste space by telling us to look at a table or figure;
for example, don't say, "A comparison of the cases with the controls is given
in Table 1." Instead, put the name of the table in parenthesis when this is
relevant. For example: "Cases were more often younger than 5 years compared
with controls (Table 1)." Use the past tense since what you did is in the
past: for example, "boys were taller than girls."
P values are rarely of interest.3, 4
We almost always want to see point estimates and 95% confidence intervals.
For example, if you studied the outcomes of patients treated with drug A compared
with drug B, tell us the proportion in each group that had the outcome of
interest, then estimate the difference or ratio of these proportions, and
provide a 95% confidence interval for that estimate. Don't pontificate about
your results in this section; take the Jack Webb approach and just give the
facts.
Comment
This section is too long in most submitted papers. Make it short and
to the point. Start with a 1-paragraph summary of your key findings. Review
relevant past studies and explain how your results compare with those of others;
give numerical comparisons when possible. If your results disagree with prior
work, discuss why this might have occurred. Describe the limitations of your
work and how they may have influenced your results. No study is perfect. You
and your coauthors will be (or should be) aware of some of these problems.
Peer reviewers may point out others. Good peer reviewers and knowledgeable
readers will appreciate a forthright discussion of limitations. Don't be boastful.
If your study was the first, simply note that you cannot compare your study
with previous work because there isn't any. Be thoughtful about your final
conclusions. Don't make trite statements like "more research is needed." Above
all, don't make statements that are not supported by your data. The concluding
sentence from Watson and Crick's paper was the ultimate in understatement:
"It has not escaped our notice that the specific pairing we have postulated
immediately suggests a possible copying mechanism for the genetic material."2
Tables and Figures
Often, the main message of a paper can be presented in 1 or 2 well-formatted
tables. Give your tables clear titles so that a reader who does not have the
manuscript can still understand the table. Avoid abbreviations if possible,
and provide footnotes that explain any abbreviations used. A good plan is
to emulate the tables in studies similar to yours that have been published
in the ARCHIVES or JAMA. Pie charts are rarely of use, and the information
in most bar graphs can be better presented in a table or in the text. Make
sure that numbers in tables agree with those in the text.
REFERENCES
Use enough references to support your statements, but not so many that
they become distracting or take up a great deal of space. Scientific articles
published in journals are not doctoral dissertations aimed at a complete review
of the literature. Make sure that the references are both accurate and up-to-date.
For example, don't cite vital statistics data that are 5 years old when more
recent information is easily available on the Internet. Do yourself a favor
by using bibliographic software. This will make it easy for you to format
your in-text citations and list of references. Software can make it almost
painless to redraft your paper and cut, add, or reorder your citations.
COMMON PROBLEMS
There are some recurring errors that make editors and reviewers less
likely to look favorably on a paper. Failure to read and follow the instructions
to authors for the ARCHIVES leaves the impression that the authors were either
careless or arrogant.5, 6 If some
of our instructions seem a bit arbitrary, keep in mind that we review nearly
700 manuscripts each year. Having a common format makes this task easier.
Use your software's spell-checker. Then proofread your paper; your software
won't know if you meant "from" instead of "form" or "there" instead of "their."
Avoid cliches and superlatives. Words such as "very," "extremely," or "enormous"
are subjective and usually superfluous. Be precise in your use of words. Don't
use the word "rate" unless you mean the number of events per person-time.
Reserve the word "significance" for discussion of statistical significance.
Don't say there was "no difference" based on a P
value. There is "no difference" only when the difference is zero.7 Don't say "we compared the groups with regard to demographic
variables"; say that you compared them with regard to age, sex, income, or
whatever the case may be.
SUBMITTING THE MANUSCRIPT
We accept manuscripts in either electronic or paper format. Make sure
that all the copyright and conflict of interest forms have been signed by
all authors. We can't process your work until we have these.
USEFUL REFERENCES FOR AUTHORS
Here we list a few publications that we recommend to authors:
- Browner WS. Publishing and Presenting Clinical
Research. Baltimore, Md: Lippincott Williams & Wilkins; 1999. [Often
humorous, a bargain for the price.]
- Huth EJ. Writing and Publishing in Medicine. 3rd ed. Baltimore, Md: Lippincott Williams & Wilkins; 1999. [Full
of practical information. The chapter about revisions is superb.]
- Lehner A. An antidote to "The Elements of Style." The New Yorker. June 4, 2001:43. [Humorous examples of bad writing.]
- Rothman KJ. Writing for epidemiology. Epidemiology. 1998;9:333-341. [Blunt advice.]
- Tufte ER. The Visual Display of Quantitative
Information. Cheshire, Conn: Graphics Press; 1983. [Read the discussions
of "chartjunk" and data-ink.]
- Williams JM. Style: Ten Lessons in Clarity and
Grace. 6th ed. New York, NY: Addison Wesley Longman; 1999. [In-depth,
witty discussion. The author suggests that we emulate the clear style of Mark
Twain.]
