Plan the manuscript and the order of drafting it

What a research manuscript will say can be settled before the study, on a prepublication worksheet, and before the first draft, in a problem statement of a few sentences and in outlines that grow from a skeleton into the manuscript without its prose. The sections can then be drafted in more than one order.

The prepublication worksheet and the zeroth draft

One planning sequence runs: complete a prepublication worksheet, make an outline, write a "zeroth" draft that is more than an outline and less than a first draft, then use the worksheet and the zeroth draft to get coauthors and collaborators to commit to the plan. ¹ A comprehensive prepublication worksheet forces investigators to make sure all the needed data are collected. ¹ Once it is complete, the outline should be easy to develop. ²

One such worksheet for clinical and animal studies covers ten sections. ^{1, 2}

1. Administration: the title of the study, the principal author or investigator, the departments and institutions, the coauthors and co-investigators, the database of interest, the proposed journal, institutional review board or institutional animal care and use committee approval with all necessary forms, and the deadline.
2. The target audience.
3. The hypotheses.
4. The background.
5. The study design, covering the type of study and the reasons for choosing it, the key messages, the sample size, the eligibility criteria, the interventions, matching criteria for matched studies, outcome variables, predictor variables, the method of randomisation and blinding, the statistics, and the limitations.
6. A signature and the potential impact on practice or research.
7. The anticipated diagrams, tables and charts.
8. The key references.
9. A checklist reminding authors to review the appropriate checklist both while writing the report and before submitting it.
10. The budget or funding source.

^{1, 2}

The zeroth draft, written after the outline, holds a fairly complete introduction, a protocol turned into a methods section, a hypothetical results section and a potential discussion. ^{2, 3} Writing one can help the writer and the authors get organised, and it works best for research that does not depend on sequential studies to prove a hypothesis; it also works well for grant applications. ³ Using the worksheet, the outline and the zeroth draft can save writers time and decrease headaches. ³

The problem statement

Getting started writing a manuscript, or any other document, can be difficult. ⁴ Manuscripts are large projects, sometimes taking hundreds of hours and many months, and a writer faced with a blank page or screen might be tempted to simply start typing and take the path of least resistance, which can lead to a poorly defined text that may even stray into tangential or irrelevant areas. ⁴ A clear, organised plan of attack can save a manuscript writer a lot of time and avoid many headaches. ⁴

New medical writers and medical writing students are often unsure how to start writing a manuscript and need help organising their thoughts. ^{5, 6}

In the experience of writers who work on manuscripts and teach medical writing, the biggest problem for most writers is clearly defining what the manuscript is about. ^{7, 8} The problem is not an inability to write well in English: the subject matter might be clear to the writers, but coming up with a clear and coherent concept for the manuscript is often challenging. ⁷ Writers need to consider more than the declared study objective, if there is one: they also need to consider what background information is essential and what the conclusions and key messages should be. ⁷ The problem statement is a simple and effective method for sorting this out. ⁷

A problem statement is two to three sentences that define the problem and what the manuscript does to address it, with the focus on the purpose of the manuscript rather than of the study. ⁹ Another account gives it as one or two sentences describing the purpose of the manuscript, directly related to the objective, aim, hypothesis or central question. ⁸ Distilling these ideas into a single statement can be difficult, especially for less experienced writers, but going through the process is an excellent way to start writing a coherent, effective and interesting text. ⁸

A problem statement should use the fewest words possible. ⁷ Its first part need not explain every aspect of the problem, only the main one: what specifically is insufficient or unacceptable. ⁷ Its second part needs to describe only the bottom line of what the manuscript is attempting, and This article describes and This article shows are useful ways to open it. ⁷ An example reads: Current influenza vaccines are not fully effective and must be given annually due to antigenic drift by influenza viruses. A more broadly protective vaccine is needed. This article describes the results of a clinical trial examining the immunogenicity and safety of a candidate broadly protective vaccine that targets the conserved stalk region of the influenza haemagglutinin protein. ⁹ Both parts are clear in it: the current situation is insufficient, because existing influenza vaccines are not good enough and an alternative is needed, and the manuscript addresses this by describing the results of a study of an alternative approach. ^{9, 10}

In two typical examples, the first sentence describes the overall problem and the second the specific problem for the manuscript: Many candidate HIV vaccines have been developed but results in animals have not been predictive of efficacy in humans. A reliable animal model for predicting the efficacy of HIV vaccines is needed. and Despite several initiatives to improve the treatment of epilepsy in low- and middle-income countries, in many countries, as many as 95% of people with epilepsy remain untreated. Clear, simple goals that can be immediately put into action are needed to reduce the epilepsy treatment gap. ⁸ Both describe the problem directly and succinctly and enable the writer to define a clear and simple process for the article. ⁸

Every part of the manuscript can be related back to the problem statement. ⁹ The introduction elaborates both of its parts: what the overall problem was and why it mattered, where things stand now and what is missing, and what specific problem the study aimed to address. ⁹ The methods describe how the problem was addressed, and the results describe what new information was added as a result of attempting to address it. ^{9, 7} The discussion says whether the objective was met, what the individual results were and what they meant, what important considerations there might be, and what the problem looks like with the new information. ^{7, 9}

Once the problem statement is defined, building the manuscript becomes much simpler: the statement converts easily to a basic outline or skeleton, and from there to a detailed outline and a first draft. ^{11, 12} In the outline, the introduction elaborates the statement into the overall problem and why it is important, the current situation and what is missing or insufficient, and what the study examined; the methods and the results take the items to include from CONSORT or another relevant reporting guideline; and the discussion explains whether the objective of the study was met, the individual detailed results and their relationship to what has been published, the considerations, strengths and limitations with rebuttals, the conclusions, and recommendations for applying the conclusions and for further work. ¹¹

The concept outline and the detailed outline

With a well-defined problem statement, outlining is the essential next step in building a manuscript. ¹³ One way of outlining uses two kinds of outline, a concept outline and a detailed one, the concept outline being the skeleton on which the detailed one is built. ¹³ The concept outline contains the main sections of the manuscript followed by one bullet point for each major subsection or point to be made; its bullet points are mostly place-fillers that help organise thoughts, and it is a good way to get past a blank screen. ¹³

The detailed outline goes into extensive detail and contains essentially all the information to be included in the final manuscript. ¹⁴ Giving it to collaborators saves substantial time: little time has yet gone into crafting and perfecting the prose, so major changes in content or organisation are easier to make at this stage than at the first draft. ¹⁴ With a good concept outline, building the detailed one is simple: for each bullet point, fill in all the details from whatever source materials are available, which can include clinical documents such as the clinical study report, protocol, statistical analysis plan, statistical output or investigator's brochure, as well as presentations, posters and abstracts. ¹⁴

The detailed outline can be written in any order, although a logical method starts with the results and continues with the methods, then the introduction and finally the discussion. ¹⁴ The results section of the outline holds the detailed findings that directly address the problem statement and the principal objective, question, aim or hypothesis. ¹⁴ It can be populated by extracting the data, by cutting and pasting, from the clinical study report, statistical output, poster or presentation, and figures and tables go directly in the sections or at the end of the detailed outline. ¹⁴

Once the results section is complete, what to include in the methods section should be obvious, because each result must have a method. ¹⁴ The methods should also conform to CONSORT or another relevant guideline. ¹⁴ Populating the methods at this point makes clear what information is missing, so that it can be requested from collaborators early in the writing process. ¹⁴ Wherever possible, this section is cut and pasted from the protocol, the clinical study report or other texts containing methodological details. ¹⁴

The introduction is the next logical part to complete. ¹⁴ It holds detailed information for each point in the concept outline, such as specific definitions, descriptions, epidemiological data and clinical data, and the writer checks it against the problem statement and, where they exist, the study objectives, to make sure it addresses them. ¹⁴ It can be populated by copying text from the introduction of a study protocol or clinical study report, or with background from a slide presentation, monograph or other printed information, but data and references in these documents are often out of date or inaccurate, so such information is best treated as preliminary and backed by the writer's own literature search for valid, up-to-date information and ideas. ¹⁴ Including references in the introduction of the detailed outline avoids having to track down sources later and lets the co-authors see whether the writer is using the references they prefer, and listing them by author rather than as numbered citations simplifies this for the writer and the co-authors. ¹⁴

The discussion is the most difficult part of the detailed outline to complete, because it is difficult to know what the co-authors want to say and the messages may change up until the last draft of the manuscript. ¹⁴ So this part is often much less detailed than the others, but it should include at least the main points the writer thinks belong in the discussion, so that the co-authors have a framework to begin considering what it should contain. ¹⁴ The last step in preparing the detailed outline is a cover page with a proposed title, the names and affiliations of the co-authors, the target journal or proposed target journals, and key information about the target journal, such as its limits on the number of words and of figures or tables. ¹⁴

The first draft is generated by connecting the individual points of the detailed outline into a text, avoiding plagiarism of any information copied from another document or reference, and formatting the manuscript according to the target journal's instructions to authors. ¹⁵ With a detailed outline, and ideally the co-authors' comments on it, writing the text is easy. ¹²

An outline built from points

One experienced scientific editor advocates resting a manuscript on a strong outline, because it is difficult to review an entire manuscript but easier and faster to review a short list of concise points and confirm that each is clear and that their sequence effectively tells the story. ¹⁶ Rather than taking the journal article's section headings and filling in a plan under them, the outline can summarise six things. ^{16, 17}

• the problem investigated
• what questions remain unanswered
• which of those the work tried to answer
• methods developed by previous researchers that the work will use
• new methods developed to solve problems that other researchers did not solve
• the details of the statistical analysis the methods require

¹⁷

Extending this rationale to the plan for the results and discussion sections can ensure that each result in the results section has a method used to produce it, and that every key interpretation in the discussion is supported by data described in the outline of the results. ¹⁷ In one medical writer's view, the same outline could serve as a check on much of a medical writer's work, because it can be all too easy to be distracted from the fundamental purpose of research by its details. ¹⁷

The referenced research outline

A detailed and referenced outline can be developed from the references that PubMed, MEDLINE, EMBASE, CDSR and other literature searches identify. ¹⁸ If a congress abstract is available, it is introduced as the abstract segment of the outline, and figures or tables from an existing congress poster are introduced into the outline after the statistician has confirmed that they are final and clean. ¹⁸

One such outline, the Gutkin 4 × 4 cogent manuscript structure outline, has four headings, Introduction, Methods, Results and Discussion, with four subheadings each. ^{18, 19} Each heading becomes a heading of the manuscript and each subheading one paragraph or more, each paragraph with a topic sentence. ¹⁹ The topic sentences provide horizontal logic: a reader of the outline or the essay should be able to progress from one topic sentence to the next and appreciate the arc of the argument. ¹⁹

1. Introduction, 250 to 500 words
   1. Disease state, definitions and basic epidemiology: one or two paragraphs
   2. Pathophysiology, aetiology, natural history and genetics: one to three paragraphs
   3. Consensus diagnostic and management guidelines: one paragraph
   4. Thesis statement: the clinical problem or educational need and the unique or incremental value in addressing it, the study rationale and the objectives: one paragraph
2. Methods, 250 to 750 words
   1. Study design, setting, participants and ethics: one to three paragraphs
   2. Interventions: one paragraph
   3. Assessments and outcome measures: one to three paragraphs
   4. Statistical methods: one to four paragraphs
3. Results, 250 to 750 words
   1. Patient disposition: one paragraph
   2. Baseline characteristics: one paragraph
   3. Efficacy outcome measures: one or two paragraphs
   4. Tolerability and safety: one or two paragraphs
4. Discussion, 500 to 1,000 words
   1. Key findings, with P < 0.05 and > 0.05, expected and unexpected, and meeting or not meeting endpoints or minimum clinically important differences, and why: one to three paragraphs
   2. Relationship to the published literature, narrowly construed as it relates to the study design and findings, with potential clinical implications and alternative explanations of the data: one to three paragraphs
   3. Potential study strengths and limitations: one or two paragraphs
   4. Conclusions, with clinical implications and needs for future research: one paragraph

¹⁹

The journal's author guidelines and online content are consulted, and reporting checklists such as CONSORT for randomised controlled trials and STROBE for observational studies adapt this basic structure to other forms of study report. ¹⁹ The outline is formulated in collaboration with the corresponding author and circulated to all potential authors, whose feedback is requested and whose input, deletions included, is recorded and dated. ¹⁹ Outlines should be referenced and include newly generated figures and tables or table shells. ¹⁹

The order of drafting the sections

In the experience of authors who have published scientific articles for more than three decades, the most efficient way to write a scientific paper starts in the middle and works outward, and this approach is adaptable to any topic, simple, reduces the number of drafts and consistently results in a clear paper. ²⁰

The approach finalises the tables and figures first, including supplementary ones, and decides their order so that, viewed alone, they tell the paper's story. ^{20, 21} A paper generally includes figures and tables that clearly communicate its main message, and they are an ideal place to start because one way of reading a paper is to go from its title, abstract and conclusions to the tables and figures before deciding whether to read it in full, and other scientists often present a paper's figures when discussing their own work. ²¹ This first step is often the most difficult and needs the most thought, but once it is done, writing the Results and then the Methods is usually straightforward. ²⁰ After those sections are proofed and polished, a quick review of them and of the tables and figures highlights the points the Discussion must make, and the Introduction and Abstract are written last, the one to set up the paper and the other to summarise it. ²⁰

Deciding on an initial order of the tables and figures eases the first draft of the Results even if the order later changes. ²¹ A clear set of tables and figures with a straightforward Results section eases the drafting of a clear and concise Methods section. ²² Once the Methods are drafted, it is worth checking that all the salient information, but no more, is presented transparently. ²² A good way to check is to print a clean copy of the Methods, Results, tables, figures and supplement and review them as if someone else had written them, which can help in finding confusing or distracting text. ²² It is also useful to have a colleague read these sections, because writing more of the paper is not worthwhile until they are polished. ²²

Combined with frequent proofreading on a medium other than the one used for writing, the middle-outward approach in the authors' experience gives a clear, uncluttered paper, completed in a minimal number of drafts and most likely to be favourably reviewed and accepted for publication. ²⁰ Writing clearly is difficult and takes practice whatever is being written, and a clear underlying structure with a systematic approach to building on it can ease the writing of a paper, ensure that the final paper communicates clearly and make it more likely to be used to advance biomedical knowledge. ²³

Another order begins with the Results, "because everything in the document is dependent on the presented results", and writes the Methods once the Results are complete. ²⁴

A third order, directed to graduate students familiar with a research project, writes the title first, as the shortest part, then the methods, in uncomplicated chronological order, and the results third. ²⁵ This order starts from one set of data for which the pertinent methods were written, and the whole article takes shape as further sets of data and their methods are written. ²⁵

To give a paper a good chance of success, the writer needs to write from the perspective of the reader, meaning the editorial reviewers and the eventual readers of the published paper, and in one view its title, abstract, conclusions and visuals are the key elements that determine its impact. ²⁶ Another good place to start is to write the conclusions as bullet points reflecting the key messages. ²⁶ A sentence leading into the points and a concluding sentence on the value of the findings can then be added, or the points can be written into a paragraph as a narrative instead. ²⁶ The conclusions can then be used to create an informative title and an introduction whose aims are linked to the points. ²⁶ A valuable second step is to decide which visuals will best emphasise the key messages and to create them so that they reflect those messages without reference to the text. ²⁶