How to Write a Manuscript

Dr. Nadim Mahmud

Foundational Research Curriculum

Introduction

At some point in your training, you will have data. Maybe you collected it yourself; maybe you inherited a dataset from a graduating fellow; maybe you ran a query on an institutional database. Whatever the origin, the challenge is the same: turning that data into a manuscript that can be shared with the broader medical community.

Writing a manuscript for the first time can feel overwhelming. There are a lot of moving parts — structure, content, tone, authorship, submission logistics — and most trainees receive little formal instruction on any of it. This guide walks through the manuscript writing process from start to finish, with a focus on the conventions that matter most for peer-reviewed medical literature.

Before You Start Writing

Before writing a single sentence, a few preparatory steps will save you significant time and frustration downstream.

Know your story. Before writing, you should be able to answer in one or two sentences: what did you study, what did you find, and why does it matter? If you cannot answer these questions concisely, the data may not yet be ready for a manuscript, or your analysis may need further refinement. Work through this with your mentor before drafting.

Choose your target journal early. Different journals have different word limits, formatting requirements, and article types. Writing toward a specific journal from the outset ensures your manuscript fits that journal's scope and style. Review the journal's "Instructions for Authors" before you begin drafting.

Build your tables and figures first. Your tables and figures are the core of your manuscript; they contain the actual data. Construct them before writing any prose. Once you know what you are presenting visually, the narrative text becomes much easier to build around them.

Work closely with your mentor. Manuscript writing is a collaborative process. Discuss authorship, the overall narrative arc, the target journal, and a realistic timeline with your mentor early. Share drafts frequently and be responsive to feedback throughout.

The IMRaD Structure

Nearly all original research manuscripts in the biomedical literature follow the IMRaD format: Introduction, Methods, Results, and Discussion. This structure exists for good reason, as it mirrors the scientific process and allows readers to quickly locate what they need.

Section	Core Question It Answers	Typical Length
Introduction	Why did you do this study?	2–3 paragraphs
Methods	How did you do it?	Varies; typically 400–800 words
Results	What did you find?	Varies by number of analyses
Discussion	What does it mean?	3–5 paragraphs

One practical tip: you do not have to write the sections in order. Many experienced writers start with the Methods (the most concrete section), move to the Results, then write the Introduction and Discussion last, once the story has fully crystallized.

Writing Each Section

Introduction

The Introduction has one job: convince the reader that your study needed to be done. It should follow a funnel structure, starting broad and narrowing to your specific research question. A strong Introduction typically includes:

Context: What is the clinical or scientific landscape? Why does the topic matter?
The gap: What is unknown, unresolved, or contested in the existing literature? Language like "however" or "despite this" signals tension and justifies the study.
The rationale: Why is this gap worth addressing? What are the limitations of prior work?
The aim: The final sentence should clearly state what your study set out to do. Keep it concise and specific.

Keep the Introduction focused and concise. It is not a literature review. Two to three paragraphs is usually sufficient; resist the urge to summarize everything that has been written about the topic.

Methods

The Methods section should be sufficiently detailed that another researcher could replicate your study. It is also the section that expert reviewers will scrutinize most carefully. A well-organized Methods section typically includes:

Study design and data source: Always the first sentence. State the design (e.g., retrospective cohort, randomized controlled trial) and the data source or setting.
Population and eligibility: Who was included and excluded, and how were they identified?
Exposure and outcome definitions: How were these measured or classified?
Covariates: What confounders or other variables were measured?
Statistical analysis: What tests or models did you use, and why? What software was used?
Ethical considerations: IRB approval status, data use agreements, or waiver of consent where applicable.

Write in the past tense throughout the Methods, as you are describing what was done. Use subheadings (e.g., Study Population, Exposure, Outcomes, Statistical Analysis) to organize longer Methods sections; this makes the section easier to navigate for reviewers.

Results

The Results section presents your findings clearly and objectively, without interpretation. It should follow the same logical order as your Methods. A typical Results section proceeds as follows:

Cohort description: How many patients were identified? How many were included after applying exclusion criteria? A flow diagram (CONSORT or STROBE-style) is often helpful here.
Baseline characteristics: Describe the study population and any differences between groups. Direct the reader to Table 1 here. Table 1 is the conventional label for the first table in a manuscript, and typically presents a summary of the study cohort's demographic and clinical characteristics — often stratified by exposure group or study arm.
Primary outcome: Present your main result first, with the effect estimate (e.g., odds ratio, hazard ratio, risk difference), 95% confidence interval, and p-value.
Secondary outcomes and sensitivity analyses: Report these after the primary outcome, in the order described in your Methods.

Let your tables and figures do the heavy lifting. The text should highlight the most important findings and guide the reader through the data, not re-state every number from a table. Avoid interpretation in the Results; save that for the Discussion.

Discussion

The Discussion is where you interpret your findings, place them in context, and explain why they matter. A well-structured Discussion typically follows this sequence:

Summary of main findings: Begin by restating your primary finding, not the aim of the study. Starting with "In this study, we aimed to..." is the single most common Discussion error.
Context and comparison with prior work: How do your findings align with or diverge from existing literature, and why?
Mechanism and clinical implications: Why might this finding be true? What does it mean for clinical practice, policy, or future research?
Limitations: Every study has them. Be specific and name the most important ones for your study design (e.g., residual confounding and inability to establish causality for observational studies; selection bias if enrollment was non-consecutive; information bias if outcomes were self-reported; limited generalizability if the cohort was narrow). Generic disclaimers are not useful, and reviewers will raise specific limitations if you do not.
Conclusion and future directions: End with a concise conclusion that reflects the weight of your evidence, and suggest meaningful directions for future work.

Tone matters in the Discussion. For observational studies, avoid causal language ("X caused Y") in favor of associative language ("X was associated with Y"). Do not overstate generalizability, and be honest about what your data can and cannot answer.

Pre-Submission Checklist

Use this checklist to review your manuscript before submitting. Check off each item as you confirm it is addressed.

Introduction0 / 4

Opened with clinical or epidemiological context that establishes the importance of the topic
Identified the gap or unresolved question in the existing literature
Explained why prior work is insufficient and why a new study is needed
Ended with a clear, concise aim statement describing what you did (not what you found)

Methods0 / 5

First sentence states the study design and data source
Inclusion and exclusion criteria are explicitly defined
Primary exposure and primary outcome are clearly defined and operationalized
Statistical approach, key covariates, and software are described
IRB approval status or waiver of consent is noted

Results0 / 5

Begins with cohort description (how many patients were identified and how many were included)
Baseline characteristics are described and Table 1 is referenced
Primary result is reported with an effect estimate, 95% confidence interval, and p-value
Secondary outcomes and sensitivity analyses are reported in the same order as the Methods
No interpretation appears in the Results section (saved for the Discussion)

Discussion0 / 5

Opens with the primary finding, not the study aim
Findings are contextualized within the existing literature
A mechanistic explanation or clinical implication is provided
Limitations are discussed specifically, not with generic disclaimers
Ends with a conclusion and meaningful future directions

Beyond the Main Text0 / 5

Title is specific and conveys the study design, population, and primary finding or question
Abstract was written last and reads as a standalone summary
All co-authors have reviewed and approved the final manuscript
Conflicts of interest and funding sources are disclosed
Journal's Instructions for Authors have been reviewed and formatting requirements are met

Beyond the Main Text

A manuscript is more than just the four IMRaD sections. Several other components require careful attention before submission.

Title. The title should be informative and specific, ideally conveying the study design, population, exposure or intervention, and primary finding or research question. Avoid vague titles that leave readers uncertain about what was studied.

Less informative

"A Study of Statin Use and Outcomes in Patients with Cirrhosis"

Vague: the study design, direction of the finding, and specific outcomes are all unclear.

✓ More informative

"Statin Use Is Associated with Reduced Hepatic Decompensation and Mortality in a National Cohort of Patients with Cirrhosis"

Specific: conveys the exposure, direction of effect, outcomes, study design, and population.

Abstract. The abstract is a standalone summary of your manuscript and is often the most widely read part of your paper. Write it last, once the full manuscript is complete. See the How to Write a Research Abstract module for detailed guidance.

Authorship. Authorship should reflect genuine intellectual contributions to the study. The ICMJE (International Committee of Medical Journal Editors) criteria require contributors to have participated in all four of the following: (1) conception or design, or data acquisition, analysis, or interpretation; (2) drafting or critically revising the work for intellectual content; (3) final approval of the version to be published; and (4) accountability for all aspects of the work. Discuss authorship with your mentor early and revisit it if the project scope changes. First authorship typically goes to the trainee who led the intellectual work and the writing.

Tables and Figures. Well-constructed tables and figures should be largely self-explanatory — a reader should understand what they show without reading the full text. Every table and figure needs a clear title and sufficient footnotes to define abbreviations and data conventions. Keep formatting consistent, follow the target journal's style guide, and use clean, uncluttered designs.

Acknowledgments and Disclosures. Acknowledge individuals who contributed meaningfully but do not meet authorship criteria (e.g., statistical consultation, data extraction support). Disclose all funding sources and any potential conflicts of interest per journal policy — this is a requirement, not optional.

Tips for Writing

1. Start with the Methods. If you are staring at a blank page, do not start with the Introduction. The Methods is the most concrete section and requires the least interpretive judgment. Once that is written, the Results and Introduction fall into place more naturally.

2. Write badly first. The goal of a first draft is to get words on a page, not to produce polished prose. Give yourself permission to write imperfectly; editing is far easier than writing from scratch. A rough draft you can revise is infinitely better than a blank page.

3. Use active voice where appropriate. Active voice ("We identified...") is generally cleaner and more readable than passive voice ("Patients were identified..."), and is now preferred or accepted by most major biomedical journals. Check your target journal's style guide for their preference.

4. Let your tables and figures tell the story. Resist the temptation to repeat every number from your tables in the text. Point the reader to the table or figure, highlight the most important values, and move on. Redundancy between prose and tables inflates word count and dilutes the key findings.

5. Share drafts early and often. Do not wait until you have a polished draft to share with your mentor. Sharing early catches structural problems before they become entrenched, and ensures your mentor's vision is aligned with yours from the start.

Common Pitfalls to Avoid

1. Starting the Discussion by restating the aim. Opening the Discussion with "In this study, we aimed to examine..." is the most common structural error in manuscript writing. The Discussion should begin with the primary finding, not the aim. The aim was already stated in the Introduction.

2. Introducing new data in the Discussion. The Discussion should interpret findings that were already presented in the Results. Do not introduce new data, statistics, or analyses in the Discussion. If a finding belongs in the paper, it belongs in the Results section.

3. Using causal language in observational studies. Observational studies establish associations, not causation. Avoid writing that exposure X "caused" or "led to" outcome Y. Use terms like "was associated with," "was linked to," or "correlated with."

4. Ignoring journal-specific formatting requirements. Word limits, reference styles, figure requirements, and supplementary data policies vary widely across journals. Failing to follow these will result in desk rejection regardless of the quality of your science. Read the Instructions for Authors carefully before finalizing your submission.

5. Being vague in the Limitations section. Limitations are not a sign of weakness; they reflect methodological self-awareness. Reviewers expect you to identify them. Being vague ("this was a single-center study") is far less useful than being specific about the direction and likely magnitude of any resulting bias.

Conclusion

Writing a manuscript is one of the most intellectually demanding tasks in research, and one of the most rewarding. It requires synthesizing data, constructing a coherent narrative, anticipating the reader's questions, and communicating scientific ideas with precision and clarity. Like any skill, it improves with practice.

Follow the IMRaD structure, build your tables and figures before you start writing, collaborate closely with your mentor, and do not let the perfect be the enemy of the good. The best manuscript is a submitted one.

Continue Learning

Explore other key topics in the Research Curriculum: