1. Separate publication order from drafting order
Introduction–Methods–Results–Discussion describes a finished IMRaD manuscript. The six steps here are a suggested route from organized records, not a drafting rule set by ICMJE or a reporting guideline. Kallestinova also suggests starting with Methods when getting started is difficult.
Before writing, check the research question, the actual state of the analysis, data-use permissions and ethics requirements. This workflow does not design a study or justify changing its question to fit favorable findings.[1][2]
2. Produce a usable output at each of six stages
Move between stages as material becomes ready. For missing information, record the source and person who can resolve it rather than hiding the gap in prose. These outputs provide practical completion checks.
On small screens, scroll the table horizontally to compare every column.
| Stage | Inputs and work | Output for the next stage |
|---|---|---|
| 1. Question, journal and reporting guideline | Check the actual question and data readiness; identify the article type, journal instructions and design-specific guideline. | A one-sentence objective, a record inventory, instructions and a checklist Observational reporting → |
| 2. Methods | Describe the protocol and actual conduct, analyses and ethics decisions, including changes. | A draft that explains participants, dates, variables and analyses Writing Methods → |
| 3. Figures, tables and Results | Compare primary and secondary findings, analysis populations, denominators and units; connect displays with text. | A Results draft with consistent numbers across text and displays Writing the Results → |
| 4. Introduction | Narrow the unresolved question from relevant evidence and connect it with the actual objective. | A visible background→gap→objective argument Writing the Introduction → |
| 5. Discussion | Compare the principal findings with existing evidence; address limitations, alternative explanations and implications. | An interpretation that stays within the design and data Discussion guide → |
| 6. Abstract, title and submission documents | Match abstract numbers and conclusions to the completed body, then finalize the title, declarations and cover letter. | A submission version consistent with the full paper Writing the Abstract → |
3. Choose starting material for the actual design
Use EQUATOR to find guidance for the actual design. A reporting guideline identifies information to report; it does not choose the study design or appraise study quality. The routes below are drafting suggestions for organizing that information. Do not use them to invent research procedures or analysis plans retrospectively.[3]
4. Observational studies: align selection and analysis units first
Gather eligibility records and variable definitions, then draft Methods alongside participant flow. For a cohort, separate recruitment, exposure, follow-up and data-collection periods. For a case-control study, check how cases and controls were selected; for a cross-sectional study, check selection and measurement timing. Use the STROBE version that fits the actual design.
Next prepare descriptive and main-result tables, comparing denominators, missing data and loss to follow-up across analyses. If you report unadjusted and adjusted estimates, readers should find the included confounders and selection rationale in Methods. Use these connections to refine the objective and limitations.[6]
- Take forward: Methods with eligibility and dates, stage-specific participant counts, and result tables connected to analysis denominators and variable definitions.
- Pause and resolve: Return to records and code if tables mix patients with visits or exclusions for missing data change denominators without explanation.
5. Randomized trials: compare plans and registration with actual conduct
Open the protocol, registration, statistical analysis plan and conduct records together. Draft allocation, intervention, comparator, blinding, outcomes and time points in Methods, distinguishing changes. Next connect allocation, intervention received, follow-up and analysis counts with the flow diagram and result tables.
For each primary and secondary outcome, check the analysis population, numbers with data at that time point, effect estimate and uncertainty. Check harms and unintended events alongside benefits. Distinguish prespecified from post hoc additional analyses, and explain discrepancies between plans and results. Use CONSORT 2025 and applicable extensions to check reporting.[7]
- Take forward: Methods checked against plans and amendments, a participant flow diagram, and tables covering primary and secondary outcomes and harms.
- Pause and resolve: Check records and analyses first if group counts do not reconcile across allocation, follow-up and analysis, or the original primary outcome disappears.
6. Systematic reviews: connect search and selection records with results
Compare the protocol, if one exists, with actual conduct. Gather search strategies, dates and sources, deduplication records, and stage-specific selection and exclusion decisions. Describe actual methods and prepare the PRISMA flow diagram and study-characteristics table. Do not claim independent duplicate screening if it was not performed.
Check extracted data and risk-of-bias assessments before organizing the actual synthesis. If no meta-analysis was performed, explain the summary or synthesis method used and why. State if registration or a protocol was absent, and document amendments and reasons. Use the records and results to refine the Introduction, Discussion and Abstract; do not describe a narrative review as using systematic procedures it did not follow.[5]
- Take forward: Search and selection methods, a flow diagram, study characteristics and risk-of-bias assessments, and results of the synthesis actually performed.
- Pause and resolve: Return to selection records if records, reports and studies are confused or exclusion reasons do not reconcile with flow counts.
7. Case reports: choose the message, then write the timeline
Choose the message and a working title, then build a timeline from clinical records. Connect presenting concerns, diagnostic assessment, interventions, follow-up and outcomes chronologically before writing the narrative. Use CARE and the journal’s case-report structure rather than forcing an original-research Methods and Results layout.
Explain the case’s relevance in the Introduction and Discussion by comparing it with literature, and finish the Abstract last. Start publication-consent, de-identification and patient-perspective checks early. Do not generalize a single course into an overall treatment effect or invent causes, doses or outcomes absent from the records.[4]
- Take forward: A narrative matching the timeline, a bounded message, and a verified account of publication consent and de-identification.
- Pause and resolve: Resolve inconsistent dates, doses or outcomes and outstanding publication consent first. De-identification does not substitute for consent.
8. Make a second pass for consistency between sections
Follow a one-sentence objective through the variables in Methods, the tables and denominators in Results, the interpretation in Discussion and the abstract conclusion. Return to a section when the population or time window changes unexpectedly. A working title can come early; align its final wording with the completed paper.
Following a drafting order does not establish complete reporting. Locate checklist items in the final manuscript, obtain coauthor checks and finish the target journal’s submission requirements.
Frequently asked questions
Must I always write Methods first?
No. Documented methods offer a useful starting point. Figures, tables or an objective outline may be ready first; adapt the route to the study type and the material available.
Should I leave the Introduction and title completely unwritten until the end?
You can make an early outline and working title. Recheck their scope once the population, analyses and findings are organized, and finish the abstract by comparing it with the body and tables.
Sources and how we checked them
Primary sources checked on November 2, 2026. The examples, workflow and completion checks are editorial suggestions, not an official research-design or reporting sequence.
- Kallestinova ED. How to Write Your First Research Paper. Yale J Biol Med. 2011;84:181–190
- ICMJE: manuscript preparation and guidance by section
- EQUATOR: finding and using reporting guidelines
- CARE: writing a case report
- PRISMA 2020: original checklist
- STROBE: original checklists by study design
- CONSORT 2025: randomized-trial checklist and flow diagram