1. Open the protocol and the records of what actually happened
The protocol is a starting point for the draft; study records and code establish what you actually did. Do not describe a later change as an original plan. Check what changed, why and when, and keep the numerical study findings in Results.
ICMJE calls for enough methodological detail for someone with access to the data to reproduce the results. The sequence below is an editorial way to build that account. Randomized trials and systematic reviews need their own reporting guidance; case reports have a separate writing workflow.[1][3]
- Collect the inputs: Use the protocol and amendments, eligibility rules, dated data coverage, variable dictionary, executed code and outputs, software versions and the ethics review decision.
- Keep unknowns visible: Record who can resolve each question and where to look. Do not fill a gap with a plausible dose, manufacturer or exclusion rule.
2. Build a subheading outline and connect it with STROBE
Follow required journal subheadings first. EQUATOR explains that reporting guidelines recommend information to include without prescribing its order. This map offers one way to organize an observational-study draft.[2][3]
On small screens, scroll the table horizontally to compare every column.
| Subheading | What to organize here | Detailed guidance |
|---|---|---|
| Study design and setting | Actual design, setting and distinct recruitment, observation, follow-up and data-collection periods | STROBE 4–5 Start with design → |
| Participants | Eligibility, exclusions, and the follow-up, control selection or matching relevant to the design | STROBE 6 Participant guidance → |
| Variables and definitions | Outcomes, exposures, confounders and classification rules | STROBE 7 Variable guidance → |
| Data sources and measurement | Each variable’s source and measurement, including comparability across groups | STROBE 8 Measurement guidance → |
| Bias | Steps actually taken to address sources of bias | STROBE 9 Bias guidance → |
| Study size | How the available records or a calculation determined the study’s size | STROBE 10 Study-size guidance → |
| Statistical analysis | Actual variable handling, adjustment, missing-data handling, loss to follow-up and sensitivity analyses | STROBE 11–12 Statistical guidance → |
| Ethics | The documented review decision and consent handling | ICMJE and journal requirements |
3. Replace vague wording with details from the records
These are hypothetical examples, not accounts of real patients or studies. Replace brackets only with verified information. Decide which details your study needs; the examples do not rank omissions by frequency.
Where drugs or equipment are relevant, check generic names, doses and units, administration routes and schedules, equipment, manufacturers and measurement procedures. Record the actual software version as well as its name. Read this detail alongside the STROBE explanations and ICMJE guidance.[1]
Dates: distinguish observation from data extraction
Wording to improve
Records from several years were reviewed.Revision or template
Eligible visits occurred between [start date] and [end date]. Follow-up ended on [date]. Records were extracted on [date].The eligibility period, follow-up endpoint and extraction date can differ. Identify the periods that actually apply to your study.
Measurement and treatment: describe what was done
Wording to improve
Patients received the standard treatment, and outcomes were assessed.Revision or template
[Generic drug] was given at [dose and unit] via [route], [frequency] for [duration]. The outcome was assessed using [definition and method] at [time point].Use this only for relevant interventions and measurements. If treatment varied, explain how it was defined and summarized in the data instead of implying a single uniform regimen.
Analysis: explain more than the program name
Wording to improve
Data were analyzed using statistical software.Revision or template
We used [method] to estimate [specified measure], adjusting for [variables actually included]. Missing data were handled by [actual approach]. Analyses used [software name, exact version].Check the purpose, analysis population, basis for variable selection and missing-data approach against the executed code and outputs. Naming a test alone is insufficient.
4. Choose planning and ethics statements from evidence
Distinguish prespecified and exploratory analyses by when the decisions were actually made. An additional subgroup analysis developed after seeing the results is not a prespecified analysis. If there is no plan or you cannot verify its timing, check with coauthors and do not conceal that uncertainty.
Do not insert a consent waiver merely because a study was retrospective. Check the review body, decision, reference, date and consent handling in the actual decision letter. If AI was used to conduct the study, also prepare the methodological description required by the journal.[1]
Conditional wording for analysis planning
Revision or template
The analysis of [question] was specified in [dated plan] before [relevant outcome analysis]. The additional analysis of [question] was developed after [stage] and is reported as exploratory.Use either sentence only when supported by the records. The example is not a reason to invent a prior hypothesis, a priori confounder selection or a sensitivity analysis.
5. Fill an English template with verified study information
This is a starting framework for an observational study, not an official submission form or a completed Methods section. Replace bracketed instructions with verified statements and delete inapplicable material. You do not have to make every sentence passive: make the actor and procedure clear.
Study design and setting
We conducted a [study design] using [data source] at [setting]. The
[recruitment/observation] period was [dates]; follow-up ended on [date].
Participants
Eligibility required [criteria]. Exclusions were [criteria and how applied].
[If matching was used, describe the variables, method and ratio actually used.]
Variables and measurement
The outcome was [definition], assessed using [rule/tool] at [time point].
The exposure was defined as [definition]. Measurements used [method,
equipment and manufacturer, where relevant].
[If relevant: generic drug, dose and unit, route, frequency and duration.]
Bias and study size
[Describe measures actually taken to address bias and remaining constraints.]
[Explain how the sample size was determined, including the source and period
if all eligible records were used. Do not invent a power calculation.]
Statistical analysis
We used [method] to [estimate/compare the specified outcome], with
[variables actually included and the basis for selecting them].
Missing data were handled using [actual method and analysis population].
Analyses were performed with [software name and exact version].
[Distinguish analyses specified before examining the relevant outcomes from
analyses developed afterwards; report changes and their timing honestly.]
Ethics
[Report the actual review body's decision, institution, reference and date.
Describe consent handling only as documented in that decision and records.]6. Compare with Results, then record checklist locations
Connect each subsection with the relevant data, code or decision record. Then enter final manuscript page and line numbers in the STROBE checklist. A location in a checklist is useful only if the required explanation is actually present there.
- Unit of analysis: Match the Methods unit with Results denominators; do not interchange patients, visits and measurements.
- Variables and time points: Check that the objective, Methods, tables and abstract refer to the same outcomes and follow-up windows.
- Unknowns and changes: Confirm that unresolved information, analyses not performed and departures from the plan are represented honestly.
Frequently asked questions
Must Methods subheadings follow STROBE’s order?
No. Follow the journal’s structure while making every required piece of information easy to locate. These subheadings are a drafting example, not an official STROBE form.
What if a retrospective study had no a priori sample-size calculation?
Describe how the study size was actually determined. If all eligible records in a fixed period were included, report the period, eligibility rules and reason for that choice. Do not add a calculation that was never performed; the adequacy and limitations of that size still need separate consideration.
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.