Writing the Methods Section of Your Medical Manuscript: What Journals Now Expect in 2026

Ask a peer reviewer which section of a clinical manuscript they find most frustrating to evaluate, and the answer is almost always the methods. Not because authors are careless, but because the methods section has quietly accumulated more obligations over the past few years than any other part of the paper. Study registration numbers, participant eligibility criteria, blinding procedures, statistical analysis plans, software versions, data availability declarations, code repositories, and reporting guideline checklists have all either appeared as new requirements or tightened in language since 2023. What was once a relatively compact summary of what you did has become, at many journals, a detailed accountability document.

The shift is intentional. The reproducibility problems that have affected clinical medicine, oncology, and preclinical research over the past decade grew partly from vague methods reporting. When a reader cannot reconstruct what you did, they cannot verify your results, identify errors, or repeat your study. PLOS Biology made code availability mandatory for all submissions as of January 2026. CONSORT 2025 added new protocol transparency items and a dedicated open science section. The STROBE-Equity extension was published in JAMA Network Open in September 2025. The trajectory is clear: journals and funders are moving from aspirational transparency language to specific, checkable requirements.

Why the Methods Section Gets More Reviewer Attention Than the Results

Most authors assume reviewers focus on the results and discussion. In reality, statisticians, methodologists, and experienced clinical reviewers often form their verdict in the methods section. The results can only be judged in the context of how they were obtained. If the methods contain a fundamental flaw, no amount of precise results reporting salvages the paper.

JAMA assigns statistical reviewers to papers where the analysis is sufficiently complex, and those reviewers focus almost entirely on the methods section first. BMJ Open states explicitly that its bar is rigor and reproducibility rather than novelty, and that the fastest path through review is a study reported to the relevant standard with its limitations stated plainly. Most major clinical journals now run the manuscript against a reporting checklist before it reaches external reviewers, which means desk rejection for an incomplete methods section is a real possibility, not a theoretical one.

The other reason methods get scrutinized is that the field is actively trying to close loopholes that enabled questionable practices in the past. Outcome switching, undisclosed subgroup analyses, post-hoc selection of statistical tests, and selective exclusion of participants are all harder to conceal when the methods section is written with specificity. Reviewers know this, and they read accordingly.

Start Before the Study: Registration and Protocol Disclosure

The methods section of a randomized controlled trial or prospective observational study should open with the study registration. This is not a new requirement, but the level of detail now expected has grown. It is no longer enough to state the ClinicalTrials.gov number; many journals now ask for the URL, the date of first participant enrollment relative to the registration date, and a statement confirming that the primary outcome registered matches the primary outcome reported. That last requirement catches outcome switching and corresponds to ICMJE guidance that has been reinforced in the January 2026 update to its Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals.

If a protocol paper was published before the study began, cite it in the methods. CONSORT 2025 introduced a new item specifically addressing protocol transparency, and SPIRIT 2025 (the companion guideline published simultaneously in The Lancet, BMJ, PLOS Medicine, and Nature Medicine in April 2025) provides the current standard for what a clinical trial protocol should contain. If your study has a published protocol, readers can compare it to the methods as reported and verify that the primary analysis matches the pre-specified plan. If there were deviations, the methods section is where you name them.

Participants: Eligibility Criteria, Setting, and Who Was Actually Excluded

The participant description is where methods sections most often under-deliver. Authors tend to list inclusion criteria and then give a single paragraph summary of the population. What reviewers need is enough detail to judge whether the findings are likely to generalize and whether the enrolled population matches the one the study was designed for.

Inclusion criteria should be specific: age range, diagnosis (with code or criteria used for diagnosis), disease stage or severity, and relevant prior treatment or comorbidity restrictions. Exclusion criteria often tell you more about a study than inclusions do, because they reveal the implicit assumptions. If you excluded patients with eGFR below 30, with concurrent immunosuppressive therapy, or with certain imaging findings, state it plainly. A STROBE-compliant observational study is expected to give eligibility criteria and the sources and methods of selection for controls, a level of detail that many authors skip.

Setting matters for generalizability. A single tertiary academic center, a network of community hospitals, a multi-country consortium, and an administrative claims database have different implications for what the results mean. Be specific about institution type, geographic region, and time frame of data collection. For retrospective database studies, name the database and its version or coverage dates. For studies using platforms like TriNetX or similar real-world data systems, describe the specific network used, the query performed, and any preprocessing steps, since a growing number of journals now ask for this level of detail following methodological concerns raised in 2025 and 2026.

The SAGER guidelines, which have been adopted by Nature Portfolio and the Lancet family among others, require that sex and gender be addressed in the methods. If your study enrolled both male and female participants, state how sex was determined (self-report, medical records, biological measurement) and specify whether your analysis plan included sex-stratified or sex-disaggregated subanalyses. If you did not include such analyses, say so and briefly explain why.

Interventions and Exposures: Enough Detail to Replicate

The intervention description for a randomized trial should be written so that a clinical team could reproduce it. For drug interventions: generic name, dose, route, frequency, duration, and any dose-adjustment rules. For surgical or procedural interventions: technique, equipment specifications, operator training requirements, and any control or sham procedure used. For behavioral or educational interventions: content, delivery mode, number of sessions, session duration, and who delivered the intervention.

For observational studies, the exposure definition needs equivalent precision. If you are studying the effect of metformin use on a clinical outcome, define how metformin use was classified (prescription dispensing, medication record, self-report), what constituted current versus past versus never use, and how you handled dose or duration variability. Vague exposure definitions are among the most common reasons a statistical reviewer flags an observational study for revision.

CONSORT 2025 recommends authors link to or describe the use of resources that facilitate replication, including validated intervention manuals, published protocols, or repositories where supplementary procedural materials can be deposited. The TIDieR (Template for Intervention Description and Replication) framework remains a useful scaffold for writing intervention descriptions, even when TIDieR is not formally required by your target journal.

Outcomes: Primary, Secondary, and What You Measured With

The outcome section of a well-written methods section states the primary outcome first and includes the name of the outcome, the measurement tool or instrument used, the timing of assessment, and who performed the assessment. For trials, the primary outcome should match the registered outcome exactly. If it does not, that is a deviation that must be explained.

Secondary outcomes should be listed explicitly. A paper that lists a primary outcome in the methods but reports a dozen secondary outcomes in the results without naming them in advance raises outcome switching concerns. Many journals now require secondary outcomes to be pre-specified, and authors who cannot point to a registration or protocol entry for each reported outcome may face revision requests or, in some cases, rejection on integrity grounds.

Validated instruments deserve their citations. If you used the PHQ-9 to screen for depression, the PROMIS pain interference scale, a Likert satisfaction scale, or a composite clinical endpoint, cite the validation study and note the version. If an instrument was adapted for your population or translated, describe that process. Reviewers working in clinical outcomes research will notice when a widely-used instrument is deployed without citation, and it reads as superficial rather than a deliberate omission.

Statistical Methods: The SAMPL Requirements and What Reviewers Actually Check

The statistical methods subsection is where many medical manuscripts fail at the detail level. The SAMPL (Statistical Analyses and Methods in the Published Literature) guidelines, first published in 2015 and widely cited by journals and statistical reviewers, provide a structured framework for what should appear in this section. A 2025 audit of 100 clinical medicine papers found that 65 percent lacked adequate descriptions of their statistical methods and 64 percent failed to report effect sizes, which suggests SAMPL adoption has been slower than the guidelines would suggest.

At a minimum, the statistical methods section should specify: the primary analysis approach and any planned sensitivity analyses, how missing data were handled (with justification for the approach chosen), how the sample size was determined and what assumptions were used, the software and version used for analysis, and any thresholds used for statistical significance. This last point deserves attention because JAMA and several other journals now ask authors to report confidence intervals alongside or instead of p-values alone, and some journals have shifted toward not using significance thresholds at all.

Propensity score methods, survival analyses, hierarchical models, and machine learning classifications each carry their own reporting expectations. If you used propensity score matching, specify the variables used to build the score, the matching algorithm, the caliper width, and how balance was assessed after matching. If you used a Cox proportional hazards model, state whether the proportional hazards assumption was tested and how. These are not optional disclosures for sophisticated reviewers. They are the specific questions a statistical reviewer will ask if you do not answer them in advance.

Code, Software, and the New Sharing Expectations

PLOS Biology made code availability mandatory for all submissions as of January 2026, requiring that analysis code be deposited in an appropriate repository and cited in the methods. Other journals have been moving in the same direction, and in early 2026 a PLOS report framed this as part of a broader transition to recognizing data, code, and protocols as first-class research outputs alongside the paper itself. For authors working in fields where R, Python, or other scripted analyses are common, this is a direct mandate: deposit the code before submission.

Even where code deposit is not yet mandatory, naming software precisely is now expected across virtually all medical journals. Accepted repositories include GitHub (often pointed to a persistent archived version via Zenodo), OSF (Open Science Framework), Dryad, Figshare, and Zenodo directly. Some institutions have their own approved repositories. A link that says "code available on reasonable request" is no longer sufficient at journals with explicit open science policies, and at several journals it has never been acceptable.

For biological wet lab research, the Resource Identification Initiative has introduced Research Resource Identifiers (RRIDs) as a way to make antibodies, cell lines, model organisms, and software tools traceable and unambiguous. Many journals in cell biology, neuroscience, and immunology now ask or require RRIDs for key resources. The RRID portal allows you to look up or register a resource and obtain a citable identifier that travels with the paper permanently.

If your study uses a commercial analytical platform, name the platform, the vendor, and if possible the version or the date of analysis. This matters particularly for AI-assisted or automated tools, where the underlying algorithm may change between versions. A study reported as using a specific EHR analytics platform without naming the network configuration, query parameters, or software version leaves reproducibility in question from the outset.

Reporting Guideline Checklists and the Methods Section They Require

One practical way to structure a methods section is to use the relevant reporting guideline as a scaffold. CONSORT 2025 covers randomized trials (and its companion SPIRIT 2025 covers protocols). STROBE covers cohort, case-control, and cross-sectional studies, with STROBE-Equity extending the framework to health equity reporting. PRISMA 2020 covers systematic reviews, with the PRISMA-S extension specifying how literature searches should be reported. ARRIVE 2.0 covers animal research. TRIPOD+AI covers clinical prediction models. All of these are available through the EQUATOR Network, which maintains the most complete catalogue of health research reporting standards.

The standard practice is to complete the relevant checklist, annotate each item with the page or line number where it appears in your manuscript, and upload it as a supplementary file at submission. Some journals also ask for this in the main text. What the checklist forces is a line-by-line review of whether you have actually reported what the standard requires. Authors who do this routinely report that it catches omissions they did not notice during writing, particularly in the participant description, the blinding details, and the statistical analysis section.

An underappreciated aspect of this process is that incomplete checklist items often point to genuine methodological gaps rather than just reporting omissions. If you cannot find where your manuscript describes how blinding was maintained during outcome assessment, it may be because blinding was not maintained. The guideline will not let you ignore that.

Ethics, Consent, and the Statements That Must Be in the Manuscript

Ethics approval statements have been required for decades, but the detail journals expect has become more specific. The committee name, the reference number, and whether the study was classified as exempt or minimal risk all belong in the methods section. For studies involving patient data that were conducted under a waiver of informed consent, the waiver itself should be mentioned along with the reason it was granted.

The ICMJE 2026 recommendations reinforce that the methods section should state whether participants provided written or oral informed consent, and if the study involved non-consented secondary use of existing clinical records, the legal or regulatory basis for that use should be named. In many European countries, this involves citing the relevant GDPR article. In the United States, studies relying on HIPAA-covered datasets require a statement of how protected health information was handled.

Patient and public involvement (PPI) is increasingly expected in clinical research. CONSORT 2025 added an item specifically addressing how patients or members of the public were involved in the design, conduct, or reporting of the trial. SPIRIT 2025 likewise requires protocols to describe PPI plans. If your study did not include PPI, most journals currently accept a statement to that effect, but the expectation is moving toward explanation rather than silence.

Common Failures That Slow or Block Peer Review

Certain methods section problems come up so consistently that experienced editors can identify them on a first read. Missing sample size justification is the most frequent: many authors note their sample size but do not explain the power calculation, the assumed effect size, the expected event rate, or the significance threshold used. Without those components, the sample size number is uninterpretable.

Vague analytic descriptions are the second most common problem. Writing "logistic regression was used" without specifying what covariates were included, how they were selected, and whether any were treated as continuous or categorical gives a reviewer no basis for judging the analysis. Similarly, writing "appropriate statistical tests were applied" or "standard statistical methods were used" appears throughout manuscripts submitted to major journals, but it does not actually describe anything. Replace those phrases with specific test names, software packages, and the reasoning behind analytical choices.

Missing blinding information is common in randomized trial methods sections. Who was blinded: participants, care providers, outcome assessors, or the data analysts? How was blinding maintained? Was it ever broken, and if so, under what circumstances and how many times? These are specific CONSORT items and specific reviewer questions. A trial labeled double-blind that does not explain the blinding procedure will be sent back for revision.

Undescribed missing data is a growing problem as journals expect more sophisticated handling. Simply stating "patients with missing data were excluded" is increasingly insufficient, particularly for longer follow-up studies or registry-based analyses where missing data rates may be high or non-random. If you used complete-case analysis, state the proportion of excluded records and any comparison of included versus excluded cases. If you used imputation, describe the imputation model and the number of imputed datasets.

The Data Availability Statement and What It Has to Match

Most medical journals now require a data availability statement, and the phrase "available on reasonable request" is no longer accepted at a growing number of publishers, including Nature Portfolio titles and an increasing portion of Elsevier and Wiley journals. The methods section and the data availability statement must be consistent with each other. If the methods say you used a de-identified administrative dataset, the data availability statement cannot claim the data cannot be shared for privacy reasons without explaining the specific obstacle.

For clinical trial data, several journals now link directly to the AllTrials initiative or require that participants be given the opportunity to request their own data. The BMJ requires clinical trial manuscripts to include a statement on patient data availability. NEJM has run a data sharing pilot for certain large clinical trials.

If your data are genuinely not shareable due to regulatory or ethical constraints, that is a legitimate position, but it must be stated specifically. Which constraint? Which regulatory body? Is an application process available to qualified researchers? A data availability statement that invites requests while specifying the conditions and the contact mechanism is far more useful to readers than a blanket declaration that data are unavailable.

A Practical Pre-Submission Methods Checklist

Before sending your manuscript out, read the methods section with the reviewer's eye rather than the author's. The author knows what was done. The reviewer only knows what is written. The gap between the two is usually where the revision requests originate.

A practical check is to cover up your methods section and try to reconstruct the study design from memory as if you were an independent investigator. Where you could not do so from the text alone, add detail. Pay particular attention to the moments where you were tempted to write a general phrase instead of a specific one: these are the places where you know what you did but have not yet translated it into words that transfer the knowledge.

One thing this list makes clear is that the methods section of a 2026 clinical manuscript involves coordination across the research team and across several external resources: the trial registry, the protocol publication, the analysis code repository, the reporting guideline checklist, and the ethics documentation. None of these should be handled in the week before submission. The methods section is not a retrospective summary of what happened. At its best, it is the written form of a study that was designed for transparency from the beginning.

Further Reading