nejm-writing

Main-Text Writing (nejm-writing)

When to trigger

• The main text is bloated, or you are unsure of the article type.
• The Discussion over-states the implications or omits limitations.
• The Introduction starts with a literature review instead of the clinical question.
• Methods detail belongs in the protocol/supplement, not the body.

Article type first

• Original Article — definitive trials / major studies; full IMRAD; structured abstract.
• Brief Report — a smaller but important finding; substantially shorter; fewer items.
• Review — commissioned/vetted clinical synthesis (different structure).

Confirm length and reference caps against the current author guidelines. For an Original Article, design for a short main text (often ~2700 words) and a limited reference list (on the order of ~40) — NEJM is deliberately terse.

IMRAD, the NEJM way

Introduction (short)

Two to three paragraphs. State the clinical problem, the gap, and the specific question the study answers. End with the objective. No exhaustive background — move detail to references.

Methods

Design, setting, participants (eligibility), intervention/comparator, randomization and blinding, outcomes (primary pre-specified, then secondary), and the statistical analysis (ITT primary; multiplicity; pre-specified subgroups). Push full procedural detail to the protocol and supplementary appendix; the body states what a clinician needs to judge validity. Reference the reporting guideline (CONSORT/STROBE).

Results

Claim-first, numbers-led. Open with enrollment and the analysis population (tie to the CONSORT flow diagram). Report the primary outcome with effect size + 95% CI, then key secondary outcomes, then safety/adverse events. Do not interpret here — that is the Discussion. Tables carry the detail; text states the headline.

Discussion

• Open with what the study found, in one or two plain sentences.
• Place it among prior evidence — without re-reviewing the field.
• State clinical implications soberly: what should change, for whom, and what should not.
• A dedicated limitations paragraph is expected (generalizability, open-label bias, follow-up duration, missing data, power).
• Do not over-state: avoid causal language for observational data; do not extrapolate beyond the population studied.

NEJM house style (terse and plain)

• Short sentences; plain words; minimal hedging stacks.
• Define each abbreviation once; avoid acronym soup.
• Past tense for what was done and found; present tense for established facts.
• No "novel", "robust", "interestingly" as filler; let the numbers carry the claim.
• Active voice where it reads naturally; first-person plural is acceptable.

Over-claiming watch (NEJM-specific)

The fastest way to lose a clinical reviewer is a Discussion that outruns the data. Match every implication sentence to the strength of the design: a single trial supports a conclusion in its population, not a universal recommendation. Surrogate outcomes do not license patient-outcome claims.

Output format

【Article type】 Original Article / Brief Report / Review
【Main-text length】 N words vs target (~2700 for Original Article) → ok / over
【Reference count】 N vs ~40 cap → ok / over
【IMRAD check】 intro=question? methods→protocol? results claim-first+CI? discussion sober?
【Limitations paragraph present】 yes/no
【Over-claiming flags】 [...] (causal language / extrapolation / surrogate→outcome)
【Next】 nejm-statistics

Anti-patterns

• Do not pad the Introduction into a mini-review — state the question and stop.
• Do not keep full procedural detail in the body when it belongs in the protocol/supplement.
• Do not interpret results inside the Results section.
• Do not omit the limitations paragraph or soften it into a throwaway sentence.
• Do not let the Discussion recommend practice changes the single study cannot support.