Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› lancet-study-design
lancet-study-design
GitHub用于锁定《柳叶刀》临床稿件的研究设计与强制前瞻性注册规范。涵盖试验注册时机、方案及统计分析计划、RCT设计严谨性（随机化、盲法等）及观察性研究选择，确保符合ICMJE标准。
Trigger Scenarios
规划或撰写临床试验/大型观察性研究时
缺失试验注册号或注册时机不明时
缺乏研究方案或预设统计分析计划时
主要终点未预设且面临审稿质疑时
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill lancet-study-design -g -y
SKILL.md
Frontmatter
{
"name": "lancet-study-design",
"description": "Use to lock study design and mandatory prospective registration for a Lancet clinical manuscript — trial registration before enrollment, protocol + statistical analysis plan, and design rigor (randomization, allocation concealment, blinding, pre-specified endpoints, power, ITT). Covers observational design choices too."
}
Study Design & Registration (lancet-study-design)
When to trigger
- Planning or writing up a clinical trial or a major observational study.
- The trial registration number is missing, or registration timing is unclear.
- There is no protocol or no pre-specified statistical analysis plan (SAP).
- A reviewer is likely to ask "was the primary endpoint pre-specified?" and there is no clear answer.
Mandatory: prospective trial registration
The Lancet follows ICMJE: clinical trials must be registered prospectively — before the first participant is enrolled — in a WHO-ICTRP primary registry or ClinicalTrials.gov.
- Registered before enrollment of the first participant (retrospective registration is a serious, often fatal, problem — flag it immediately and disclose honestly).
- Registration number reported in the abstract and Methods (e.g., ClinicalTrials.gov NCTxxxxxxxx; ISRCTN; or a WHO-ICTRP primary registry).
- Registered outcomes match the reported primary/secondary outcomes; explain and flag any change.
| Registry | Use / region |
|---|---|
| ClinicalTrials.gov | US and international |
| ISRCTN | UK / international |
| WHO ICTRP | global portal aggregating primary registries (e.g., EU-CTR, ANZCTR, CTRI, ChiCTR, PACTR) |
The Lancet often handles large international / multi-country trials — confirm registration in each relevant jurisdiction and report the lead registry.
Protocol + statistical analysis plan (SAP)
- A protocol exists and (typically) is submitted with the manuscript; published-protocol citation if available.
- A pre-specified SAP defines the primary analysis, populations (ITT / per-protocol), handling of missing data, and pre-specified subgroups before unblinding.
- Any deviation from protocol/SAP is reported transparently with dates and rationale.
Trial design rigor (RCT)
- Randomization method (sequence generation) described.
- Allocation concealment mechanism described (separate from blinding).
- Blinding of participants, clinicians, outcome assessors, and analysts — or justification if open-label.
- Pre-specified primary endpoint, clinically meaningful, with a clear hierarchy of secondary endpoints.
- Sample size / power justified for the clinically meaningful effect, with assumptions stated.
- Intention-to-treat (ITT) as the primary analysis population.
- Trial design named (parallel, cluster, crossover, factorial, adaptive, non-inferiority/equivalence — state the margin).
- Independent data and safety monitoring committee (DSMC) and any interim/stopping rules described.
Observational design choices
- Design named (cohort, case-control, cross-sectional) and justified.
- Exposure/outcome definitions, eligibility, and follow-up specified a priori.
- Confounding strategy (matching, adjustment, propensity methods) pre-specified; causal language calibrated to the design.
- Registration/protocol encouraged for major observational studies (e.g., a pre-registered analysis plan).
What Lancet editors check on design and registration
Registration timing is the most consequential design fact The Lancet checks: following ICMJE, it will decline to consider a trial not registered prospectively, and an outcome switched from the registered one is an integrity concern. Desk-reject or major-revision triggers: retrospective registration left undisclosed; a switched primary outcome; no protocol or pre-specified SAP; concealment conflated with blinding; per-protocol presented as primary for a superiority trial. Confirm any current protocol-submission requirement against the journal's author guidelines.
Worked micro-example (illustrative numbers — not real data)
A hypothetical double-blind, multi-country superiority RCT.
Design lock (illustrative):
Registry ISRCTN registered 2021-03 BEFORE first enrolment 2021-06 -> prospective, PASS
Registered primary outcome = reported primary outcome -> no switch, PASS
Randomisation 1:1 stratified by country; central web concealment; double-blind
Power: ARR 6 pp, alpha 0.05, power 90% -> n=2 040 (1 020/arm), +8% attrition
Primary population ITT; per-protocol pre-specified sensitivity; independent DSMC
The design passes: registration is prospective and dated, outcomes agree, concealment and blinding are separate, and the sample size is justified.
Reviewer-pushback patterns and the venue-specific fix
- "The trial was not registered prospectively." → Disclose exact registration and enrolment dates; if retrospective, flag it transparently — the editor may decline.
- "The primary outcome was switched." → Reconcile to the registered outcome or document the change with dates and rationale.
- "The study is underpowered / concealment is unclear." → Justify the sample size against the minimally important difference; describe sequence generation and concealment separately from blinding.
Output format
【Study type】 RCT (parallel/cluster/crossover/factorial/adaptive/NI) / observational (cohort/case-control/cross-sectional)
【Registration】 registry + number + prospective? (before first enrollment: yes/no) + in abstract? yes/no
【Protocol + SAP】 present? primary analysis & populations pre-specified? yes/no
【RCT rigor】 randomization / allocation concealment / blinding / primary endpoint / power / ITT → gaps
【Observational rigor】 design / confounding strategy / causal language calibrated → gaps
【Deviations from protocol/SAP】 [...] reported transparently? yes/no
【Next】 lancet-reporting
Anti-patterns
- Do not submit a trial that was registered after enrollment without flagging it — The Lancet/ICMJE may decline to consider it.
- Do not report a primary outcome that differs from the registered one without an explicit, dated explanation.
- Do not present a post-hoc analysis as if it were pre-specified.
- Do not rely on per-protocol as the primary analysis for a superiority trial — ITT leads.
Version History
- 1839142 Current 2026-07-05 14:02
