← AcademiaOne PhD Hub · 返回首页
AcademiaOneEDUCATION
PhD Research Proposal Guide · UK · Medicine & Health 英国医科研究计划书指南
AcademiaOneEDUCATION
Your PhD application, from idea to offer你的博士申请,从构想到录取
DOCTORAL APPLICATIONS · UNITED KINGDOM · MEDICINE & HEALTH   英国医科博士申请

PhD Research
Proposal Guide · Medicine & Health

英国医科博士研究计划书写作指南

A section-by-section model of a PhD research proposal in medicine and health for the UK, covering public health, epidemiology, nursing, clinical research and biomedical science. Each part shows the structure, a fully worked example from public health, and the difference between a weak and a stronger version, with the current requirements of UK faculties taken from their official pages.

逐部分拆解一份英国医科博士研究计划书,涵盖公共卫生、流行病学、护理、临床研究与生物医学。每一部分都包含结构说明、一个完整的公共卫生范例,以及“较弱”与“更好”写法的对比,并附英国各学院官网上的最新要求。

Start here 开始之前

What UK universities ask for英国各校要求

There is no single UK format. The school or department sets the length and the form, and funded routes set their own, so check the programme page before you cut your proposal to size. Most UK proposals fall between 1,000 and 3,000 words, and in most departments you contact or name a supervisor before you apply.

英国没有统一的计划书格式。篇幅与格式由学院或学系规定,带资助的申请通道还有各自的要求,因此在压缩篇幅之前,请先查看目标项目的官方页面。多数英国计划书在 1,000 至 3,000 词之间;多数院系要求你在申请前联系或写明导师。

UK medical and health faculties work in two ways. Laboratory and clinical-science routes often ask you to agree a supervisor and project first, with a short statement instead of your own proposal (Oxford RDM: a personal statement of up to 500 words; Imperial NHLI: the full proposal is written with the supervisor). Public health, social work and medical education usually ask for your own proposal (UCL medical education about 2,000 words; Manchester Social Work 2,500 words).

英国医学与健康学院有两种做法。实验室与临床科学路径通常要求先确定导师与项目,只交简短陈述而不是自己的计划书(牛津 RDM:不超过 500 词的个人陈述;帝国理工 NHLI:完整计划书与导师共同撰写)。公共卫生、社会工作与医学教育通常要求你自己的计划书(UCL 医学教育约 2,000 词;曼彻斯特社会工作 2,500 词)。

Medicine & Health: what each programme asks for医科:各项目要求

UNIVERSITY 大学WHAT THE PROGRAMME ASKS FOR 项目要求SUPERVISOR CONTACT 联系导师OFFICIAL PAGE 官方页面
Imperial
帝国理工学院
QS 2027: =2
NHLI: agree a supervisor and project outline first (or apply to a studentship); full proposal written with supervisor
Proposed supervisor cannot be a referee; apply ≥2-3 months before start; no fixed deadline
References: 2
Usually agree a supervisor first (NHLI, Earth Science); Computing: list 3-4 potential supervisorsimperial.ac.uk
Oxford
牛津大学
QS 2027: 4
DPhil Medical Sciences (RDM): no proposal; personal statement ≤500 words; agree a willing supervisor (up to 3) before applying
≥2 refs before assessment, 3rd completes file; page shown is for Oct 2025 entry; shortlisted candidates interviewed
References: 3
Varies by course: Archaeology encourages contact; Medical Sciences requires a willing supervisor firstrdm.ox.ac.uk
Cambridge
剑桥大学
QS 2027: 6
Public Health & Primary Care PhD: statement of interest ≤2,500 characters + CV; named supervisor needed; proposal only for Gates applicants
Referees academic or professional; funding deadline 8 Dec 2026; no AI in statements/CVs
References: 2
Name proposed supervisor(s); Public Health rarely accepts without onepostgraduate.study.cam.ac.uk
UCL
伦敦大学学院
QS 2027: =8
Medical education: about 2,000 words; Medicine: supervisor arranged first
At least one academic reference
References: 2
Often required first (e.g. Risk & Disaster Reduction, Experimental Psychology)ucl.ac.uk
Edinburgh
爱丁堡大学
QS 2027: 35
Identify a supervisor and project before applying (Psychiatry)
Formal interview
References: 2
Name a suitable supervisor for most programmesstudy.ed.ac.uk
KCL
伦敦国王学院
QS 2027: 37
IoPPN: personal statement + letter of support from supervisors
References: 2
Optional; Education centre director matches you if none namedkcl.ac.uk
Manchester
曼彻斯特大学
QS 2027: =40
Social Work: proposal 2,500 words + personal statement ~1,000
References: 2
Name a proposed supervisor; check fit and funding with thembmh.manchester.ac.uk
Bristol
布里斯托大学
QS 2027: 57
Dental School: email a supervisor with CV and project first; typed proposal
References: 2
Advised, not required (Education)bristol.ac.uk
Warwick
华威大学
QS 2027: =68
Medicine projects: one-page proposed methodology + CV
References: 2
Most departments advise contact; CS requires a supervisor in placewarwick.ac.uk
Birmingham
伯明翰大学
QS 2027: =68
Medical & Dental Sciences: ~2-page double-spaced proposal, ~5,000-character statement, 2-page CV; contact a supervisor first
Rolling; shortlisted applicants interviewed; IELTS 6.5, 6.0 each
References: 2
Identify and contact a supervisor via staff profilesbirmingham.ac.uk
Leeds
利兹大学
QS 2027: =77
Faculty of Medicine & Health: outline proposal or area of interest + CV; nominate and contact at least one supervisor
Place 'unlikely' without a matching supervisor; references count not stated (Leeds-wide: ≥2)
Must identify available supervisorsmedicinehealth.leeds.ac.uk
Glasgow
格拉斯哥大学
QS 2027: 80
Public Health (MVLS): discuss proposal with an MVLS supervisor before applying; CV + MVLS cover letter
IELTS 6.5, no sub-test <6.0; no deadline listed
References: 2
Contact a supervisor to discuss the proposal before applyinggla.ac.uk
Sheffield
谢菲尔德大学
QS 2027: =82
School of Medicine and Population Health: contacting potential supervisors before applying is advantageous (PhD Enquiry form or supervisor database); no proposal length on page
Central rule: research interests in supporting statement; Population Health IELTS 7.0 (6.5 each)
References: 2 (central rule)
Contact the school before applyingsheffield.ac.uk
Nottingham
诺丁汉大学
QS 2027: =97
Health sciences: base the proposal on existing research; contact supervisors first
References: 2
Some schools require contact firstnottingham.ac.uk
Queen Mary University of London
伦敦玛丽女王大学
QS 2027: 103
Institute of Dentistry: short 1,000-word proposal + CV to supervisor first; full proposal 2,000 to 3,000 words + personal statement; apply only after supervisor's positive feedback
Funded places usually close late January; possible interview; IELTS 6.5 overall
References: 2
Encouraged: email a potential supervisor before applying; Business requires supervisor agreement firstqmul.ac.uk
University of Southampton
南安普顿大学
QS 2027: =111
Medicine: own proposal or short statement for advertised projects + CV; email potential supervisors well before deadline
Project deadlines e.g. 4 Dec 2026, 10 Jan 2027; apply 6 months ahead if visa/ATAS; IELTS 6.5, 6.0 each
References: 2
Yes: speak to the supervisor before applying and name themsouthampton.ac.uk
University of Bath
巴斯大学
QS 2027: 125
Health: own-project proposal 1,000 words + personal statement; must name 1 to 2 lead supervisors
Intl deadline 3 months before start; IELTS 7.0, 6.5 each
References: 2
Yes for own projects: must name 1 to 2 lead supervisors, contact strongly recommended; not needed for defined projectsbath.ac.uk
University of Exeter
埃克塞特大学
QS 2027: 136
Medicine/Health: write a proposal and approach supervisors informally before applying
Self-funded 30 June intl for Sept; IELTS 6.5, 6.0 each (MD 7.0)
Strongly recommended to approach supervisors with proposal; Business and Law need a willing-supervisor letterexeter.ac.uk
University of Liverpool
利物浦大学
QS 2027: 139
Public Health: proposal + personal statement + CV + 2 proposed supervisors; contact supervisors before applying recommended
Year round; needs master's; IELTS 7.0, 6.0 each
References: 2
Recommended, not compulsory; form asks for 2 proposed supervisors (Management: optional)liverpool.ac.uk
University of York
约克大学
QS 2027: =158
Health Sciences: proposal ~500 to 1,000 words, up to 10 references; contact with staff encouraged (Hull York Medical School 500 to 1,000)
IELTS 7.0, 6.5 each
References: 2
Varies: Psychology requires contact first; others encourage or do not specifyyork.ac.uk
Lancaster University
兰卡斯特大学
QS 2027: 164
Health Research PhD: viable research proposal; degree must include research methods module and dissertation
IELTS 6.5, 6.0 each; no deadline listed
Not required: name preferred staff in application; contact optionallancaster.ac.uk
Queen's University Belfast
贝尔法斯特女王大学
QS 2027: =174
Medicine, Dentistry and Biomedical Sciences: own project needs CV max 2 pages + motivation statement (interests, funding, start date); may be asked for a proposal outline; name one supervisor and contact them first
IELTS 6.5, no component below 5.5; studentships have own deadlines (sample page shows 2025 round, 1 Oct start)
References: Not stated (central rule 2)
Optional: name a supervisor or Queen's matches you; many Schools suggest contact first (SSESW requires feedback first)qub.ac.uk
Cardiff University
卡迪夫大学
QS 2027: =179
Medicine: personal statement or covering letter on how you would contribute; contact potential supervisors for unfunded projects
Apply any time; some interviewed; IELTS 6.5, 5.5 each
Self-funded projects: find and discuss with a supervisor; Business requires named supervisor contacted in advancecardiff.ac.uk
University of Reading
雷丁大学
QS 2027: =196
Pharmacy (no medical school): proposal, read college guidance; contacting supervisors recommended
Start any time, mostly Sept; IELTS 6.5 (central)
References: 2
Make an initial enquiry; contacting a supervisor recommended, not essential (IoE: name 1 to 2)reading.ac.uk

Requirements are quoted from the official page linked in each row, in English as published. The supervisor column is the university-wide rule. 表中要求按各行所附官方页面的英文原文摘录;“联系导师”一栏为全校通用规则。

Golden rule 核心原则

The programme page always beats this guide. Use this guide to decide what to write; use the programme page to decide how long it is and what form it takes.

项目官网的要求永远优先于本指南。用本指南决定“写什么”,用项目页面决定“写多长、用什么格式”。

Funding dates come first 资助日期更早

Funding deadlines are often earlier than the programme deadline, so work backwards from the funding date. Examples for 2027 entry: Cambridge’s funding deadline is 8 December 2026 and LSE’s main funding deadline is 13 January 2027; UKRI studentships mostly close between November and February for October entry.

资助截止日期通常早于项目申请截止日期,请以资助日期倒推时间表。以 2027 年入学为例:剑桥资助截止日期为 2026 年 12 月 8 日,LSE 主要资助截止日期为 2027 年 1 月 13 日;UKRI 资助的博士名额多在 11 月至次年 2 月截止(10 月入学)。

A named supervisor 写明导师

Cambridge Public Health and Primary Care needs a named supervisor, Imperial NHLI asks you to agree a supervisor and project outline first (and the proposed supervisor cannot be a referee), and KCL IoPPN asks for a letter of support from your supervisors.

剑桥公共卫生与初级保健需要写明导师;帝国理工 NHLI 要求先与导师商定项目大纲(拟定导师不能担任推荐人);KCL IoPPN 要求导师出具支持信。

Table built from 24 official programme pages (each checked on the programme page itself; data checked 2026-11-18). Rules change each cycle; the programme page always wins. 表格依据 24 个官方项目页面整理(均在项目页面本身核对;数据核对日期 2026-11-18)。规则每年可能调整,以项目页面为准。

Start here 开始之前

Two routes into a health PhD两种申请路径

Health PhDs reach students by two routes. In the laboratory sciences the project usually exists before the student: a supervisor has funding, a question and equipment, and advertises the post. In public health, nursing and health services research you more often bring the question yourself and write a proposal. Decide which route each programme on your list uses before you write anything.

医科博士有两种申请路径。在实验室学科中,通常先有项目后有学生:导师已有经费、研究问题和设备,然后公开招人。在公共卫生、护理与卫生服务研究领域,更常见的是你自带研究问题并撰写研究计划。动笔之前,先弄清申请名单上每个项目走的是哪条路径。

ROUTE 路径WHAT YOU WRITE 你要写什么WHERE YOU MEET IT 常见于
A · Advertised project
申请已发布的项目
A fit statement or personal statement (often 500-1,000 words, or less) showing that your skills match the project, plus a CV. You do not design the project; you show you can do it and have ideas for it.
匹配陈述或个人陈述(常为 500-1,000 词,甚至更短),证明你的技能与项目相符,并附简历。你不需要设计项目,而是证明你能完成它,并对它有自己的想法。
Biomedical and laboratory science; Oxford RDM (personal statement up to 500 words), Imperial NHLI studentships, Warwick listed projects, Southampton advertised projects.
生物医学与实验室学科;牛津 RDM(不超过 500 词的个人陈述)、帝国理工 NHLI 奖学金项目、华威列名项目、南安普顿已发布项目。
B · Own proposal
自拟研究计划
A research proposal of about 2 pages to 3,000 words: problem, evidence gap, aims, design, data, ethics, feasibility and fit. Usually developed with a supervisor before you submit.
约 2 页至 3,000 词的研究计划:问题、证据空白、目标、设计、数据、伦理、可行性与匹配度。通常在提交前与导师共同完善。
Public health, epidemiology, nursing, health services research and global health; UCL medical education, Manchester Social Work, Birmingham Medical and Dental Sciences, Glasgow Public Health, Leeds Medicine and Health, Liverpool Public Health.
公共卫生、流行病学、护理、卫生服务研究与全球健康;UCL 医学教育、曼彻斯特社会工作、伯明翰医学与牙科学院、格拉斯哥公共卫生、利兹医学与健康学部、利物浦公共卫生。

Clinical and non-clinical applicants临床背景与非临床背景申请者

Clinical 临床Clinical applicants

Use your clinical experience to show why the question matters and how you will reach patients or records. If the project involves patient contact in the host country, ask whether you need local professional registration; without it, base the project on data or work alongside clinical staff.

用临床经历说明问题为何重要、你将如何接触患者或病历。如果项目在目标国家涉及接触患者,请询问是否需要当地执业注册;若无法取得,就以数据为基础设计项目,或与临床人员合作完成。

Non-clinical 非临床Non-clinical applicants

Show the methods you already use (statistics, laboratory or social science) and name the clinical collaborator or service that gives you access to patients or data.

展示你已掌握的方法(统计、实验室或社会科学),并写明为你提供患者或数据渠道的临床合作者或医疗服务机构。

Not covered: clinical doctorates 本指南不涉及:临床博士

Clinical doctorates such as the MD(Res) for medically qualified doctors and the DClinPsy professional training in clinical psychology are separate routes with their own entry rules and application systems. This guide covers the PhD (and research master’s) only.

以医学博士学位(MD(Res),面向已有医学资格的医生)和临床心理学专业博士(DClinPsy)为代表的临床博士是独立的申请路径,有各自的入学要求与申请系统。本指南只讲 PhD(及研究型硕士)。

Before you apply 申请之前

Supervisor first, then the application先找导师,再正式申请

Many UK medical and health faculties expect a supervisor to agree before the formal application, and some will not process an application without one. The supervisor then helps you shape the proposal, so the first contact matters as much as the proposal itself.

许多英国医学与健康学院要求在正式申请前取得导师同意,部分学院没有导师就不受理申请。之后导师会帮你完善研究计划,所以第一次联系与研究计划本身同样重要。

STEP 步骤WHAT YOU DO 你要做什么EXAMPLES 示例
1 · Find
寻找导师
Choose two or three supervisors whose recent papers use your kind of data or design.
选两三位近期论文使用与你相同类型数据或设计的导师。
Leeds: a place is unlikely without a matching supervisor.
利兹:没有匹配的导师,基本不会录取。
2 · Email
发邮件
A short email with your question, data and a one-page summary (sample below).
一封简短邮件,写明研究问题、数据,并附一页摘要(见下方范例)。
Birmingham MDS and Nottingham health sciences: contact a supervisor first.
伯明翰医学与牙科学院、诺丁汉健康科学:先联系导师。
3 · Talk
面谈
An online meeting or interview; agree the question and who holds the data.
线上会面或面试;商定研究问题以及数据由谁持有。
Glasgow Public Health: discuss the proposal with a supervisor before applying.
格拉斯哥公共卫生:申请前先与导师讨论研究计划。
4 · Agreement or support
导师同意或支持
The supervisor, and sometimes the school, sign a form or write a letter of support.
导师(有时还有学院)签署表格或写支持信。
Imperial NHLI: agree a supervisor and project outline. KCL IoPPN: a letter of support from supervisors.
帝国理工 NHLI:与导师商定项目大纲。KCL IoPPN:导师支持信。
5 · Full application
正式申请
Submit the proposal you finalised with the supervisor, with CV and referees.
提交与导师共同定稿的研究计划,以及简历与推荐人信息。
Queen Mary Dentistry: apply only after the supervisor’s positive feedback, with a full proposal of 2,000-3,000 words.
玛丽女王牙科:获得导师积极反馈后再申请,提交 2,000-3,000 词的完整计划书。
Worked example · First email to a potential supervisor (under 150 words) 范例 · 第一封联系导师的邮件(150 词以内)

Subject: Prospective PhD applicant: early-life PM2.5 exposure and childhood asthma

Dear Professor A,

1I read your recent paper on residential PM2.5 estimates and children’s lung function (Authors, Year), in particular the finding that exposure in the first year mattered more than later exposure.

2I am planning a PhD on whether early-life PM2.5 exposure increases the risk of childhood asthma, using an existing birth cohort of about 5,000 children in eastern China, where I worked as a research assistant. 3The cohort team has agreed in principle to share pseudonymised data for this analysis.

4I hold an MPH from a UK university, with a dissertation using survival analysis in R. Would you consider supervising this project for the next intake? I attach my CV and a one-page summary.

Best wishes,
[Name]

Why it works 写法解析
1Their work, specifically 具体提到对方研究

One paper and one finding, so the email cannot be mass-sent.

一篇论文、一个发现,表明不是群发邮件。

2Question and data 问题与数据

One sentence each.

各用一句话。

3Data access 数据获取

Supervisors in health ask first about data; answer before they ask.

医科导师最先问的就是数据,提前回答。

4Readiness and a clear ask 准备度与明确请求

Methods you already use, then a direct question.

先写已掌握的方法,再直接提出请求。

Start here 开始之前

How long, and how the words are shared篇幅与字数分配

Write one master version of about 2,500 words, then cut it to size for each programme, from about 2 pages to 3,000 words. Health panels spend most of their time on methods, so methods get the most words, and ethics and data access get their own section. In the UK the department sets the limit, so check whether references and appendices count before you cut.

先写一份约 2,500 词的母版,再按各项目要求裁剪,篇幅从约 2 页到 3,000 词不等。医科评审把大部分时间花在研究方法上,所以方法部分字数最多;伦理与数据获取单独成节。在英国,字数上限由院系规定;裁剪之前请确认参考文献与附录是否计入字数。

200350450850300200150
0 Title & abstract1 Background2 Gap & aims3 Methods4 Ethics, data & PPI5 Impact & fit6 Timeline

If you are applying in another health field其他医科方向

FIELD 方向WHAT PANELS LOOK FOR 评审关注
Public health · Epidemiology
公共卫生 · 流行病学
A defined population, exposure and outcome; a named data source; a plan for confounding and bias.
明确的人群、暴露与结局;具体的数据来源;控制混杂与偏倚的方案。
Nursing
护理
A practice problem, access to a clinical setting, and often mixed methods or an intervention developed with patients and staff.
一个护理实践问题、进入临床场景的渠道,常用混合方法,或与患者和医护人员共同开发的干预。
Biomedical science
生物医学
A mechanism, a model system (cells, animals, human samples), the techniques and the controls.
一个机制、一个模型系统(细胞、动物、人体样本)、所用技术与对照设置。
Clinical research
临床研究
A clinical question in PICO form, a feasible design (trial, cohort, diagnostic study), recruitment numbers and a sample-size calculation.
用 PICO 表述的临床问题、可行的设计(试验、队列、诊断研究)、招募数量与样本量计算。
Health services research · Global health
卫生服务研究 · 全球健康
A service or policy, routinely collected or survey data, partners in the setting, and who will use the findings.
一项服务或政策、常规收集数据或调查数据、当地合作方,以及谁会使用研究结果。

The five questions every health panel asks医科评审必问的五个问题

#PANEL QUESTION 评审问题WHERE YOU ANSWER IT 在哪里回答
1Does the problem matter for health, and is there a real evidence gap? 这个问题对健康重要吗?证据空白真实存在吗?Background · Gap 背景 · 空白
2Is the question specific enough to answer (population, exposure or intervention, outcome)? 问题是否足够具体(人群、暴露或干预、结局)?Aims & objectives 目标
3Can the design and data answer it without serious bias? 设计和数据能否在没有严重偏倚的情况下回答它?Methods 方法
4Is it ethical and feasible, with data access in place? 它是否合乎伦理、可行,数据获取是否落实?Ethics & data · Timeline 伦理与数据 · 时间安排
5Why this group, and is this applicant ready? 为什么是这个团队?申请人准备好了吗?Impact & fit 影响与匹配度
Before you write 动笔之前

The golden thread逻辑主线

A health proposal works when each part follows from the one before: a health problem leads to a gap in the evidence, the gap becomes a specific question, and the design and data answer it. If you cannot write these four sentences, narrow the topic before drafting anything else.

好的医科计划书前后衔接:健康问题引出证据空白,空白变成具体的研究问题,再由研究设计和数据来回答。如果写不出这四句话,请先聚焦课题再动笔。

PROBLEM健康问题

Childhood asthma is common in Chinese cities, and children there have grown up with high but falling levels of fine particulate matter (PM2.5).

GAP证据空白

Few cohort studies in high-exposure settings have measured exposure in pregnancy and early childhood separately, so the critical window is unclear.

QUESTION研究问题

Is PM2.5 exposure in pregnancy, the first year or ages 1-3 associated with incident asthma by age 8?

DESIGN研究设计

Secondary analysis of an existing birth cohort of about 5,000 children, linked to residential PM2.5 estimates, using survival models.

The worked example used in this guide本指南使用的范例

Example applicant 范例申请人

Background 背景
Bachelor of Medicine in Preventive Medicine (Mainland China); MPH (UK), with a dissertation using survival analysis in R; research assistant for two years on a birth cohort study in eastern China, cleaning follow-up data and coding questionnaires.

Applying to 申请
PhD in epidemiology or population health in the UK; non-clinical applicant (no clinical registration).

Working title 暂定题目
Breathing Early: Early-Life PM2.5 Exposure and Incident Asthma in a Chinese Birth Cohort

How to use the example 如何使用范例

The same public-health proposal appears in every section so you can watch one project move from problem to timeline. Do not copy the topic. Study the moves, the numbered notes beside each extract, then make the same moves with your own field, data and supervisor. The cohort is fictional; citations shown as (Authors, Year) are placeholders, and in your own proposal every claim needs a real source you have read. The examples are in English only, because your application will be written in English.

同一份公共卫生计划书贯穿每个部分,方便你看到一个项目如何从健康问题一路走到时间安排。不要照搬课题,要学习写法:先看每段旁边的编号解析,再用你自己的方向、数据和导师做同样的事。范例中的队列是虚构的;(Authors, Year) 为占位符,你自己的计划书里每个论断都必须有你读过的真实文献。范例只用英文,因为你的申请材料也用英文写。

Section 0 第 0 部分
0

Title & Abstract题目与摘要

Typical length 典型篇幅 title 10-20 words · abstract 150-250 words

Health titles name the exposure or intervention, the outcome, the population and often the design. A reader should know from the title alone what is measured in whom. The abstract follows the structure of a journal abstract: background, aim, methods, expected contribution.

医科题目要写出暴露或干预、结局、研究人群,常常还包括研究设计。读者仅凭题目就应知道在什么人群中测量什么。摘要沿用期刊摘要的结构:背景、目标、方法、预期贡献。

Anatomy of a title题目结构

Breathing Early: Hook (optional) 开头(可选)
Early-Life PM2.5 Exposure and Incident Asthma Exposure + outcome 暴露 + 结局
in a Chinese Birth Cohort Population + design 人群 + 设计

Weak vs stronger titles较弱与更好的题目

✕ Weak 较弱“Air Pollution and Children’s Health in China”

No specific exposure, outcome or design. It could be a textbook chapter.

没有具体的暴露、结局或设计,像是教科书的一章。

✓ Stronger 更好“Early-Life PM2.5 Exposure and Incident Asthma: A Chinese Birth Cohort Study”

Names the exposure, the outcome, the population and the design.

写出暴露、结局、人群与设计。

✕ Weak 较弱“How Air Pollution Causes Asthma in Chinese Children”

Assumes a causal answer that an observational study cannot prove alone.

预设了因果结论,而观察性研究本身无法单独证明。

✓ Stronger 更好“Is Early-Life PM2.5 Exposure Associated with Childhood Asthma? Evidence from a Chinese Birth Cohort”

Neutral wording that matches what the design can show.

用词中立,与研究设计能证明的内容相符。

Nursing 护理 A Nurse-Led Telephone Follow-Up Programme for Older Adults after Hip Fracture Surgery: A Feasibility Study
Biomedical 生物医学 Oxidative Stress and Airway Epithelial Barrier Function after Fine Particle Exposure in Primary Human Cells
Health economics 卫生经济学 Cost-Effectiveness of Community Diabetes Screening in Rural China: A Markov Model Analysis
Section 0 continued 第 0 部分(续)

The abstract摘要

Blueprint结构

1Move 1 · Problem

The health burden, with a source.

健康负担,附出处。

2Move 2 · Gap

What the evidence does not yet show.

现有证据尚未说明的内容。

3Move 3 · Aim

The aim in one sentence.

用一句话写目标。

4Move 4 · Methods

Design, population, data, analysis.

设计、人群、数据、分析。

5Move 5 · Contribution

Who uses the result.

谁会使用研究结果。

Worked example · Abstract 范例 · 摘要

1Asthma is one of the most common chronic diseases of childhood, and children in many Chinese cities have grown up with high, though falling, concentrations of fine particulate matter (PM2.5) (Authors, Year). 2Studies from lower-exposure settings link early-life PM2.5 to asthma, but few cohorts in high-exposure settings have measured exposure separately in pregnancy and early childhood, so it is unclear which period matters most.

3This project aims to estimate the association between PM2.5 exposure in pregnancy, the first year of life and ages 1-3 and incident asthma by age 8. 4It will analyse an existing birth cohort of about 5,000 children in eastern China, linking residential addresses to published satellite-based PM2.5 estimates, and will use survival models adjusted for confounders identified with a directed acyclic graph. 5The findings will show when protection from air pollution matters most for children’s respiratory health, which can inform guidance for pregnant women and families with young children.

Why it works 写法解析
1Problem 问题

Burden and setting in one sentence, with a citation.

一句话写出疾病负担与情境,附引用。

2Gap 空白

Names the missing evidence: exposure windows in high-exposure settings.

点明缺失的证据:高暴露地区的暴露窗口期。

3Aim 目标

One aim with exposure, outcome and age.

一个目标,包含暴露、结局与年龄。

4Methods 方法

Data, exposure assessment and analysis in one sentence.

用一句话写数据、暴露评估与分析。

5Contribution 贡献

Ends on who can use the result.

以谁能使用这一结果收尾。

Write it last 最后再写

Write the abstract after the rest of the proposal is finished, then check that every sentence matches a section. Some faculty forms and templates have their own summary box, sometimes limited to 100 words, so keep a 100-word version ready.

摘要最好在整份计划书完成后再写,并核对每句话都能对应到某一部分。部分学院表格和模板设有固定的摘要栏,有时限 100 词,请准备一个 100 词版本。

Section 1 第 1 部分
1

Background & Rationale研究背景与依据

Typical length 典型篇幅 300-450 words · about 14% 约占 14%

Health proposals open with the size of the problem: how many people it affects, how serious it is, and why the setting you study matters. Give numbers from a cited source, then move quickly to what is known and what is not. Leave the detailed review of studies to Section 2.

医科计划书一开头就写问题有多大:影响多少人、有多严重、你研究的地方为什么值得关注。数据要有出处,然后很快转到哪些已经知道、哪些还不知道。具体的文献综述留到第 2 部分。

Blueprint结构

1Para 1 · Burden

How common and how serious, with a source.

有多常见、有多严重,附出处。

2Para 2 · Setting

Why this population or place.

为何是这一人群或地区。

3Para 3 · What is known

The main finding of existing research.

现有研究的主要发现。

4Para 4 · Why it matters now

Policy, practice or a new data source.

政策、实践或新的数据来源。

Worked example · Background (paragraphs 1-4) 范例 · 研究背景(第 1-4 段)

1Asthma is one of the most common chronic diseases in children worldwide, and its prevalence among children in Chinese cities has risen over recent decades (Authors, Year). It is a leading cause of school absence and emergency visits in this age group (Authors, Year).

2China offers an unusual setting for studying air pollution and child health. PM2.5 concentrations in many cities were high in the early 2010s and have fallen since the national Air Pollution Prevention and Control Action Plan of 2013 (Authors, Year), so children born a few years apart in the same city had exposures that differed widely.

3Cohort studies, mostly from Europe and North America, have linked early-life PM2.5 exposure to later asthma (Authors, Year), but exposure levels in those studies were much lower, and it is not known whether the same associations hold at higher concentrations.

4Birth cohorts established in China in the past decade now follow children to school age, which makes it possible, for the first time in this setting, to separate exposure in pregnancy from exposure in early childhood.

Why it works 写法解析
1Burden with a source 有出处的负担

States how common and how serious, with a citation for each claim.

写出常见程度与严重程度,每个论断都有引用。

2Why this setting 为何是此情境

Explains why the Chinese setting helps the science, which is more than naming a place.

说明中国这一情境为什么对研究有利,而不是只交代地点。

3What is known 已知内容

One sentence summary; detail comes in Section 2.

一句话概括,细节放在第 2 部分。

4Why now 为何是现在

A new data source makes the study possible now.

新的数据来源让研究现在得以实现。

Check every fact about your setting 核实每一个情境事实

Prevalence figures, pollution levels and policy dates must come from published studies or official sources, not from news articles or memory. Panels in health read numbers closely.

患病率、污染水平和政策日期必须来自已发表的研究或官方资料,而不是新闻或记忆。医科评审会仔细核对数字。

Section 1 continued 第 1 部分(续)

Making the case for the problem把问题的重要性讲清楚

✕ Weak 较弱“Air pollution is a serious problem in China and it is very harmful to people’s health.”

No numbers, no source, no population. It could open any proposal.

没有数据、出处与人群,可以放在任何计划书开头。

✓ Stronger 更好“Asthma prevalence among children in Chinese cities has risen over recent decades (Authors, Year), and it is a leading cause of school absence in this age group (Authors, Year).”

Specific, sourced and about one population.

具体、有出处,聚焦一个人群。

✕ Weak 较弱“Many studies have shown that PM2.5 causes asthma.”

Overstates the evidence and leaves no gap for the project.

夸大了证据,也没有为本项目留下空白。

✓ Stronger 更好“Cohort studies from lower-exposure settings link early-life PM2.5 to asthma (Authors, Year), but it is not known whether these associations hold at higher concentrations.”

Accurate about the evidence and points straight to the gap.

准确描述证据,并直接指向空白。

Common background mistakes研究背景常见错误

Textbook opening 教科书式开头

Defining asthma or PM2.5 at length. The panel knows the definitions.

长篇解释哮喘或 PM2.5 的定义,评审早已知道。

Numbers without sources 数据无出处

Prevalence or mortality figures with no citation, or from a website.

患病率或死亡率数据没有引用,或来自普通网站。

Too broad 范围过大

Covering all air pollutants and all respiratory diseases. Choose one exposure and one outcome.

涵盖所有污染物和所有呼吸系统疾病。请选定一个暴露和一个结局。

Nursing 护理 Open with the clinical problem in a service, for example readmissions after hip fracture surgery among older adults, and its cost to patients and wards.
Biomedical 生物医学 Open with the disease and the unexplained mechanism, then the model system that can test it.
Health economics 卫生经济学 Open with the cost of the condition to the health system and the decision a funder has to make.
Section 2 第 2 部分
2

Evidence Gap, Aims & Objectives证据空白、目标与具体目标

Typical length 典型篇幅 400-550 words · about 18% 约占 18%

Review the evidence by theme, not paper by paper, and end each theme with what it leaves unanswered. Where a systematic review exists, cite it and say what it concluded. Then state one overall aim and three or four objectives, each of which matches an analysis in Section 3. Many health proposals also state a hypothesis for each objective.

按主题而不是逐篇综述文献,每个主题都以“尚未回答的问题”收尾。如果已有系统综述,请引用并说明其结论。然后写出一个总目标和三到四个具体目标,每个具体目标都对应第 3 部分中的一项分析。许多医科计划书还会为每个具体目标写出假设。

Blueprint结构

1Theme 1 · Main association

What cohorts have found so far.

已有队列研究的发现。

2Theme 2 · Timing

What is known about exposure windows.

关于暴露窗口期的已知内容。

3The gap

One sentence on what is missing.

一句话写出缺失的内容。

4Aim and objectives

One aim, three objectives, each testable.

一个总目标、三个可检验的具体目标。

Worked example · Evidence gap and aims 范例 · 证据空白与目标

1A systematic review and meta-analysis of cohort studies found that higher early-life PM2.5 exposure was associated with a higher risk of childhood asthma, although estimates varied between studies (Authors, Year). Most included cohorts came from Europe and North America, where average exposures were a fraction of those in Chinese cities.

2Studies that examined timing have reached different conclusions: some point to pregnancy, others to the first year of life (Authors, Year; Authors, Year). Few could separate the two, because exposure in the same home changes little from one year to the next. 3No cohort in a high-exposure setting with steeply falling pollution has yet separated these windows.

4Aim: to estimate the association between early-life PM2.5 exposure and incident asthma by age 8 in a Chinese birth cohort. Objective 1: estimate the association between average PM2.5 from conception to age 3 and incident asthma. Objective 2: identify the window (pregnancy, first year, ages 1-3) in which exposure is most strongly associated with asthma. Objective 3: examine whether associations differ by sex and by household second-hand smoke exposure. Hypothesis for Objective 2: exposure in the first year of life shows the strongest association.

Why it works 写法解析
1Cite the review 引用系统综述

Starting from a systematic review shows you know the state of the evidence.

从系统综述出发,表明你了解证据现状。

2Theme, not list 按主题综述

Timing is treated as one theme with conflicting results.

把暴露时间当成一个主题来写,这个主题下的结果相互矛盾。

3One-sentence gap 一句话空白

The gap is specific: setting plus exposure windows.

空白很具体:情境加暴露窗口期。

4Testable objectives 可检验的目标

Each objective names an exposure, an outcome and a comparison that Section 3 analyses.

每个具体目标都写出暴露、结局与比较,第 3 部分会逐一分析。

Section 2 continued 第 2 部分(续)

Writing aims a health panel can judge写出评审能评判的目标

✕ Weak 较弱“Aim: to explore the impact of air pollution on children’s health.”

“Explore” and “impact” promise nothing measurable, and “children’s health” has no outcome.

“探索”与“影响”无法测量,“儿童健康”也没有具体结局。

✓ Stronger 更好“Aim: to estimate the association between early-life PM2.5 exposure and incident asthma by age 8.”

Exposure, outcome, age and the kind of result are all named.

暴露、结局、年龄和结果类型都写得很清楚。

✕ Weak 较弱A review that lists twenty studies with their sample sizes and results.

A catalogue, not an argument; the gap does not emerge.

这是目录而不是论证,空白无从显现。

✓ Stronger 更好Two themes, each ending in what it leaves unanswered, then one sentence on the gap.

Synthesis that leads straight to the objectives.

综合各主题,直接引出具体目标。

Frame the question with PECO or PICO用 PECO 或 PICO 界定问题

ELEMENT 要素WORKED EXAMPLE 范例IN A TRIAL (PICO) 在试验中
P · Population
人群
Children in a birth cohort in eastern ChinaAdults with type 2 diabetes in community clinics
E or I · Exposure or intervention
暴露或干预
PM2.5 in pregnancy, first year, ages 1-3A smartphone self-management programme
C · Comparison
对照
Lower versus higher exposure (per 10 µg/m³)Usual care
O · Outcome
结局
Doctor-diagnosed asthma by age 8Change in HbA1c at 12 months
Nursing 护理 Aim: to assess the feasibility and acceptability of a nurse-led telephone follow-up programme for older adults after hip fracture surgery.
Biomedical 生物医学 Aim: to test whether oxidative stress mediates the loss of airway epithelial barrier function after fine particle exposure.
Health economics 卫生经济学 Aim: to estimate the incremental cost per quality-adjusted life year of community diabetes screening compared with no screening.
Section 3 第 3 部分
3

Methods: Design, Data & Analysis研究方法:设计、数据与分析

Typical length 典型篇幅 700-1,000 words · about 34% 约占 34%

This is where a health panel decides whether the study can answer the question. Name the design, the population and how it was sampled, how exposure and outcome are measured, which confounders you will adjust for and why, and the statistical model. State the sample size, or show that the existing data give enough power.

医科评审读到这一部分,会判断研究到底能不能回答问题。写明研究设计、人群及其抽样方式、暴露与结局如何测量、调整哪些混杂因素及原因,以及统计模型。写出样本量,或证明现有数据有足够的统计效能。

Blueprint结构

11 · Design & population

Design, data source, sample, follow-up.

设计、数据来源、样本、随访。

22 · Exposure & outcome

How each is measured and validated.

各自如何测量与验证。

33 · Confounding & analysis

DAG, model, missing data.

因果图、模型、缺失数据。

44 · Power & sensitivity

Sample size and checks on bias.

样本量与偏倚检验。

Worked example · Methods 范例 · 研究方法

1Design and population. This is a secondary analysis of an existing prospective birth cohort in a city in eastern China. About 5,000 mother-child pairs were recruited in pregnancy between 2015 and 2017 and followed at ages 1, 3, 6 and 8 years, with residential addresses recorded at each visit.

2Exposure and outcome. Monthly PM2.5 at each residential address will be taken from a published satellite-based model at 1 km resolution (Authors, Year) and averaged over pregnancy, the first year and ages 1-3, allowing for changes of address. The outcome is doctor-diagnosed asthma reported by parents at ages 3, 6 and 8 using the ISAAC core questions; in a subsample, diagnoses will be checked against hospital records.

3Confounding and analysis. Confounders will be chosen with a directed acyclic graph and include parental asthma, maternal education, household income, maternal smoking, second-hand smoke and season of birth. Cox proportional hazards models with age as the time scale will estimate hazard ratios per 10 µg/m³. Exposure windows will be modelled together in a distributed lag model so that each is adjusted for the others. Missing covariates will be handled by multiple imputation.

4Power and sensitivity. A preliminary calculation based on the incidence observed at age 6 (Appendix 1) suggests the cohort can detect a hazard ratio of about 1.15 per 10 µg/m³. Sensitivity analyses will exclude children who moved city, use wheeze as an alternative outcome, and test for unmeasured confounding with a negative-control exposure (PM2.5 after the outcome).

Why it works 写法解析
1Design in one line 一句话写设计

Names the design, the source, the period and the follow-up points.

写出设计、来源、时间段与随访节点。

2Measurement 测量

Exposure method and outcome tool are named, with a validation step.

写明暴露方法与结局工具,并有验证步骤。

3Confounding 混杂

A DAG explains why each confounder is there; the model is named.

用因果图解释每个混杂因素的依据,并写明模型。

4Power and bias 效能与偏倚

Shows the sample is large enough and anticipates the panel’s objections.

证明样本量足够,并预先回应评审的质疑。

Section 3 continued 第 3 部分(续)

Making the methods convincing让研究方法更有说服力

✕ Weak 较弱“I will collect data from hospitals and use SPSS to analyse the relationship between pollution and asthma.”

No design, no source, no measures, no model. The software is the only specific detail.

没有设计、来源、测量和模型,唯一具体的是软件名称。

✓ Stronger 更好“Secondary analysis of a birth cohort of about 5,000 children, with residential PM2.5 from a 1 km satellite model and Cox models adjusted for confounders chosen with a DAG.”

A supervisor could start planning the analysis from this sentence.

导师凭这一句就能开始规划分析。

✕ Weak 较弱“We will control for all possible confounders.”

Not possible, and it shows no thinking about which ones matter.

做不到,也看不出你思考过哪些因素重要。

✓ Stronger 更好“Parental asthma, maternal education, household income and second-hand smoke are adjusted for, based on a DAG (Appendix 2).”

Named confounders with a stated reason for each.

写明混杂因素及其依据。

Methods checklist for health proposals医科方法检查清单

CHECK 检查项WHAT THE PANEL WANTS TO SEE 评审希望看到
Design 设计Named design (cohort, case-control, trial, qualitative, mixed methods) and why it fits the question. 写明设计类型(队列、病例对照、试验、质性、混合方法)及其适用原因。
Measures 测量Validated instruments or standard definitions for exposure and outcome. 暴露与结局使用经过验证的工具或标准定义。
Bias 偏倚Selection, information and confounding bias, and what you will do about each. 选择偏倚、信息偏倚与混杂偏倚,以及各自的应对方法。
Numbers 数字Sample size or power calculation; for trials, recruitment rate and attrition. 样本量或效能计算;试验需写明招募速度与失访率。
Reporting 报告规范The guideline you will follow, such as STROBE for observational studies, CONSORT for trials, COREQ for qualitative work. 遵循的报告规范,如观察性研究用 STROBE,试验用 CONSORT,质性研究用 COREQ。
Skills 技能Software and methods you already use, and the training that fills the gaps. 已掌握的软件与方法,以及弥补不足的培训。
Nursing 护理 Feasibility study: recruitment and retention rates, intervention fidelity, and semi-structured interviews with patients and nurses analysed thematically.
Biomedical 生物医学 Primary cells from named donor groups, exposure doses, outcome assays, controls, number of biological replicates and the statistical test.
Health economics 卫生经济学 Model structure (decision tree or Markov), time horizon, perspective, sources for costs and utilities, and probabilistic sensitivity analysis.
Section 4 第 4 部分
4

Ethics, Data Access & Public Involvement伦理、数据获取与公众参与

Typical length 典型篇幅 200-400 words · about 12% 约占 12%

Any study with human participants, their samples or their records needs ethics approval before data collection or analysis begins. Health panels also ask who holds the data, whether you have permission to use it, and how patients or the public have shaped the project. Answering these in the proposal shows the project can start on time.

任何涉及人类参与者、其样本或其记录的研究,都必须在数据收集或分析开始之前获得伦理审批。医科评审还会问:数据由谁持有?你是否已获准使用?患者或公众如何参与塑造了这个项目?在计划书里先回答这些问题,评审就能看出项目可以按时开始。

Blueprint结构

11 · Approvals

Which committees must approve, and when.

需要哪些委员会审批、何时审批。

22 · Consent & data

What participants agreed to; how data are protected.

参与者同意了什么;数据如何保护。

33 · Access route

Who holds the data and how you reach it.

数据由谁持有、你如何获取。

44 · Public involvement

How patients or the public take part.

患者或公众如何参与。

Worked example · Ethics, data access and involvement 范例 · 伦理、数据获取与参与

1Approvals. The cohort was approved by the research ethics committee of the hospital that hosts it. I will apply for approval from the university’s research ethics committee for this secondary analysis in the first three months, and the cohort’s data access committee will review the analysis plan.

2Consent and data protection. Parents gave written consent at recruitment that covers follow-up research on child health and the environment. Residential addresses are identifiable, so cohort staff will geocode addresses and link them to PM2.5 estimates on the cohort’s secure server in China. I will receive a pseudonymised dataset with exposure values but no addresses.

3Access route. The cohort team has agreed in principle to share these data; a data-sharing agreement will be signed before analysis. Because cross-border transfer of personal information from China is regulated (Authors, Year), analysis will take place through the cohort’s remote secure access system, and only aggregate results will leave it.

4Public involvement. A group of six to eight parents of children with asthma, recruited through the hospital’s paediatric respiratory clinic, will meet three times online: to comment on which outcomes matter to families, on the interpretation of results, and on a plain-language summary in Chinese. Members will be paid for their time.

Why it works 写法解析
1Named committees 写明委员会

Separates the original approval from the new one this study needs.

区分队列原有的审批与本研究需要的新审批。

2Consent covers the use 同意书覆盖用途

Shows the original consent covers this secondary use.

说明原始知情同意覆盖本次二次使用。

3A real access route 真实的获取途径

Who holds the data, what agreement is needed, and where analysis happens.

谁持有数据、需要什么协议、在哪里分析。

4Involvement with a purpose 有目的的参与

Says which parts of the project parents will shape, beyond a promise to consult them.

写明家长会影响项目的哪些部分,不只是一句“会征求意见”。

Section 4 continued 第 4 部分(续)

Approvals and data access by data type按数据类型看审批与数据获取

The approvals you need depend on where the data come from. The table describes the usual pattern in general terms; the exact committees and forms differ by country and institution, so confirm them with your supervisor.

所需的审批取决于数据来源。下表概括一般情况;具体的委员会和表格因国家和机构而异,请与导师确认。

DATA TYPE 数据类型WHAT IT USUALLY NEEDS 通常需要WHAT TO WRITE 计划书中写什么
New participants
招募新参与者
University ethics approval; if recruited through hospitals or health services, the health system’s ethics committee as well (in England, an NHS research ethics committee and Health Research Authority approval).
大学伦理审批;如果通过医院或医疗服务机构招募,还需医疗系统的伦理委员会审批(在英格兰为 NHS 研究伦理委员会与卫生研究管理局 HRA 批准)。
Who is recruited, where, how consent is taken, and how long approval may take.
招募谁、在哪里招募、如何取得知情同意、审批大约需要多久。
Existing cohort data
已有队列数据
The cohort’s data access committee, a data-sharing agreement, and usually university approval for secondary analysis.
队列的数据获取委员会、数据共享协议,通常还需大学对二次分析的审批。
That consent covers your use, that the team has agreed in principle, and where the data will be analysed.
知情同意覆盖你的用途、团队已原则上同意、数据在哪里分析。
Hospital records
医院病历记录
Hospital or joint university-hospital ethics review (an institutional review board, IRB, in many places), permission from the record holder, and rules for de-identification.
医院或大学与医院联合的伦理审查(许多地方称为机构审查委员会 IRB)、病历持有方的许可,以及去标识化规则。
A named clinical collaborator who can request access, and how identifiers are removed.
一位能申请数据访问的临床合作者,以及如何去除身份信息。
Routinely collected data
常规收集数据
An application to the data custodian (for example a national registry or health insurance database), often with a fee and a waiting period; analysis often in a secure environment.
向数据管理方(如国家登记系统或医保数据库)提交申请,常需付费并等待;分析通常在安全环境中进行。
The custodian, the application timeline and a fallback if approval is late.
数据管理方、申请时间线,以及审批延误时的备选方案。
Human samples
人体样本
Ethics approval for tissue or blood use, material transfer agreements, and in some countries specific rules on human genetic material.
使用组织或血液的伦理审批、材料转移协议,部分国家对人类遗传材料另有专门规定。
Where samples come from, the consent they carry, and the lab’s existing approvals.
样本来源、附带的知情同意,以及实验室已有的审批。
Patient and public involvement (PPI) 患者与公众参与

PPI means doing research with patients and the public, not on them: they help choose outcomes, review materials and interpret results. UK funders and many health faculties expect a PPI plan in the proposal. It is different from recruiting participants, and it does not usually need ethics approval itself, though your faculty may advise otherwise.

PPI 指与患者和公众一起做研究,而不是以他们为研究对象:他们参与选择结局指标、审阅研究材料并解读结果。英国的资助机构和许多医科学院都希望在计划书中看到 PPI 方案。它不同于招募研究参与者,通常本身不需要伦理审批,但请以学院的意见为准。

Nursing 护理 Ward or service manager permission as well as ethics approval; a PPI group of patients who have used the service.
Biomedical 生物医学 The lab’s existing tissue or animal licences and approvals, and whether new work falls under them.
Health economics 卫生经济学 Access to cost and utilisation data from the payer or registry; public input on which outcomes to value.
Section 5 第 5 部分
5

Impact & Fit影响与匹配度

Typical length 典型篇幅 150-300 words · about 8% 约占 8%

Say what the findings add to the evidence and who will use them: clinicians, public-health agencies, guideline writers or patients. Then say why this research group, which methods or data link you to them, and what you already bring.

说明研究结果为证据增加了什么、谁会使用:临床医生、公共卫生机构、指南制定者或患者。然后说明为什么是这个研究团队、哪些方法或数据把你和他们联系起来,以及你已具备的基础。

Blueprint结构

1Contribution to evidence

What the field learns.

这一领域将获得什么认识。

2Use in practice

Who acts on the result.

谁会依据结果采取行动。

3Group fit

Two named staff, a specific link each.

两位老师,各有具体联系。

4Readiness

Data, skills, prior work.

数据、技能、前期工作。

Worked example · Impact and fit 范例 · 影响与匹配度

1The project will provide the first estimates from a high-exposure setting that separate exposure in pregnancy from exposure in early childhood, adding to systematic reviews that so far draw mainly on lower-exposure cohorts (Authors, Year). 2If one window carries most of the risk, advice on reducing exposure, such as the use of air purifiers at home, can be aimed at that period.

3I am applying to this group because Professor A’s work on satellite-based exposure models (Authors, Year) underpins the exposure assessment, and Dr B’s research on life-course epidemiology of respiratory disease (Authors, Year) matches the timing question. The school’s training in causal inference would strengthen the analysis.

4My MPH dissertation used Cox models in R, and as a research assistant on this cohort I cleaned follow-up data and know its variables, so I can begin the analysis plan in my first month.

Why it works 写法解析
1Evidence gap filled 填补的空白

Links back to the gap in Section 2.

回应第 2 部分的空白。

2Practical use 实际用途

One concrete use of the result.

一个具体的成果用途。

3Specific fit 具体匹配

One person for the exposure method, one for the question.

一位对应暴露方法,一位对应研究问题。

4Readiness 准备度

Familiarity with the data is the strongest evidence of feasibility.

熟悉数据是可行性最有力的证据。

✕ Weak 较弱“I chose this university because it has an excellent medical school with a high ranking.”

Says nothing about research or people.

完全没有涉及研究或老师。

✓ Stronger 更好“Professor A’s satellite-based exposure models (Authors, Year) underpin my exposure assessment.”

Names a person, a piece of work and its exact role in the project.

点名一位老师、一项研究及其在项目中的具体作用。

Section 6 第 6 部分
6

Timeline & Feasibility时间安排与可行性

Typical length 典型篇幅 100-200 words or one table · about 6% 约占 6%

Health timelines must allow for approvals. Ethics review and data-sharing agreements can take months, and recruitment through health services is often slower than planned. Put approvals in the first months, plan work that does not depend on them, and name a fallback.

医科研究的时间表必须给审批留出时间。伦理审查和数据共享协议可能要几个月,通过医疗机构招募也常常比计划慢。把审批放在最初几个月,安排不依赖审批的工作,并写明备选方案。

Worked example · Timeline (three-year PhD) 范例 · 时间安排(三年制博士)

PERIOD 时间ACTIVITY 活动OUTPUT 产出
Months 1-6Ethics application; data-sharing agreement; systematic review of exposure windows; DAG and analysis planApprovals; review protocol registered
Months 7-12Exposure linkage by cohort staff; data checks; first-year reviewLinked dataset; review submitted
Year 2Objective 1 and 2 analyses; parent advisory group meetings 1 and 2Paper on main association and windows
Year 3Objective 3; sensitivity analyses; plain-language summary; thesis writingThird paper; thesis submission

2Risk: if the data-sharing agreement is delayed, the systematic review and a pre-registered analysis plan form the first chapter, and analysis starts when access is granted.

Why it works 写法解析
1Approvals first 审批先行

Ethics and data agreements are planned in the first six months.

伦理与数据协议安排在前六个月。

2Work that does not wait 不必等待的工作

The systematic review keeps the project moving during approvals.

系统综述让项目在审批期间继续推进。

Programmes with coursework 含课程的项目

Where the PhD includes coursework or a qualifying examination, add these to Year 1 and move the first analysis later.

如果博士包含课程或资格考试,请把它们加入第一年,并相应推迟第一项分析。

Section 7 第 7 部分
7

References参考文献

Usually excluded from the word count 通常不计字数 15-30 core sources

Cite systematic reviews, the main cohort or trial papers, the methods papers behind your exposure model or estimator, and the reporting guideline you follow. Health faculties often use Vancouver (numbered) style; check the faculty page and use one style throughout.

引用系统综述、主要的队列或试验论文、你的暴露模型或估计方法的出处,以及所遵循的报告规范。医学院常用温哥华(编号)格式;请查看学院页面,全文统一一种格式。

Blueprint结构

Reviews

Systematic reviews and meta-analyses first.

先引用系统综述与荟萃分析。

Primary studies

The cohorts or trials closest to yours.

与你最相近的队列或试验。

Methods

Exposure model, statistical method, reporting guideline.

暴露模型、统计方法、报告规范。

One style

Vancouver or the faculty’s style, used consistently.

温哥华格式或学院格式,前后统一。

Worked example · Reference list (excerpt, Vancouver style) 范例 · 参考文献节选(温哥华格式)

1. Author AA, Author BB. Early-life exposure to fine particulate matter and childhood asthma: a systematic review and meta-analysis. Journal Name. Year;00(0):000-000.

2. Author CC, Author DD, Author EE. Satellite-based estimates of daily PM2.5 concentrations at 1 km resolution in China. Journal Name. Year;00:000-000.

3. Author FF. Distributed lag models for identifying windows of susceptibility. Journal Name. Year;00(0):000-000.

Why it works 写法解析
1Real sources only 只用真实文献

In your own proposal every reference must be a paper you have read.

你自己的计划书里,每条参考文献都必须是你读过的论文。

2Include the methods 引用方法来源

Citing the exposure model and estimator shows you know their assumptions.

引用暴露模型与估计方法,表明你了解其前提假设。

News and websites 新闻与网站

Citing news reports or health websites instead of studies and official statistics.

引用新闻报道或健康网站,而不是研究论文和官方统计。

No recent reviews 缺少近期综述

Missing a systematic review published in the last few years.

缺少近几年发表的系统综述。

Broken citations 引用不匹配

Numbers in the text that do not match the list.

正文编号与参考文献列表对不上。

Putting it to work 实际应用

Adapting to each university按校调整

Keep the 2,500-word master version and cut from it. Never write a new proposal for each university; change the fit section, follow the faculty form, and trim.

保留 2,500 词母版,在此基础上裁剪。不要为每所学校重写,只需调整匹配部分、按学院表格填写并删减。

TARGET 目标WHAT TO DO 做法
Advertised project (Route A)
已发布项目(A 路径)
Oxford RDM (personal statement up to 500 words); Imperial NHLI studentships; Southampton advertised projects 牛津 RDM、帝国理工 NHLI、南安普顿
Write a fit statement from the methods and readiness paragraphs, matched to the project description. Drop your own design.
根据方法与准备度段落写匹配陈述,与项目说明逐条对应。删去你自己的设计。
Short statement of interest
简短兴趣陈述
Cambridge Public Health (statement of interest up to 2,500 characters plus CV); Cardiff Medicine (personal statement or covering letter) 剑桥公共卫生、卡迪夫医学
Keep the problem, the question, the data in one line and a fit paragraph with your named supervisor.
保留问题、研究问题、一行数据说明,以及写明导师的匹配段落。
Short proposal or faculty form (1-2 pages)
短篇计划书或学院表格(1-2 页)
Birmingham Medical and Dental Sciences (about 2 pages, double-spaced); Warwick Medicine (one-page methodology); York Health Sciences 500-1,000; Bath 1,000 伯明翰、华威、约克、巴斯
Background in one paragraph, aim and objectives, a methods paragraph, one ethics and data sentence, a short timeline. Move fit to the personal statement.
一段背景、目标与具体目标、一段方法、一句伦理与数据说明、简短时间安排。匹配度移到个人陈述中。
Full proposal (2,000-3,000 words)
完整计划书(2,000-3,000 词)
UCL medical education about 2,000 words; Manchester Social Work 2,500; Queen Mary Dentistry 2,000-3,000 UCL 医学教育、曼彻斯特社会工作、玛丽女王牙科
Use the master version; add a figure of the DAG or the timeline if space allows.
使用母版;篇幅允许时可加入因果图或时间安排图。
Supervisor and project first
先确定导师与项目
Queen Mary Dentistry (1,000-word proposal to the supervisor first); Edinburgh Psychiatry and Bristol Dental (supervisor and project first) 玛丽女王牙科、爱丁堡精神病学、布里斯托牙科
Send a one-page summary (title, aim, data, methods in brief, proposed supervisors); write the full proposal with the supervisor once they support it.
先发送一页摘要(题目、目标、数据、简要方法、拟定导师);导师表示支持后,再与其共同完成完整计划。

The email before the proposal计划书之前的那封邮件

Where a supervisor must agree first, your first email is part of the application. Give the supervisor enough detail to judge the fit: one specific link to their work, your question and method in one sentence each, one easy question, and your CV plus a short summary attached. Check the department’s rule first, because a few say there is no need to contact anyone. Write the first draft yourself; your academic will check the tailored paragraph.

在需要先取得导师同意的项目中,第一封邮件就是申请的一部分。写得足够详细,让导师能判断是否匹配:联系对方一项具体研究,用一句话写研究问题、一句话写方法,只问一个容易回答的问题,并附简历和一份简短摘要。先核对院系规定,少数院系说明无须联系。第一稿由你自己撰写,导师只负责核对针对导师的那一段。

Before you submit 提交之前

Read it like a panel用评审的眼光重读

Reread your proposal as a faculty member with ten minutes and forty others to read. Score each question 1 (weak), 2 (adequate) or 3 (strong), and fix the lowest score first.

像一位只有十分钟、还有四十份要读的老师那样重读。每项打 1(弱)、2(一般)、3(强)分,先修改得分最低的一项。

#CHECK 检查SELF-SCORE 自评
1Can I state the question in one sentence, with population, exposure and outcome? 我能用一句话说出包含人群、暴露与结局的研究问题吗?1 2 3
2Does every figure about the health problem have a source? 关于健康问题的每个数字都有出处吗?1 2 3
3Does each objective match an analysis in the methods? 每个具体目标都对应方法中的一项分析吗?1 2 3
4Have I named the confounders or biases and how I deal with them? 我写明了混杂与偏倚及其应对方法吗?1 2 3
5Have I named the approvals, the data holder and the access route? 我写明了审批、数据持有方和获取途径吗?1 2 3
6Is the fit section specific to named staff, and is the supervisor agreed where the faculty requires it? 匹配部分是否针对具体老师?学院要求时,导师是否已同意?1 2 3
7Is it the right document type (fit statement, statement, form or proposal) and length? 文件类型(匹配陈述、陈述、表格或计划书)与篇幅是否正确?1 2 3

Three tests三个测试

The reviewer test

Imagine a journal reviewer reads your methods. What is their first objection about bias? Answer it in the text.

想象期刊审稿人读你的方法部分,他关于偏倚的第一个质疑是什么?在正文中回答它。

The ethics test

Could an ethics committee approve this as written: who, what consent, what data, where stored? If not, add it.

伦理委员会能否按现有写法批准:研究谁、什么同意、什么数据、存在哪里?不能就补上。

The two-minute test

Explain the question, the data and the main analysis in two minutes. Where you stumble, the writing is unclear.

用两分钟讲清研究问题、数据和主要分析。卡壳的地方就是写得不清楚的地方。