模块0:召唤

一些人建造了房子。其他人确保基础稳固。

你有没有想过谁检查检查器?

谁确保医疗决策背后的方法是正确的?

隐形建筑师

桥梁的故事检查员:

A city built a magnificent bridge. Engineers celebrated. Citizens crossed daily. But one woman spent her days beneath, checking welds, measuring stress, documenting cracks.

"Why don't you build bridges?" they asked.

"Because," she said, "I ensure every bridge built will stand."

方法学家是医学证据的检查员。他们不进行试验——他们确保试验告诉我们真相。
$2.6B
Saved when RECOVERY trial methods proved dexamethasone works
15K+
在方法审查之前因有缺陷的抑肽酶研究导致大量死亡
85%
Of medical research has methodological limitations

方法工作并不光彩。但这是治愈药物和造成伤害的药物之间的区别。

想想道格·奥尔特曼博士的故事...

Doug Altman (1948-2018): The Scandal of Poor Medical Research

In 1994, a statistician published an editorial in BMJ titled "The Scandal of Poor Medical Research." He wrote:

“我们需要更少的研究,更好的研究,并且出于正确的理由进行研究。”

道格·奥尔特曼没有发现新药。他没有进行手术。他的职业生涯致力于制定报告指南(CONSORT、STROBE、PRISMA),这些指南改变了世界呈现证据的方式。

他的引用次数:500,000+。他的工具被各大期刊使用。他没有建造桥梁,但现在每座桥梁都使用他的检查标准。
Altman DG. The scandal of poor medical research. BMJ 1994;308:283-4. Citations: 6,000+

What They Do

• 设计严格的研究方法

• Develop reporting guidelines

• 创建偏差风险工具

• Advance statistical techniques

• 培训下一代

What They Don't Do

• 通常不是主要数据收集

• 通常没有耐心

• 很少出现在头条新闻

• 通常不是“第一作者”

• 不是突破的面孔

方法学家是证据的建筑师,而不是个人研究的建设者。

的故事约翰·斯诺的霍乱地图:

1854 年,伦敦医生相信霍乱通过“瘴气”(恶劣的空气)传播。他们的地图很漂亮,显示了风向和海拔。

Dr.约翰·斯诺绘制了一张简单的地图——只有一些点,显示了布罗德街水泵周围聚集的死亡人数。他的简单、方法论健全的地图将受污染的水确定为来源。

瘴气地图很优雅,但错误。斯诺的普通地图拯救了生命。

方法论胜过演示。

您是 1854 年的伦敦医生。霍乱正在蔓延。您选择什么?

Path A: 遵循瘴气理论——创建漂亮的风型和空气质量地图。
>→ 当您研究空气时,患者会死亡。疾病继续不受控制。
Path B: 系统地绘制死亡图——绘制病例并寻找数据模式。
>→ 找到 Broad Street 水泵。拆下手柄。通过方法论思维拯救生命。
+340%
Growth in HTA positions since 2015
$95K-$180K
Salary range for senior methodologists
72%
Report high job satisfaction

需求正在上升。监管机构、制药公司和卫生系统都需要了解证据质量的人员。

对于那些愿意走这条路的人来说,道路是清晰的。

模块 1:方法学家做什么

模块 1:方法学家做什么

他们不要建造这艘船。他们确保它不会下沉。

想象一下生活中的一天......

Not one life—many lives, in many settings.

The Methodologist Who Changed Medicine: Archie Cochrane

In 1972, a physician named Archie Cochrane published 有效性和效率,认为大多数医疗决策缺乏对照试验的证据。他向整个医疗机构提出挑战,要求证明治疗方法确实有效。

在 Cochrane 倡导之前,绝大多数药物治疗方法从未在随机对照试验中进行过测试。到 1993 年 Cochrane 协作组织成立时,系统评价已成为循证医学的黄金标准。

一位方法学家对严谨性的坚持创造了现代医学现在依赖的基础设施。

您是 1975 年的年轻研究员。一位资深教授驳回了您进行系统评价的建议,他说:“我们已经知道什么源于临床经验。”

Path A: 服从权威——接受临床经验就足够了,放弃系统证据综合的想法。
>→ 未经测试的治疗继续不受挑战。患者因“人人都知道有效”的无效或危险护理而受到伤害。
Path B: 遵循 Cochrane 的愿景——坚持系统地收集和评估证据,即使证据不受欢迎。
>→ 您帮助建立改变全球医学的方法论。系统评价成为全球临床指南的基础。

Setting 1: Academic Medical Center

Dr. Sarah Chen, Associate Professor of Biostatistics
McMaster University, Hamilton, Canada

8:00 AM: 审查博士生的系统评价方案草案。发现了一个致命的缺陷:他们计划进行单次筛选。推荐试点校准练习。

10:00 AM: 向 40 名临床医生教授 GRADE 研讨会。一半人对“间接性”感到困惑。她使用了阿司匹林治疗头痛与心脏病的例子。灯泡继续。

2:00 PM: 关于网络元分析不一致方法论文的合著者会议。主张采用新的可视化方法。

4:00 PM: Reviews grant application as methodologist. Flags sample size calculation error that would doom a $2M trial.

影响:防止一项有缺陷的试验,培训 40 名未来证据使用者,推进该领域的发展。

Setting 2: Health Technology Assessment Agency

Ahmed Al-Rashid, Senior HTA Analyst
Saudi Food and Drug Authority, Riyadh

8:30 AM: 收到制造商的新糖尿病药物档案。 847 页。他的工作:在 3 周内评估证据质量。

10:00 AM: Spots a problem: the pivotal trial used a surrogate endpoint (HbA1c) but claims mortality benefit. Surrogate ≠ patient-relevant outcome.

1:00 PM: 与临床专家会面。 “这种药有效,”他们说。 “但这对病人重要吗?”他问道。已确定证据差距。

3:00 PM: 评估报告草案。建议根据现实世界证据收集要求有条件批准。

影响:国家的药物覆盖决策现在基于严格的方法,而不是营销声明。

Setting 3: Pharmaceutical Industry

Dr. Maria Santos, Director of Evidence Strategy
Global Pharmaceutical Company, Basel

9:00 AM: 设计 III 期试验方案。争取务实的设计而不是解释——更难招募,但结果将适用于真实的患者。

11:00 AM: Reviews competitor's published trial. Notes: open-label, subjective outcomes, high attrition. Prepares competitive landscape briefing.

2:00 PM: 在 FDA 提交前会议上指导临床团队。 “他们会询问您的主要终点。准备好敏感性分析。”

4:30 PM: 加入针对 HTA 提交的网络元分析全球呼吁。就传递性假设进行 90 分钟的辩论。

影响:更好设计的试验、更清晰的监管路径、为患者和企业服务的证据。

Setting 4: WHO/Global Health

Dr. Kwame Asante, Technical Officer
World Health Organization, Geneva

8:00 AM: 与非洲地区的晨会。需要更新疟疾指南。审查哪些快速审查可以调整,哪些需要新的证据。

10:30 AM: 指南制定小组会议。 14位专家不同意干预。他的职责是:确保客观地呈现证据,无论谁的声音最大。

1:00 PM: 起草 HIV 检测建议的 GRADE 证据简介。 “非常低的确定性”并不意味着“不推荐”——他写了简单语言的摘要。

4:00 PM: 实时系统评论仪表板的评论。自上个月以来进行了三项新的随机对照试验。触发更新协议。

Impact: Guidelines used by 194 member states. Methods decisions affect millions.

Design

Creating study protocols, sample size calculations, endpoint selection

Appraise

Risk of bias assessment, GRADE ratings, critical appraisal

Synthesize

Meta-analysis, network meta-analysis, qualitative synthesis

Develop

New tools, guidelines, reporting standards, software

Teach

Training researchers, clinicians, policymakers

Advise

Consultations on specific studies, grants, submissions

Consider what happens when methods fail...

抑肽酶灾难

1993-2007: A drug that should have been stopped years earlier

抑肽酶用于心脏手术以减少出血。早期研究表明它有效。但方法学家注意到了问题:

• Observational studies had severe confounding
• RCT 对于安全结果而言太小
• Meta-analyses pooled incompatible populations

2007: BART trial finally showed increased mortality. Drug withdrawn.

Estimate: 15,000-22,000 excess deaths while methodological concerns were ignored.
Fergusson et al. NEJM 2008. Shaw et al. BMJ 2014.

RECOVERY 胜利

2020: Methods done right, lives saved

RECOVERY 试验从第一天起就使用了严格的方法:

• Pragmatic design embedded in routine care
• Adequate sample size (thousands, not dozens)
• 预先指定的终点和分析计划
•独立数据监测委员会

Result: Dexamethasone proven in 100 days. Estimated 1 million lives saved globally.

设计该方案的方法学家是我们这个时代最有影响力的科学家之一,但很少有人知道他们的名字。
RECOVERY Collaborative Group. NEJM 2021. Horby P, Landray M (co-chief investigators).

A methodologist's primary role is to:

Scenario
抑肽酶事件估计导致 15,000-22,000 人死亡。主要的方法论失败是:

选择最准确的答案:

方法不是抽象的。方法决定生死。

模块 2:技能堆栈

模块 2:技能堆栈

A methodologist is not one thing. A methodologist is many things woven together.

哪些技能将方法学家与研究人员区分开来?

Not depth in one area—but bridges between many.

1。统计和定量方法

• 回归、生存分析

• Meta-analysis (pairwise, network)

• 贝叶斯方法(越来越多)

• 缺失数据、敏感性分析

2。流行病学与研究设计

• Bias recognition (selection, information, confounding)

• RCT 设计和 CONSORT

• 观察性研究局限性

• Causal inference frameworks

3. Domain Expertise

• Clinical knowledge (enough to ask right questions)

• Understanding healthcare systems

• 政策和决策背景

• Patient-relevant outcomes

4. Communication

• Translating complexity to clarity

• 为不同受众写作

• 教学和研究指导

• Diplomatic disagreement

DNA 发现的故事:

In 1953, discovering DNA's structure required four disciplines:

Rosalind Franklin's X-ray crystallography revealed the helix shape.
Erwin Chargaff's chemistry showed base-pairing rules.
Linus Pauling's modeling expertise suggested the spiral.
沃森和克里克的生物学见解将这些片段组合在一起。

No single specialist could have solved it alone. The double helix emerged only when disciplines converged.

方法学家是专家之间的翻译者。在任何一个领域都不是最深入的,但在所有领域都足够流畅。

您拥有有关 DNA 结构的重要 X 射线数据。您选择什么?

Path A: 独立工作——独自完善您的分析,小心保护您的数据。
>→其他人通过反向渠道看到您的数据,请先发布。您的贡献已被载入史册。
Path B: 跨学科合作 - 分享见解,与生物学家和化学家互动。
>→ 集体发现。突破来自综合。科学共同进步。

Core Competencies (Must Have)

Regression fundamentals: Linear, logistic, interpretation of coefficients

Survival analysis: Kaplan-Meier, Cox regression, hazard ratios

Meta-analysis: Fixed/random effects, heterogeneity (I², τ²), forest plots

Missing data: MCAR/MAR/MNAR, multiple imputation basics

Growth Areas (Increasingly Valuable)

• Network meta-analysis and indirect comparisons

• Bayesian approaches (priors, credible intervals)

• Machine learning for prediction models

• Causal inference methods (IPW, G-methods)

John Ioannidis 的故事

从临床医生到方法革命

John Ioannidis trained as an infectious disease physician. He could have spent his career treating patients.

相反,他在 2005 年发表了“为什么大多数发表的研究结果都是错误的”——这是大多数人认为的统计论点声称研究结果是错误的。

该论文已被引用超过 15,000 times。它改变了期刊、资助者和研究人员对证据的看法。

他不需要成为世界上最好的统计学家。他需要充分理解统计数据,才能提出其他人没有问过的问题。
Ioannidis JPA. PLoS Medicine 2005;2(8):e124. Most-accessed article in journal history.

Selection Bias

谁进入研究 ≠ 目标人群

Example: Trials excluding elderly patients, then applied to nursing homes

Information Bias

Measurement error, recall bias, detection bias

Example: Unblinded outcome assessors rating subjective endpoints

Confounding

Third variable creates spurious association

Example: Vitamin D and COVID—sick people stay indoors

方法论者的问题

“哪些偏见可以解释这一发现 - 他们需要解释多少?”

咖啡和咖啡的故事心脏病:

几十年来,观察性研究表明喝咖啡的人患心脏病的比例更高。指南警告不要喝咖啡。

然后方法学家注意到:喝咖啡的人也抽烟更多。当研究针对吸烟进行调整时,这种关联就消失了。

后来的研究,采用更好的方法,表明咖啡实际上可能具有保护作用。

方法论者的工作不是宣布真相,而是问:“还有什么可以解释这一点?”

50%
Of methodologist time spent communicating with non-methodologists
1
Slide that explains GRADE better than 100 papers
Impact multiplier when clinicians understand methods
The test: 您能用一句话向患者解释“非常低确定性的证据”吗?

“我们对这些结果没有信心——真正的效果可能与研究显示的结果大不相同。”
决策树:您能向临床医生解释置信区间吗?如果是,您能在观察性研究中找出 3 个偏差吗?如果两者都是,那么您就已经准备好基础了。如果您无法识别偏见,请参加研究设计课程。如果您无法解释置信区间,请先学习生物统计学课程。

方法学家技能的四大支柱是:

Scenario
一项研究表明,吃有机食品的人患癌症的几率较低。该研究为观察性研究,有 50,000 名参与者参与。方法学家的第一个问题应该是什么?

选择最佳方法学问题:

技能是基础。现在,您将在哪里构建?

模块 3:职业道路

模块 3:职业道路

条条大路通向方法。每条道路都会塑造您的未来。

Where do methodologists work?

答案可能会让您感到惊讶。

Academia

Universities, research institutes

$80K-$180K

HTA Agencies

NICE, CADTH, IQWIG, regional authorities

$70K-$140K

Cochrane/JBI

证据综合组织

$65K-$120K

WHO/Global Health

International organizations, NGOs

$90K-$160K

Pharmaceutical

行业证据团队、CRO

$100K-$220K

Consulting

Independent or boutique firms

$120K-$300K+

Salary ranges approximate, USD, vary by region and seniority. Source: Industry surveys 2023-2024.

Path 1: Academia

The Profile

Day-to-day: Teaching, research, supervision, grant writing, committee work
Autonomy: 高(您选择您的问题)
Stability: Variable (tenure track vs. soft money)
Impact timeline: Long (publications take years)

Who thrives here: 那些热爱教学、想要研究自由、能够容忍慢节奏和学术政治的人

注意: Publish-or-perish pressure, grant uncertainty, limited practical application
Academic Methodologist, 15 years
"I've published 300 papers. Maybe 10 changed practice. But the students I've trained? They're changing practice every day."

Path 2: HTA Agencies

The Profile

Day-to-day: Reviewing manufacturer submissions, writing assessment reports, guideline development
Autonomy: Medium (structured processes)
Stability: High (government/quasi-government)
Impact timeline: Immediate (decisions affect coverage)

Who thrives here: 那些想要直接政策影响、享受结构化工作、乐于承担责任的人

注意: Political pressure, tight deadlines, less research creativity
NICE Senior Analyst, UK
"Every report I write might determine whether 100,000 patients get a treatment. That focus keeps me sharp."

Path 3: WHO & Global Health

The Profile

Day-to-day: Guideline development, evidence reviews, technical assistance to countries
Autonomy: Medium (diplomatic constraints)
Stability: Medium (contract-based, but renewable)
Impact timeline: Variable (guidelines can take years; emergency response is immediate)

Who thrives here: Those who want global impact, enjoy cross-cultural work, can navigate bureaucracy

注意: Slow decision-making, political considerations, frequent travel
WHO Technical Officer, Geneva
"Our malaria guideline is used by 40 countries. But getting 15 experts to agree on one sentence can take six months."

Path 4: Pharmaceutical Industry

The Profile

Day-to-day: Trial design, HTA submissions, competitive intelligence, medical affairs support
Autonomy: Medium (business objectives)
Stability: Medium (restructuring common)
Impact timeline: Medium (product lifecycle)

Who thrives here: 那些想要高报酬、享受快节奏、适应商业的人背景

注意: 利益冲突认知、“找到正确答案”的压力、较少的出版自由
Director of Evidence Strategy, Global Pharma
“是的,我们希望我们的药物获得批准。但是失败的第三阶段试验设计会花费5亿美元。严谨是好生意。”
Decision tree: Is direct policy impact more important than research freedom? If impact, choose between global scope (WHO/NGO) or national focus (HTA agency). If freedom, choose between compensation priority (Pharma/Consulting) or autonomy priority (Academia).

三位方法论者的故事:

Doug Altman 在牛津工作了 40 年,慢慢建立了 CONSORT、PRISMA 和报告标准——他的影响跨越了数十年。大卫·萨克特离开学术界,直接在汉密尔顿的患者身上实践循证医学——立即发挥临床作用。 Hans Rosling 利用他的流行病学培训创建了 Gapminder,通过数据可视化覆盖数百万人。

Same training, three paths.

Altman 峡谷(深刻、缓慢的影响)、Sackett the mill(日常临床使用)、Rosling 三角洲(广泛的公众影响力)。

您已经发展了证据合成方面的专业知识。选择您的路径:

Path A (Academia): Like Altman—build standards, shape methodology over 40 years.
→ CONSORT, PRISMA, EQUATOR. Deep, slow, lasting impact. You change how science is reported.
Path B (Clinical Practice): Like Sackett—bring evidence directly to patient care.
>→ 立即临床实用。每个患者都受益。您每天都会看到影响。
Path C (Public Communication): 像 Rosling 一样,翻译数以百万计的数据。
>→ Gapminder、TED 演讲、全球影响力。您改变了世界看待数据的方式。

Years 1-5: Foundation

Graduate degree (MPH, MSc, PhD). First job often academic or junior HTA role.

Years 5-10: Specialization

Develop niche expertise. May move sectors. Build publication record and network.

Years 10-15: Leadership

高级角色。经常在旁边咨询。邀请专家提供指导。

Years 15+: Influence

塑造领域。写教科书。培训下一代。

Most senior methodologists have worked in 2-3 sectors. Mobility is an asset.

哪种职业道路通常提供最高的薪酬,但需要适应商业环境?

选择该路径。现在,您如何变得可见?

模块 4:构建您的投资组合

模块 4:构建您的投资组合

In a field of invisible architects, how do you become known?

A methodologist without visibility is a methodologist without impact.

Your work must speak—but first, people must hear it.

未知统计学家的故事

The Methodologist Who Changed Medicine—Anonymously:

In 1996, a statistician named David Moher led a group to create CONSORT—guidelines for reporting randomized trials.

He wasn't famous. He wasn't at Harvard. He was a methodologist at a Canadian research institute.

Today, CONSORT is required by 600+ journals. It's cited 35,000+ times. It's arguably saved more lives than most drugs—by ensuring we know which drugs actually work.

Moher built his reputation not through brilliance alone, but through relentless, visible contribution to a problem everyone faced.

1. Publications

• 优秀期刊中的方法论文

• Collaborative applied papers (methods role)

• 评论和信件

• 教程和操作指南

2. Tools & Resources

• Software packages (R, Stata)

• 检查表和报告指南

• Teaching materials (open access)

• 计算器和决策帮助

3. Presence

• Conference presentations

• 研讨会和培训

• Social media (X/Twitter, LinkedIn)

• 博客文章和时事通讯

4. Network

• Cochrane/Campbell involvement

• Guideline panel memberships

• Journal editorial boards

• Mentorship relationships

工具的力量

metafor
R package by Wolfgang Viechtbauer. 10,000+ citations.
ROB 2
偏见工具的风险。全球 Cochrane 评价标准。
GRADE
Framework by Guyatt et al. Used by 100+ organizations.
工具量表。一篇论文读一次。一个工具会被使用数千次。

如果您能够解决重复出现的问题并将其很好地打包,那么您的影响会成倍增加。

方法论文公式

Problem: Identify a recurring methodological challenge

Solution: Propose a clear, implementable approach

Demonstration: 将其应用于实际数据或模拟

Accessibility: 提供代码、工具或模板

Validation: Show it works better than status quo

Target journals: BMJ、JAMA、Annals of Internal Medicine(方法部分)、医学统计、研究综合方法、临床流行病学杂志

协作乘数

How Georgia Salanti Built a Methods Empire:

Georgia Salanti 不仅仅撰写有关网络荟萃分析的文章,她还与需要的临床团队合作

每次合作都会产生:(1)一篇解决临床问题的应用论文,(2)她下一篇方法论文的方法见解,(3)一位感激的同事将她推荐给其他人。

她的网络呈指数级增长。现在,她领导着欧洲最有成效的证据综合小组之一。

教训:不要等待被问到。向有重要临床问题的团队提供您的方法专业知识。
Salanti G, University of Bern. 500+ publications. Developed CINeMA framework for NMA confidence.

What Works

• Sharing useful content (tutorials, explainers)

• 尊重地参与方法辩论

• Celebrating others' work

• Making complex ideas accessible

• 展示您的学习历程

What Doesn't

• Self-promotion without value

• 公开攻击他人的方法

• Jargon-heavy posts

• Inconsistent presence

• Ignoring engagement

Best methodologist accounts: @EpiEllie, @statsepi, @CochraneCollab, @GRADEWorkingGrp

罗莎琳德·富兰克林的故事:

罗莎琳德·富兰克林的 X 射线照片(照片 51)对于发现 DNA 结构至关重要。但她发表文章十分谨慎,并没有自我推销。

沃森和克里克看到了她的数据,首先发表,并获得了诺贝尔奖。富兰克林于 37 岁时去世,很大程度上没有得到证实。

几十年后,她的笔记本显示她距离自己解决这个结构还有几周的时间。精彩的工作,如果过于安静地分享,就会黯然失色。

Impact requires not just discovery, but communication.

You've made a crucial discovery. What do you choose?

Path A: 谨慎发布——等待完美数据,避免自我推销。
→ Others see your work, publish first. You're remembered as a footnote, vindicated only decades later.
Path B: Share findings actively—engage the community, present at conferences, collaborate openly.
→ Risk being scooped, but ensure your contribution is visible. Credit follows communication.

方法学家提高影响力的最可扩展的方法是:

Visibility without substance is noise. Substance without visibility is silence.

模块 5:工具包

模块 5:工具包

工匠的好坏取决于他们的工具以及对工具的掌握。

What software do methodologists use?

What resources shape their thinking?

R (Essential)

metafor: 荟萃分析

netmeta: Network meta-analysis

robvis: Risk of bias visualization

tidyverse: Data manipulation

Free, open source, reproducible

Stata (Common)

metan/admetan: Meta-analysis

network: NMA

• 生存分析能力强

• Preferred by some HTA agencies

Commercial license required

RevMan (Cochrane)

• 官方Cochrane 工具

• User-friendly interface

• Limited customization

• 适合标准评审

免费进行Cochrane 评审

Python (Growing)

PyMeta: Meta-analysis

• 机器学习集成筛选

• Text mining capabilities

• Automation scripts

免费,有利于AI集成

Covidence

Screening, extraction, collaboration

$$

Rayyan

AI-assisted screening, free tier

Free

ASReview

Active learning screening

Free

EPPI-Reviewer

Full workflow, text mining

$$

DistillerSR

Enterprise, AI features

$$$

SysRev

Open source option

Free

Essential Resources

方法学家书架:

Cochrane Handbook: The gold standard reference. Free online. Read chapters 6-10 closely.

Higgins & Green → Higgins et al.: 定期更新。任何 SR 问题的参考。

Borenstein et al. "Introduction to Meta-Analysis": Best statistical foundation.

Guyatt et al. JAMA Users' Guides: Clinical epidemiology essentials.

IOM "Finding What Works in Health Care": 系统评价标准。

Read one chapter per week. In two years, you'll have a graduate education in methods.

Cochrane

Largest SR producer. Training. Methods groups. Volunteer opportunities.

Campbell Collaboration

Social science SRs. Growing. Less competitive to join.

JBI (Joanna Briggs)

Nursing/allied health focus. Strong training program.

GRADE Working Group

Certainty assessment. Influential. Membership by invitation.

EQUATOR Network

Reporting guidelines hub. Guideline development opportunities.

HTAi

HTA professional society. Annual conference. Interest groups.

弗洛伦斯·南丁格尔图形的故事:

弗洛伦斯·南丁格尔受过基本的统计训练——远远少于她那个时代的大学数学家。但她掌握了一种工具:极地区域图。

她的“花花公子”图表显示,士兵死于感染的人数多于死于战斗受伤的人数。忽略数字表格的议会立即理解了她的图形。

她没有最好的工具。她掌握了她所掌握的知识,并永远改变了军事医学。

您的数据显示士兵死于感染,而不是战斗受伤。议会必须采取行动。您选择什么?

Path A: 呈现数字表格——数据说明一切,显示原始统计数据。
>→ 议会不理会您。对于忙碌的政治家来说,没有可视化的数字是看不见的。
Path B: 创建可视化图表 - 将数据转换为一目了然的图片。
>→ 议会立即理解。政策变化。军队医院进行改革。生命得到拯救。

用于元分析的最广泛使用的 R 软件包是:

学习工具。现在,您将专注于哪些领域?

模块 6:寻找您的利基

模块 6:找到您的利基

通才对一切都有所了解。专家了解某事的一切。方法学家知道要问关于这两种方法的哪些问题。

您无法掌握所有方法。

但您可以成为其中一种方法的世界专家。

Network Meta-Analysis

Indirect comparisons, complex interventions

High demand

Diagnostic Test Accuracy

Sensitivity/specificity, SROC curves

Specialized

IPD Meta-Analysis

Individual patient data, advanced stats

Growing

Living Systematic Reviews

Continuous updating, automation

Emerging

Qualitative Synthesis

Meta-ethnography, framework synthesis

Underserved

偏差工具的风险

工具开发,验证

Influential

Reporting Guidelines

CONSORT, PRISMA extensions

High impact

AI/ML in Evidence Synthesis

Automation, screening, extraction

Frontier

Deep Dive: Network Meta-Analysis

为什么有这个利基?

The problem: Policymakers need to compare 10 drugs. We have trials of A vs B, B vs C, C vs D... but not A vs D directly.

NMA solves: Borrows strength across the network to estimate all comparisons.

The opportunity: HTA agencies require NMA for reimbursement decisions. Pharma needs NMA experts. Few people truly understand it.

The barrier: Requires Bayesian statistics, graph theory, and deep understanding of transitivity. 2-3 year learning curve.

Key names: Salanti, Caldwell, Dias, Welton. Key resource: NICE DSU Technical Support Documents.

深入研究:证据合成中的人工智能/机器学习

为什么有这个利基?

The problem: 手工筛选 50,000 份摘要需要数月时间。从 200 篇论文中提取数据既繁琐又容易出错。

AI/ML offers: 主动学习进行筛选(工作量减少 50-90%)。用于数据提取的 NLP。用于协议起草的 GPT。

The opportunity: 字段是新的。标准还不存在。编写验证框架的人塑造了这个领域。

The barrier: 需要 ML 知识和深入的 SR 方法理解。大多数 ML 人员不了解 SR;大多数 SR 人员不了解 ML。

Key names: Marshall, Wallace, Thomas. Key tools: ASReview, RobotReviewer, Trialstreamer.
决策树:您更喜欢统计复杂性还是流程改进吗?如果是统计学,您对贝叶斯方法感到满意吗?如果是,请考虑 NMA 或 IPD 荟萃分析。如果还没有,请考虑 DTA 或预测模型。如果注重流程,您对技术/编码感兴趣吗?如果是,请考虑人工智能/机器学习或实时系统评论。如果不是,请考虑指南或偏差工具的风险。

Archie Cochrane 的故事:

Archie Cochrane 本来可以继续担任普通医生。相反,他花了几十年的时间问一个问题:“证据在哪里?”

当同事们相信临床直觉时,他支持随机试验。他被嘲笑、被忽视、被称为痴迷。到 1988 年他去世时,他的“痴迷”已成为循证医学的基础。

Cochrane 协作组织以他的名字命名。在一个问题上的深厚专业知识永远改变了医疗保健。

您的同事相信临床直觉而不是试验。您认为证据应该指导实践。您选择什么?

Path A: 接受现状——重视临床经验,遵循既定实践。
>→ 医学仍然以意见为基础。在没有证据的情况下,治疗仍在继续。患者接受的护理是基于传统,而不是证据。
Path B: 坚持要求证据——不断问“证据在哪里?”尽管受到嘲笑。
>→ 被称为痴迷数十年。最终改变医学。全球合作以您的名字命名。

Network meta-analysis is particularly valuable when:

选择利基。现在,向那些走过的人学习。

模块 7:真实旅程

模块 7:真实旅程

Every methodologist was once a beginner who didn't quit.

How did they actually get here?

路径很少是笔直的。

Journey 1: The Clinician Who Asked Why

Gordon Guyatt: From Physician to GRADE Creator

Gordon Guyatt 是一名执业内科医生。 20 世纪 80 年代。他注意到资深医生做出了自信的声明,但他们的证据很薄弱。

转折点: He coined "Evidence-Based Medicine" in 1991. But he didn't stop there. He realized clinicians needed frameworks to assess evidence quality.

The path: MD → Clinical practice → Frustration with dogma → Methodological training → GRADE working group leadership

Today: GRADE 是确定性评估的全球标准。被 WHO、Cochrane、100 多个组织使用。

Lesson: Clinical frustration can become methodological innovation.

Journey 2: The Statistician Who Saw Chaos

Doug Altman: From Numbers to Standards

Doug Altman 接受过统计学家而非临床医生的培训。但当他查看医学期刊时,他看到了统计混乱——错误的测试、误报的结果、不可能的数字。

转折点: His 1994 "Scandal of Poor Medical Research" editorial in BMJ. He could have complained. Instead, he built solutions.

The path: Statistics degree → Medical statistics unit → Frustration with reporting → CONSORT, STROBE, PRISMA, EQUATOR Network

Today: His reporting guidelines changed scientific communication. Cited 500,000+ times across his work.

Lesson: Identify a widespread problem, then systematically solve it.

Journey 3: The Outsider Who Found a Gap

From PhD Student to NMA Pioneer

Georgia Salanti 是希腊的一名博士生,当时网络元分析还只是一种小众好奇心。很少有人理解它。很少有人能教授它。

转折点: 她认识到 NMA 将成为 HTA 的必需品。她将自己定位在前沿。

The path: Statistics PhD → Postdoc at Cambridge → Deep NMA focus → Own research group at Bern → CINeMA framework developer

Today: One of the most cited NMA methodologists globally. Consulted by WHO, NICE, pharmaceutical companies.

课程:找到每个人都需要的新兴方法,并在他们之前掌握它。

旅程 4:区域先驱

在中东建设方法能力

2010 年,系统审查方法中东和北非地区的培训很少。大多数方法学家在欧洲或北美接受过培训。

The pioneers: Clinicians and researchers who sought training abroad, then returned to build local capacity.

The path: Local medical degree → International methods fellowship → Return home → Establish training programs → Become regional experts

Today: Jordan, Lebanon, Iran, Saudi Arabia, Oman have growing evidence synthesis capacity. Regional Cochrane branches emerging.

课程:方法专业知识是可移植的。把它带回家可以倍增你的影响力。

They All Had...

• 挫折变成了使命

• Willingness to learn continuously

• Collaborators who believed in them

• Persistence through rejection

• 随着时间的推移而加深的关注

They Didn't Have...

• Perfect pedigrees

• 从一开始就清晰的道路

• Instant recognition

• 免于挫折

• All skills from day one

Ignaz Semmelweis 的故事:

1847 年,Ignaz Semmelweis 发现洗手可以将孕产妇死亡率从 18% 降低到 2%。医疗机构拒绝了他。

他被解雇、嘲笑,最终被送进精神病院,47 岁时去世。又过了 20 年,细菌理论才证明他的正确性。

塞梅尔维斯没有看到他的影响。但今天每一位进行擦洗的外科医生都遵循他的道路。在他死后,河流流入大海。

You've discovered handwashing saves lives, but colleagues reject it. What do you choose?

Path A: Give up—accept rejection, stop fighting the establishment.
>→ 被遗忘并心碎而死。死后得到平反,但永远看不到你的影响。
Path B: 记录一切——尽管被拒绝,仍坚持下去,确保证据保存下来。
>→ 即使在你有生之年被拒绝,你记录的证据仍然存在。后代继承了真理。

Gordon Guyatt's path to creating GRADE began with:

Best Entry Points

Research Assistant 系统审查小组

Junior HTA Analyst at national/regional agencies

Cochrane/JBI Fellow (competitive but career-defining)

Postdoc 方法学家

What Employers Want

• Evidence of SR/MA experience (even one published)

• R or Stata proficiency

• GRADE or risk of bias training certificates

• Strong writing samples

Pro tip: 自愿担任 Cochrane 综述的筛选员。这是免费培训和出版积分。

The Reading List: Priority Order

Start Here (Month 1-3):
1. Cochrane Handbook — Chapters 6-10 (free online)
2. Borenstein "Introduction to Meta-Analysis" — Statistical foundation

Then (Month 4-6):
3. Guyatt's JAMA Users' Guides — Clinical epidemiology
4. GRADE Handbook — Certainty assessment (free online)

Specialize (Month 7-12):
5. NICE DSU Technical Support Documents — For NMA
6. Ioannidis papers collection — Meta-research classics

One chapter per week = graduate-level education in 2 years.

Regional Pathway: MENA & Global South

在海湾和中东建设方法能力:

The Challenge: Limited local training programs. Most methodologists trained abroad.

The Opportunity: Growing demand. Saudi FDA, Qatar PHCC, UAE DOH all building HTA capacity. First-mover advantage is real.

Pathway:
• Complete international training (McMaster, Oxford, JBI)
• 带着认证和网络返回
• 与区域机构合作
• Become the local expert others consult

Regional conferences: ISPE Middle East, Dubai Health Forum, Gulf HTA Network meetings.

McMaster GRADE/EBM

Gold standard. Online + in-person options.

Essential

Cochrane Training

Free online modules. Certificate tracks available.

Free

JBI Programs

Comprehensive SR training. Strong in qualitative.

Respected

Oxford CEBM

Short courses and MSc options.

Prestigious

Coursera/edX

Epidemiology, Biostatistics foundations.

Affordable

AHRQ EPC Fellowships

US-based. Competitive but career-defining.

Elite

您已经看到了这些旅程。现在,您准备好开始您的了吗?

Final Assessment

模块 8:最终评估

您已经游历了方法的世界。

方法学家的五项原则

1。严格性不是可有可无的——它是信任的基础。

2. Invisible work has visible consequences.

3. Depth in one area enables breadth in impact.

4.工具为方法服务;方法服务于真理。

5.方法学家问道:“还有什么可以解释这一点?”

Doug Altman's CONSORT guidelines have been cited over:

The aprotinin case (15,000-22,000 excess deaths) illustrates:

在评估一项显示咖啡饮用者患有更多心脏病的观察性研究时,方法学家的第一个问题应该是:

对于方法学家来说,扩大其影响力的最可扩展的方式是:

通常提供最高薪酬但需要适应商业环境的职业道路是:

致力于您的发展。完成每个里程碑时勾选它:

您的进度会自动保存到此浏览器。

您已经完成了旅程。

继续前进,为其他人奠定基础。

Remember:
方法是不可见的。影响则不然。
工具是学来的。判断是赢得的。
人迹罕至的路径不那么拥挤是有原因的。无论如何走吧。