证据边界 · Evidence Boundary
“发现了机制”“观察到相关”和“证明对特定人群有效”是不同层级的陈述。
细胞实验可以帮助研究机制，动物实验可以观察复杂生物系统中的变化，人体研究才开始直接回答与人有关的问题。即使进入人体研究，参与人数、对照方式、随访时间和研究终点仍会影响结论的可靠程度。
因此，“在细胞中出现变化”不能直接改写成“能够治疗人类疾病”，“与较好结果相关”也不能自动改写成“导致较好结果”。
研究通常描述一个群体中的平均变化或风险差异。个人的年龄、基础疾病、药物、遗传背景和生活环境可能与研究参与者不同。即使总体结果有意义，也不能保证每个人都获益，更不能忽略不良反应与机会成本。
阅读结果时应同时寻找绝对变化、置信区间、研究限制和不良事件，而不是只看标题里的相对百分比。
WHO 的研究与证据工作强调从研究、证据综合和专家评估到指南与实施的路径。科研发现要进入实际医疗，通常还需要可重复性、质量标准、监管审查、适应证和临床专业判断。
负责任的健康内容应说明证据处于哪一阶段，并把未知项保留下来，而不是用肯定语气填补空白。
Evidence Boundary
Finding a mechanism, observing an association, and demonstrating benefit in a defined population are different levels of claim.
Cell studies can help investigate mechanisms. Animal studies can observe changes in a more complex biological system. Human research begins to answer questions directly related to people. Even then, sample size, comparison groups, follow-up, and chosen outcomes shape how much a study can establish.
A change in cells cannot simply become a claim that something treats human disease. An association with a better outcome does not automatically establish that one factor caused it.
Research often describes an average change or risk difference in a population. An individual's age, medical conditions, medication, genetics, and environment may differ from those of the study participants. A meaningful group result cannot guarantee benefit for everyone or remove the need to consider adverse effects and opportunity costs.
Look for absolute change, uncertainty, study limitations, and adverse events—not only a relative percentage in a headline.
WHO's work on research and evidence describes a path from research and evidence synthesis through expert appraisal, guidance, and implementation. Before findings become routine care, they commonly require reproducibility, quality standards, regulatory review, a defined indication, and professional clinical judgment.
Responsible health content should state the stage of evidence and preserve what remains unknown instead of filling gaps with certainty.