生物学评价里的文献资料怎么用

结论：文献能不能用，取决于它与你的产品的相关性能否论证

生物学评价中，文献是可以利用的证据来源之一。用得好可以减少试验、缩短周期、降低成本。

但文献不是随便引用的。 关键问题是：这篇文献研究的对象与你的产品有多相关？材料是否相同？接触条件是否可比？研究质量是否可靠？

这几个问题回答不了，引用就站不住。 实务中常见的做法是在报告里列几篇文献然后写「综上，该材料安全」，这样的引用基本上不会被接受。

文献适用的场合

文献比较适合用在这几种情况：

材料的一般毒理学信息。 某种化学物质的毒理学数据，用于风险评估。

材料的长期使用经验。 某类材料在同类应用中的使用历史。

方法学的依据。 说明某个评价方法或思路的合理性。

补充说明。 对试验结果的解释提供支持。

不太适合单独用文献解决的： 产品特有的风险（与具体配方、加工、灭菌相关的）；需要产品本身数据的终点；以及涉及特定使用条件的问题。

一个实用的区分是：越接近「这种化学物质的毒性如何」，文献越有用；越接近「你这个产品安全吗」，文献越不够用。

检索的做法

文献引用要建立在系统检索之上，而不是找到支持自己观点的几篇就停。建议：

明确检索问题。 要回答什么问题，围绕这个问题设计检索。

确定检索来源。 使用哪些数据库、检索哪些类型的资料。

设定检索词与策略。 记录使用的检索词、组合方式、限定条件。

记录检索日期。 文献会更新，检索日期是资料时效的依据。

记录检索结果数量与筛选过程。 检出多少、排除多少、为什么排除、最终纳入多少。

这个过程要可重复。 别人按你记录的策略重新检索，应当得到相近的结果。记录不完整的检索，等于结论不可验证。

筛选与评价

检出的文献要筛选，筛选的标准应当预先设定：

「一致性」这一项要特别注意。 如果检出的文献中有结论不一致的，不能只引用支持自己的那些。应当说明存在不同结论，并解释为什么采信其中一方。 选择性引用是评价质量问题，被发现后对整体可信度的影响很大。

引用时的论证

引用文献时，报告中应当说明：

文献研究的对象。 什么材料、什么形态、什么接触条件。

与本产品的比较。 相同之处、不同之处。

差异的影响分析。 不同之处是否影响结论的适用性。

结论如何支撑本产品的判断。 具体到哪个终点。

局限性。 文献不能覆盖的方面如何处理。

第三步是论证的核心。 文献研究的材料与你的材料总有差异——牌号不同、加工不同、添加剂不同。这些差异是否影响结论，必须说明。

常见的引用问题

只列不论。 附了文献清单但没有说明如何支撑结论。

相关性不足。 文献研究的材料与产品材料只是类别相同。

选择性引用。 只引用有利的结论。

检索不系统。 没有说明检索策略，无法判断是否遗漏。

时效性问题。 引用很旧的文献而未说明当前认识是否改变。

混淆研究类型。 把动物试验的结论直接推到人体，或者把体外结论推到体内，未做必要的说明。

来源不可靠。 引用非同行评议的资料而未说明其局限。

与其他证据的结合

文献很少单独构成完整的证据，通常需要与其他证据结合：

与材料信息结合。 文献说明某物质的毒性，材料信息说明产品中是否含有该物质。

与化学表征结合。 表征说明释放了什么，文献提供这些物质的毒理学数据。这是最常见也最有效的组合。

与使用历史结合。 文献提供一般性信息，使用历史提供本产品或类似产品的实际经验。

与试验数据结合。 文献提供背景，试验提供产品特有的数据。

第二项组合是当前实务中应用较多的路径 ——表征找出物质，文献提供毒理学数据，两者结合做风险评估。

文献的保存与更新

引用的文献应当保存原文，并在技术文件中可追溯。另外：

定期更新检索。 认识会变化，长期在市场上的产品建议定期复核。

关注不利信息。 如果后续出现关于所用材料的不利研究，应当评估影响。

纳入上市后监督。 把材料相关的文献监测纳入上市后的信息收集。

第二项是主动合规的体现。 材料的安全性认识会随研究进展而变化，被动等待监管方提出，不如主动跟踪。

使用历史作为证据

与文献相关的另一类证据是材料的使用历史——某种材料在同类应用中已有长期安全使用记录。这类证据的使用同样需要论证：

材料是否真的相同。 牌号、供应商、配方。

应用是否可比。 接触性质、接触时间、使用人群。

历史数据的来源。 是本企业的记录、行业的公开信息，还是推测。

是否有不良事件。 使用历史中出现过的问题要一并说明。

时间跨度与使用量。 用了多久、用了多少，决定了证据的分量。

本企业自己的使用历史是较有力的证据，因为材料、工艺、应用都可追溯。引用行业的一般经验则较弱，因为具体条件无法核实。

论证的写法示例

一个可用的表述结构是：先说明所引用资料研究的是什么材料、在什么条件下、得到什么结论；再说明本产品使用的材料与之的异同；然后分析差异对结论适用性的影响；最后说明该资料支撑本产品的哪个终点，以及仍需通过其他方式覆盖的部分。

关键是「差异分析」那一段不能省。 只写相同之处而回避差异，论证是不完整的，审查方一眼就能看出来。

我们的做法

在讨论评价方案时，如果委托方计划用文献支撑某些终点，我们会先看文献与产品的相关性。相关性不足的，我们会说明可能的风险 ——引用被质疑之后还是要补试验，时间成本更高。

对于走「化学表征加文献数据」路径的产品，我们会把表征做得尽量充分，因为这条路径的可行性取决于物质是否被完整识别。识别不全，文献数据就无从对应。

如果你在准备生物学评价，想判断哪些终点可以用文献支撑，可以把材料信息和计划发过来一起讨论，或者直接联系：132 4819 8029。能力范围见服务介绍，送检要求见送检要求，流程见检测流程。

English version

Conclusion. Literature is one usable source of evidence in biological evaluation, and used well it reduces testing, shortens timelines and lowers cost. It cannot be cited casually, however. The key questions are how closely the study's subject relates to your product, whether the materials are the same, whether contact conditions are comparable, and whether the study quality is sound. Where those cannot be answered, the citation does not stand. A common practice is listing several papers and concluding that the material is safe, and citations of that kind are essentially never accepted.

Where literature suits. It suits general toxicological information on a material, such as toxicological data for a chemical used in risk assessment. It suits history of use, covering a material class in comparable applications. It suits methodological justification, explaining why an evaluation approach is sound. And it suits supplementary explanation, supporting the interpretation of test results. It suits less well as the sole basis for product-specific risks tied to formulation, processing or sterilisation; for endpoints requiring data on the product itself; and for questions bound to particular conditions of use. A practical distinction: the closer the question is to how toxic a given chemical is, the more useful literature becomes; the closer it is to whether your product is safe, the less sufficient it is.

Searching. Citation must rest on a systematic search rather than stopping once a few supportive papers are found. Define the question the search is to answer and design the search around it. Determine the sources, meaning which databases and which types of material. Set search terms and strategy, recording terms, combinations and limits. Record the search date, since literature changes and the date establishes currency. And record the number of results and the screening process, covering how many were found, how many excluded, why, and how many finally included. The process must be reproducible: someone repeating your recorded strategy should obtain comparable results. A search recorded incompletely leaves the conclusion unverifiable.

Screening and appraisal. Screening criteria should be set in advance. Relevance asks whether the study's material and application relate to the product. Study quality asks about method reliability, sample size and controls. Publication asks about the reliability of the source. Currency asks whether it reflects present understanding. And consistency asks whether multiple papers agree. Consistency deserves particular attention: where retrieved papers disagree, citing only those that support your position is not acceptable. State that differing conclusions exist and explain why one is relied upon. Selective citation is a quality failing, and once noticed it damages the credibility of the whole evaluation.

Justifying a citation. State what the study examined, covering material, form and contact conditions. Compare it with your product, setting out similarities and differences. Analyse the effect of the differences on whether the conclusion transfers. Explain how the conclusion supports your judgement, identifying the specific endpoint. And state the limitations and how aspects not covered are addressed. The third step is the heart of it: the material studied always differs from yours in grade, processing or additives, and whether those differences affect the conclusion must be explained.

Common problems. Listing without reasoning, attaching a bibliography without explaining how it supports the conclusion. Insufficient relevance, where the studied material shares only a general class with the product. Selective citation of favourable conclusions. Unsystematic searching, with no stated strategy so that omissions cannot be judged. Currency problems, citing old work without addressing whether understanding has changed. Confusing study types, extrapolating animal conclusions to humans or in vitro conclusions to in vivo without the necessary explanation. And unreliable sources, citing non-peer-reviewed material without noting the limitation.

Combining with other evidence. Literature rarely constitutes complete evidence alone. Combined with material information, literature gives a substance's toxicity while material information establishes whether the product contains it. Combined with chemical characterisation, characterisation shows what is released and literature supplies the toxicological data, which is the most common and most effective combination. Combined with history of use, literature gives general information and history gives actual experience with this or similar products. And combined with test data, literature gives background and testing gives product-specific results. The second combination is the route most used in current practice: characterisation identifies substances, literature supplies toxicological data, and the two together support risk assessment.

Retention and updating. Retain the full text of cited works, traceable within the technical documentation. Repeat the search periodically, since understanding changes and products long in the market warrant periodic review. Attend to adverse information, assessing the effect if unfavourable research on the materials used subsequently appears. And build literature monitoring on materials into post-market information gathering. Attending to adverse information demonstrates active compliance: understanding of material safety evolves with research, and tracking it is better than waiting for a regulator to raise it.

How we handle it. When discussing evaluation plans, where a client intends to support endpoints with literature we look first at relevance to the product. Where relevance is weak we explain the risk, since a citation that is challenged still leads to testing later at greater cost in time. For products taking the characterisation plus literature route we make characterisation as thorough as possible, because the route depends on substances being completely identified; incomplete identification leaves nothing for the literature data to attach to.

Send us the material information and your plan and we will discuss which endpoints literature can support. Phone or WeChat: +86 132 4819 8029.