FDA 对第三方实验室报告的接受口径

结论：看的是方法、数据和可追溯性，不是机构的知名度

「我们的报告 FDA 认不认」是常见问题。答案不取决于实验室的名气，也不取决于它在哪个国家——取决于方法是否适用、数据是否可信、以及是否可追溯。

换句话说，判断标准是技术性的。一家不知名但方法正确、记录完整的实验室，出具的报告可能比一家知名机构的一份内容单薄的报告更站得住。

接受口径的关键点

第一行是最基本也最常出问题的。 方法选错了，后面做得再规范也没用。常见情形包括：用了已经被替代的方法版本；用了适用于其他产品类别的方法；或者方法本身适用，但产品的某个特征超出了方法的适用范围。

报告内容的要求

一份能被接受的报告，内容上通常要能回答这些问题：

测的是什么样品——型号、批号、数量、状态、接收日期；用的什么方法——标准编号及版本，或者自建方法的详细描述；在什么条件下测的——环境、设备、参数设定；数据是什么——实测值而不只是合格结论；怎么判定的——判定规则与限值来源；谁做的、谁审核的。

「实测值而不只是合格结论」这一条值得强调。 只写合格的报告，审查时无法判断余量有多大，也无法与其他数据比对。带实测值的报告说服力明显更强。

常见被质疑的情形

样品代表性不足。 送检的样品与申报产品在型号、生产工艺或批次上存在差异，而报告里没有说明这种差异如何处理。

方法偏离未说明。 实际执行时对标准方法做了调整（比如样品尺寸不符合标准要求而做了变通），但报告里没有记录偏离及其影响。

条件记录不足。 报告只给结论和数值，没有试验条件，无法判断数据的可比性。

自建方法缺少确认。 使用非标准方法时，没有提供方法确认的证据。

报告与其他资料矛盾。 报告上的产品参数与技术文件、说明书不一致。

这五类里，前两类是技术性的，后三类是文档性的，而后三类其实更容易避免——只要在委托和出报告时多确认一次。

委托前该确认什么

把项目清单给实验室之后，建议确认这几件事：

哪些项目在其能力范围内，哪些需要分包；使用什么方法及版本；报告会包含哪些内容，能否提供实测值；报告的语言，是否需要英文版；样品需要多少、什么状态；以及大致周期。

这些问题在委托前问，成本是一次沟通；在报告出来后发现不合适，成本是重做。

另外值得问一句：这家实验室做过类似产品吗。做过的话，对方法适用性的判断会更准，也更清楚哪些细节容易被追问。

自建方法的处理

有些性能没有现成的标准方法，需要自建。这种情况下报告要额外提供方法确认的证据：方法的原理与步骤；为什么这个方法适合测这个性能；准确度、精密度等性能参数的验证数据；以及与已知方法或参考物质的比对（如果可能）。

自建方法不是不能用，但举证责任在你这边。 提供充分的确认证据，自建方法同样可以被接受。

报告语言与格式

如果报告是中文出具，用于英文申报时需要翻译。翻译件的处理方式建议提前确认——有些情况下需要说明翻译的准确性。

更省事的做法是直接出具英文报告或中英对照。 这个需求要在委托时提出，报告出具后再补英文版，等同于重新出一份报告。

分包情况的披露

如果报告中部分项目由分包方完成，这一点应当在报告中明确，并说明分包方的资质。隐去分包信息的报告，一旦在审查中被发现，影响的不只是那一项，而是整份报告的可信度。

委托方也应当主动了解分包情况，因为最终对报告负责的是出具方，但申报时被问到实际完成方，你需要答得上来。

报告的时效性

检测报告本身通常不标有效期，但审查时会关注数据是否仍能代表当前产品。距今较久的报告，如果期间产品或工艺发生过变化，其代表性会被质疑。

建议在提交前核对一遍：报告出具之后，产品有没有发生过变更？执行的方法版本是否仍然现行？两者中任一有变化，都要评估是否需要补充数据。

多份报告之间的一致性

一个产品往往有多份报告，来自不同时间、不同实验室。提交前应当横向核对：同一参数在不同报告里的数值是否一致；样品标识是否连贯；方法版本有没有冲突。

不一致的地方要么解释，要么统一。 审查方看到同一参数两个数值而没有说明，一定会问。

语言之外的本地化

报告除了语言，表述习惯也可能造成理解障碍。比如判定用语、量的表达方式、日期格式，不同地区的习惯不同。这些细节不影响技术内容，但会增加阅读成本。

面向境外申报的报告，建议采用国际通行的表述方式，日期用不易歧义的格式、单位用国际单位制、判定用语明确。这些小事做到位，报告的专业感和可读性都会提升。

我们的做法

我们出具的报告会包含方法依据与版本、完整的试验条件、实测数据、设备信息和样品标识。这些不是形式要求，而是报告能不能被接受的实质要件。

对于面向境外申报的委托，我们会在委托阶段确认报告语言和需要体现的信息，避免报告出具后再调整。如果项目中有需要分包的部分，也会事先说明分包方及其能力范围，因为这一点在审查中会被看。

如果你不确定计划的检测方案能否支撑申报，或者对报告形式有特定要求，可以在委托前一起确认，或者直接联系：132 4819 8029。能力范围见服务介绍，流程见检测流程，其他问题见常见问题。

English version

Conclusion. Whether a report will be accepted does not depend on the laboratory's reputation or on which country it sits in. It depends on whether the method is applicable, whether the data are credible and whether everything is traceable. The criteria are technical. A little-known laboratory using the correct method with complete records may produce a report that stands up better than a thin report from a well-known one.

Key points. Method applicability: the method suits the product and the property and is current. Completeness of test conditions: conditions are recorded sufficiently for the work to be reproduced. Traceability of data: equipment, personnel, dates and raw records can be reached. Sample identification: the relationship between what was tested and what is being registered is clear. A supported conclusion: the acceptance rule is explicit and the conclusion follows from the data. And demonstrated laboratory capability for the item in question.

Method applicability is both the most basic and the most frequently mishandled. If the method is wrong, nothing done afterwards rescues it. Typical cases include using a superseded version, using a method intended for a different product category, or using an applicable method on a product whose characteristics fall outside its stated scope.

What the report must contain. An acceptable report generally answers: what was tested, covering model, batch, quantity, condition and receipt date; what method was used, by standard number and version or a full description for an in-house method; under what conditions, covering environment, equipment and parameter settings; what the data are, meaning measured values rather than a bare conformity statement; how the judgement was made, with the acceptance rule and the source of the limit; and who performed and reviewed the work.

The point about measured values deserves emphasis. A report stating only that the item conforms leaves a reviewer unable to judge margin and unable to compare against other data. Reports carrying measured values are noticeably more persuasive.

Frequent grounds for challenge. Inadequate sample representativeness, where the tested sample differs from the registered product in model, process or batch without the difference being addressed. Unexplained method deviation, where the standard method was adapted in practice, for instance because sample dimensions did not suit it, without the deviation and its effect being recorded. Insufficient condition records, leaving comparability unassessable. In-house methods without validation evidence. And reports contradicting other documents, where product parameters differ from the technical file or instructions. The first two are technical; the remaining three are documentary and are actually easier to avoid, needing only one more round of confirmation when commissioning and when the report is issued.

What to confirm before commissioning. After providing the item list, establish which items fall within the laboratory's capability and which will be subcontracted; which method and version will be used; what the report will contain and whether measured values will be included; the report language and whether an English version is needed; sample quantity and condition; and approximate lead time. Asked beforehand these cost one conversation; discovered afterwards they cost a repeat of the work. It is also worth asking whether the laboratory has worked on comparable products, since that improves its judgement of method applicability and its awareness of which details attract questions.

In-house methods. Where no standard method exists, an in-house method may be used, but the report must carry validation evidence: the principle and procedure, why it suits the property, verification of accuracy and precision, and comparison against a known method or reference material where possible. In-house methods are usable; the burden of demonstration simply sits with you.

Language and format. A report issued in Chinese will need translation for an English-language submission, and how the translation is handled is worth confirming in advance, since its accuracy may need to be attested. Issuing in English or bilingually from the outset is simpler, but the requirement must be raised when commissioning, since adding an English version afterwards amounts to reissuing the report.

How we handle it. Our reports carry the method basis and version, complete test conditions, measured data, equipment information and sample identification. These are not formalities; they are the substantive conditions for a report being accepted. For work aimed at overseas submissions we confirm report language and required content at the commissioning stage. Where part of the work will be subcontracted we state the subcontractor and its scope in advance, since that is examined during review.

Send us your planned approach and we will review whether it will support the submission. Phone or WeChat: +86 132 4819 8029.