结论：证据决定了召回范围，而范围决定了成本和风险

发现产品存在质量问题时，最关键的判断是：问题涉及哪些产品。

范围判小了，有问题的产品留在市场上，风险继续存在；范围判大了，不必要的产品被召回，成本和影响都放大。

而范围的判定需要证据： 问题的根本原因是什么、哪些批次受影响、影响的程度如何。这些问题都要靠检测数据来回答。

没有数据的范围判定，只能凭推测，而推测的结果通常是「宁可扩大」，代价很高。

证据在各环节的作用

环节 需要的证据
问题确认 确认问题真实存在及其表现
原因分析 找到根本原因
范围判定 确定受影响的产品范围
风险评估 评估对使用者的风险程度
措施验证 验证纠正措施有效
效果确认 确认问题不再发生

「范围判定」和「措施验证」是最依赖检测数据的两环。

需要的数据类型

失效样品的分析数据。 出问题的产品，失效机理和原因。

同批产品的检测数据。 同批次的其他产品是否有同样问题。

其他批次的对比数据。 问题是否只限于特定批次。

原材料的检测数据。 如果原因指向材料。

过程记录与检验数据。 生产环节的参数和检验结果。

设计验证数据。 如果原因指向设计。

整改后产品的验证数据。 措施是否有效。

「其他批次的对比数据」是判定范围的直接依据。 如果问题只在某个时间段生产的批次中出现，而前后批次正常，范围就可以收窄到该时间段。

范围判定的方法

从原因反推。 原因是什么，什么条件下会产生这个原因，符合这些条件的产品就在范围内。

按批次追溯。 通过材料批次、生产批次、时间段来界定。

抽样验证。 对边界批次抽样检测，确认范围的边界。

逐台排查。 数量不多且风险高时，可以逐台检查。

保守处理。 界限不清楚时按较大范围处理。

「对边界批次抽样」是收窄范围的有效做法。 怀疑某个时间段的产品有问题，就对该时间段前后的批次抽样检测——如果边界外的产品正常，范围就可以确定下来，而不必扩大。

根本原因的确定

范围判定依赖原因，而原因的确定需要严谨：

区分直接原因与根本原因。 零件断了是直接原因，为什么断才是根本原因。

排除其他可能。 不只找到一个能解释的原因，还要排除其他可能。

验证原因。 如果可能，用试验复现问题，验证原因成立。

多因素的处理。 如果是多个因素共同作用，各因素的贡献要分清。

「用试验复现」是验证原因的有力方法。 按推断的原因条件做试验，如果能复现同样的失效，原因的成立就有了直接证据。

风险评估

失效的后果。 可能造成什么伤害。

失效的概率。 受影响产品中有多大比例会失效。

可探测性。 使用者能否在失效造成后果前发现。

暴露的人群。 多少使用者在使用、使用者的特征。

已发生的事件。 是否已经造成伤害。

风险评估的结论决定措施的力度 ——是召回、是现场整改、是发布通知、还是加强监测。

措施的验证

采取措施之后，要验证措施有效：

整改产品的验证。 整改后的产品是否解决了问题。

新生产产品的验证。 改进后的工艺或设计是否有效。

长期验证。 某些问题需要一段时间才能确认是否解决。

监测的持续。 措施实施后继续监测同类问题的发生。

「长期验证」对涉及耐久或老化的问题很必要。 改进了材料或工艺，短期测试正常，但问题本来就是长期使用后才出现的，需要耐久试验来验证。

记录与沟通

记录的完整性。 从发现问题到关闭的全过程记录。

证据的保存。 失效样品、检测报告、分析记录。

决策的依据。 每个决策（范围、措施）基于什么证据。

对外沟通的准确性。 向监管方、客户、使用者的说明要准确。

内部的知识沉淀。 问题和处理经验纳入设计和工艺的改进。

「决策的依据」要记录清楚。 将来被问到为什么这样判定范围时，能拿出依据。

预防的角度

从事件中学到的东西应当用于预防：

设计改进。 针对根本原因的设计变更。

工艺改进。 制造环节的控制加强。

检验加强。 在检验中增加相应项目。

风险文件更新。 新发现的失效模式纳入分析。

追溯能力的改进。 如果本次追溯困难，改进批次管理。

「追溯能力」是事件中常暴露的短板。 批次记录不完整、材料批次与产品批次对应不清，都会导致范围判定困难而被迫扩大范围。

时效上的考虑

质量事件的处理有时效压力，检测安排上要相应调整：

分阶段出结果。 先给出初步判断，不必等完整分析。

优先做关键项。 先做能支撑范围判定的项目。

并行安排。 多个样品、多个项目同时进行。

及时沟通。 中间结果及时反馈，便于决策。

保留复核的可能。 初步结论后续可能需要修正，样品和数据要保留。

「先给初步判断」在事件处理中价值明显，因为决策不能等，而初步判断能支撑第一步的行动。

与供应链的配合

如果问题指向外购件，需要供方配合：

供方的同批次留样检测；供方的过程记录；供方的材料证明；以及供方的其他客户是否有类似问题。

「供方留样」的价值在于它是未经使用的同批产品，与失效样品对比能说明问题是出厂时就有还是使用中产生。

我们的做法

承接质量事件相关的检测委托时，我们理解这类工作的时效要求——范围判定拖延一天，风险和成本都在累积。 我们会优先安排，并在有初步结果时及时沟通，不必等完整报告。

范围判定方面，我们建议对边界批次做抽样对比测试，在相同条件下同时测量不同批次的样品，差异直接反映批次之间的区别，这比分别测试再比较更可靠。

如果你遇到质量事件需要检测支持，建议先保全样品和记录，然后联系：132 4819 8029。检测能力见服务介绍，产品分类见手动轮椅检测与电动轮椅检测，案例见案例。

English version

Conclusion. When a quality problem emerges, the critical judgement is which products are affected. Judge the scope too narrowly and affected product remains in the market with the risk continuing; judge it too broadly and unaffected product is recalled, magnifying cost and consequence. Determining scope requires evidence: what the root cause is, which batches are affected and to what degree. All of those are answered by test data. Without data, scope is judged by supposition, and supposition generally errs towards wider, which is expensive.

Where evidence is needed. Problem confirmation needs evidence that the problem genuinely exists and how it manifests. Cause analysis needs evidence of the root cause. Scope determination needs evidence of which products are affected. Risk assessment needs evidence of the degree of risk to users. Measure verification needs evidence that corrective action works. And effectiveness confirmation needs evidence that the problem no longer occurs. Scope determination and measure verification depend most heavily on test data.

Data types needed. Analysis of failed samples, establishing the mechanism and cause. Testing of other units from the same batch, establishing whether the problem is shared. Comparative data from other batches, establishing whether the problem is confined to particular ones. Raw material test data where the cause points to materials. Process records and inspection data covering production parameters and results. Design verification data where the cause points to design. And verification of remediated product, establishing whether the measure works. Comparative data from other batches is the direct basis for scope: where the problem appears only in batches made during a certain period while those before and after are sound, the scope narrows to that period.

Determining scope. Reason back from the cause: what the cause is, what conditions produce it, and which products met those conditions. Trace by batch, delimiting through material batches, production batches and time periods. Sample the boundary, testing batches at the edges to confirm where the boundary lies. Inspect individually where quantities are small and risk is high. And handle conservatively where the boundary is unclear. Sampling the boundary is the effective way to narrow scope: suspecting product from a certain period, test batches immediately before and after, and if those are sound the scope is established rather than expanded.

Establishing root cause. Since scope depends on cause, the cause must be established rigorously. Distinguish immediate from root cause: a part fractured is the immediate cause, and why it fractured is the root. Exclude other possibilities rather than settling for one explanation that fits. Verify the cause, reproducing the problem experimentally where possible. And separate contributions where several factors combine. Experimental reproduction is a strong verification: testing under the conditions the inferred cause implies and reproducing the same failure gives the cause direct evidential support.

Risk assessment. Consider the consequence of failure, meaning what harm may result. Consider the probability, meaning what proportion of affected product will fail. Consider detectability, meaning whether users can notice before harm results. Consider exposure, meaning how many users and what their characteristics are. And consider events already occurred, meaning whether harm has been caused. The risk assessment determines the strength of the measure: recall, field remediation, notice, or enhanced monitoring.

Verifying measures. After acting, verify effectiveness. Verify remediated product, confirming the problem is resolved. Verify newly produced product, confirming the improved process or design works. Verify over time, since some problems need a period before resolution can be confirmed. And continue monitoring for recurrence after implementation. Long-term verification is necessary for durability and ageing problems: an improved material or process may test normally in the short term while the problem only appeared after prolonged use, so durability testing is needed.

Records and communication. Keep complete records from discovery to closure. Preserve evidence including failed samples, test reports and analysis records. Record the basis for each decision on scope and measures. Communicate accurately to regulators, customers and users. And capture the learning internally, feeding it into design and process improvement. Record decision bases clearly: when asked later why the scope was judged as it was, you need to produce the reasoning.

The preventive perspective. Use what the event teaches. Improve the design against the root cause. Improve the process with tighter control. Strengthen inspection with additional items. Update the risk file with newly identified failure modes. And improve traceability if tracing proved difficult this time. Traceability is a weakness events commonly expose: incomplete batch records and unclear correspondence between material and product batches make scope determination difficult and force it wider.

How we handle it. For quality event testing we understand the urgency: every day scope determination is delayed, risk and cost accumulate. We prioritise such work and communicate preliminary results as they emerge rather than waiting for a complete report. On scope we suggest comparative sampling across boundary batches, measuring samples from different batches under the same conditions at the same time, since the difference then reflects the batches directly and is more reliable than separate testing followed by comparison.

If you face a quality event needing testing support, preserve samples and records first and then get in touch: +86 132 4819 8029.