体外诊断试剂的性能评价项目

结论：性能指标回答的是「结果可信到什么程度」

体外诊断试剂的作用是给出检测结果，而临床依据这个结果做判断。所以性能评价的核心问题是：这个结果可信到什么程度。

这个问题被拆解成若干具体的指标：

重复测同一个样本，结果一致吗？ 精密度。

测出来的值与真值接近吗？ 准确度。

在整个测量范围内都准吗？ 线性与可报告范围。

能测到多低的浓度？ 检出限与定量限。

其他物质会不会干扰？ 干扰与交叉反应。

放一段时间后还准吗？ 稳定性。

不同批次之间一致吗？ 批间差异。

这些指标合起来，描述了结果的可信程度。

精密度的评价

重复性。 同一条件下短时间内重复测量的一致性。

中间精密度。 不同日期、不同操作者、不同批次试剂条件下的一致性。

再现性。 不同实验室之间的一致性。

浓度水平。 应当在多个浓度水平评价，特别是医学决定水平附近 ——这些浓度是临床判断的分界点，精密度在这里尤其重要。

样本类型。 用与实际检测相同类型的样本。

「医学决定水平附近的精密度」是评价的重点。 结果在决定水平附近波动，可能导致临床判断的翻转，而在远离决定水平的区域，同样的波动没有临床意义。

准确度的评价

与参考方法比较。 与公认的参考方法对比。

与已上市产品比较。 与同类产品的结果比对。

回收试验。 加入已知量的被测物，看回收率。

使用标准物质。 有证标准物质的测量。

样本量与浓度分布。 覆盖测量范围，样本量足够。

方法比对的统计分析。 线性回归、偏差分析。

「方法比对的样本应当覆盖整个测量范围」，只在中间浓度比对，无法说明高低浓度区域的一致性。

线性与测量范围

线性范围。 在此范围内，测量值与真值成线性关系。

可报告范围。 可以直接报告结果的浓度范围。

稀释线性。 超出范围的样本稀释后测量的准确性。

钩状效应。 极高浓度时结果反而降低的现象，这是危险的失效模式，因为高浓度样本可能报出低值。

「钩状效应」必须验证，特别是免疫类检测。漏检的高浓度样本可能被报为正常，临床后果严重。

检出限与定量限

空白限。 空白样本测量结果的上限。

检出限。 能可靠检出的最低浓度。

定量限。 能以规定精密度定量的最低浓度。

三者的关系。 定量限高于检出限，检出限高于空白限。

「定量限」是实际报告的下限，低于定量限的结果只能定性报告，不能给出具体数值。

干扰与交叉反应

内源性干扰。 样本中常见的干扰物质，比如溶血、脂血、黄疸。

外源性干扰。 药物、抗凝剂、防腐剂。

交叉反应。 结构相似的物质被误测。

评价方式。 在样本中加入干扰物，看结果的偏差。

干扰的可接受限。 偏差多大算有临床意义的干扰。

「溶血、脂血、黄疸」是最常见的内源性干扰，实际样本中经常遇到，应当作为基本的干扰评价项目。

样本相关的要求

样本类型。 声称适用的样本类型都要评价。

抗凝剂的影响。 不同抗凝剂可能影响结果。

样本的稳定性。 采集后到检测前的储存时间和条件。

样本的处理。 离心、分离、稀释的要求。

样本量。 所需的最小样本量。

「样本的储存稳定性」应当评价并在说明书中给出，因为临床中样本不一定立即检测，储存条件不当会影响结果。

稳定性与批间差异

效期稳定性。 在声称的有效期内性能保持。

开瓶稳定性。 开封后的可用时间。

运输稳定性。 运输条件下的性能保持。

机载稳定性。 在仪器上放置的可用时间（自动化检测适用）。

批间差异。 不同生产批次之间的结果一致性。

校准品与质控品的稳定性。 配套品的稳定性同样要评价。

「批间差异」对临床的影响是：换了批号之后结果出现系统性偏移，可能被误读为患者状况的变化。所以批间一致性要严格控制并验证。

性能评价项目一览

这张表可以作为评价方案的检查清单，逐项确认是否已覆盖。

与说明书的对应

评价得到的性能数据应当在说明书中体现：测量范围、精密度、检出限与定量限、已知的干扰物质及限值、样本要求与稳定性、以及储存与有效期。

「已知的干扰物质」要具体列出，实验室在遇到异常结果时需要据此排查。

配套仪器的相关项目

试剂通常与仪器配套使用，仪器相关的项目包括：电气安全与电磁兼容；温控精度（很多反应对温度敏感）；光学系统的稳定性；加样精度；以及软件相关的验证。

「温控精度」对结果影响直接，反应温度偏差会改变反应速率，进而影响测量值。

「加样精度」同样关键，加样量偏差直接导致浓度偏差。

我们的做法

体外诊断试剂的性能评价涉及生物化学和免疫学方法，与我们承担的物理和电气检测不是同一类工作。但配套的仪器设备（分析仪、读数仪）的电气安全、电磁兼容、以及机械环境项目，检测逻辑与其他医用电气设备一致。

我们承接这类仪器时，关注的是基本性能的定义是否包含了测量结果的准确性 ——因为抗扰度测试中，仪器可能不停机但测量结果已经偏移，这一条应当写进判定准则。

如果你有体外诊断相关的仪器设备需要安排电气与环境方面的检测，可以把产品资料发过来一起讨论，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，其他问题见常见问题。

English version

Conclusion. An in vitro diagnostic reagent produces a result on which clinical judgement rests, so the central question in performance evaluation is how far that result can be trusted. The question decomposes into specific measures. Does repeated measurement of the same sample agree, which is precision. Is the measured value close to the true value, which is trueness. Is it accurate across the whole range, which is linearity and reportable range. How low a concentration can be measured, which is limit of detection and quantitation. Do other substances interfere, which is interference and cross-reactivity. Does it remain accurate after storage, which is stability. And do batches agree, which is lot-to-lot variation. Together these describe how far the result can be trusted.

Evaluating precision. Repeatability covers agreement among replicates under the same conditions within a short period. Intermediate precision covers agreement across days, operators and reagent lots. Reproducibility covers agreement between laboratories. Concentration levels should span several points, particularly around medical decision levels, which are the thresholds on which clinical judgement turns and where precision matters most. And sample types should match those actually tested. Precision near medical decision levels is the focus: variation there can flip a clinical judgement, whereas the same variation far from a decision level carries no clinical significance.

Evaluating trueness. Comparison with a reference method compares against a recognised procedure. Comparison with a marketed product compares results against a similar product. Recovery experiments add known quantities of analyte and measure recovery. Certified reference materials provide traceable values. Sample numbers and concentration distribution should span the measuring range with adequate numbers. And statistical analysis of method comparison covers regression and bias analysis. Method comparison samples should span the whole measuring range: comparing only at mid-range says nothing about agreement at high and low concentrations.

Linearity and measuring range. The linear range is where measured and true values relate linearly. The reportable range is where results can be reported directly. Dilution linearity covers accuracy after diluting samples above the range. And the hook effect, where very high concentrations produce lower readings, is a dangerous failure mode because high-concentration samples may be reported low. The hook effect must be verified, particularly for immunoassays: a missed high-concentration sample reported as normal has serious clinical consequences.

Detection and quantitation limits. The limit of blank is the upper value obtained from blank samples. The limit of detection is the lowest concentration reliably detected. The limit of quantitation is the lowest concentration quantified with specified precision. The three relate in order, with quantitation above detection above blank. The limit of quantitation is the practical lower reporting limit: below it, results can only be reported qualitatively rather than as a value.

Interference and cross-reactivity. Endogenous interference covers substances commonly present, including haemolysis, lipaemia and icterus. Exogenous interference covers drugs, anticoagulants and preservatives. Cross-reactivity covers structurally similar substances being measured in error. Evaluation adds interferents to samples and measures the resulting bias. And an acceptance limit defines how much bias constitutes clinically significant interference. Haemolysis, lipaemia and icterus are the commonest endogenous interferents, encountered frequently in real samples, and belong in basic interference evaluation.

Sample requirements. Every claimed sample type must be evaluated. Anticoagulant effects must be assessed, since different anticoagulants can affect results. Sample stability covers storage time and conditions between collection and testing. Sample handling covers centrifugation, separation and dilution requirements. And sample volume covers the minimum required. Sample storage stability should be evaluated and stated in the instructions, since clinical samples are not always tested immediately and inappropriate storage affects results.

Stability and lot variation. Shelf life stability covers performance retention through the claimed period. In-use stability covers the period after opening. Transport stability covers performance under shipping conditions. On-board stability covers the usable period on an analyser. Lot-to-lot variation covers agreement between production batches. And calibrator and control stability must be evaluated for the accompanying materials. Lot variation matters clinically because a new lot may shift results systematically, which can be misread as a change in the patient's condition, so lot consistency must be controlled and verified rigorously.

How we handle it. Performance evaluation of diagnostic reagents involves biochemical and immunological methods and is not the same work as the physical and electrical testing we undertake. The accompanying instruments, however, including analysers and readers, follow the same testing logic as other medical electrical equipment for electrical safety, electromagnetic compatibility and mechanical and environmental items. For such instruments we attend to whether the essential performance definition encompasses measurement accuracy, because during immunity testing an instrument may continue running while its results have already shifted, and that belongs in the acceptance criteria.

Send us the product documentation and we will discuss the electrical and environmental scope. Phone or WeChat: +86 132 4819 8029.