本期文章讨论了满足欧洲数据可靠性要求的实践。

A HPLC laboratory for pharmaceutical quality control is used as an example in which data for the batch release of a finished medicinal product is generated. A schematic representation of the underlying process is shown in Figure 1.

以用于药品质量控制的HPLC实验室为例子，其中产生了药品的批次放行数据。底层流程的示意图如图1所示。

Figure 1 Quality control and batch release

图1 质量控制和批放行

What types of data (in accordance with the WHO guidance) are collected in a chromatography laboratory? The most important types are listed below:

色谱实验室需要收集哪些类型的数据（根据WHO指南）？最重要的类型如下：

● data from the tested batch and data on personnel who carry out and control the testing process

● 测试批次的数据以及执行和控制测试过程的人员的数据

● sampling and sample storage data, records and observations

● 取样和样品储存数据、记录和观察

● weighing and sample preparation, standards and reagents used

● 称重和样品制备，使用的标准品和试剂

● qualification and calibration data for the pipettes and balances used

● 所使用的移液器和天平的确认和校准数据

● qualification data for all of the devices used

● 所有使用的设备的确认数据

● instrument control data (detector wavelength range, flow, temperature, etc.)

● 仪器控制数据（检测器波长范围、流量、温度等）

● sequence data in full

● 序列数据完整

● data to be recorded (e. g. data rate, integration parameters, etc.)

● 要记录的数据(例如：数据速率、积分参数等)

● chromatography testing data (initial electronic data, peak areas)

● 色谱检测数据（初始电子数据、峰面积）

● processed data from chromatography testing (processed electronic data)

● 从色谱测试中处理的数据（处理的电子数据）

● measuring process and device-specific calibrations

● 测量过程和设备特定的校准

● device-specific calculations

● 特定于设备的计算

● peak areas after integration

● 积分后的峰面积

● HPLC calibration data

● HPLC校准数据

● calculation data (software-based or completed manually)

● 计算数据（基于软件或手工完成）

● trend analyses

● 趋势分析

● all system suitability test results

● 所有系统适用性测试结果

● reports generated from electronic data (sample-list printouts, chromatograms, etc.)

● 从电子数据生成的报告（样本列表打印输出、色谱图等）

● audit trail data and all deviations and changes

● 审计追踪数据和所有偏差和变更

● documented observations

● 记录到的观察

● if applicable, calculations carried out using external software (LIMS, Excel) = derived data, results (reportable result), evaluation (with OOS, OOE, OOT)

● 如果适用，使用外部软件（LIMS, Excel）进行计算=导出的数据、结果（可报告的结果）、评估（使用OOS、OOE、OOT）

All of this data should comply with the ALCOA principles – and, in an ideal situation, the ALCOA plus principles

所有这些数据都应符合ALCOA原则——在理想情况下，还应符合ALCOA+原则

Figure 2 ALCOA principles of data integrity

图2 ALCOA数据可靠性原则

A number of these requirements are already contained in the EUGMP Guidelines, i.e. they were introduced before the recent publication of data integrity specifications.

其中一些要求已经包含在欧盟GMP指南中，即它们是在最近发布数据可靠性规范之前提出的。

Key aspects of data integrity during the generation of data are examined below. It will be shown that the definition of data is of major importance for a project. The difference between raw data and metadata is not discussed here because the differentiation is difficult. The general term data is used instead. This approach is also taken by the FDA.

下面将研究数据生成过程中数据可靠性的关键方面。这表明，数据的定义对于一个项目是非常重要的。这里不讨论原始数据和元数据之间的区别，因为很难区分。取而代之的是通用术语数据。FDA也采用了这种方法。

A number of conflict situations that often arise in relation to data integrity requirements are examined below.

下面将审查与数据可靠性要求有关的一些经常出现的冲突情况。

● Access to the key functions of the control and evaluation software (e.g. switching off the audit trail, changing the system time) must be restricted, e.g. limited to IT personnel.

● 必须限制访问控制和评估软件的关键功能（如关闭审计跟踪、更改系统时间），如仅限于IT人员。

● Access to control and evaluation software must be based on individual login accounts. Group access or anonymous logins should not be possible. User privileges should be limited to the individual job profile.

● 访问控制和评估软件必须基于个人登录帐户。组访问或匿名登录应该避免。用户权限应该仅限于单个岗位职责。

● A complete qualification and validation of all computers and the control and evaluation systems is absolutely essential.

● 所有计算机、控制和评估系统的完全确认和验证是绝对必要的。

● A review of audit trails should be carried out before data-based decisions are made. This must be completed before the project is concluded and/or the collected data is used for batch release. Actions must be defined and/or put in place for deviations so that an appropriate investigation can be initiated and completed.

● 在做出基于数据的决策之前，应进行审计追踪的审查。这必须在项目结束和/或收集的数据用于批次放行之前完成。必须定义和/或对偏差采取行动，以便启动和完成适当的调查。

● Test runs and the generation of data for testing purposes are not permitted during testing prior to batch or raw material release if these processes are not carried out in accordance with defined protocols during qualification, validation or the system suitability test.

● 如果在确认、验证或系统适用性测试期间，这些过程没有按照规定的方案进行，则在批次或原料放行前的测试期间，不允许测试运行和测试数据的生成。

● There has to be a reason for every subsequent modification, and the reason must be completely and transparently documented. This applies in particular to reintegration and generally to manual integration and changes to manual data entries such as calculation factors, weights and quantities.

● 每一个后续的修改都必须有一个原因，并且这个原因必须被完整且透明地记录下来。这特别适用于重新积分，一般适用于手动积分和手动数据条目的更改，如计算因子、重量和数量。

● There has to be a reason for every repeated analysis, and the reason must be completely and transparently documented. This type of situation is limited to the investigation of OOS, OOE and OOT results as well as deviations, e.g. failed injections. Repeats can also be indicated when a root cause analysis is carried out. For all repeats, prospectively defined processes must be in place and documentation must be mandatory.

● 每一个重复的分析都必须有一个原因，并且原因必须完整和透明地记录下来。这种情况仅限于对OOS、OOE和OOT结果以及偏差的调查，例如进样失败。当执行根本原因分析时，还可以进行重复。对于所有重复，预先定义的过程必须到位，文档必须是强制性的。

● If data lacks robustness and accuracy, e.g. due to an incomplete, unsuitable or old validation or calibration, it may not be used. This can be the case when response factors are used that represent device-specific values, but have not been established as such, or when response factors are not checked and corrected when a device is replaced.

● 如果数据缺乏稳健性和准确性，例如，由于不完整、不合适或旧的验证或校准，它可能不被使用。这种情况可能发生在响应因子被用来表示设备特定的值，但还没有建立成这样的情况下，或者当设备被替换时响应因子没有被检查和校正。

 

 

来源：GMP干货

关键词： QC实验室 数据可靠性