临床研究统计学SOP(英文原稿+中文翻译)

Standard Operating Procedure:
Statistics
Purpose
This standard operating procedure (SOP) describes the statistical procedures involved in clinical research conducted within the University of Birmingham (UoB).
Scope
This SOP applies to all clinical research where the UoB is sponsor or takes on sponsor responsibilities for the statistical procedures. Where clinical research is (co-)sponsored by another institution, this procedure should be followed as far as possible, and in line with the contractual agreement between the UoB and the other institution. This SOP also applies to clinical research approved by a UoB Research Ethics Committee (REC) that is required to follow the UoB Principles of Good Clinical Practice (GCP) for Clinical Research (UoB-GCP-POL-001). This SOP may be used as a guidance document in all other cases.
Implementation plan
This SOP will be implemented directly after its effective date for all clinical trials and any clinical studies that are in set up. For existing clinical research that is already set up and in the recruitment phase (or further) at the time of implementation, this SOP will be implemented within three months of the effective date. Stakeholders
Note that where the UoB takes on the sponsor responsibility for statistics, the UoB will delegate the majority of these duties to the chief investigator (CI) and/or to a clinical trials unit (CTU), who may delegate these duties further to their research team(s). All delegation of duties will be documented e.g. using either the CI declaration and/or the Clinical Trials Task Delegation Log (UoB-SPO-QCD-001).
☐CI: The CI may delegate activities to members of their research team, although evidence of CI involvement and approval is still expected and may not be delegated where ‘no delegation allowed’ is indicated. This SOP will state where delegation is possible. For clinical research approved by a UoB REC, the role of CI may be referred to as the principal investigator (PI), or the supervisor for the postgraduate research student. ☐Statistician: The statistician must possess the relevant statistical knowledge and experience. This may be the CI (or delegate) where they have sufficient knowledge and experience to advise on and/or perform the required statistical elements of the clinical research.
Statistical advisor: Where a statistical advisor is utilised, they may contribute to some, but not all, aspects of the clinical research design, conduct, analysis and reporting. The statistical advisor may take on the responsibility of the ‘statistician’ for specific statistical aspects of the clinical research that they will
perform and will adhere to the relevant sections of this SOP.
Background and rationale
For the purposes of this SOP the terms ‘clinical research’ or ‘project’ will cover clinical trials of investigational medicinal products (CTIMPs), other interventional trials (e.g. surgical trials, device trials and non-CTIMP trials, and any other projects deemed to be ‘interventional’ by the sponsor), clinical studies.
The purpose of this document is to describe the statistical procedures involved in clinical research conducted within the UoB. Statistics is a fundamental component of clinical research. The design, conduct, analysis of the resultant data and reporting must be scientifically sound in order to ensure that the results are reliable, minimise bias, and address the research question.
The statistician takes responsibility for the statistical methods, which may include (but not limited to): the design of the clinical research, protocol development, case report form (CRF) development, database design, input into the data management process, development of the statistical analysis plan (SAP), research project conduct, performance of interim analyses and the final analyses, interpretation of the results and contribution to the publication(s).
* A statistical advisor may take on the responsibility of the ‘statistician’ for specific statistical aspects of the clinical research that they will perform as documented in a memorandum of understanding or collaboration agreement.
Identify statistical support
1.The CI (or delegate) will seek appropriate advice on project design and statistical methodology at the initial
conceptualisation of the clinical research (e.g., at the grant application phase).
o This may be the CI where they have sufficient knowledge and experience to take on the statistician’s role as detailed below under ‘statistician’. In the small number of cases, where statistical review for
determining the sample size is not deemed necessary, the CI (or delegate) will provide suitable
justification for this in the protocol e.g. with reference to other published research using a similar
sample size.
o Where appropriate, the CI (or delegate) will include appropriate funding for statistical support in the grant application.
o The CI may choose to work with a statistical advisor to contribute to some, but not all, aspects of the project design, conduct, analysis and reporting. The statistical advisor may take on the responsibility of the ‘statistician’ for specific statistical aspects of the clinical research that they will perform as
detailed below.
2.The CI (or delegate) will ensure that where the statistical advisor is not, or will not be, a member of the
research team, their responsibilities are clearly detailed in an agreement.
o For internal staff, this will be set out in a non-legally binding memorandum of understanding (e.g. CTU statistician advising an independent Investigator)
o For staff working in different organisations, this will be set out in a legally binding collaboration agreement (e.g. statistician from another organisation).
3.The CI (or delegate) will file evidence that the statistician and/or statistical advisor possesses relevant
statistical knowledge and experience to fulfil their role in the project. See the Training SOP (UoB-TRN-SOP-001).
o It is expected that for clinical trials undertaken to support marketing authorisation applications, or large clinical trials where their publications may change prescribing practice or the standard care, an appropriate qualified and experienced statistician should take responsibility for the statistical aspects of the clinical trial.
4.The CI (or delegate) will document the level of ongoing input required from a statistician. For example, this
may be documented in the protocol, a project-specific risk assessment, or local policies/processes. Project design and protocol development
5.Where applicable to the project, the CI (or delegate) will seek input from a statistician on the following
aspects of the research:
o research design (e.g. observational, randomised clinical trial, inclusion of a feasibility/pilot study) o formulating clinical research objectives
o suitability of outcome measures for meeting the objectives of the clinical research
o methods to minimise bias
o sample size
o randomisation
o stratification/minimisation variables
o blinding
o statistical stopping guidelines (e.g. futility, superiority)
o analysis methods (final and interim)
o the clinical research reporting requirement for registries. See also the Project Closure SOP (UoB-CLO-SOP-001).
6.The CI (or delegate) will maintain documented evidence of the statistician’s involvement in
project/protocol design as applicable (e.g. emails, meeting minutes, comments on proposals/protocols).
o It is recommended that a second check of the statistical section of the protocol is performed prior to finalisation, in particular to check the sample size calculation (if performed), and where appropriate this should be performed by a second statistician.
SAP development
7.The trial statistician will document the pre-specified statistical methodology for the project, either directly
in the protocol and/or in a separate document, typically referred to as a SAP, which will be version-
controlled. See also Protocol Template (CTIMPs)(UoB-ESD-QCD-001) and Protocol Template (non-CTIMPs and studies) (UoB-ESD-QCD-002).
8.The statistician will, as a minimum, include the following details for the primary and secondary outcomes
(or equivalent terms) in the SAP:
o how the outcome will be measured
o appropriate statistical method(s) which will be used to analyse the data.
9.Where appropriate, the statistician will include details on these, and other, analytical aspects in the SAP:
o handling missing data
o handling protocol deviations
o adjustments for multiple testing or multiple comparisons
o pre-specified sub-groups
o rules for calculation of derived variables
o use of baseline data as co-variates in adjusted analysis
o levels of confidence/ statistical significance
o sensitivity analyses
o treatment interactions, particularly for factorial clinical trials.
10.The statistician will approve (each version of) the SAP, and the CI (or delegate) will retain documentation
of this approval (e.g., emails, meeting minutes). It is expected that the first version of the SAP is in place prior to:
o data collection for an open label project
o unblinding the data for a blinded project.
CRF design and data management activities
11.Where appropriate (as per the risk assessment), the CI (or delegate) will document the statistician’s
involvement in developing the CRF, where this is being used to collect data, and the clinical research database. See the Case Report Form Development SOP (UoB-CRT-CRF-SOP-001).
o It is expected for CTIMPs that the statistician will review and formally approve the final version of the CRF, and any amendments which then impact on data items. However, this procedure (where
appropriate) could also be applied in non-CTIMPs/clinical studies to support best practices.
12.Where appropriate (as per the risk assessment), the CI (or delegate) will document the statistician’s
involvement in data management activities. See also the Data Management SOP (UoB-DMA-SOP-001).
o If applicable, it is recommended that the statistician will review the validation process (validation specification), which describes the checks to be performed as part of the data validation plan.
13.The CI (or delegate) will work with the statistician to document all protocol non-compliances and collate
them prior to statistical analysis (where appropriate, prior to unblinding).
14.The statistician will review all significant protocol non-compliances for assessment of the impact on the
analysis.
Amendments
15.The CI (or delegate) will highlight amendments to the project that are likely to have an impact on the
design and analysis of the clinical research, and the statistician will review and document these (as
appropriate as per the risk assessment). This includes:
o review of the impact of the amendment on CRFs and project systems
o review of the impact of the amendment on the SAP
o whether the amendment introduces any risk of bias.
16.Where changes to the project necessitates an update to the SAP, the statistician will amend the SAP
ensuring appropriate version control is employed and changes justified.
Statistical programming, analysis and publication
17.Where appropriate, the CI (or delegate) will perform quality checks, data cleaning and data validation prior
to analysis. For clinical trials, see the Data Management SOP (UoB-DMA-SOP-001) for further details. 18.The statistician will ensure that user-written code/macros and programs used during the analysis process
are fit for purpose (validated). This will be on a risk-based approach with the detail and level of checking varying depending on the item begin checked. For example, output related to the primary objective of the clinical research should be checked.
19.It is expected that hardcoding during the analysis process is avoided. In extreme circumstances where hard
coding is considered necessary, the statistician will retain documentation of its use and provide clear justification for this. For example, use of hard coding in free text fields for adverse event data.
20.Once data collection has been completed, the CI (or delegate) will lock the data from any further changes.
This includes the procedures for unlocking the data for extreme circumstances where changes to the data are required. For qualitative data, an example of this may include converting transcripts to PDF or printing and signing finalised versions of the transcripts. For clinical trials, see the Data Management SOP (UoB-DMA-SOP-001) for further details.
21.The statistician will ensure an appropriate audit trail in the statistical analysis process, in order to allow the
data analysis to be reconstructed from start to finish. An appropriate audit trail means that the final
clinical research results for publication/reports can be traced back to the statistical program output and the datasets used. For example, a table of results including mean difference and 95% confidence interval should be verifiable against information that shows who undertook the analysis, and how and when this analysis occurred. This can be documented, for example, by email or a printout that has been signed and dated.
22.The statistician will apply version control to all outputs of the statistical process.
23.For clinical trials, where appropriate, the statistician will perform interim analyses at the intervals specified
in the protocol (e.g. for data monitoring committee (DMC) meetings) and following the SAP.
o For this, a frozen copy (snapshot) of the dataset will be made for analyses, ensuring the frozen dataset is preserved so that the analysis can be reproduced if required at a later date.
o Where a decision is taken to cancel interim analysis (e.g. due to insufficient participant numbers), this must be documented.
o If the project is blinded, and interim unblinded analysis is required, ensure the clinical research team cannot gain access to unblinded data or the randomisation schedule. See the Randomisation and
Blinding SOP(UoB-RND-SOP-001) for further information.
24.During final analysis, the statistician will ensure statistical packages remain unchanged e.g. by disabling
automatic updates. Where updates have occurred e.g. if analysis undertaken over a period of time, these packages may require re-validation (see point 18 above).
25.The statistician will perform and document checks to ensure that the output of the statistical analysis
process is accurate, and ensure that the SAP has been followed.
26.The statistician will prepare the analytical section of project reports (e.g. DMC, end of project report,
reports to other third parties e.g. ethics, funder etcetera).
27.The statistician will review publications (including DMC reports) prior to submission, ensuring the results
have been correctly transferred into the publication and that they accurately reflect the analysis data, and retain evidence of this review. For randomised clinical trials, the statistician will follow The CONSORT (Consolidated Standards of Reporting Trials) Statement guideline for publication. Also see Extensions of the CONSORT Statement for “non-standard” randomised clinical trial with specific designs, data and
interventions.
o For clinical trials, it is expected that all significant non-compliances that occurred during the trial and how these contributed to the analysis, and any changes to the planned statistical analysis with
justification for these changes are reported.
Essential documentation and archiving
28.The CI (or delegate) will retain adequate documentation to allow verification that data has been accurately
handled and reported in accordance with the UoB Principles of GCP for Clinical Research (UoB-GCP-POL-001). This includes (but is not limited to):
o a SAP (where separate from the protocol)
o an audit trail (see previous section for details).
29.The CI (or delegate) will archive documentation related to statistical processes with the rest of the
study/trial master file (S/TMF) at the end of the clinical research. See the Archiving SOP (UoB-ARC-SOP-001) for further information.
List of expected outputs
☐Documented evidence that the statistician has the relevant statistical knowledge and experience to perform the required tasks.
☐Roles and responsibilities of the statistician are documented, with evidence of the level of ongoing input required and contracts and agreements to be in place (where appropriate).
☐Evidence of the statistician involvement in the project design and protocol development. Where appropriate (as per the risk assessment), documented evidence of the statistician involvement in CRF design, clinical research database and data management activities.
☐Evidence of a process to record and collate all protocol non-compliances and documented evidence that all significant protocol non-compliances have been reviewed by the statistician.
☐Evidence of the pre-specified statistical methodology for the project in the protocol or in a separate document (SAP). As a minimum, it will detail how the primary and secondary outcomes will be measured and what statistical method(s) will be used.
☐Documented evidence of the statistician’s approval of (each version of) the SAP.
☐Where amendments to the project are likely to have an impact on the design and analysis of the clinical research: evidence that the impact of these have been reviewed by the statistician, with any changes justified and appropriate version control applied.
☐Evidence of quality check, data cleaning and data validation processes. For clinical trials, this will be in accordance with the Data Management SOP (UoB-DMA-SOP-001).
☐Evidence that the data has been locked (or a snapshot has been created) prior to the start of the statistical analysis. For clinical trials, this will be in accordance with the Data Management SOP (UoB-DMA-SOP-001). ☐An audit trail in the statistical analysis process in order to allow the data analysis to be reconstructed from start to finish, with appropriate version control applied.
☐Evidence of appropriate checks by the statistician to ensure the outputs of the statistical analysis process is accurate.
☐Evidence that the SAP has been followed.
☐Evidence that the statistician has reviewed publications prior to submission.
☐Evidence that the documentation related to statistical processes has been stored as per the Archiving SOP (UoB-ARC-SOP-001).
Related documents
☐UoB-ARC-SOP-001 Archiving
☐UoB-CLO-SOP-001 Project Closure
☐UoB-CRT-CRF-SOP-001 Case Report Form Development
☐UoB-DMA-SOP-001 Data Management
☐UoB-GCP-POL-001 UoB Principles of GCP for Clinical Research
☐UoB-RND-SOP-001 Randomisation and Blinding
☐UoB-SPO-QCD-001 Clinical Trials Task Delegation Log
☐UoB-TRN-SOP-001 Training
UoB QMS documents can be found on the CRCT website. Internal work instructions can be obtained from the CRCT (XXX) and/or from the RGT (XXX).
References and frameworks
☐The CONSORT (Consolidated Standards of Reporting Trials) Statement guideline
☐Extensions of the CONSORT Statement
☐Medicines and Healthcare products Regulatory Agency (MHRA). Good Clinical Practice Guide, London: The Stationery Office, 2012.
Abbreviations and definitions:
See also the Glossary of Terms.
标准操作程序：
统计学
目的
本标准操作程序（SOP）描述了在伯明翰大学（UoB）内进行的临床研究中涉及的统计程序。

范围
本 SOP 适用于 UoB 作为申办者或承担统计程序申办者责任的所有临床研究。

如果临床研究由另一个机构（共同）赞助，则应尽可能遵循此程序，并符合 UoB 与其他机构之间的合同协议。

本 SOP 也适用于UoB 研究伦理委员会（REC）批准的临床研究，该委员会需要遵循UoB 临床研究良好临床实践原则（GCP）（UoB-GCP-POL-001）。

本 SOP 可用作所有其他情况下的指导文件。

实施计划
该 SOP 将在生效日期后立即对所有临床试验和任何正在建立的临床研究实施。

对于在实施时已经建立并处于招募阶段（或进一步）的现有临床研究，本 SOP 将在生效日期后三个月内实施。

利益相关者
请注意，当 UoB 承担统计的申办者责任时，UoB 会将大部分这些职责委托给首席研究员（CI）和/或临床试验单位（CTU），后者可以将这些职责进一步委托给他们的研究团队。

所有职责委派都将被记录在案，例如使用 CI 声明和/或临床试验任务委派日志（UoB-SPO-QCD-001）。

CI ：CI 可以将活动委托给其研究团队的成员，但仍然需要 CI 参与和批准的证据，并且在表明“不允许委托”的情况下不得委托。

此 SOP 将说明可以委派的位置。

对于 UoB REC 批准的临床研究，CI 的角色可以称为首席研究员（PI），或研究生的导师。

统计学家：统计学家必须具备相关的统计知识和经验。

这可能是 CI（或代表），他们有足够的知识和经验来就临床研究的必要统计元素提供建议和/或执行。

SOP：统计
统计顾问：如果使用统计顾问，他们可能会为临床研究设计、实施、分析和报告的某些方面做出贡献，但不是全部。

统计顾问可以承担“统计学家”的责任，负责他们将要进行的临床研究的特定统计方面，并将遵守本 SOP 的相关部分。

背景和理由
就本 SOP 而言，术语“临床研究”或“项目”将涵盖研究性医药产品（CTIMP）的临床试验、其他干预性试验（例如外科试验、器械试验和非 CTIMP 试验，以及申办方认为“干预性”的任何其他项目）、临床研究。

本文件的目的是描述在 UoB 内进行的临床研究所涉及的统计程序。

统计学是临床研究的基本组成部分。

结果数据的设计、实施、分析和报告必须具有科学依据，以确保结果可靠、最大限度地减少偏见并解决研究问题。

统计学家负责统计方法，其中可能包括（但不限于）：临床研究的设计、方案制定、病例报告表（CRF）的开发、数据库设计、数据管理过程的输入、统计分析计划（SAP）的制定、研究项目的实施、中期分析和最终分析的执行、结果的解释和对出版物的贡献。

* 统计顾问可以承担“统计学家”的责任，负责他们将按照谅解备忘录或合作协议中的记录进行的临床研究的特定统计方面。

确定统计支持
30.CI （或代表）将在临床研究的初始概念化（例如，在资助申请阶段）寻求有关项目设计和统计方法
的适当建议。

这可能是他们有足够的知识和经验来承担统计学家角色的 CI，如下文“统计学家”部分所述。

在少数情况下，认为没有必要进行统计审查来确定样本量，CI （或代表）将在方案中提供适当的理
由，例如参考使用类似样本量的其他已发表的研究。

在适当的情况下，CI （或代表）将在资助申请中提供适当的资金，用于统计支持。

CI 可以选择与统计顾问合作，为项目设计、实施、分析和报告的某些（但不是全部）方面做出贡献。

统计顾问可以承担“统计学家”的责任，负责他们将执行的临床研究的特定统计方面，如下所述。

31.CI （或代表）将确保在统计顾问不是或不会是研究团队成员的情况下，他们的职责在协议中明确详
细说明。

对于内部员工，这将在不具有法律约束力的谅解备忘录中规定（例如 CTU 统计师为独立调查员提供建议）
对于在不同组织工作的员工，这将在具有法律约束力的合作协议中规定（例如来自其他组织的统计人员）。

32.CI （或代表）将提交证据，证明统计师和/或统计顾问拥有相关的统计知识和经验，以履行他们在
项目中的角色。

请参阅培训 SOP （UoB-TRN-SOP-001）。

对于为支持上市许可申请而进行的临床试验，或其出版物可能改变处方实践或标准护理的大型临床试验，应由适当的合格且经验丰富的统计学家负责临床试验的统计方面。

33.CI（或代表）将记录统计师所需的持续输入水平。

例如，这可能记录在协议、项目特定的风险评估
或当地政策/流程中。

项目设计和协议开发
34.在适用于项目的情况下，CI （或代表）将就研究的以下方面寻求统计学家的意见：
研究设计（例如观察性、随机临床试验、纳入可行性/试点研究）
制定临床研究目标
结果测量对满足临床研究目标的适用性
减少偏倚的方法
样本量
随机化
分层/最小化变量
致盲
统计停止指南（例如 funutility、superiority）
分析方法（最终和临时）
注册机构的临床研究报告要求。

另请参见项目关闭 SOP （UoB-CLO-SOP-001）。

35.CI（或代表）将保留统计师参与项目/协议设计的书面证据（如适用）（例如电子邮件、会议记录、
对提案/协议的评论）。

建议在最终确定之前对方案的统计部分进行第二次检查，特别是检查样本量计算（如果执行），并且在适当的情况下，应由第二位统计学家执行。

SAP 开发
36.试验统计学家将直接在方案中和/或单独的文档（通常称为 SAP）中记录项目的预先指定的统计方
法，该文档将进行版本控制。

另见方案模板（CTIMP）（UoB-ESD-QCD-001）和方案模板（非CTIMP 和研究）（UoB-ESD-QCD-002）。

37.统计师至少将在 SAP 中包括主要和次要结果（或等效术语）的以下详细信息：
如何衡量结果
将用于分析数据的适当统计方法。

38.在适当的情况下，统计师将在 SAP 中包括有关这些和其他分析方面的详细信息：
处理缺失数据
处理协议偏差
针对多个测试或多个比较的调整
预先指定的子组
派生变量的计算规则
在调整后的分析中使用基线数据作为协变量
置信水平/统计显著性
敏感性分析
治疗相互作用，特别是对于析因临床试验。

39.统计师将批准 SAP（每个版本），CI （或代表）将保留此批准的文档（例如，电子邮件、会议记
录）。

预计 SAP 的第一个版本在以下时间之前到位：
Open Label 项目的数据收集
取消隐蔽项目的数据。

CRF 设计和数据管理活动
40.在适当的情况下（根据风险评估），CI（或代表）将记录统计学家参与开发 CRF 的情况，用于收集
数据，以及临床研究数据库。

请参阅病例报告表开发 SOP （UoB-CRT-CRF-SOP-001）。

对于 CTIMP，统计师应审查并正式批准 CRF 的最终版本，以及随后影响数据项的任何修订。

然而，该程序（在适当的情况下）也可以应用于非 CTIMPs/临床研究，以支持最佳实践。

41.在适当的情况下（根据风险评估），CI（或代表）将记录统计师参与数据管理活动的情况。

另请参
阅数据管理 SOP （UoB-DMA-SOP-001）。

如果适用，建议统计人员查看验证过程（验证规范），该流程描述了作为数据验证计划的一部分要执行的检查。

42.CI（或代表）将与统计师合作，记录所有协议不合规情况，并在统计分析之前（在适当的情况下，
在揭盲之前）对其进行整理。

43.统计学家将审查所有重要的协议不合规情况，以评估对分析的影响。

修订
44.CI（或代表）将突出显示对项目可能对临床研究的设计和分析产生影响的修改，统计师将审查和记
录这些修改（根据风险评估酌情）。

这包括：
检讨修订对 CRF 和项目系统的影响
审查修订对 SAP 的影响
该修正案是否引入了任何偏倚风险。

45.如果对项目的更改需要对 SAP 进行更新，统计师将修改 SAP，确保采用适当的版本控制并证明更改
的合理性。

统计编程、分析和发布
46.在适当的情况下，CI（或代表）将在分析前执行质量检查、数据清理和数据验证。

有关临床试验，
请参阅数据管理 SOP （UoB-DMA-SOP-001）了解更多详情。

47.统计学家将确保在分析过程中使用的用户编写的代码/宏和程序适合用途（已验证）。

这将基于基于
风险的方法，检查的细节和级别因开始检查的项目而异。

例如，应检查与临床研究的主要目标相关的输出。

48.预计会避免在分析过程中进行硬编码。

在认为有必要进行硬编码的极端情况下，统计学家将保留其
使用记录并提供明确的理由。

例如，在不良事件数据的自由文本字段中使用硬编码。

49.数据收集完成后，CI（或委托人）将锁定数据，防止任何进一步的更改。

这包括在需要更改数据的
极端情况下解锁数据的过程。

对于定性数据，这方面的一个例子可能包括将成绩单转换为 PDF 或打印并签署成绩单的最终版本。

有关临床试验，请参阅数据管理 SOP （UoB-DMA-SOP-001）了解更多详情。

50.统计学家将确保在统计分析过程中进行适当的审计跟踪，以便从头到尾重建数据分析。

适当的审计
跟踪意味着出版物/报告的最终临床研究结果可以追溯到统计程序输出和使用的数据集。

例如，包括均值差和 95% 置信区间的结果表应该可以根据显示谁进行分析以及分析如何以及何时发生的信息进行验证。

这可以通过电子邮件或已签名并注明日期的打印输出进行记录。

51.统计人员将对统计过程的所有输出应用版本控制。

52.对于临床试验，在适当的情况下，统计学家将按照方案中规定的时间间隔（例如数据监测委员会
（DMC）会议）并遵循 SAP 进行中期分析。

为此，将制作数据集的冻结副本（快照）用于分析，确保保留冻结的数据集，以便以后需要时可以复制分析。

如果决定取消中期分析（例如由于参与者人数不足），则必须记录在案。

如果项目是盲法的，并且需要临时非盲分析，请确保临床研究团队无法访问非盲数据或随机化计划。

有关更多信息，请参阅随机化和盲法 SOP （UoB-RND-SOP-001）。

53.在最终分析期间，统计师将确保统计包保持不变，例如禁用自动更新。

如果发生了更新，例如，如
果在一段时间内进行了分析，这些包可能需要重新验证（参见上文第 18 点）。

54.统计师将执行和记录检查，以确保统计分析过程的输出准确无误，并确保遵循 SAP。

55.统计师将准备项目报告的分析部分（例如 DMC、项目结束报告、向其他第三方报告，例如道德、资
助者等）。

56.统计学家将在提交之前审查出版物（包括 DMC 报告），确保结果已正确转移到出版物中，并且它
们准确反映了分析数据，并保留此审查的证据。

对于随机临床试验，统计学家将遵循CONSORT（报告试验综合标准）声明指南进行发布。

另请参阅具有特定设计、数据和干预措施的“非标准”随机临床试验的 CONSORT 声明的扩展。

对于临床试验，预计会报告试验期间发生的所有重大不合规情况以及这些不合规情况对分析的影响，以及对计划统计分析的任何更改以及这些更改的理由。

基本文档和存档
57.CI（或代表）将保留足够的文件，以验证数据是否已根据临床研究 GCP 的 UoB 原则（UoB-GCP-POL-
001）准确处理和报告。

这包括（但不限于）：
一个 SAP （与协议分开）
审计跟踪（有关详细信息，请参阅上一节）。

58.CI（或代表）将在临床研究结束时将与统计过程相关的文档与研究/试验主文件（S/TMF）的其余部
分存档。

有关更多信息，请参阅存档 SOP （UoB-ARC-SOP-001）。

预期输出列表
☐统计师具有执行所需任务所需的相关统计知识和经验的书面证据。

☐统计人员的角色和职责被记录在案，并证明所需的持续输入水平以及要签订的合同和协议（在适当的情况下）。

☐统计学家参与项目设计和协议开发的证据。

在适当的情况下（根据风险评估），统计人员参与 CRF 设计、临床研究数据库和数据管理活动的书面证据。

☐记录和整理所有方案不合规情况的过程证据，以及统计师已审查所有重大方案不合规情况的书面证据。

☐协议或单独文档（SAP）中为项目预先指定的统计方法的证据。

作为最低限度，它将详细说明如何测量主要和次要结果以及将使用哪些统计方法。

☐统计师批准 SAP（每个版本）的书面证据。

☐对项目的修改可能对临床研究的设计和分析产生影响的地方：有证据表明这些修改的影响已经由统计学家审查，任何更改都是合理的，并应用了适当的版本控制。

☐质量检查、数据清理和数据验证过程的证据。

对于临床试验，这将符合数据管理 SOP （UoB-DMA-SOP-001）。

☐在统计分析开始之前数据已被锁定（或已创建快照）的证据。

对于临床试验，这将符合数据管理SOP （UoB-DMA-SOP-001）。

☐统计分析过程中的审计跟踪，以便从头到尾重建数据分析，并应用适当的版本控制。

☐统计师进行适当检查的证据，以确保统计分析过程的输出准确无误。

☐跟踪 SAP 的证据。

☐统计学家在提交之前已审查出版物的证据。

☐与统计过程相关的文档已按照存档SOP （UoB-ARC-SOP-001）存储的证据。

相关文档
☐UoB-ARC-SOP-001 存档
☐UOB-CLO-SOP-001 项目关闭
☐UOB-CRT-CRF-SOP-001 病例报告表开发
☐UOB-DMA-SOP-001 数据管理
☐UOB-GCP-POL-001 用于临床研究的 GCP 的 UoB 原则
☐UOB-RND-SOP-001 随机化和盲法
☐UOB-SPO-QCD-001 临床试验任务委派日志
☐UOB-TRN-SOP-001 培训
UoB QMS 文件可以在 CRCT 网站上找到。

内部工作指导书可以从 CRCT （XXX）和/或 RGT （XXX）获得。

参考资料和框架
☐CONSORT（报告试验综合标准）声明指南。

☐CONSORT 语句的扩展。

☐药品和保健产品监管局（MHRA）。

良好临床实践指南，伦敦：文具办公室，2012 年。

缩写和定义：。