Clinical Trial Protocol Checklist: What to Include, What Regulators Require, and Where Teams Fall Short

Seventy-six percent of Phase I through IV trials now require at least one protocol amendment, up from 57% in 2015, according to a 2024 analysis of 950 protocols by the Tufts Center for the Study of Drug Development ^[1]. Amendment drivers range widely: regulatory authority requests, evolving study strategy, and emerging safety signals account for a significant portion. But a meaningful share also traces back to design decisions left unresolved before the protocol was finalized, including imprecise eligibility criteria, underspecified endpoints, and assessment schedules that sites cannot operationally execute ^[1].

A structured protocol checklist, applied during drafting and before IND submission, is one tool for catching those resolvable gaps early. This article covers what a complete clinical trial protocol must contain under current FDA and ICH requirements, how international standards like SPIRIT 2025 and ICH M11 CeSHarP structure these obligations, and where protocol teams most often leave gaps.

Why Protocol Completeness Matters More Than It Did a Decade Ago

Protocol complexity has grown faster than many teams' protocol-review processes. Since 2005, the number of procedures per protocol has increased 139%, the number of endpoints has risen 214%, and data points collected have surged approximately 600%, according to IQVIA analysis drawing on Tufts CSDD benchmarking data ^[2]. Phase III protocols now average nearly 5.96 million data points collected per study, up from 930,000 in 2012 ^[3].

More content does not equal better protocols. A 2025 study by Tufts CSDD and 14 TransCelerate member companies found that roughly one-third of procedures in clinical trial protocols do not directly support primary objectives or key secondary endpoints ^[4]. That non-core data accounted for 25 to 30 percent of total burden on sites and participants. In non-oncology trials, the figure reached 40% of patient-level data ^[4].

The consequences run downstream across the entire study. Tufts CSDD data presented at the 2024 SCOPE Summit showed that from 2010 to 2020, substantial amendments rose 113.3% across the industry, enrollment duration extended by 36.9%, and protocol-approval-to-first-patient timelines lengthened by 27.2% ^[5]. Kenneth Getz, Executive Director of Tufts CSDD, observed that protocol complexity correlates directly with more deviations, more amendments, slower enrollment, and higher dropout rates ^[6].

This is the context in which a protocol checklist operates. It is not an administrative convenience. It is a quality-by-design tool.

The Regulatory Framework Behind Protocol Requirements

FDA: 21 CFR Part 312

In the United States, clinical trial protocols submitted as part of Investigational New Drug (IND) applications must satisfy the content requirements specified under 21 CFR 312.23(a)(6) ^[7]. The regulations are explicit that Phase 1 protocols may be less detailed than Phase 2 and Phase 3 protocols, but they must still include an estimate of the patient population, a description of safety exclusions, a dosing plan, and detailed specifications for any elements critical to participant safety, including monitoring of vital signs and blood chemistries ^[7].

For Phase 2 and 3 studies, 21 CFR 312.23(a)(6)(iii) requires considerably more: a statement of objectives and purpose, a description of the study design (including control groups), the methods for minimizing bias, the procedures for participant selection and exclusion, a description of the observations to be made, a description of clinical procedures, laboratory tests, or other measures to monitor participants for safety, and a description of how data will be collected and managed ^[7].

The FDA's guidance on Phase 1 IND content reinforces that even early-phase protocols must specify stopping rules, dose adjustment criteria, and toxicity-based monitoring requirements ^[8].

ICH E6(R3): Good Clinical Practice

ICH E6(R3), finalized by the ICH Assembly in December 2024 and published by FDA in September 2025 with the EMA adopting it for effect from July 23, 2025, introduces a quality-by-design framework intended to shape how protocols are constructed from the outset ^[9]. The guideline recommends that sponsors apply risk-proportionate thinking during protocol development, identifying factors critical to trial quality before the document is drafted rather than as a corrective exercise after problems emerge ^[9],[11].

ICH E6(R3) explicitly recommends that the protocol and supporting documents, including the statistical analysis plan and data management plan, be clear, concise, and operationally feasible ^[10]. That language has direct implications for checklist practice. A protocol checklist aligned with E6(R3) does not simply verify that sections exist; it asks whether each element is written in a way that sites can actually execute.

ICH M11 CeSHarP: The Digital Protocol Standard

The ICH M11 Clinical Electronic Structured Harmonised Protocol (CeSHarP) template was adopted at Step 4 by the ICH Assembly in November 2025, with final Step 5 adoption by EMA/CHMP on December 11, 2025, and came into effect on June 11, 2026 ^[11]. FDA published the final M11 Template guidance in the Federal Register in May 2026 ^[12]. The template provides standardized content structure, formatting, common headers, and instructions for every section of an interventional clinical trial protocol. It is designed to work across all phases and therapeutic areas and includes both required and optional components ^[13].

The significance of CeSHarP for protocol checklists is structural. Because M11 specifies where each element should appear within a harmonized document, a protocol authored against the CeSHarP template is organized in a way that regulatory reviewers at FDA, EMA, and other ICH-region agencies expect to navigate. Sponsors using the template know in advance which fields are mandatory and which are conditional on study type. M11 does not eliminate regional or submission-specific requirements; it provides a common structural and data-exchange foundation designed to reduce reformatting burden across authority submissions ^[13].

SPIRIT 2025: Updated Academic and Industry Standard

The SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) 2025 statement, published in Nature Medicine in April 2025 following a Delphi consensus process involving 317 individuals and 30 experts, updated the original 2013 guideline into a 34-item checklist for randomized trial protocols ^[14]. Notable additions include a new open science section, expanded requirements for harm assessment and description of interventions and comparators, and a new item on patient and public involvement in trial design and conduct ^[14].

SPIRIT 2025 builds on the 2013 statement, which EQUATOR Network reported was endorsed by over 150 medical journals and widely recognized by national funders ^[15]. The 2025 update expanded the checklist to 34 items, adding open science, enhanced harm reporting, and patient and public involvement requirements. Growing expectations from journals, research ethics committees, and funders for pre-registered, transparent protocols make alignment with SPIRIT a practical consideration for sponsors preparing for publication.

The Protocol Checklist: Section by Section

The following checklist consolidates obligations from 21 CFR 312.23, ICH E6(R3), ICH M11 CeSHarP, and SPIRIT 2025. Each section represents a functional area of the protocol, not merely a document heading.

1. Administrative and Identification Information

• Descriptive title identifying the study design, population, and interventions ^[14]
• Protocol version number and date ^[14]
• Trial registration identifier and registry name (ClinicalTrials.gov or equivalent WHO-recognized registry) ^[14]
• Sponsor name and contact information ^[14]
• Protocol contributor names, affiliations, and roles ^[14]
• Funding sources and types of support ^[14]
• Roles and responsibilities of sponsor, funder, steering committee, data management team, and endpoint adjudication committee if applicable ^[14]

2. Background and Scientific Rationale

• Summary of the condition under study, including prevalence, clinical burden, and existing treatment options
• Summary of available preclinical and clinical evidence for the investigational product, including relevant pharmacology and toxicology data ^[7]
• Justification for the study design, including the choice of control, endpoints, and comparator
• Risk-benefit assessment at the time of protocol finalization, consistent with ICH E6(R3) quality-by-design principles ^[9]

Failure to ground the rationale in current evidence is a common protocol weakness. Review committees, IRBs, and regulatory agencies will scrutinize whether the scientific justification supports the proposed risk to participants.

3. Objectives and Endpoints

• Primary objective, stated in terms that directly correspond to the primary endpoint
• Primary endpoint, with a precise definition, measurement method, time point, and analysis population
• Secondary objectives and secondary endpoints, each with the same level of precision as the primary
• Exploratory or tertiary endpoints, if applicable, clearly labeled as such
• Hierarchy of endpoints and pre-specified multiplicity adjustments for trials with multiple endpoints; ICH E9(R1) provides a framework for precisely aligning endpoint definitions with the estimand for each objective ^[16],[22]

The ICH E9(R1) addendum on estimands and sensitivity analyses, adopted at Step 4 by the ICH Assembly in November 2019 and subsequently implemented by FDA and EMA ^[22], provides a framework for precisely defining the treatment effect to be estimated in a trial ^[16]. Protocols that leave endpoint definitions misaligned with their stated estimands create statistical analysis plan ambiguity that surfaces only after the protocol is locked.

4. Study Design

• Study phase (Phase 1, 2, 3, or 4)
• Overall design description: randomized or non-randomized, controlled or uncontrolled, open-label or blinded, parallel or crossover, single-center or multicenter
• Method of randomization: sequence generation, allocation concealment procedure, implementation ^[14]
• Blinding arrangements: who is blinded, mechanism for maintaining the blind, conditions for unblinding, and unblinding procedures ^[14]
• Adaptive elements, if applicable, with pre-specified decision rules and stopping criteria
• Duration of participant involvement: screening window, treatment period, follow-up window, and total study duration

5. Population: Eligibility Criteria

• Inclusion criteria, written as concrete, operationally testable requirements
• Exclusion criteria, with each criterion tied to a specific participant safety concern or scientific rationale
• Criteria for early withdrawal or discontinuation from study treatment
• Criteria for study discontinuation at the site or trial level

Eligibility criteria are a leading driver of protocol amendments. Getz et al. 2024 identified changes to inclusion and exclusion criteria as a consistently common trigger for post-approval amendments ^[1]. Overly restrictive criteria slow enrollment; overly broad criteria may admit participants outside the intended population. Both problems become expensive after IND submission.

6. Investigational Product and Study Treatments

• Name, formulation, dose, dose route, and dosing schedule for each investigational product
• Permitted and prohibited concomitant medications
• Dose modification or interruption rules, tied to specific toxicity grades or clinical criteria
• Storage, handling, and accountability procedures for the investigational product
• Placebo description and manufacturing controls, if a placebo-controlled trial ^[7]
• Provisions for investigational product supply to participants in decentralized elements, if applicable

7. Study Procedures and Schedule of Assessments

• Complete schedule of assessments in tabular format, with visit windows for each time point
• Description of each assessment: what is measured, how it is measured, by whom, and under what conditions
• Procedures for managing missed visits, delayed assessments, or out-of-window visits
• Source data verification expectations for each assessment category

This section requires operational scrutiny. Every assessment in the schedule of assessments adds to site and participant burden. Under ICH E6(R3) quality-by-design principles, sponsors should verify that each assessment directly supports a protocol objective or a safety monitoring requirement before it is included ^[9]. The TransCelerate/Tufts CSDD data found that nearly one-third of collected data does not serve this purpose ^[4].

8. Statistical Considerations

• Sample size calculation, including assumptions for effect size, variability, and power
• Primary analysis method, specified in advance, with the analysis population (intent-to-treat, per-protocol, or modified intent-to-treat)
• Pre-specified handling of missing data
• Statistical analysis plan reference or summary, including planned sensitivity analyses
• Interim analysis plans with decision rules and alpha-spending approach, if applicable
• Stopping rules for efficacy, futility, or safety

For Phase 2 and Phase 3 studies, 21 CFR 312.23(a)(6)(iii) requires that the protocol describe the observations to be made and provides the framework within which a statistical analysis plan should operate. The level of analytical detail required expands substantially for studies submitted in support of marketing applications, where the primary efficacy endpoint, analysis population, and handling of missing data must be pre-specified ^[7].

9. Safety Monitoring and Reporting

• Definition of adverse events (AEs), serious adverse events (SAEs), adverse events of special interest (AESIs), and suspected unexpected serious adverse reactions (SUSARs)
• Reporting timelines for each category: FDA requires expedited reporting of unexpected serious adverse reactions within 7 or 15 calendar days depending on fatality or life-threatening status under 21 CFR 312.32 ^[17]
• Site-level deviation reporting plan: who is notified, on what timeline, through what system
• Data Safety Monitoring Board (DSMB) or Safety Review Committee charter, if applicable, including meeting frequency, access to unblinded data, and voting rules
• Stopping rules for individual participants based on safety signals

10. Ethical and Regulatory Compliance

• IRB or IEC review process, consistent with 21 CFR Part 56
• Informed consent process: how and when consent will be obtained, documentation requirements, re-consent triggers
• Privacy and data protection provisions: HIPAA compliance (US), GDPR provisions (EU), or applicable national regulations
• Trial registration obligations on ClinicalTrials.gov for applicable clinical trials under FDAAA 801 ^[23]; sponsors conducting trials under other jurisdictions should confirm registration requirements with the applicable WHO-recognized registry
• Compliance with the Declaration of Helsinki ethical principles for medical research involving human participants ^[25]

11. Amendments and Protocol Deviations

• Definition of major and minor protocol deviations, with examples relevant to the specific study design
• Deviation reporting timeline: study-specific deviation reporting windows vary by protocol; sponsors should define the expected notification timeline to the CRO or sponsor and require sites to notify the IRB/IEC per local regulatory requirements
• Amendment process: criteria for substantive versus administrative amendments, IND amendment submission requirements under 21 CFR 312.30
• Roles and responsibilities for deviation corrective action plans

12. Data Management

• Data collection methods: electronic data capture (EDC) system identification, paper backup procedures if applicable
• Source data definition: what constitutes source data in the study, consistent with ICH E6(R3) principles ^[9]
• Data cleaning and query resolution timelines
• Audit trail requirements under 21 CFR Part 11 for electronic records ^[18]
• Data transfer and archival requirements at study close

13. Quality Assurance and Monitoring

• Monitoring approach: on-site, remote, or risk-based monitoring per ICH E6(R3) ^[9]
• Monitoring frequency and triggers for increased monitoring intensity
• Quality control procedures at the site level
• Audit procedures and sponsor's right to audit

ICH E6(R3) specifically recommends that monitoring plans be proportionate to the risks identified in the risk assessment conducted during protocol development ^[9]. Sponsors who apply uniform monitoring intensity regardless of study risk profile may not be aligned with E6(R3) expectations and face greater regulatory scrutiny during inspections.

14. Publication and Data Sharing

• Authorship and publication policy
• Plans for data sharing, where applicable (addressed under ICH E6(R3) transparency principles ^[9] and as a new open science item under SPIRIT 2025 ^[14])
• Plans for registering results on ClinicalTrials.gov within 12 months of primary completion, per FDAAA 801 and 42 CFR 11.44 requirements for applicable clinical trials ^[23],[24]

SPIRIT 2025 added an explicit open science section to the protocol checklist, reflecting growing funder and regulatory expectations for pre-registration, protocol publication, and results sharing ^[14].

Regulatory and Documentation Considerations

M11 CeSHarP and Electronic Protocol Submission

The M11 CeSHarP Technical Specification, revised in March 2025 by the ICH Assembly and released for public comment by FDA in June 2025 ^[19], recommends an open, nonproprietary standard for electronic exchange of protocol content. This means the structured data fields within a protocol are intended to be machine-readable and interoperable across regulatory systems.

Sponsors preparing for M11 adoption should verify that their protocol authoring tools can output protocol content tagged to the M11 data element schema. Protocols that exist only as PDFs or Word documents will require manual re-tagging to meet the technical specification. Early adoption of structured authoring, even in advance of enforcement, reduces the rework burden at the point of formal M11 adoption.

Regulatory Language and Binding Status

Protocols submitted to FDA as part of an IND are subject to binding regulatory requirements under 21 CFR Part 312. The ICH guidelines (E6(R3), M11, E9(R1)) function as guidance documents: they are not binding in the sense of statute or regulation, but FDA and EMA have both adopted them into their frameworks and generally expect alignment or documented scientific justification for departures ^[9]. In practice, deviation from ICH guidance without that justification creates regulatory risk.

The SPIRIT checklist is neither a regulatory document nor a guidance document in the FDA sense. It is a reporting standard: the 2013 version was endorsed by over 150 medical journals and recognized by national funders ^[15], and SPIRIT 2025 builds on that foundation. For IND submissions, the governing requirements are those in 21 CFR Part 312. SPIRIT and M11 provide the structure that makes an IND protocol easier to review, harder to miss checklist items from, and better aligned with the expectations of multiple international regulatory authorities simultaneously.

Common Protocol Checklist Gaps and Their Downstream Effects

Certain omissions appear with regularity across amendments. The Getz et al. 2024 study analyzed 2,188 amendments across 950 protocols and found that design deficiencies and changes in eligibility criteria were consistently among the most common triggers ^[1].

AI and Automation in Protocol Checklist Review

AI-assisted protocol review is a genuine and growing capability, though its maturity varies considerably across applications. Generative AI tools can perform rapid checklist verification: cross-referencing regulatory requirement lists against protocol text, flagging missing sections, identifying inconsistencies between the endpoints section and the statistical analysis plan, or detecting eligibility criteria that conflict internally.

What AI cannot reliably do, at least within current validated systems, is assess whether a protocol's scientific rationale is sound, whether the endpoint definitions are operationally feasible given the site population, or whether the risk-benefit balance is appropriate for the enrolled population. Those are clinical and scientific judgments that require human expert review.

The SPIRIT-AI extension to the SPIRIT checklist, published in Nature Medicine in 2020 and since incorporated into ongoing reporting standard development, provides specific guidance for protocols involving AI interventions and addresses transparency requirements for AI-specific trial elements ^[20]. This matters for sponsors evaluating AI-driven diagnostics or therapeutic algorithms as investigational products.

A practical AI application that adds operational value is cross-document consistency checking: verifying that eligibility criteria stated in the protocol match those entered in the EDC build and in the Informed Consent Form, or that endpoint definitions in the protocol are identical to those in the Statistical Analysis Plan. These consistency failures are common, costly to correct, and well within the scope of automated document review.

Kitsa's KScribe platform is designed to support regulatory document generation across the protocol development lifecycle, helping teams produce internally consistent documents and reduce the manual burden of checklist compliance at the drafting stage. For teams managing multiple concurrent protocols across therapeutic areas, consistency across documents is a nontrivial operational problem.

Key Takeaways

• A 2024 Tufts CSDD analysis found 76% of trials require at least one protocol amendment ^[1]; established Tufts CSDD benchmarking on amendment cost impact has estimated direct implementation costs in the range of $141,000 to $535,000 per amendment ^[21], though the exact figure varies by study phase, scope, and sponsor, and some amendment drivers can be reduced through earlier protocol review and structured quality-by-design work.
• Protocol requirements under 21 CFR 312.23 differ meaningfully by phase: Phase 1 protocols require safety-focused outlines, while Phase 2 and 3 protocols must address endpoints, eligibility, statistical design, and monitoring in detail ^[7].
• ICH E6(R3), finalized in December 2024 and published by FDA in September 2025, sets out expectations for sponsors to apply quality-by-design and risk-based thinking during protocol development rather than retrospectively ^[9].
• ICH M11 CeSHarP, which came into effect on June 11, 2026, provides a harmonized digital template that supports consistent protocol structure and electronic data exchange across FDA, EMA, and other ICH-region regulators ^[12],[13].
• SPIRIT 2025, published in Nature Medicine in April 2025, updated the minimum 33-item protocol checklist to 34 items, adding an open science section, expanded harm assessment requirements, and patient and public involvement provisions ^[14].
• Approximately one-third of procedures in current Phase II and III protocols do not directly support primary or key secondary endpoints, creating unnecessary site and participant burden that can be identified and reduced during protocol design ^[4].
• AI-assisted checklist review is useful for identifying missing sections and cross-document inconsistencies, but clinical and scientific judgment remains irreplaceable for evaluating protocol quality.

FAQ

References

1. [1]Getz K, Smith Z, Botto E, Murphy E, Dauchy A. "New Benchmarks on Protocol Amendment Practices, Trends and Their Impact on Clinical Trial Performance." Therapeutic Innovation & Regulatory Science. 2024;58(3):539-548. doi:10.1007/s43441-024-00622-9. PMID: 38438658. https://pubmed.ncbi.nlm.nih.gov/38438658/
2. [2]IQVIA. "Assessing Protocol Complexity and Its Impact on Trial Outcomes." IQVIA Blog, January 2026. https://www.iqvia.com/blogs/2026/01/assessing-protocol-complexity-and-its-impact-on-trial-outcomes
3. [3]Smith Z, Galuchie L, et al. (TransCelerate / Tufts CSDD). "TransCelerate and Tufts CSDD Uncover Opportunities to Rethink Data Collection and Optimize Protocol Design." PR Newswire, September 15, 2025. https://www.prnewswire.com/news-releases/transcelerate-and-tufts-csdd-uncover-opportunities-to-rethink-data-collection-and-optimize-protocol-design-302556373.html
4. [4]Tufts CSDD / TransCelerate BioPharma. "Optimizing Data Collection." TransCelerate Initiative. Accessed June 2026. https://www.transceleratebiopharmainc.com/initiatives/optimizing-data-collection/
5. [5]Getz K. "New Insights on the Clinical Trial Industry." Presented at the 2024 SCOPE Summit. Reported by Clinical Trial Vanguard, March 21, 2024. https://www.clinicaltrialvanguard.com/conference-coverage/tufts-csdd-new-insights-on-the-clinical-trial-industry/
6. [6]Getz K, Bechtel J. "Endpoints, Procedures, and Deviations Are on the Rise: Why?" Clinical Leader, January 20, 2026. https://www.clinicalleader.com/doc/endpoints-procedures-and-deviations-are-on-the-rise-why-0001
7. [7]U.S. Food and Drug Administration. "21 CFR 312.23: IND Content and Format." Electronic Code of Federal Regulations. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-D/part-312/subpart-B/section-312.23
8. [8]U.S. Food and Drug Administration. "IND Content and Format for Phase 1 Studies." FDA Guidance. https://www.fda.gov/media/72057/download
9. [9]U.S. Food and Drug Administration. "E6(R3) Good Clinical Practice." FDA Guidance for Industry, September 2025. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/e6r3-good-clinical-practice-gcp
10. [10]International Council for Harmonisation. "ICH E6(R3) Guideline for Good Clinical Practice, Step 5." European Medicines Agency, adopted December 12, 2024; effective July 23, 2025. https://www.ema.europa.eu/en/documents/scientific-guideline/ich-e6-r3-guideline-good-clinical-practice-gcp-step-5_en.pdf
11. [11]European Medicines Agency. "ICH M11 CeSHarP Template, Step 5." Adopted December 11, 2025; effective June 11, 2026. EMA/CHMP/ICH/778801/2022. https://www.ema.europa.eu/en/documents/template-form/ich-m11-clinical-electronic-structured-harmonised-protocol-cesharp-template-step-5_en.pdf
12. [12]U.S. Food and Drug Administration. "M11 Clinical Electronic Structured Harmonised Protocol (CeSHarP)." Federal Register, May 22, 2026. https://www.federalregister.gov/documents/2026/05/22/2026-10295/m11-clinical-electronic-structured-harmonised-protocol-cesharp-international-council-for
13. [13]European Medicines Agency. "ICH M11 Guideline on Clinical Study Protocol Template and Technical Specifications." EMA/CHMP/ICH/778799/2022. https://www.ema.europa.eu/en/ich-m11-guideline-clinical-study-protocol-template-technical-specifications-scientific-guideline
14. [14]Dunn G, Chan AW, Tetzlaff JM, et al. "SPIRIT 2025 Statement: Updated Guideline for Protocols of Randomized Trials." Nature Medicine. 2025;31:1784-1792. doi:10.1038/s41591-025-03668-w. PMID: 40294521. https://www.nature.com/articles/s41591-025-03668-w
15. [15]EQUATOR Network. "SPIRIT 2013 Statement Just Published." January 8, 2013. https://www.equator-network.org/2013/01/08/spirit-2013-statement-just-published/
16. [16]Kahan B, Hindley J, Edwards M, Cro S, Morris T. "The Estimands Framework: A Primer on the ICH E9(R1) Addendum." BMJ. 2024;384:e076316. doi:10.1136/bmj-2023-076316. PMID: 38262663. https://pubmed.ncbi.nlm.nih.gov/38262663/
17. [17]U.S. Food and Drug Administration. "21 CFR 312.32: IND Safety Reporting." Electronic Code of Federal Regulations. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-D/part-312/subpart-B/section-312.32
18. [18]U.S. Food and Drug Administration. "21 CFR Part 11: Electronic Records; Electronic Signatures." Electronic Code of Federal Regulations. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-A/part-11
19. [19]U.S. Food and Drug Administration. "M11 Technical Specification: Clinical Electronic Structured Harmonised Protocol." Federal Register, June 6, 2025. https://www.federalregister.gov/documents/2025/06/06/2025-10359/m11-technical-specification-clinical-electronic-structured-harmonised-protocol-international-council
20. [20]Cruz Rivera S, Liu X, Chan AW, Calvert MJ, Dirks I, et al.; SPIRIT-AI and CONSORT-AI Working Group. "Guidelines for Clinical Trial Protocols for Interventions Involving Artificial Intelligence: The SPIRIT-AI Extension." Nature Medicine. 2020;26(9):1380-1391. https://www.nature.com/articles/s41591-020-1037-7
21. [21]Getz KA, Stergiopoulos S, Kaitin KI. "Evaluating the Impact of Protocol Amendments on Clinical Trial Performance and Cost." Therapeutic Innovation & Regulatory Science. 2016;50(4):436-441. doi:10.1177/2168479016632271. https://link.springer.com/article/10.1177/2168479016632271
22. [22]International Council for Harmonisation. "ICH E9(R1): Addendum on Estimands and Sensitivity Analysis in Clinical Trials." Adopted Step 4, November 2019. https://database.ich.org/sites/default/files/E9-R1_Step4_Guideline_2019_1203.pdf
23. [23]U.S. Food and Drug Administration. "FDAAA 801 and the Final Rule: Requirements for Submitting Results of Applicable Clinical Trials to ClinicalTrials.gov." FDA Guidance, January 2017. https://www.fda.gov/science-research/science-and-research-special-topics/fdaaa-801-and-final-rule
24. [24]U.S. Government Publishing Office. "42 CFR 11.44: When Must Results Information Be Submitted?" Electronic Code of Federal Regulations. https://www.ecfr.gov/current/title-42/chapter-I/subchapter-A/part-11/subpart-C/section-11.44
25. [25]World Medical Association. "WMA Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Subjects." Adopted 1964; most recent revision 2024. https://www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for-medical-research-involving-human-subjects/
26. [26]U.S. Food and Drug Administration. "21 CFR 312.42: Clinical Holds and Requests for Modification." Electronic Code of Federal Regulations. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-D/part-312/subpart-C/section-312.42