Executive Summary
The Benefit-Risk Action Team (BRAT) Framework, developed through a collaboration between pharmaceutical industry stakeholders and regulatory scientists, provides a structured approach to characterizing and communicating the benefit-risk profile of medical products. Originally published in 2009 and refined through extensive real-world application, the BRAT Framework has become one of the most widely adopted methodologies for structured benefit-risk assessment in the pharmaceutical industry.
This guide provides a step-by-step approach to implementing the BRAT Framework, describes how ArcaScience's platform automates key elements of the process, and offers best practices for regulatory submissions to the FDA, EMA, and other health authorities.
1. What Is the BRAT Framework?
The BRAT Framework is a systematic, iterative process for organizing and summarizing benefit-risk information throughout the product lifecycle. Unlike purely quantitative approaches, BRAT provides a flexible structure that can accommodate both qualitative and quantitative analyses, making it applicable across development phases and therapeutic areas.
The framework was developed through the work of the Benefit-Risk Action Team, an industry initiative that included representatives from major pharmaceutical companies, academic institutions, and regulatory bodies. Its design reflects practical lessons learned from attempting to implement structured BRA across diverse development programs.
Core principles of the BRAT Framework include:
- Transparency: All assumptions, data sources, and decision criteria are explicitly documented
- Iterative refinement: The assessment evolves as new data emerge throughout the product lifecycle
- Stakeholder inclusivity: The framework accommodates perspectives from patients, healthcare providers, regulators, and payers
- Visual communication: Key tools such as value trees and summary tables facilitate clear communication of complex information
- Proportionality: The depth and complexity of analysis is proportional to the decision at hand
2. Step-by-Step Implementation Guide
The BRAT Framework consists of six iterative steps. While presented sequentially, in practice teams often cycle between steps as understanding deepens and new information becomes available.
Clearly articulate the decision to be made, the target population, the comparator(s), and the timeframe. This step establishes the scope and boundaries of the assessment. Key questions include: What is the therapeutic indication? What is the unmet medical need? Who are the relevant decision-makers? What are the available treatment alternatives?
Develop a comprehensive list of potential benefit and risk outcomes relevant to the decision context. This typically begins with a broad evidence review and is refined through clinical expert input. Outcomes are then prioritized based on clinical significance, frequency, patient relevance, and regulatory importance. The result is a focused set of key outcomes that will form the basis of the assessment.
For each selected outcome, identify the best available data sources and extract relevant efficacy and safety data. Sources may include pivotal clinical trials, pooled safety analyses, meta-analyses, observational studies, patient registries, and published literature. Data quality, relevance, and potential biases should be documented.
Tailor the assessment structure to the specific decision context. This may involve constructing a value tree that organizes outcomes hierarchically, selecting appropriate data presentation formats, and determining whether quantitative analysis (such as MCDA) is warranted. The level of analytical sophistication should match the decision's complexity and importance.
Synthesize the evidence for each outcome, characterize uncertainty, and create summary displays. The BRAT Framework emphasizes visual summary tools, including the key benefit-risk summary table, forest plots for comparative data, and value trees showing the overall assessment structure. Quantitative analyses such as weighted scoring or probabilistic modeling may be applied at this stage.
Draw conclusions about the benefit-risk balance, articulate remaining uncertainties, and prepare tailored communications for different audiences. Regulatory submissions require specific formatting and detail, while internal decision committees may benefit from interactive presentations. The assessment should clearly state its conclusions while acknowledging limitations.
3. Decision Model Construction
3.1 Building the Value Tree
The value tree is a hierarchical representation of the benefits and risks being evaluated. At its highest level, it separates outcomes into benefit and risk categories. These are further decomposed into specific outcome measures, each linked to quantifiable endpoints.
A well-constructed value tree for a typical oncology submission might include:
- Benefits: Overall survival, progression-free survival, objective response rate, duration of response, patient-reported quality of life
- Risks: Treatment-related mortality, Grade 3+ adverse events, immune-related adverse events, hepatotoxicity, cardiovascular events, treatment discontinuation due to adverse events
3.2 Selecting Appropriate Quantitative Methods
The BRAT Framework does not mandate a specific quantitative approach, allowing teams to select methods appropriate to their context. Common quantitative extensions include:
| Method | Complexity | When to Use |
|---|---|---|
| Descriptive comparison tables | Low | Clear-cut benefit-risk profiles, early development decisions |
| Number needed to treat/harm (NNT/NNH) | Low-Medium | Binary outcomes with clinical trial data |
| MCDA with weighted scoring | Medium-High | Multiple criteria with stakeholder preferences |
| Probabilistic modeling (stochastic MCDA) | High | Significant uncertainty, need for probabilistic conclusions |
| Bayesian network models | High | Complex interdependencies between outcomes |
4. Data Source Integration
A robust BRAT assessment draws on multiple data sources to characterize benefits and risks comprehensively. Key data sources and their roles include:
- Pivotal clinical trials: Primary source for efficacy data and controlled safety data
- Pooled safety databases: Broader safety characterization across the development program
- Post-marketing surveillance: Real-world safety signals and long-term outcomes
- Published literature: Class effects, comparator data, epidemiological context
- Patient registries: Natural history data, real-world effectiveness
- Claims and EHR databases: Real-world treatment patterns, healthcare utilization
- Patient preference studies: Discrete choice experiments, surveys, qualitative research
5. Regulatory Agency Expectations
5.1 FDA Perspective
The FDA's Benefit-Risk Framework, formalized through PDUFA V and subsequent legislative updates, encourages sponsors to provide structured benefit-risk assessments as part of their regulatory submissions. The FDA's framework organizes assessment around five dimensions: Analysis of Condition, Current Treatment Options, Benefit, Risk, and Risk Management. While the FDA has not mandated any specific methodology, they have expressed strong support for structured approaches and have specifically referenced the BRAT Framework in regulatory science publications.
5.2 EMA Perspective
The EMA has been at the forefront of promoting structured benefit-risk assessment through its Benefit-Risk Methodology Project. The agency developed the PrOACT-URL framework and has encouraged sponsors to use structured approaches in their applications. Since 2015, the CHMP has applied its own effects table approach to all product evaluations, creating a clear expectation for structured data presentation. The EMA's guidance recommends that Marketing Authorization Applications include a structured benefit-risk section using standardized presentation formats.
5.3 Other Regulatory Bodies
Health Canada, PMDA (Japan), TGA (Australia), and other regulatory agencies are increasingly aligned with the ICH framework for benefit-risk assessment. The ICH M4E(R2) guideline on the Common Technical Document includes specific provisions for benefit-risk summaries, creating a globally harmonized expectation for structured assessment.
6. How ArcaScience Automates the BRAT Framework
ArcaScience's platform maps directly to the six steps of the BRAT Framework, automating manual processes while preserving the scientific rigor and regulatory compliance that the framework demands:
| BRAT Step | Manual Process | ArcaScience Automation |
|---|---|---|
| 1. Decision context | Workshop facilitation, document drafting | Structured templates with regulatory intelligence pre-populated |
| 2. Outcome identification | Literature review, expert consultation | NLP-driven outcome extraction from literature and clinical protocols |
| 3. Data extraction | Manual data collection across sources | Automated ingestion from trial databases, safety systems, and RWE platforms |
| 4. Customization | Excel-based models, manual value tree creation | Interactive model builder with drag-and-drop value tree construction |
| 5. Assessment | Statistical analysis, manual visualization | Automated MCDA, Monte Carlo simulation, dynamic visualizations |
| 6. Communication | PowerPoint decks, Word documents | Auto-generated regulatory-ready documents and interactive dashboards |
7. Best Practices Checklist
Use this checklist to ensure comprehensive BRAT Framework implementation:
Planning Phase
- Define clear decision question and scope before beginning assessment
- Identify all relevant stakeholder groups and their information needs
- Establish a cross-functional BRAT team (clinical, regulatory, safety, medical affairs, HEOR)
- Determine appropriate level of quantitative rigor based on decision context
- Create a project timeline aligned with regulatory submission milestones
Execution Phase
- Conduct systematic outcome identification using predefined criteria
- Document all data sources, extraction methods, and quality assessments
- Apply consistent analytical methods across all outcomes
- Include patient perspective data (preferences, PROs, quality of life)
- Perform sensitivity analyses to characterize uncertainty
- Validate results with independent clinical experts
Communication Phase
- Prepare audience-specific presentations (regulatory, internal, advisory committee)
- Include clear visual summaries (value trees, summary tables, forest plots)
- Explicitly state conclusions, limitations, and remaining uncertainties
- Document methodology thoroughly for regulatory transparency
- Plan for iterative updates as new data become available
Quality Assurance
- Ensure traceability from source data to final conclusions
- Conduct independent quality review of all data inputs
- Verify consistency between BRAT assessment and other submission documents
- Archive all working documents and analytical outputs for inspection readiness
Implement the BRAT Framework with Confidence
ArcaScience's platform automates the BRAT Framework while maintaining full regulatory compliance.