FDA Guidance on Real-World Evidence from EHRs and Medical Claims for Regulatory Decisions
On July 24th, US FDA (Food and Drug Administration) finalized the guidance (initially published in September 2021) for evaluating and using electronic health records (EHR) and medical claims data for real-world evidence (RWE) studies and for using these for regulatory approvals.
In recent years, there has been a significant shift in the healthcare industry towards utilizing real-world data (RWD) to improve regulatory decision-making. This shift is driven mainly by the 21st Century Cures Act, which aims to speed up medical product development and bring innovations quickly to patients in need. The FDA’s new guidance on assessing EHRs and medical claims data underscores the importance of this movement.
Real-world evidence (RWE) derived from RWD, such as EHRs, medical claims data, real-world imaging data (RWiD) and others, provides valuable insights into the effectiveness and safety of medical products. This information, collected during routine clinical practice, offers a comprehensive overview of patient health statuses and healthcare delivery, which can substantially influence regulatory decisions.
Understanding FDA’s Position
The FDA has issued guidance on utilizing EHRs and medical claims for R&D which can further support regulatory decision-making. The guidance is part of a broader effort to encourage real-world evidence (RWE) in the drug development and R&D processes. By leveraging real-world datasets, like EHRs and claims data, pharmaceutical companies can reduce R&D costs while the FDA aims to create a better decision-making protocol. The guidance emphasizes three crucial aspects:
- Selection and validation of appropriate datasets,
- Ascertainment and validation of study design elements, and
- Ensuring data quality.
Key Points of the FDA Guidance
The FDA’s guidance highlights several important considerations for using EHRs and claims data in regulatory submissions:
Selection and validation of appropriate data sources
1. Data relevance and reliability:
There is a major emphasis by the FDA regarding the relevance and reliability of the data chosen - the data should accurately represent the clinical environment of the study & population in question and should be collected & maintained to ensure its integrity. The differences in medicinal practices worldwide and between health systems can affect the relevance of data. Since the data capture systems were not built for this purpose, it should be noted whether the data sources provide all the information required for the study and the population.
2. Comprehensiveness and completeness of the data:
Data usually gets captured when there is an interaction between the patient and the health care system. Hence this data may contain only some of the information necessary for the study. Study sponsors should make sure and demonstrate that the data to be used in the study has the necessary information and how they have been / will be acquired. FDA recommends the usage of multimodal datasets through data linkage and synthesis (with quality control in place) to ensure the data is complete and robust. The document also provides guidance on the usage of distributed data networks with common data models and how they can help with the standardization and interoperability of data sources across sites. Using computable phenotypes can help with the efficient selection of study populations and ascertainment of outcomes of interest / other study variables. For data that are in unstructured form, it is prudent to use a range of existing and emerging technologies to identify and obtain the necessary information.
3. Considerations for missing data:
The document provides guidance regarding the need to understand the reason for missing data and if possible, to try to identify a variable that can act as a proxy for the data. It emphasizes that the protocol should be developed considering the reasons for the presence and absence of information. The underlying assumptions need to be presented, along with the type of missing data and its implications on the study, while submitting the protocol.
4. Data Validation:
Studies using RWD should have both conceptual and operational definitions for key study variables such as study population (with inclusion and exclusion criteria), exposure, outcomes and covariates. The definitions should reflect current medical and scientific thinking regarding the variable of interest. Validation of the key variables is important due to potential misclassification that can occur due to imperfections with operational definitions. These misclassifications may lead to bias. Sponsors should understand how potential misclassification of a variable of interest can impact the association and interpretation of results and ensure that validation methods are performed to reduce bias due to those variables.
Study Design Elements
1. Defining time-period:
FDA recommends defining various periods pertinent to the study design in the protocol, which should also demonstrate the data availability, accuracy and completeness for the proposed periods with potential impact on study validity.
2. Selection of study population:
The guidance recommends that the protocol should include a detailed description of methods determining the inclusion and exclusion criteria, that will be implemented to identify appropriate patients from the data sources. The protocol should specify the operational definitions for inclusion and exclusion criteria, to identify eligible populations from the data source. It must include quantitative approaches to demonstrate whether and how misclassification of inclusion and exclusion criteria might impact study findings.
3. Exposure ascertainment and validation:
The protocol should contain the definition of exposure, regarding the product/regimen / any other dosing of interest for both the study and comparator population. This should include information on dose, formulation, strength, route, timing, frequency and duration of use, manufacturer if necessary, and a note on other variables that could affect the study's outcome. Sponsors should demonstrate the ability to identify the exposure to the product of interest (and about the coding system in the selected data source). While using structured and unstructured data, it is essential to report operational definitions. Also, to note, is that while using medical claims data, there could be a chance that it hasn’t captured all the relevant exposures for the study (in case they have obtained prescriptions not covered by insurance – like low-cost generics, drugs through discount programs, samples by pharmaceutical companies, out of pocket purchases etc. Validation is needed to compare exposure classification from the proposed data source with a reference data source, to produce estimates of misclassification that can be used for qualitative or quantitative assessments of the impact on study validity.
4. Outcome ascertainment and validation:
The protocol must provide information on how it will capture the outcome of interest. And how it will determine/identify outcomes occurring outside medical care. It should specify the definition of outcomes, how it is ascertained (from structured and unstructured data), and have a sensitivity & specificity analysis for the defined outcome.
5. Covariate ascertainment and validation:
The guidance document provides information related to 2 types of covariates – confounders and effect modifiers. FDA recommends a proper description of covariates and methods to capture them (including linking multiple datasets) to be included in the protocol. If exact information is unavailable, proxies can be used with appropriate justifications. The FDA recommends providing and justifying the validity of operational definitions in the protocol and the study report for all covariates.
Data Quality
The quality of the datasets being used is paramount. The FDA advises sponsors to thoroughly assess the data quality during the accrual, curation and transformation phases. This includes evaluating the completeness, accuracy, and consistency of the data. Sponsors should also consider potential biases that may arise from the way the data is collected or recorded. The integration of multimodal datasets can enhance the robustness of the analysis. The study protocol and analysis plan should specify the traceability and describe how these procedures could affect the integrity and validity of the study.
Challenges and Considerations
While the FDA’s guidance provides a clear framework for using RWD (focusing on EHRs and claims data) in regulatory submissions, this is only the first step. Sponsors while planning and implementing a RWE study can face challenges with variability in data quality and completeness across different EHR systems and claims databases. Additionally, sponsors must be mindful of potential biases that could impact the validity of their findings. Invariably there is a need to integrate EHR and claims datasets with other types of datasets to reduce the biases and errors that occur with reports.
Segmed can support you navigate these challenges. Our curated fit-for-purpose regulatory grade datasets – multimodal and longitudinal can support your studies. Segmed’s dedicated in-house team of Subject matter experts consisting of medical doctors, RWD/RWE and regulatory experts make sure that the datasets delivered are regulatory grade and can be directly used for study purposes. Our expertise in integrating imaging datasets with other types like EHRs and claims through tokenization, provides a comprehensive and complete picture of the study population.
The Future of RWD in Regulatory Decisions
The FDA’s guidance represents a significant step forward in integrating real-world data into the regulatory decision-making process. By providing a clear framework for the use of EHRs and claims data, this can further be extrapolated to other types of datasets. The FDA is encouraging sponsors to explore innovative approaches to drug development and post-market surveillance.
As the healthcare landscape continues to evolve, the role of RWD in regulatory decisions is likely to expand. Sponsors who adhere to the FDA’s guidance will be well-positioned to navigate the complexities of the regulatory environment and contribute to the development of safe and effective medical products. Segmed is here to support the sponsors through their journey of RWE and regulatory approvals.
Conclusion
The FDA’s guidance on using RWD marks a pivotal moment in integrating real-world evidence into regulatory decision-making. FDA’s recommendations on data relevance, quality, transparency, and documentation provide a path for sponsors to leverage these data sources to support their regulatory submissions effectively. As the industry continues to evolve, the importance of real-world data in shaping the future of healthcare cannot be overstated.
Get in touch with Segmed to understand our solutions and how they can support your studies.
Connect with us to explore how our datasets have supported various trials and to learn how we have helped organizations navigate the complex web of RWE.
