Real-World Data Are Everywhere
Useful Real-World Evidence Takes Planning
For many sponsors, real-world data can be both an opportunity and a risk. The right data source can reduce burden, speed up evidence generation, and inform a regulatory or commercial decision. However, the wrong data source can result in even more uncertainty than before and provide evidence that is neither reliable nor useful for regulators, clinicians, or payers. That is why a successful real-world evidence (RWE) strategy should begin with these simple questions: what decision are we trying to make, and how robust does the evidence need to be?
Thanks to electronic health records (EHRs), registries, claims databases, and digital health technologies, providing access to clinical data is easier than ever before. However, access to data is one thing while generating usable evidence is another. Sponsors still need to determine whether the available real-world data (RWD) can be used to generate RWE that is relevant, reliable, and fit for the intended regulatory, clinical, or market-access purpose. The use of RWD to conduct post-market studies has been encouraged by RWD stakeholders, and FDA and other regulatory bodies increasingly recognize the importance of well-designed RWE to facilitate decision-making throughout the product lifecycle.
The Sponsor Problem: Having Data Does Not Equal Having Evidence
Many sponsors already have access to RWD, but they are not sure whether those data can support a regulatory submission, publication, reimbursement discussion, or internal clinical decision. A database may include thousands of patients and still be missing the endpoints, follow-up duration, confounders, source documentation, or data quality controls needed to answer the question at hand.
This is where early RWE planning becomes important. The study design, data source, endpoint definitions, statistical analysis plan, and bias mitigation strategies should be thoughtfully developed before any data analysis begins. Practically speaking, the data need to be fit for purpose and able to address the research question.
Where RWE Can Help
Conventional clinical trials are still necessary, but they may not always be the most feasible solution to every problem. Clinical trials can be expensive and time-consuming, and inclusion and exclusion criteria may further restrict the applicability of the findings because results do not always reflect routine clinical practice. Enrollment can also be particularly challenging when patients have rare diseases, complex or multiple comorbidities, travel constraints, or limited access to major research sites.
In the right setting, RWD may allow sponsors to assess product performance within routine clinical practice. It can also decrease the burden on patients, shorten timelines, and decrease research expenses by using data that are already being captured as part of routine care.
Retrospective studies using existing data may qualify for exempt IRB review under 45 CFR 46.104(d)(4), depending on the type of data, whether the information is identifiable, HIPAA considerations, and applicable institutional and regulatory requirements. When this pathway is appropriate, it can reduce study duration and cost while maintaining appropriate protections for human subjects.
Irregular timepoints and missing data: Registry patients may not attend regular visits compared with patients in traditional clinical studies, and timepoints are not always standardized. Clinical study staff are trained to encourage follow-up adherence and may offer flexible scheduling, transportation, travel reimbursement, or modest financial incentives. These supports are not usually available in real-world practice. Patients may also seek a different provider if their condition is not improving or stop attending visits once they feel better. | Define acceptable visit windows, follow-up requirements, missing-data methods, sensitivity analyses, and rules for excluding or censoring records before reviewing outcomes. |
Bias: Retrospective analyses can be vulnerable to selection bias, information bias, confounding, and data-driven decision-making if study methods are developed after reviewing outcomes. | Develop the protocol and statistical analysis plan before accessing outcome results. Independent feasibility assessments, blinded data characterization, prespecified analytic methods, and independent review can help reduce bias and strengthen credibility. |
This is where experienced clinical and regulatory planning can make a meaningful difference. By pressure-testing the data source, defining the research question, and aligning the protocol and analysis plan before the study begins, sponsors can avoid spending time and money on an RWE strategy that will not be able to inform their decision.
Regulatory Pathways
FDA centers for drugs, biologics, and devices have published guidance on using RWD and RWE in regulatory decision-making. For medical devices, FDA’s updated guidance describes how the Agency evaluates whether RWD are of sufficient quality to generate RWE for submissions such as 510(k), De Novo, PMA, IDE, HDE, expanded indications, post-approval studies, and post-market surveillance. Across FDA programs, the message is consistent: sponsors should be able to explain why the selected data are relevant and reliable for the specific regulatory question. FDA also recommends early engagement when sponsors plan to use RWE to support regulatory decision-making, including discussion of the proposed protocol, data sources, and statistical analysis plan.
Not sure whether your real-world data can support your clinical or regulatory strategy?
MED helps sponsors move from “we have data” to “we have evidence we can use.” Our clinical and regulatory teams can evaluate RWD sources, design fit-for-purpose RWE studies, develop protocols and analysis plans, identify data-quality risks, draft articles for publication, and align evidence-generation strategies with regulatory and business objectives. and align evidence-generation strategies with regulatory and business objectives. MED offers a variety of clinical, regulatory, and scientific communications services and is a full-service CRO. We have more than 40 years of experience designing and executing clinical studies, ranging from early feasibility studies to multinational controlled pivotal trials and post-market registries. Our studies are conducted to meet applicable requirements for Good Clinical Practice (GCP), ISO 14155, and national regulations, including applicable U.S. CFR and HIPAA requirements.
RWE can also help answer a question that matters to many sponsors: How is this product likely to be used in the real world? EHR and registry data may provide details that are difficult to capture in a narrowly controlled trial, such as disease severity, anatomy, procedural technique, clinician assessment, care pathways, and health economic considerations.
Overcoming Common RWE Limitations
RWD can create real efficiencies, but it also brings practical limitations. The key is to identify those limitations early and decide whether they can be addressed well enough for the intended use.
Limitations | Proposed Solution |
Not clinically comprehensive: RWD may not provide relevant confounders that tell the whole patient story. | Start with the research question, then choose the data source. If a key variable, such as smoking status, disease severity, or prior treatment history, is not captured in a reliable way, the RWD pathway may need supplemental data collection or may not be the right approach. |
Integration: Patients may have information recorded by different providers in different systems that are not integrated. When multiple institutions or databases are combined, duplicate patient records may occur, especially in countries such as the United States, where there is no single national patient identification number. | Use standardized terminology, common data models, deduplication methods, and documented data-linkage procedures so the data can be converted into a consistent format and traced back to the source. |
Unstructured data: EHR systems vary and rely heavily on free text comments. Therefore, RWD may not be documented consistently or reliably. | Define key data elements before extraction begins, establish abstraction rules, and use trained reviewers when information must be pulled from free text. Natural language processing may help, but it should be validated and paired with appropriate human oversight. |
Contact us today to start your project discussion!
855.463.1633 | askmed@medinstitute.com | medinstitute.com.
OUR COMMITMENT
We are committed to consistently performing services with high quality, that deliver exceptional results, and add value to the client’s business.
For client surveys sent since 2024, we received ratings of 4.99/5 (16).