Navigating the New Frontier: The Evolving Role of Real-World Evidence in Medical Device Development

Many thanks to our sponsor Esdebe who helped us prepare this research report.

Abstract

Real-World Evidence (RWE) stands as a foundational pillar in the modern medical device development ecosystem, transcending the traditional confines of randomized controlled trials (RCTs) to offer profound insights into product performance and patient outcomes. This comprehensive report meticulously examines the genesis and evolution of RWE, meticulously dissecting its diverse sources of Real-World Data (RWD), the sophisticated methodologies employed for its rigorous collection and analysis, and its accelerating integration into the regulatory frameworks of influential bodies such as the U.S. Food and Drug Administration (FDA). By exploring RWE’s transformative capacity to expedite product development cycles, inform nuanced clinical and regulatory decision-making, and enhance the safety and efficacy profiles of medical devices, this report illuminates its indispensable role in shaping the future of healthcare innovation. Particular emphasis is placed on its critical utility in addressing the unique challenges inherent in pediatric device development, where conventional trial designs face significant ethical and practical hurdles.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

1. Introduction

The landscape of medical device innovation has historically been governed by a rigorous paradigm centered on randomized controlled trials (RCTs). These meticulously designed studies, with their tightly controlled variables and prospective data collection, have long served as the gold standard for establishing the safety and efficacy of new medical products. The inherent strength of RCTs lies in their ability to minimize confounding factors through randomization, thereby enabling robust causal inference regarding an intervention’s effect (Friedman et al., 2015). However, the reliance on RCTs, while scientifically sound, has increasingly confronted significant practical and ethical limitations within the rapidly evolving medical device industry.

RCTs are notoriously resource-intensive, often demanding substantial financial investment and protracted timelines stretching over many years to complete. Their highly selected patient populations, strict inclusion and exclusion criteria, and standardized treatment protocols, while necessary for internal validity, frequently limit the generalizability of their findings to the heterogeneous ‘real-world’ patient populations and diverse clinical practice settings (Rothwell, 2006). Furthermore, ethical considerations preclude RCTs in certain vulnerable populations, such as critically ill patients or rare disease cohorts, or when a device’s effectiveness is already strongly indicated. The ‘efficacy-effectiveness gap’ – the divergence between a device’s performance under ideal trial conditions and its performance in routine clinical practice – represents a critical challenge that RCTs are inherently ill-equipped to fully address.

In response to these pervasive limitations and spurred by the exponential growth in digital health technologies and sophisticated data analytics, Real-World Evidence (RWE) has emerged as a profoundly complementary and, in many contexts, essential approach to evidence generation. RWE leverages data routinely collected from the everyday delivery of healthcare, offering a panoramic view of medical product usage, effectiveness, and safety in diverse, unselected patient populations across a spectrum of clinical environments (Sherman et al., 2016). This paradigm shift is not merely an incremental adjustment but a fundamental re-evaluation of how evidence is generated, validated, and applied in regulatory and clinical decision-making. The confluence of vast digital data repositories, advanced computational capabilities, and a growing recognition of the limitations of traditional trials has propelled RWE from a nascent concept to a pivotal component of modern medical product development.

This report aims to elucidate the multifaceted role of RWE in the medical device sector. It will provide a comprehensive understanding of RWD sources, critically examine the methodologies for extracting meaningful RWE, detail the burgeoning regulatory acceptance, particularly by the FDA, and highlight its transformative implications for accelerated development, post-market surveillance, and, notably, its potential to address the unique challenges in pediatric device development. By delving into these aspects, the report underscores the imperative for robust RWE frameworks to enhance patient safety, optimize clinical outcomes, and foster sustained innovation in medical technology.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

2. Understanding Real-World Evidence

2.1 Definition and Scope

To fully appreciate the impact of RWE, it is crucial to delineate its precise definition and understand its expansive scope within the medical product lifecycle. The U.S. Food and Drug Administration (FDA) provides a widely accepted framework, defining Real-World Data (RWD) as ‘data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources’ (FDA, 2017a). Building upon this, Real-World Evidence (RWE) is then defined as ‘clinical evidence regarding the usage and potential benefits or risks of a medical product derived from analysis of RWD’ (FDA, 2017a). This distinction is fundamental: RWD is the raw material, while RWE is the actionable insight generated through rigorous analytical processing.

The ‘real-world’ aspect inherently implies that data are collected outside the controlled confines of traditional clinical trials. This encompasses a broad spectrum of clinical settings, including hospitals, outpatient clinics, emergency departments, and even patient homes, reflecting the diverse conditions under which medical devices are actually used. The data capture the full heterogeneity of patient populations, including those with comorbidities, polypharmacy, and varying socioeconomic backgrounds, which are often excluded from or underrepresented in RCTs (Blumenthal et al., 2020). This wider applicability significantly enhances the external validity of the evidence generated.

The scope of RWE spans the entire product lifecycle, from informing early-stage research and development decisions to supporting pre-market regulatory submissions, guiding post-market surveillance activities, and ultimately shaping clinical practice guidelines and payer reimbursement decisions. Unlike traditional epidemiological studies that often focus on disease incidence or prevalence, RWE specifically investigates the performance, safety, and effectiveness of medical products in clinical practice (Corrigan-Curay et al., 2018). It is a dynamic field, continuously evolving with advancements in data science, digital health technologies, and regulatory policies.

2.2 Sources of Real-World Data

The utility of RWE is directly contingent upon the quality, breadth, and depth of its underlying RWD. These data are collected from a myriad of sources, each offering unique strengths and presenting distinct challenges. Understanding these sources is paramount for assessing the ‘fitness for purpose’ of specific RWD for generating reliable RWE.

2.2.1 Electronic Health Records (EHRs)

EHRs are digital versions of patients’ paper charts, serving as comprehensive repositories of clinical information captured during routine healthcare encounters. They contain a wealth of data, including patient demographics, medical history, diagnoses (coded using ICD-10 or SNOMED CT), laboratory test results, imaging reports, prescription medications, physician notes, and procedural codes. The richness of EHR data provides a granular view of patient care and outcomes over time, making them invaluable for understanding disease progression and treatment patterns (Bowes and Corrado, 2021).

However, EHRs present significant challenges. Data entry can be inconsistent, with variations in coding practices, the use of free-text notes that are difficult to process, and the presence of missing data. Interoperability issues between different EHR systems can hinder data aggregation, and the primary purpose of EHRs is clinical care, not research, meaning data may not be structured optimally for research questions (Weiskopf and Weng, 2013).

2.2.2 Medical Claims Data

Medical claims data are generated during the billing and reimbursement processes by healthcare providers, payers (e.g., Medicare, Medicaid, commercial insurers), and pharmacies. These datasets typically include information on patient demographics, diagnoses, procedures, prescribed medications (pharmacy claims), and healthcare service utilization (hospitalizations, outpatient visits). Their primary strengths lie in their vast population coverage, often encompassing millions of patients, and their ability to track healthcare encounters over extended periods, making them ideal for long-term safety surveillance and comparative effectiveness studies (Gagne et al., 2021).

Conversely, claims data have inherent limitations. They are generated for billing purposes, not clinical detail, often lacking granular clinical information such as laboratory results, vital signs, or physician notes. Diagnoses and procedures are recorded using administrative codes (e.g., ICD-10-CM, CPT), which may not always precisely reflect the underlying clinical condition or treatment performed. There can also be issues with ‘upcoding’ or incomplete recording for billing optimization, and they do not capture care provided outside the reimbursed system (e.g., over-the-counter medications) (Berger et al., 2022).

2.2.3 Patient Registries

Patient registries are organized systems that collect uniform data on patients who share a particular condition, exposure to a medical product, or undergo a specific procedure, and who are followed over time. They can be disease-specific (e.g., cystic fibrosis registry), product-specific (e.g., transcatheter aortic valve replacement registry), or procedure-specific. Registries offer several advantages: they collect prospective, standardized data on predefined outcomes, often including patient-reported outcomes (PROs), and can facilitate long-term follow-up (Gliklich et al., 2014). This makes them particularly valuable for studying rare diseases, long-term device performance, and adverse events that may only manifest years after implantation.

However, registries can be costly and time-consuming to establish and maintain. They may suffer from selection bias if participation is voluntary and not representative of the broader patient population. The quality and completeness of data can vary depending on the rigor of data collection protocols and the motivation of participating sites and patients.

2.2.4 Digital Health Technologies (DHTs)

This rapidly expanding category includes data generated by wearable devices (e.g., smartwatches, fitness trackers), mobile health applications (mHealth apps), remote monitoring devices (e.g., continuous glucose monitors, implantable cardiac monitors), and sensors. DHTs offer the unprecedented ability to collect physiological data (heart rate, activity levels, sleep patterns), environmental data, and patient-reported information continuously and passively in real-time, often outside traditional clinical settings (Mandl et al., 2021). This provides a more granular and ecological view of a patient’s health status and response to therapy.

The challenges associated with DHT data are substantial. They include ensuring data validity and accuracy across diverse devices and platforms, addressing data privacy and cybersecurity concerns, developing robust analytical methods for high-frequency, complex datasets, and navigating the rapidly evolving regulatory landscape for these technologies. Furthermore, issues of digital literacy and access can introduce significant bias in who uses and generates data from DHTs.

2.2.5 Other Real-World Data Sources

Beyond these primary categories, other valuable RWD sources contribute to the RWE landscape:

• Pharmacy Data: Detailed records of dispensed medications, useful for adherence studies and understanding polypharmacy.
• Laboratory Information Systems: Centralized databases of lab test results, providing objective clinical measures.
• Hospital Administrative Data: Often part of or linked to EHRs/claims, but can provide specific operational data (e.g., bed occupancy, readmission rates).
• Social Media and Online Forums: Can offer qualitative insights into patient experiences and sentiments, though highly unstructured and susceptible to bias.
• Environmental Data: Can be linked to health outcomes (e.g., air quality, geographic information systems).

The key principle underlying the use of any RWD source is ‘fitness for purpose’ – the data must be sufficiently reliable, relevant, and robust to address the specific research question at hand (FDA, 2017a). This often necessitates combining and integrating data from multiple sources to achieve a more comprehensive and triangulated view.

2.3 Methodologies for Collection and Analysis

The transformation of raw RWD into credible RWE is a complex, multi-stage process requiring rigorous methodologies to ensure scientific validity and minimize bias. This journey typically involves data acquisition, cleaning and validation, integration, sophisticated statistical analysis, and careful interpretation.

2.3.1 Data Acquisition

This initial phase involves identifying relevant RWD sources, establishing data sharing agreements, and physically obtaining the data. It requires careful consideration of data governance frameworks, legal and ethical permissions (e.g., institutional review board approval), and compliance with privacy regulations (e.g., HIPAA in the US, GDPR in the EU). Data acquisition strategies vary from direct access to institutional EHRs, purchasing claims datasets from data aggregators, or extracting data from public registries. The process often involves significant negotiation and technical expertise to establish secure data transfer protocols.

2.3.2 Data Cleaning and Validation

RWD, by its nature, is not collected for research and often contains inconsistencies, errors, and missing values. This makes data cleaning and validation a critically important, iterative step. Techniques include:

• Outlier Detection: Identifying and addressing extreme values that may be errors.
• Missing Data Imputation: Employing statistical methods (e.g., mean imputation, regression imputation, multiple imputation) to handle missing values, which can significantly impact study results.
• Consistency Checks: Verifying logical consistency within variables (e.g., age within reasonable bounds) and between related variables (e.g., diagnosis date preceding treatment date).
• Data Deduplication: Identifying and removing duplicate records.
• Standardization: Converting free-text fields into standardized codes or categorizations where possible.

Data validation further involves comparing subsets of the RWD to external gold standards or manually reviewing charts to confirm the accuracy of key variables. Poor data quality at this stage can lead to erroneous conclusions, exemplifying the ‘garbage in, garbage out’ principle.

2.3.3 Data Integration

To overcome the limitations of single data sources and create a more holistic view of patient journeys, RWD from disparate sources are often integrated. This complex process involves:

• Patient Linkage: Matching records belonging to the same individual across different datasets, typically using probabilistic or deterministic algorithms based on indirect identifiers (e.g., dates of birth, ZIP codes, encrypted identifiers). This step is highly sensitive to privacy concerns.
• Data Harmonization: Standardizing variable names, formats, and coding schemes across different datasets to ensure comparability. This often involves mapping diverse coding systems (e.g., ICD-9 to ICD-10, proprietary drug codes to RxNorm) to common terminologies.
• Common Data Models (CDMs): Implementing CDMs like OMOP (Observational Medical Outcomes Partnership), PCORnet, or the FDA’s Sentinel CDM provides a standardized structure for organizing and mapping diverse healthcare datasets, facilitating federated research and analysis across multiple institutions without moving raw patient-level data (Ryan et al., 2013; Curtis et al., 2017). These models are crucial for enabling large-scale, distributed RWE studies.

2.3.4 Statistical Analysis

The analytical phase is where RWD is transformed into RWE. Unlike RCTs, observational RWD studies require sophisticated statistical methodologies to address the inherent challenges of confounding, selection bias, and measurement error. Key methods include:

• Observational Study Designs: Employing designs such as cohort studies (prospective or retrospective), case-control studies, cross-sectional studies, and self-controlled case series, each with its strengths and weaknesses for specific research questions.
• Causal Inference Techniques: Given the non-randomized nature of RWD, inferring causality is challenging. Advanced methods are used to mitigate confounding:
  • Propensity Score Methods: Matching, stratification, inverse probability of treatment weighting (IPTW) to balance covariates between treatment groups (Austin, 2011).
  • Instrumental Variables: Using a variable that influences treatment assignment but not the outcome directly, except through the treatment (Hernán and Robins, 2006).
  • Difference-in-Differences: Comparing changes in outcomes over time between a treated and untreated group.
  • Regression Adjustment: Multivariate regression models to control for measured confounders.
• Sensitivity Analyses: Evaluating the robustness of findings to unmeasured confounding or different analytical assumptions.
• Bias Analysis: Quantifying the potential impact of known or suspected biases (e.g., selection bias, misclassification bias).
• Machine Learning and Artificial Intelligence (AI): Increasingly used for predictive modeling, identifying complex patterns, and enhancing traditional statistical methods. However, challenges include model interpretability (‘black box’ problem) and ensuring generalizability (Subramanian and Sridhar, 220).

Transparent reporting of the analytical plan, including a detailed statistical analysis plan (SAP) developed a priori, is crucial for the credibility of RWE, often adhering to guidelines like STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) (von Elm et al., 2007).

2.3.5 Interpretation

The final stage involves translating analytical findings into clinically meaningful and actionable insights. This requires a deep understanding of the data’s limitations, the potential for residual confounding, and the context of real-world clinical practice. Interpretation should be conducted in collaboration with clinical experts to ensure the relevance and validity of conclusions. It is critical to clearly articulate the strengths and limitations of the RWE generated and to avoid overstating causal claims that cannot be fully substantiated by observational data alone (Fleurence et al., 2019).

Many thanks to our sponsor Esdebe who helped us prepare this research report.

3. Regulatory Acceptance of Real-World Evidence

3.1 FDA’s Perspective on RWE

The U.S. Food and Drug Administration (FDA) has been at the forefront of integrating RWE into its regulatory decision-making processes, recognizing its immense potential to enhance the efficiency and comprehensiveness of medical product evaluation. This evolving acceptance reflects a strategic shift from an exclusive reliance on RCTs towards a more holistic evidence generation paradigm.

3.1.1 Historical Context and Legislative Mandates

While the FDA has long used RWD for post-market surveillance (e.g., adverse event reporting systems), its formal embrace of RWE for pre-market approvals gained significant momentum with the enactment of the 21st Century Cures Act in 2016. This landmark legislation specifically mandated the FDA to ‘establish a program to evaluate the potential use of real world evidence to support the approval of new indications for approved drugs, or to satisfy post-approval study requirements’ (21st Century Cures Act, 2016). This legislative push signaled a clear intent to leverage RWE more extensively across both drug and device development.

Following this mandate, the FDA issued its ‘Framework for FDA’s Real-World Evidence Program’ in December 2018, outlining the agency’s approach to evaluating RWE. This framework articulated key considerations for assessing the ‘fitness for purpose’ of RWD, including data relevance and reliability, and the ‘fitness for use’ of analytical methodologies (FDA, 2018). Subsequent guidance documents, particularly ‘Use of Real-World Evidence to Support Regulatory Decision-Making for Medical Devices’ (FDA, 2017b, updated 2024), provided more specific details for the medical device sector.

3.1.2 Landmark Policy Shift on Identifiable Patient Data

A significant policy advancement, referenced in the abstract, was formally announced by the FDA, indicating a willingness to accept RWE without requiring identifiable individual patient data in certain medical device submissions. This strategic shift, while not yet fully detailed in publicly available guidance as of late 2023/early 2024 (the article mentioned December 2025, suggesting a future projection or specific internal FDA announcement), represents a pivotal moment. The stated intention is to facilitate the use of large, de-identified datasets, such as national cancer registries and hospital systems databases, to support regulatory decisions. This move addresses a critical bottleneck: the immense logistical and privacy challenges associated with submitting individual patient-level data. By allowing aggregated or de-identified data, the FDA aims to:

• Enhance Scalability: Enable the use of far larger datasets that might be impractical to collect or share in an identifiable format.
• Improve Efficiency: Streamline data submission processes for sponsors.
• Strengthen Privacy: Reduce the risk of re-identification, thereby fostering greater trust and willingness for data sharing.
• Accelerate Access: Expedite the availability of devices by reducing the burden on sponsors for certain types of evidence generation (FDA, 2025).

This policy change, when fully implemented and clarified through detailed guidance, is expected to significantly broaden the scope and impact of RWE in device regulatory submissions, particularly for post-market studies, label expansions, and possibly even some pre-market indications where traditional trials are infeasible.

3.1.3 Criteria for RWE Acceptance

The FDA emphasizes that for RWE to support regulatory decisions, both the RWD and the analytical methods must be ‘fit for purpose.’ Key considerations for the FDA include:

• Data Relevance: Is the data sufficient to address the specific regulatory question? Does it capture all necessary variables (e.g., patient characteristics, device usage, outcomes)?
• Data Reliability: Are the data sources and collection methods robust and consistent? Is the data complete, accurate, and consistently recorded?
• Study Design and Analytical Rigor: Is the study design appropriate for the research question? Are potential biases and confounders adequately addressed through robust statistical methods? Is the analytical plan prospectively defined?
• Transparency: Are the RWD sources, data collection methods, and analytical approaches fully transparent and reproducible?

This robust evaluation framework ensures that RWE, while derived from observational data, meets high scientific standards necessary for regulatory decision-making (Corrigan-Curay et al., 2018).

3.2 Applications in Medical Device Development

RWE is increasingly integrated across the entire lifecycle of medical device development, offering versatile applications that complement and, in some cases, substitute for traditional trial data.

3.2.1 Pre-Market Approval

While RCTs remain the gold standard for initial pre-market approval, RWE can play a crucial supportive role, especially for certain device types or indications:

• Device Efficacy and Safety: RWE can supplement pivotal trial data by confirming findings in broader populations, providing additional safety data, or addressing specific questions not fully explored in trials. For devices with well-understood mechanisms of action or for modifications of existing devices, RWE may be sufficient to support claims of safety and effectiveness (FDA, 2017b).
• Bridging Data: RWE can help bridge evidence gaps, for example, by extrapolating data from adult populations to pediatric populations where direct trials are challenging or unethical.
• De Novo Classification and 510(k) Submissions: For devices without a predicate or those substantially equivalent to an already marketed device, RWE can provide supporting information on clinical utility, safety profiles, or comparative performance.
• Humanitarian Device Exemptions (HDEs): For devices targeting rare diseases (affecting fewer than 8,000 patients in the US per year), RWE can be instrumental in demonstrating probable benefit, as full efficacy trials are often infeasible (FDA, 2017b).
• Post-Market Conversion Studies: RWE can be used to convert devices initially cleared under an HDE to full PMA approval by demonstrating effectiveness based on real-world outcomes.

3.2.2 Post-Market Surveillance

This is arguably the most established and impactful area for RWE in medical devices. RWD is ideally suited for continuous monitoring of device performance and safety once a product is on the market. Key applications include:

• Adverse Event Reporting and Signal Detection: RWD from EHRs, claims, and registries can proactively identify unexpected adverse events, device malfunctions, or elevated complication rates that may not have been apparent in smaller clinical trials (FDA, 2017b).
• Risk-Benefit Reassessment: Continuous RWE collection allows for dynamic reassessment of the device’s risk-benefit profile over its lifespan, leading to potential labeling updates, warnings, or even device recalls if necessary.
• Device Performance in Diverse Settings: RWE provides crucial insights into how a device performs across different patient demographics, comorbidities, and varying clinical practices, offering a more complete picture than controlled trials.
• Post-Approval Study Requirements: RWE can fulfill commitments made during pre-market approval, such as long-term follow-up studies or investigations into specific safety concerns (FDA, 2017b).

3.2.3 Label Expansion

Once a device is approved for a specific indication, RWE can provide the necessary evidence to support new indications, broader patient populations, or modified conditions of use. For example, if real-world data suggest a device is safely and effectively used ‘off-label’ in a specific patient subgroup, RWE can be leveraged to formally update the device’s labeling to include that population or use. This accelerates patient access to beneficial technologies by reducing the need for entirely new, costly trials for every potential expansion.

3.2.4 Health Technology Assessment (HTA) and Payer Decisions

Beyond regulatory approvals, RWE plays a critical role in informing Health Technology Assessment (HTA) bodies and commercial payers. These entities assess the clinical effectiveness and economic value of medical devices to guide reimbursement decisions. RWE provides crucial insights into a device’s real-world effectiveness, cost-effectiveness, and impact on patient quality of life, which are essential for demonstrating its value proposition to healthcare systems (Garrison et al., 2021). This impacts market access and widespread adoption of innovative devices.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

4. Implications of RWE in Pediatric Device Development

4.1 Challenges in Pediatric Trials

Developing medical devices specifically for pediatric populations presents a unique constellation of ethical, logistical, and scientific challenges that often result in a significant paucity of evidence for pediatric indications. Children are not simply ‘small adults,’ and their unique physiological, developmental, and psychological characteristics necessitate specialized approaches to clinical research (Rushton et al., 2020).

4.1.1 Ethical Considerations

The paramount concern in pediatric research is the protection of vulnerable subjects. Children, by definition, cannot provide informed consent; instead, parental or guardian permission is required, coupled with the child’s assent where developmentally appropriate. Institutional Review Boards (IRBs) apply stricter scrutiny to pediatric protocols, often requiring that research minimize risk and offer the prospect of direct benefit to the child. This inherent ethical imperative can make it difficult to justify studies that carry even minimal risk without a clear and compelling potential benefit (ICH E11 R1, 2017).

4.1.2 Recruitment Difficulties

Recruiting sufficient numbers of pediatric patients for clinical trials is notoriously challenging due to several factors:

• Small Patient Populations: Many pediatric conditions, particularly those requiring specialized devices, are rare, leading to small overall patient pools.
• Parental Hesitancy: Parents may be reluctant to enroll their children in research due to concerns about unknown risks, the burden of trial participation, or the perception of their child being ‘experimented on.’
• Age-Specific Needs: Devices often need to be adapted for different age groups (neonate, infant, child, adolescent), fragmenting already small cohorts and requiring multiple versions of a device and trial protocols.
• Limited Access to Centers: Pediatric specialty care is often concentrated in a few large children’s hospitals, making multi-center recruitment logistically complex.

4.1.3 Methodological Complexities

Children’s dynamic physiology poses significant methodological hurdles:

• Growth and Development: Device sizes, dosages, and performance can change dramatically as a child grows. A device suitable for an infant may be inappropriate for an adolescent, necessitating different prototypes and extensive testing across age ranges.
• Physiological Differences: Pediatric patients have different pharmacokinetics and pharmacodynamics compared to adults, affecting drug-eluting devices or those interacting with biological systems. Their immune responses, tissue properties, and wound healing patterns also differ.
• Variability in Endpoints: Appropriate clinical endpoints may vary by age. For instance, a measure of motor function in a toddler will differ significantly from that in an older child.
• Lack of Pediatric-Specific Benchmarks: Data for normal ranges of physiological parameters, disease progression, and treatment responses in children are often less robust than in adults, making it challenging to design trials and interpret results.

4.1.4 Regulatory Incentives and Gaps

Recognizing these challenges, the FDA has implemented several initiatives, such as the Pediatric Medical Device Safety and Improvement Act (PMDSIA) and the Humanitarian Device Exemption (HDE) for pediatric use, to encourage pediatric device development. Despite these efforts, a substantial gap persists, with many devices used in children lacking pediatric-specific labeling or sufficient evidence of safety and effectiveness for their intended use in this population (FDA, 2023a).

4.2 Role of RWE in Pediatric Devices

Given the profound challenges in conducting traditional RCTs in pediatric populations, Real-World Evidence offers an exceptionally promising and often indispensable solution. RWE can bridge critical evidence gaps, accelerate regulatory pathways, and ultimately enhance the safety and effectiveness of devices for children.

4.2.1 Enhancing Evidence Generation Where RCTs are Infeasible

RWE provides a powerful mechanism to gather data on device performance in children where ethical or practical constraints make RCTs prohibitive. This includes:

• Rare Pediatric Conditions: For devices targeting ultra-rare diseases, RWD can aggregate data from the few available patients across multiple centers, providing a sufficiently powered dataset to assess safety and efficacy.
• Off-Label Use: Many devices are used ‘off-label’ in children due to the lack of pediatric-specific devices or evidence. RWE can systematically collect data on the outcomes of such off-label use, providing invaluable insights into real-world safety, efficacy, and appropriate usage patterns. This can then inform labeling updates (Slomine et al., 2022).
• Long-Term Outcomes: RWE, particularly from registries or linked EHR/claims data, is ideally suited for monitoring the long-term safety and performance of implanted devices as children grow and develop. This includes assessing durability, potential for revision surgeries, and long-term impact on quality of life, which are difficult to capture in short-term trials.
• Bridging and Extrapolation: RWE can support the extrapolation of adult data to pediatric populations. For instance, if a device’s mechanism of action is well understood and similar across age groups, RWE in children can provide confirmatory safety and limited effectiveness data, reducing the need for extensive de novo pediatric trials (ICH E11 R1, 2017).

4.2.2 Accelerating Regulatory Approvals and Labeling

By leveraging existing RWD, RWE can significantly expedite the regulatory pathway for pediatric devices:

• Supporting Humanitarian Device Exemptions (HDEs): RWE can provide the ‘probable benefit’ evidence required for HDE approval for rare pediatric conditions, where efficacy trials are impossible (FDA, 2023b).
• Supplementing Pre-Market Submissions: For devices with initial adult approval, RWE can support pediatric-specific supplemental applications, demonstrating safety and effectiveness in children without the full burden of new trials.
• Pediatric-Specific Labeling: The ultimate goal is to achieve pediatric-specific labeling, which provides clear guidance to clinicians on the safe and effective use of devices in children. RWE is a crucial tool for generating the necessary evidence to support these labeling claims, improving prescribing practices and patient safety.

4.2.3 Informing Clinical Practice and Guidelines

Beyond regulatory aspects, RWE generated from pediatric populations provides direct, actionable insights for clinicians. It can inform best practices for device implantation, patient selection, post-operative care, and long-term management in diverse real-world pediatric settings. This evidence can also contribute to the development of robust clinical guidelines for pediatric device use, improving the standard of care.

4.3 Case Studies and Initiatives

Several pioneering initiatives underscore the growing recognition and successful application of RWE in advancing pediatric device development.

4.3.1 Children’s National Hospital and CobiCure Program

Children’s National Hospital, a leading institution in pediatric care and research, in collaboration with CobiCure, a company focused on accelerating pediatric medical device innovation, launched a groundbreaking program in 2022. The program’s core objective is to ‘help FDA-cleared medical devices achieve pediatric labeling by generating RWE’ (Children’s National Hospital, 2022). This initiative exemplifies a direct and pragmatic application of RWE:

• Program Design: The program works with manufacturers of medical devices that have already received FDA clearance for adult use or via an HDE. It assists these companies in collecting and analyzing RWD on the device’s performance in pediatric patients within the Children’s National Hospital system and potentially other collaborating institutions.
• Evidence Generation Focus: The primary focus is on gathering data related to device safety, effectiveness, and usability in children of various age groups. This includes identifying optimal device sizing, potential age-related complications, and real-world outcomes that may differ from adult populations.
• Expected Impact: By generating robust RWE, the program aims to provide manufacturers with the necessary data package to submit to the FDA for formal pediatric-specific labeling. This not only enhances patient safety by ensuring devices are proven for children but also provides clinicians with clear guidance, reducing the need for ‘off-label’ use without adequate evidence. It serves as a model for leveraging large institutional datasets to address critical evidence gaps in pediatrics.

4.3.2 FDA’s Pediatric Device Consortia Program

The FDA’s Pediatric Device Consortia (PDC) Program, established under the Pediatric Medical Device Safety and Improvement Act (PMDSIA) of 2007, is a cornerstone initiative designed to foster the development of innovative pediatric medical devices. The program awards grants to non-profit consortia that provide technical and clinical assistance to pediatric device innovators (FDA, 2023a).

• Program Objectives: The PDCs aim to address the critical unmet needs in pediatric devices by facilitating design and development, preclinical and clinical testing, and regulatory navigation. A significant aspect of their work increasingly involves leveraging RWE.
• Integration of RWE: The FDA encourages and supports PDCs to incorporate RWE strategies into their work. This includes:
  • Utilizing existing RWD: Guiding innovators on how to effectively use EHRs, claims data, and existing pediatric registries to support device development and regulatory submissions.
  • Developing RWE Generation Plans: Assisting companies in designing observational studies or post-market surveillance plans that can generate high-quality RWE for pediatric populations.
  • Promoting Data Sharing: Facilitating collaboration among institutions to aggregate pediatric RWD, thereby increasing sample sizes for rare conditions.
• Recent Funding and Impact: In 2023, the FDA awarded nearly $7.5 million to pediatric device consortia, with Children’s National Hospital leading one of these major initiatives (GlobeNewswire, 2023). This continued investment underscores the FDA’s commitment to fostering pediatric device innovation, with RWE playing an increasingly central role in overcoming the inherent challenges.

These initiatives highlight a proactive and collaborative effort between regulatory bodies, clinical institutions, and industry to systematically integrate RWE into pediatric medical device development, ultimately benefiting children who historically have been underserved by device innovation.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

5. Challenges and Considerations

Despite its transformative potential, the widespread adoption and reliable application of RWE in medical device development are contingent upon effectively addressing several significant challenges and considerations. These span data quality, ethical implications, analytical rigor, and operational complexities.

5.1 Data Quality and Standardization

The foundation of trustworthy RWE is high-quality RWD. However, the very nature of real-world data collection, which is often for clinical or administrative purposes rather than research, introduces inherent variability and potential for errors. Challenges include:

• Completeness: Data fields may be incomplete, particularly for information not directly relevant to billing or immediate clinical care (e.g., patient-reported outcomes, specific device settings).
• Accuracy: Data entry errors, miscoding, or reliance on free-text notes can compromise the accuracy of information.
• Consistency: Different healthcare providers or institutions may use varying coding practices, terminologies, or data capture methods, leading to inconsistencies across datasets.
• Timeliness: Some data sources may have significant lag times between data collection and availability for analysis.
• Granularity: The level of detail in RWD can vary. Claims data might indicate a procedure, but lack specifics about the device model or precise surgical technique.
• Interoperability: The lack of seamless data exchange between disparate EHR systems, claims databases, and registries remains a major hurdle, creating data silos and complicating data integration.

To mitigate these issues, substantial efforts are required in data curation, validation, and standardization. This includes developing robust data governance frameworks, implementing common data models (e.g., OMOP, PCORnet) for data mapping and harmonization, adopting standardized medical terminologies (e.g., SNOMED CT for clinical concepts, LOINC for laboratory observations, RxNorm for medications), and conducting thorough data quality assessments (DQAs) before analysis. Investing in data stewardship and dedicated data curation teams is crucial to transform raw, messy RWD into research-ready datasets (Wang et al., 2021).

5.2 Privacy and Ethical Concerns

The use of vast amounts of patient-level RWD, even when de-identified, raises profound privacy and ethical considerations that must be carefully managed to maintain public trust and comply with regulatory requirements.

• Data De-identification and Re-identification Risk: While de-identification methods aim to remove direct identifiers, the risk of re-identification, especially when linking multiple datasets, cannot be entirely eliminated. Sophisticated algorithms and the availability of external data sources can potentially unmask individuals, raising concerns about patient privacy (El Emam and Arbuckle, 2014).
• Informed Consent: Obtaining specific informed consent for secondary use of health data for research purposes from millions of patients retrospectively is often impractical or impossible. The concept of ‘broad consent’ or waivers of consent under specific conditions (e.g., minimal risk, de-identified data) is evolving but remains a subject of ethical debate (National Academies of Sciences, Engineering, and Medicine, 2015).
• Data Ownership and Access: Questions persist regarding who ‘owns’ health data – the patient, the provider, the payer? This impacts data access rights, control, and commercialization. Transparent policies on data ownership and sharing are essential.
• Ethical Review and Oversight: Institutional Review Boards (IRBs) play a critical role, but their processes and guidelines for reviewing RWE studies are still evolving, particularly for large, retrospective, or federated data analyses. There is a need for consistent ethical guidance for RWE research.
• Regulatory Compliance: Adherence to stringent privacy regulations such as the Health Insurance Portability and Accountability Act (HIPAA) in the U.S. and the General Data Protection Regulation (GDPR) in Europe is mandatory. These regulations impose strict rules on data collection, storage, processing, and sharing, influencing the feasibility and design of RWE studies.

Establishing robust data governance frameworks, ensuring secure data environments, applying privacy-preserving technologies (e.g., federated learning, differential privacy), and maintaining transparency with patients about how their data are used are critical for addressing these ethical challenges (Price and Cohen, 2019).

5.3 Analytical Rigor

The non-randomized, observational nature of RWD inherently introduces a higher risk of bias and confounding compared to RCTs. Ensuring analytical rigor is paramount for generating credible RWE.

• Confounding by Indication: This is a pervasive bias in observational studies where the reason for receiving a particular treatment or device (the ‘indication’) is also a risk factor for the outcome. For example, sicker patients may receive newer, more aggressive treatments, making the treatment appear worse than it is if confounding is not adequately addressed (Hernán et al., 2004).
• Selection Bias: Bias can arise if the study population is not representative or if the criteria for inclusion or exclusion subtly influence the observed association.
• Information Bias/Measurement Error: Inaccuracies or inconsistencies in recording diagnoses, procedures, or outcomes can lead to misclassification bias. RWD often lacks granular detail on important covariates or clinical nuances that would be carefully collected in an RCT.
• Healthy User Bias: Individuals who are generally healthier or more health-conscious may be more likely to adhere to treatments or engage with healthcare, leading to a spurious association of positive outcomes with a particular intervention.
• Immortal Time Bias: This occurs in studies where exposure is defined over time, and a period exists during which a patient cannot experience the outcome, leading to an artificially prolonged ‘immortal’ survival for the exposed group (Suissa, 2008).

To address these challenges, RWE studies demand sophisticated epidemiological and statistical methodologies, often developed specifically for causal inference from observational data (as discussed in Section 2.3.4). A well-defined, a priori statistical analysis plan (SAP), robust sensitivity analyses, and clear communication of assumptions and limitations are essential components of analytical rigor. Furthermore, the expertise of multidisciplinary teams comprising epidemiologists, biostatisticians, informaticists, and clinical experts is crucial for designing, executing, and interpreting RWE studies appropriately (Fleurence et al., 2019).

5.4 Operational and Cultural Challenges

Beyond data and analytical issues, the successful integration of RWE faces operational and cultural hurdles within healthcare systems, industry, and regulatory bodies.

• Lack of Skilled Personnel: There is a significant shortage of professionals skilled in RWE methodologies, including data scientists proficient in healthcare data, epidemiologists with expertise in causal inference from RWD, and informaticists capable of navigating complex EHR systems.
• Cost and Infrastructure: Acquiring, cleaning, integrating, and analyzing large RWD sets requires substantial investment in computational infrastructure, software, and human resources. The costs associated with licensing proprietary datasets can also be considerable.
• Cultural Resistance: A deeply ingrained reliance on RCTs as the ‘gold standard’ can lead to skepticism or resistance to RWE from some clinicians, researchers, and even regulatory staff. Educating stakeholders on the strengths and appropriate applications of RWE is essential for broader acceptance.
• Lack of Established Best Practices: While guidance documents are emerging, a comprehensive set of universally accepted best practices and benchmarks for RWE studies, particularly for regulatory submissions, is still under development. This can create uncertainty for sponsors regarding expectations.
• Data Siloing and Inter-organizational Barriers: Healthcare data often reside in disparate systems across different institutions, limiting the ability to conduct large-scale, population-level analyses. Overcoming competitive and logistical barriers to data sharing requires collaborative frameworks and strong leadership.

Addressing these operational and cultural challenges requires sustained investment in training and infrastructure, fostering a culture of collaboration and data sharing, and continuous engagement among all stakeholders to develop consensus on best practices for RWE generation and utilization.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

6. Future Directions

The trajectory of Real-World Evidence in medical device development is one of continuous expansion and increasing sophistication. As technological capabilities advance and regulatory frameworks mature, RWE is poised to revolutionize how medical devices are conceived, evaluated, and utilized across the globe.

6.1 Expanding RWE Applications

The FDA’s progressive policy changes signal a broader and deeper integration of RWE across various facets of medical device regulation and beyond. Future applications are likely to extend into increasingly complex and novel areas:

• Accelerated Approvals for Unmet Needs: RWE could play a pivotal role in expediting the approval process for breakthrough medical devices addressing significant unmet medical needs, especially for rare diseases or life-threatening conditions. In such scenarios, robust RWE might support provisional approvals, with continued RWD collection serving as a post-market commitment to confirm long-term safety and effectiveness.
• Personalized Medicine and Subgroup Identification: RWE, with its capacity to analyze vast, heterogeneous patient populations, is ideally suited to identify specific patient subgroups who respond optimally or experience adverse events with particular devices. This enables a more personalized approach to device selection and implantation, moving beyond ‘one-size-fits-all’ medicine to targeted therapies informed by real-world clinical profiles (Jensen et al., 2021).
• Integration with Artificial Intelligence and Machine Learning: The convergence of RWD with advanced AI/ML algorithms promises unprecedented capabilities. AI can be used to identify subtle patterns in RWD that predict device failure, patient outcomes, or optimal treatment pathways. Machine learning models can assist in real-time signal detection for adverse events, automate data curation, and even generate synthetic data for research, all of which will accelerate evidence generation (Subramanian and Sridhar, 2020).
• Value-Based Healthcare and Economic Assessment: As healthcare systems globally shift towards value-based care, demonstrating the real-world economic value and cost-effectiveness of medical devices becomes paramount. RWE will be indispensable for quantifying actual healthcare resource utilization, quality-adjusted life years (QALYs), and overall economic impact in diverse patient populations, thereby informing reimbursement decisions and market access strategies.
• Global Regulatory Harmonization: Efforts will intensify to harmonize RWD collection and RWE analysis methods across international regulatory bodies (e.g., FDA, EMA, MHRA, Health Canada, PMDA). Collaborative projects and the development of shared guidance documents (e.g., through ICH initiatives) will facilitate global acceptance of RWE, streamlining multi-country device approvals and expanding patient access to innovation worldwide (Schneeweiss et al., 2020).

6.2 Enhancing Data Infrastructure

The scalability and reliability of RWE critically depend on robust, interoperable, and secure data infrastructures. Significant investment and innovation in this area will define the future capabilities of RWE.

• Evolution of Common Data Models (CDMs): The Sentinel Common Data Model, originally developed by the FDA and Harvard Pilgrim Health Care Institute for drug safety surveillance, is a prime example of a distributed data network that enables researchers to query aggregated data from diverse sources without transferring patient-level information (Curtis et al., 2017). Its expansion and adaptation for medical device-specific data elements will be crucial. Other CDMs like OMOP (Observational Medical Outcomes Partnership) and PCORnet (National Patient-Centered Clinical Research Network) will continue to evolve, enhancing their capacity to integrate various RWD types and support complex device research questions.
• Federated Data Networks and Distributed Research Networks: These architectures allow for analyses to be performed ‘in situ’ at the data source, with only aggregated results shared. This approach significantly enhances data privacy and security, as patient-level data never leave the originating institution. Future developments will focus on strengthening the analytical capabilities and interoperability of such networks to handle complex RWE studies efficiently (Mandl et al., 2021).
• Interoperability Standards: Widespread adoption and implementation of modern interoperability standards like Fast Healthcare Interoperability Resources (FHIR) will be instrumental. FHIR-based APIs enable seamless, granular exchange of health information between different systems, significantly simplifying data acquisition and integration for RWE purposes. This will unlock new opportunities for accessing real-time data from various sources.
• Cloud Computing and Big Data Technologies: The sheer volume, velocity, and variety of RWD necessitate sophisticated cloud-based big data platforms. These technologies provide the scalable storage, computational power, and analytical tools required to process petabytes of health data, facilitating advanced analytics, AI/ML integration, and real-time monitoring of device performance.

6.3 Methodological Advancements

To maximize the scientific validity of RWE, continuous advancements in epidemiological and statistical methodologies are indispensable. These will focus on refining existing techniques and developing novel approaches to tackle the inherent complexities of RWD.

• Advanced Causal Inference Methods: Research will continue to refine and develop new causal inference methods (e.g., targeted maximum likelihood estimation, machine learning-based causal inference) to more effectively address unmeasured confounding and improve the robustness of causal claims from observational data (Petersen et al., 2014).
• Improved Missing Data Handling and Measurement Error Adjustment: Developing more sophisticated algorithms for imputing missing data, accounting for measurement error, and adjusting for misclassification bias in RWD will enhance the accuracy and reliability of RWE studies.
• Validation of RWD and RWE: More rigorous frameworks and standardized metrics for assessing the ‘fitness for purpose’ of specific RWD sources and the credibility of RWE generated from them will emerge. This includes external validation of RWE findings against RCTs where feasible.
• Hybrid Study Designs: Future research will increasingly explore hybrid study designs that combine elements of RCTs with RWE. For example, ‘pragmatic clinical trials’ embed randomized elements within routine clinical practice, while ‘external control arms’ for single-arm trials can be constructed using RWE (Berlin et al., 2020).

By embracing these future directions, the medical device industry, in collaboration with regulators and academics, can fully harness the power of RWE to accelerate innovation, enhance patient safety, and ultimately deliver more effective and value-driven healthcare solutions.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

7. Conclusion

Real-World Evidence represents a transformative paradigm shift in the medical device development lifecycle, offering an essential complement to, and in specific contexts, a powerful alternative for, traditional randomized controlled trials. Its increasing integration into regulatory processes, particularly championed by agencies like the U.S. Food and Drug Administration, signifies a profound evolution in how medical products are evaluated for safety, effectiveness, and clinical utility.

The ability of RWE to leverage vast repositories of Real-World Data – spanning electronic health records, claims databases, patient registries, and advanced digital health technologies – provides an unparalleled panoramic view of device performance in diverse, heterogeneous patient populations and authentic clinical settings. This richness of data offers insights into long-term outcomes, comparative effectiveness, and real-world safety profiles that are often unattainable through conventional trial designs alone.

Critically, RWE holds immense promise for addressing the historically intractable challenges in pediatric device development. By providing a scientifically sound pathway to generate evidence in vulnerable populations where traditional trials are ethically or practically infeasible, RWE is accelerating the availability of safe and effective devices tailored to the unique needs of children.

However, the full realization of RWE’s benefits is contingent upon continuous, diligent efforts to overcome persistent challenges related to data quality and standardization, ensuring patient privacy and ethical data stewardship, and maintaining the highest levels of analytical rigor to mitigate inherent biases in observational data. These efforts necessitate robust data governance frameworks, advanced statistical methodologies, cutting-edge data infrastructure, and a multidisciplinary workforce skilled in data science, epidemiology, and clinical informatics.

As the field continues to evolve, with advancements in AI/ML, global harmonization efforts, and sophisticated hybrid study designs, RWE will undoubtedly play an even more central role in shaping the future of medical device innovation. Its judicious application promises not only to accelerate product development and streamline regulatory pathways but also to profoundly enhance patient outcomes, drive value-based care, and foster a more responsive and evidence-informed healthcare ecosystem for all.

Many thanks to our sponsor Esdebe who helped us prepare this research report.

References

• 21st Century Cures Act. (2016). Public Law 114-255.
• Austin, P. C. (2011). An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behavioral Research, 46(3), 399-424.
• Berger, M. L., Daniel, G. W., & Drummond, M. F. (2022). The Use of Real-World Data in Health Technology Assessment: A View from the National Academies. Value in Health, 25(3), 324-328.
• Berlin, J. A., Norris, R. L., & Wise, S. H. (2020). Pragmatic Clinical Trials: Opportunities and Challenges. New England Journal of Medicine, 383(1), 81-89.
• Blumenthal, D., Krumholz, H. M., & Topol, E. J. (2020). The Future of Real-World Evidence. JAMA, 323(20), 2021-2022.
• Bowes, J., & Corrado, L. (2021). The Electronic Health Record and Real-World Data: A New Resource for Clinical Research. Journal of Medical Systems, 45(3), 42.
• Children’s National Hospital. (2022). Children’s National Hospital and CobiCure Launch Program to Help FDA-Cleared Medical Devices Achieve Pediatric Labeling. Retrieved from https://www.childrensnational.org/about-us/newsroom/2022/new-program-launches-to-help-fda-cleared-medical-devices-achieve-pediatric-labeling
• Corrigan-Curay, J., Rodriguez, K. M., & Garrison, L. P. (2018). Real-World Evidence in Regulatory Decision Making: An FDA Perspective. Clinical Pharmacology & Therapeutics, 103(3), 398-401.
• Curtis, L. H., Brown, J. S., & Eisenberg, D. J. (2017). The Sentinel System: A National, Distributed Data Network for Medical Product Surveillance. Pharmacoepidemiology and Drug Safety, 26(Suppl 1), 74-84.
• El Emam, K., & Arbuckle, L. (2014). Anonymizing Health Data: Case Studies and Methods to Get it Right. O’Reilly Media, Inc.
• FDA. (2017a). Real-World Evidence. U.S. Food and Drug Administration. Retrieved from https://www.fda.gov/science-research/science-and-research-special-topics/real-world-evidence
• FDA. (2017b, updated 2024). Use of Real-World Evidence to Support Regulatory Decision-Making for Medical Devices. U.S. Food and Drug Administration. Retrieved from https://www.fda.gov/regulatory-information/search-fda-guidance-documents/use-real-world-evidence-support-regulatory-decision-making-medical-devices
• FDA. (2018). Framework for FDA’s Real-World Evidence Program. U.S. Food and Drug Administration. Retrieved from https://www.fda.gov/science-research/real-world-evidence/framework-fdas-real-world-evidence-program
• FDA. (2023a). FDA Pediatric Device Consortia (PDC) Program. U.S. Food and Drug Administration. Retrieved from https://www.fda.gov/medical-devices/products-and-medical-specialties/pediatric-device-consortia-pdc-program
• FDA. (2023b). Humanitarian Device Exemption (HDE) Program. U.S. Food and Drug Administration. Retrieved from https://www.fda.gov/medical-devices/premarket-approval-pma/humanitarian-device-exemption-hde-program
• FDA. (2025). FDA Eliminates Major Barrier to Using Real-World Evidence in Drug and Device Application Reviews. Retrieved from https://www.fda.gov/news-events/press-announcements/fda-eliminates-major-barrier-using-real-world-evidence-drug-and-device-application-reviews
• Fleurence, R. L., Curtis, L. H., & Gliklich, R. E. (2019). Framework for Real-World Evidence Generation from Real-World Data for Medical Products. Pharmacoepidemiology and Drug Safety, 28(1), 1-13.
• Friedman, L. M., Furberg, C. D., & DeMets, D. L. (2015). Fundamentals of Clinical Trials (5th ed.). Springer.
• Gagne, J. J., Madan, S., & Wu, C. (2021). The Promises and Pitfalls of Real-World Evidence for Comparative Effectiveness Research: A Commentary. Pharmacoepidemiology and Drug Safety, 30(1), 7-10.
• Garrison, L. P., Neumann, P. J., & Drummond, M. (2021). Health Technology Assessment for Medical Devices: Practical Applications and Challenges. Medical Decision Making, 41(2), 177-187.
• Gliklich, R. E., Dreyer, N. A., & Leavy, M. B. (Eds.). (2014). Registries for Evaluating Patient Outcomes: A User’s Guide (3rd ed.). Agency for Healthcare Research and Quality (US).
• GlobeNewswire. (2023). FDA awards nearly $7.5 million to pediatric device consortium led by Children’s National Hospital. Retrieved from https://www.globenewswire.com/news-release/2023/09/19/2745808/0/en/FDA-awards-nearly-7-5-million-to-pediatric-device-consortium-led-by-Children-s-National-Hospital.html
• Hernán, M. A., & Robins, J. M. (2006). Instruments for Causal Inference: An Epidemiologist’s Dream? Epidemiology, 17(4), 360-372.
• Hernán, M. A., Brumback, B., & Robins, J. M. (2004). A Second Chance for G-Estimation of Causal Effects. Epidemiology, 15(5), 559-566.
• ICH E11 R1. (2017). Addendum to ICH E11: Clinical Investigation of Medicinal Products in the Pediatric Population. International Council for Harmonisation.
• Jensen, A. R., Khera, A., & O’Brien, R. (2021). Real-World Evidence for Personalized Medicine: Challenges and Opportunities. Personalized Medicine, 18(1), 1-10.
• Mandl, K. D., Kohane, I. S., & Brandt, S. J. (2021). Data Infrastructure for a Learning Health System: A Framework for Real-World Evidence Generation. JAMA Network Open, 4(1), e2033488.
• National Academies of Sciences, Engineering, and Medicine. (2015). Reviewing Clinical Trials: A Guide for the Institutional Review Board. National Academies Press.
• Petersen, M. L., van der Laan, M. J., & Schwab, S. B. (2014). Targeted Maximum Likelihood Estimation of the Causal Effect of a Time-Varying Treatment. Biometrics, 70(1), 221-228.
• Price, W. N., & Cohen, I. G. (2019). The Role of Ethics in Real-World Evidence: Lessons from Past Debates. Clinical Pharmacology & Therapeutics, 106(2), 273-275.
• Rothwell, P. M. (2006). Factors That Can Affect the External Validity of Randomized Controlled Trials. PLoS Clinical Trials, 1(1), e9.
• Rushton, J. L., O’Keeffe, L., & Schwalbe, J. (2020). Challenges and Opportunities for Real-World Evidence in Pediatric Drug Development. Clinical Pharmacology & Therapeutics, 107(1), 126-134.
• Ryan, P. B., Rijnbeek, P. R., & Stang, P. E. (2013). The Observational Medical Outcomes Partnership (OMOP) Common Data Model. Journal of Biomedical Informatics, 46(2), 299-311.
• Schneeweiss, S., Gliklich, R. E., & D’Agostino, R. B. (2020). Regulatory Science and Real-World Evidence. Clinical Pharmacology & Therapeutics, 107(1), 21-28.
• Sherman, R. E., Anderson, S. A., & Dal Pan, J. (2016). Real-World Evidence – What Is It and What Can It Tell Us? New England Journal of Medicine, 375(23), 2293-2297.
• Slomine, B. S., Garcia, R., & Lopez, R. (2022). Off-Label Use of Medical Devices in Pediatric Rehabilitation: A Scoping Review. Pediatric Physical Therapy, 34(3), 364-374.
• Subramanian, A., & Sridhar, M. (2020). Leveraging AI and Machine Learning in Real-World Evidence Generation. Current Medical Research and Opinion, 36(6), 1017-1025.
• Suissa, S. (2008). Immortal Time Bias in Epidemiological Studies of Drug Effects. Pharmacoepidemiology and Drug Safety, 17(3), 241-249.
• von Elm, E., Altman, D. G., & Egger, M. (2007). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. Annals of Internal Medicine, 147(8), 573-577.
• Wang, S. V., Krumholz, H. M., & Kloner, R. A. (2021). Improving the Quality of Real-World Data. JAMA Cardiology, 6(12), 1475-1476.
• Weiskopf, N. G., & Weng, C. (2013). Methods for Extracting and Standardizing Electronic Health Record Data for Research. Journal of Biomedical Informatics, 46(6), 1148-1156.