FDA & EMA RWD Compliance
in Obesity Trials:
The Direct-to-Patient Action Plan

How clinical operations leaders are replacing manual site-based models with DTPDirect-to-Patient (DTP): A study model in which patients participate remotely through digital channels, including EHR-based data collection and ePRO surveys, without traveling to a physical clinical site. Also referred to as decentralized clinical trial (DCT) design. chart review to execute 5,000+ patient obesity megatrials, meeting strict regulatory mandates while dramatically compressing timelines and budgets.

Executive summary

The case for a Direct-to-Patient RWE model in obesity

Over 2,000 active obesity and GLP-1 studies are competing for the same patients^[8] while regulators have published clear frameworks defining what qualifies as acceptable real-world evidence. The traditional site-based model (built for interventional trials) cannot scale to 5,000-patient registries without breaking budgets and introducing the manual data errors the FDA explicitly flags as disqualifying. This guide presents the operational architecture for a DTP study model: how it connects to patient EHRElectronic Health Record (EHR): A digital version of a patient’s medical history maintained across healthcare providers, and the primary source for real-world evidence studies. records via TEFCATrusted Exchange Framework and Common Agreement (TEFCA): A U.S. nationwide framework standardizing health information exchange, allowing patients and authorized parties to access records across systems using a single set of technical and policy requirements., how it prevents identity fraud, where it eliminates cost, and how to evaluate whether a technology partner actually meets the traceability standard the FDA and EMA require.

The saturated market for obesity trials

As of early 2026, the clinical trials landscape holds over 2,000 active obesity, GLP-1, and weight-management studies globally.^[8] The demand for long-term real-world evidence to prove cardiovascular outcomes, therapy durability, and muscle-mass retention continues to grow at pace. The market is flooded with competing trials, and the competition for eligible patients is fierce. You are not just fighting for data. You are fighting over the right patients.

Relying on traditional clinical sites to find these patients and manually abstract their data is no longer viable. Manual abstraction of complex metabolic histories takes multiple hours per patient record. It introduces a documented 7% error rate, burdens site coordinators, and triggers significant costs for SDVSource Data Verification (SDV): Comparing EDC data against original source documents to confirm accuracy. Manual on-site SDV is one of the largest cost drivers in traditional RWE studies., a critical line item for rigorous large-scale registrational and safety studies.^[3]

Why retrospective database studies fall short

Many sponsors confronting the cost and complexity of prospective RWE studies turn to retrospective database studies instead. The appeal is clear: existing claims or EHR datasets are cheaper to access, timelines compress from years to months, and no patient recruitment is required. For hypothesis generation, safety signal detection, and comparative effectiveness, retrospective databases serve a real purpose.

But for sponsors who need prospective longitudinal data – particularly for GLP-1 outcomes where durability, discontinuation patterns, and patient-reported quality of life matter — retrospective sources have significant blind spots:

Retrospective database studies

No protocol control over data collection; endpoints depend on what was captured during routine care
No ePRO or patient-reported outcomes; quality of life and symptom burden are absent or inconsistent
Traceability is limited; source document linkage is difficult or impossible to establish
Cannot capture future events; data cut-offs mean no prospective safety follow-up
Selection bias from existing datasets; no ability to define inclusion criteria prospectively

DTP prospective RWE model

Protocol-driven data collection with defined endpoints from day one
Native ePRO integration captures patient-reported outcomes longitudinally
Full source audit trail from EHR to EDC satisfies FDA/EMA traceability requirements
Ongoing EHR re-pulls capture long-term safety events that patients may not self-report
Prospective consent and defined inclusion criteria control for selection bias

Retrospective databases remain valuable for early signal detection and hypothesis generation. When the study objective requires longitudinal patient engagement, prospective endpoint capture, or regulatory-grade source documentation, the DTP model is the stronger architecture.

The regulatory reality check: FDA & EMA directives

Global regulators have finalized their frameworks on real-world data, and the direction is clear: properly structured RWD now has a defined pathway to regulatory acceptance. The FDA has taken meaningful steps to enable the use of real-world data in regulatory submissions. The FDA’s Final RWD Guidance^[1] establishes the requirements that qualify RWD for regulatory use, with rigorous patient-level traceability as the foundational standard. The EMA‘s Data Quality Framework (DQFData Quality Framework (DQF): The EMA’s framework specifying standards for data quality, provenance, and governance in real-world data used to support EU regulatory submissions.)^[2] aligns on these same principles.

For sponsors, this means one thing: studies that can demonstrate a clear, auditable connection between EDC data points and source medical records are positioned to take full advantage of this expanding regulatory pathway. Those that cannot are not.

If your study cannot provide a clear, auditable connection linking an EDC data point back to the source medical record, your evidence is at risk of regulatory rejection, regardless of study size or scientific rigor.

Identity verification & scam prevention

A direct side effect of the booming obesity market, and the lucrative compensation often tied to trial participation, is the rise of professional patients and duplicate enrollments. Relying on self-reported screening questionnaires leaves sponsors highly vulnerable to fraudulent data.

By shifting to an EHR-driven model, you natively solve this. Retrieving a patient’s historical medical records acts as a direct, tamper-proof identity verification step. If a patient claims to have a verified obesity diagnosis, the direct API connection to their healthcare provider’s system validates their clinical history and identity instantly, locking out bad actors.

TEFCA integration & Fasten Health reduces patient friction

The interoperability landscape is evolving with the rollout of TEFCA. Through Castor’s partnership with Fasten Health, we use TEFCA to drastically reduce patient onboarding friction.

Instead of navigating complex portals, patients log in using their familiar MyChart or local EMRElectronic Medical Record (EMR): A digital record of a patient’s medical history within a single healthcare organization. Often used interchangeably with EHR, though EHR typically refers to records shared across organizations. credentials. This direct connection also acts as a secure identity verification check, combining access and authentication into a single step.

The trap: data brokers vs. unified platforms

As DTP models gain traction to solve the recruitment and data-entry crisis, clinical operations teams face a critical architectural choice: buy a dataset from a Data Broker, or build the study on a unified eClinical platform?

Aggregators and Data Brokers offer a seductive pitch: “We will collect the records from patients and sell you the curated dataset.” But for a sponsor running a prospective obesity registry, this creates massive downstream failures.

The Data Broker approach (e.g., siloed aggregators)	The Castor Catalyst integrated platform (Catalyst + unified EDC)
You rent a static dataset. They curate existing data and sell you access to their proprietary cohort.	You own a living study. You control the protocol, the patient relationship, and the 21 CFR Part 11 electronic data capture system infrastructure.
Prospective disconnect. You cannot trigger an ePRO diary based on a specific clinical event (e.g., GI distress) found in the retrospective chart.	Native ePRO integration. Retrospective EMR extraction automatically triggers protocol-specific, longitudinal ePROs via SMS in one system.
Black box curation. It is incredibly difficult for sponsors to perform independent, cost-effective SDV to satisfy FDA/EMA audits.	Direct traceability. Extracted data points are visually linked to their source documents within the EDC, streamlining remote SDV.

Step-by-step collaborative workflow

Transitioning from a site-heavy manual registry to an automated DTP study requires careful alignment. Below is the exact operational framework showing how your clinical operations team partners with Castor to deploy the study.

Phase 1: Clinical Operations Strategy (Sponsor + Castor)

Central IRBInstitutional Review Board (IRB): An independent ethics committee that reviews and approves research involving human subjects. A centralized IRB covers multiple sites under a single approval, eliminating local IRB submissions at each site. submission, protocol alignment & patient recruitment strategy

Instead of managing 50 to 100 local IRBs, your team works directly with Castor to design study documents for a single, central IRB submission. We define the DTP framework, consent language, and data release authorization materials together.

Critically, recruitment strategy is defined at this stage. DTP studies recruit through digital channels, such as social media campaigns, healthcare provider referrals via EHR alerts, and patient community partnerships. Your team defines the target population profile, the digital outreach channels, and the pre-screening logic that will filter eligible patients before they enter the consent workflow. This is where the conversion funnel from initial contact to enrolled participant is designed — not after launch.

Map the ingestion strategy to the protocol

Your clinical leads map out exactly which data points are required to satisfy the registry’s endpoints. Together, we establish the hybrid ingestion plan: structured labs via FHIRFast Healthcare Interoperability Resources (FHIR): An international standard for exchanging healthcare information electronically, enabling structured clinical data (labs, diagnoses, medications) to be shared across systems via API. APIs and unstructured clinician notes via HIPAAHealth Insurance Portability and Accountability Act (HIPAA): U.S. legislation setting national standards for the protection of individually identifiable health information, governing how covered entities and business associates handle patient data. authorizations. You also define the schedule for longitudinal EHR re-pulls throughout the study duration.

Define data review workflows & dashboards

Your team defines the required level of data oversight. You decide which fields require strict HITLHuman-in-the-Loop (HITL): A model in which qualified human reviewers are embedded in an automated workflow to review and approve system outputs before data is committed. In RWE studies, HITL review is performed by medically trained clinical staff. clinical review by qualified clinical staff before committing to the EDC.

Phase 2: Platform Deployment & Patient Workflow (Castor Executed)

Deploying the digital front door

Castor builds and hosts the secure patient landing page. This includes customized pre-screening logic, multimedia study materials, 21 CFR Part 11 compliant eConsent, and digital release of information forms.

Automated TEFCA/EHR authentication

Patients securely authenticate into their provider portals (e.g., MyChart) using Castor’s TEFCA integration. This step natively handles identity verification to prevent fraudulent enrollments while unlocking the historical medical record.

Longitudinal execution at scale

Once baseline data is extracted, Castor automatically schedules and deploys the selected ePRO surveys to the patient’s phone. Because the infrastructure is centralized, you can scale recruitment nationally without site contracts or site initiation visits.

EHR data collection does not stop at baseline. Castor schedules recurring EHR re-pulls throughout the study duration, capturing long-term safety events, hospitalizations, lab changes, and new diagnoses that the patient may not self-report via ePRO. This longitudinal passive data collection is especially important for GLP-1 cardiovascular outcomes studies where adverse events can occur months or years after baseline enrollment.

Proven at scale: Castor's metabolic footprint

Castor’s credibility spans dedicated obesity studies and a broad ecosystem of metabolic and endocrine programs supporting biotech and pharma clinical trials globally.

937

Metabolic & endocrine studies hosted

54

Commercial metabolic/obesity trials

915

Avg. ePRO data points per patient (bariatric)

In pure obesity, Castor hosts 37 dedicated academic studies, including major bariatric registries and weight management interventions. The Aruba National Bariatric Surgery Registry tracks 1,405 patients longitudinally. Another major bariatric cohort on the platform averages 915 ePRO survey data points per participant, proving our capability for heavy, longitudinal patient-reported outcomes required in modern obesity trials.

Where the platform’s footprint truly deepens is in the broader metabolic medicine landscape. We host 937 broader metabolic studies (54 commercial). 61% of our academic obesity studies utilize integrated ePRO for dietary tracking, QoL, and symptom monitoring.^[4]

Editorial note (remove before distributing): Two versions of the sponsor paragraph appear below. Use the unblinded version only for restricted internal distribution and named-account outreach. Use the blinded version for general distribution, website publication, and any channel where sponsor name disclosure requires prior approval.

Unblinded — restricted distribution only

Our registry metadata includes the programs defining the field: Novo Nordisk (including the PILLAR Wegovy RWE study and AMAZE 1), Boehringer Ingelheim (survodutide Phase 2 & Phase 3 CV outcomes), AstraZeneca (MEDI0382), and Boston Scientific (Orbera365).

Blinded — for general distribution

Our registry metadata spans the leading programs defining the field, including major GLP-1 receptor agonist programs with Phase 3 cardiovascular outcomes data, leading weight management interventional programs, and medical device manufacturers in the metabolic space.

The ROI of automation: a financial blueprint

For an obesity RWE study tracking GLP-1 outcomes across 1,000 patients over two years, shifting from a multi-site model to a direct-to-patient framework reduces total cost of ownership by roughly half, even accounting for higher recruitment and platform investment. The savings are driven by eliminating most site management and monitoring overhead, reducing manual SDVSource Data Verification (SDV): The process of comparing EDC data against original source documents. Manual on-site SDV is one of the largest cost drivers in traditional RWE studies; the DTP model replaces it with targeted remote review. through targeted remote review, and replacing manual data review and query workflows with automated edit checks. This is one of the core value propositions of decentralized clinical trial infrastructure at scale.

Cost center (N=1,000, 2-year)	Traditional site model	Castor Catalyst
Infrastructure & startup
Site selection & activation 20 sites @ $15K ea vs 1 virtual site	$300,000	$5,000
Site management & monitoring $1K/mo/site × 24 mo vs remote oversight	$480,000	$75,000
IRB management 20 local IRBs vs 1 central IRB	$80,000	$10,000
Recruitment & patient engagement
Recruitment activities Site-based referral vs DTP digital campaigns	$50,000	$165,000
Patient stipends Office visit compensation vs remote (DTP lower)	$100,000	$60,000
eCOA / ePRO software Patient-reported outcomes software licensing	$110,000	$110,000
Data operations & quality
Data abstraction / ingestion Manual CRAClinical Research Associate (CRA): A professional responsible for monitoring clinical trial sites to verify data quality and protocol compliance. CRA loaded rates are a primary input in site monitoring cost estimates. @ ~$375/pt vs automated ingestion	$375,000	$270,000
Source Data Verification On-site CRA visits vs targeted remote SDV	$200,000	$55,000
Data review & query resolution Manual listing reviews, DVMs, DMPs vs automated edit checks. Includes data review, query generation (minimized via automation), and data validation monitoring	$80,000	$15,000
EDC / platform licensing Standard EDC base vs EDC + Castor Catalyst	$75,000	$90,000
Data management & biostatistics CDISC mapping, statistical programming, TLFs, DMP	$150,000	$150,000
Total estimated TCOTotal Cost of Ownership (TCO): The complete financial cost of operating a system or process over its full lifecycle, including startup, execution, and overhead.	$2,000,000	$1,105,000

~50%

Estimated net savings

Reduction in total cost of ownership

*Estimates model a 1,000-patient, 20-site, 2-year GLP-1 obesity outcomes study. Traditional costs assume standard CRA loaded rates and manual abstraction site grants. Catalyst costs include full platform licensing, DTP recruitment, and continuous clinical data review.^[5]

Evaluating RWE automation technology

When evaluating a technology partner to execute this action plan, use these objective criteria to ensure the platform complies with global data provenance mandates and avoids the limitations of simple data brokers.

Source data traceability

Does the system provide a clear, auditable connection linking the final data point in the EDC back to its original location in the native source document?

How Castor Catalyst does it
Castor provides a secure visual audit trail that actively links extracted data in the EDC directly to the source document location, enabling rapid, remote SDV.

Native hybrid data ingestion & identity

Does the system provide a clear, auditable connection linking the final data point in the EDC back to its original location in the native source document?

How Castor Catalyst does it
Castor utilizes a dual-engine approach, parsing rich FHIR clinical notes while handling heavy legacy PDFs. Patient credentials integration acts as a direct, tamper-proof identity verification check.

Human-in-the-Loop (HITL) workflows

Are there native UI workflows for medically trained staff to review, override, and approve automated extractions before the data is committed to the database?

How Castor Catalyst does it
Catalyst includes a native, intuitive clinical review interface where medically trained clinical staff review and approve every extracted value before it is committed. No black box automated commits.

Unified eClinical ecosystem

Can the EMR data ingestion engine interact with the EDC and ePRO without custom API work?

How Castor Catalyst does it
Because Catalyst is built directly on Castor’s core eClinical suite, a new diagnostic code extracted from the EMR can instantly trigger a protocol-specific quality-of-life ePRO survey to the patient’s phone. No custom integration required.

Patient onboarding journey: from recruited participant to source data

How many steps does it take for a recruited participant to go from first contact to successfully consented and with their EHR data in the EDC? What is the typical drop-off rate at each step, and does the platform provide tools to recover participants who abandon the process?

How Castor Catalyst does it
Castor’s patient onboarding workflow is designed to minimize friction at every step: digital pre-screening, single-click TEFCA authentication via familiar MyChart credentials, eConsent with multimedia study materials, and automated data extraction in one continuous flow. Incomplete onboarding sessions are tracked in the study dashboard, allowing your team to identify and re-engage participants who did not complete the process. The goal is a single, uninterrupted session from initial contact to verified enrollment.

When this approach won't work

The DTP model is not appropriate for every study. Before committing to this architecture, clinical operations teams should assess the following constraints.

EHR coverage gaps
TEFCA enables connectivity to major health systems, but coverage is not universal. Patients receiving care from small independent practices, rural providers without certified EHR systems, or federally qualified health centers with legacy infrastructure may not be reachable through standard FHIR APIs.

Highly interventional endpoints
DTP models are optimized for observational and registry studies where data already exists in the EHR. If your protocol requires frequent in-person procedures, physical sample collection, or hands-on assessments, a hybrid or fully site-based model will be more appropriate.

Patients with low digital access
Enrollment through EHR portals and ePRO via smartphone assumes a baseline level of digital literacy and device access. Studies targeting older populations or communities with limited smartphone penetration should plan supplemental support channels alongside the DTP architecture.

International data residency requirements
TEFCA applies to U.S. health systems. For international studies, sponsors must assess local health information exchange frameworks, data residency laws, and consent requirements in each jurisdiction before assuming a comparable DTP architecture is achievable.

Novel endpoint validation
If your study captures a novel endpoint that does not map to existing FHIR data elements, structured extraction will require custom field mapping or manual abstraction for those specific variables. The cost savings on standard data collection remain, but novel endpoint capture adds development time.

Stop renting data.

Own the study.

Castor Catalyst is one of the few platforms that combines programmatic DTP EMR acquisition, automated curation, and a 21 CFR Part 11 compliant eClinical suite in a single integrated environment. Connect with our team to walk through the platform, explore what it costs for your study, and see where the biggest savings are relative to your current model.

Frequently Asked Questions

What does the FDA require for real-world data to qualify for regulatory submissions?

The FDA’s Final RWD Guidance requires patient-level traceability: a clear, auditable connection linking every data point in the EDC back to its original source medical record. Studies that cannot demonstrate this chain of custody are at risk of regulatory rejection regardless of sample size or scientific rigor. The FDA has moved to enable RWD use in submissions, but traceability remains the non-negotiable qualifying standard. Studies built on a unified eClinical platform with native source-linking are structured to meet this bar from day one.

What is a Direct-to-Patient model in clinical trials and how does it work?

A Direct-to-Patient (DTP) model places the patient at the center of the study through digital channels, replacing the physical site as the primary data collection hub. Instead of traveling to a clinic, patients authenticate through their EHR portal (e.g., MyChart via TEFCA), grant consent digitally, and have their historical medical records extracted automatically. Longitudinal follow-up combines scheduled ePRO surveys and ongoing EHR re-pulls to capture safety events the patient may not self-report. A single central IRB replaces the network of local IRBs, and all data flows into one EDC with a full source audit trail.

How much can sponsors save by switching from a traditional site model to a DTP framework for a 1,000-patient RWE study?

For a 1,000-patient, 2-year obesity RWE study tracking GLP-1 outcomes, a traditional multi-site model carries an estimated total cost of ownership of $2.0 million. The DTP model using Castor Catalyst reduces that to approximately $1.0 million (an estimated 50% reduction). The largest savings come from reduced site management and monitoring, lower patient stipends, lower SDV costs through targeted remote review, and dramatically lower data review and query resolution through automated edit checks. Note that DTP digital recruitment runs higher than site-based referral, and Catalyst’s EDC licensing is modestly higher than a standard EDC base. Learn more about Castor’s real-world evidence platform.

References

U.S. Food and Drug Administration (FDA). (2024). Real-World Data: Assessing Electronic Health Records and Medical Claims Data To Support Regulatory Decision-Making for Drug and Biological Products. Final Guidance for Industry. fda.gov/science-research/real-world-evidence [Date requires verification against source document]
European Medicines Agency (EMA). (2024). Data Quality Framework for EU medicines regulation: application to Real-World Data. Draft Guidelines. ema.europa.eu [Date requires verification against source document]
Garza JP, et al. (2024). Error Rates of Data Processing Methods in Clinical Research: A Systematic Review and Meta-Analysis. JAMA Network Open; 7(1):e2351486.
Castor. (2026). Internal Platform Analytics: Firestore Study Metadata and Parquet Enrollment Data. [Data on file, February 2026].
Bentley, C., et al. (2019). The Cost of Clinical Trials and the Impact of Decentralized Workflows. Clinical Therapeutics. (Financial benchmarking models adapted for 2026 loaded labor rates).
Castor white papers on real-world evidence and eClinical data strategies.
Learn more about Castor Catalyst and decentralized clinical trials. castoredc.com/catalyst-ai-rwe/
ClinicalTrials.gov. (2026). Active study count: obesity, GLP-1 receptor agonist, and weight management studies (interventional and observational). clinicaltrials.gov [Search accessed February 2026. Count subject to change as studies complete or register.]

FDA & EMA RWD Compliance in Obesity Trials: The Direct-to-Patient Action Plan

On this page