Understanding eSource in the New Era of Clinical Trial Design

How organizations capitalize on accelerated adoption of eSource to further improve outcomes of future trials.

Today’s clinical trial dashboards alert sponsors and sites to any developing data trends, highlight compliance issues, and provide insight into all study data, including lab and sensor data. This type of immediate alerting and reporting is available when collecting source data close to real-time. The Food & Drug Administration (FDA) defines electronic source data (eSource) as electronic source data initially recorded in an electronic format 1. In decentralized and hybrid clinical trials today, the participants may not visit the clinic or site for research-related data collection. The patient data is gathered electronically from various sources that do not need to be transcribed, verified, and double-checked. The participant may need to visit a lab or imaging center, and the resulting data from those facilities end up in a Decentralized Clinical Trial (DCT) platform. Digital endpoints are available in close to real-time. The required data provenance, or record of the origination and description of how and why the data ended up at its present location, indicates that no human has had to decipher handwriting, transcribe anything, or perform optical character recognition. Data streams from various sources such as sensors, electronic records, apps on a participant’s device, labs, and electronic investigation assessments are aggregated in a single platform. This central data repository is a result of utilizing eSource data, and the only barrier to this is not technology; it is resistance to change.

As an industry, we have graduated from paper diaries, and paper case report from data collection to acceptance of electronic case report form data (eCRF), Clinical Outcome Assessments (eCOA) including Patient Reported Outcomes (ePRO), Clinical Reported Outcomes (ClinRO), caregiver-reported outcomes (ObsRO), performance outcomes (PerfO), electronic patient diaries (eDiary), and electronic Device Reported Outcomes from sensors and wearables (eDRO). We can also include data obtained directly from the electronic health/medical record (EHR/EMR), and we have seen Real World Evidence (RWE) supporting a
clinical claim.

Clinical trials endured the COVID-19 pandemic, with some trials adapting to the travel and contact limitations by embracing technology that provides virtual visits via telehealth and data capture directly from the patient. During the pandemic, we saw trials on hold or delayed indefinitely. Stalled and delayed trials are problematic and cost the industry copious amounts of unrealized revenue; but most importantly, they delay lifesaving treatment to patients in need.

The trials that have thrived during the pandemic have been those not dependent on visits to a clinical site or institution.

The trials that thrived during the pandemic were those that were not dependent on visits to a clinical site or institution. Collecting data directly from a patient via their own device, having clinical staff use telehealth technologies to capture required data, and using home health services, as needed for assessments requiring a physical presence, are changing the way we conduct clinical research. All of these are examples of data collection in the moment, rather than recording information to be transcribed at a later date. This is called a decentralized approach, and it is supported and encouraged by regulatory authorities and eliminates unnecessary costly activities such as on-site source data verification and transcription.

The FDA has supported and encouraged the use of eSource as seen in the FDA eSource guidance released in 2013 (2). During the pandemic, both the FDA and the European Medicines Agency (EMA) released regulatory statements supporting electronic data sources. The FDA issued the FDA Guidance on Conduct of Clinical Trials of Medical Products during COVID-19 Public Health Emergency in March 2020 and has provided several updates (3). This guidance allowed alternative methods for safety assessments such as telehealth or virtual visits and alternative assessment types such as remote clinical outcome assessments (ClinRO) and remote Performance Outcome assessments (PerfO).

The FDA recommended using electronic informed consent in place of paper-based consent. Also recommended was optimizing the use of central and remote monitoring programs to maintain oversight of clinical sites.

The European Medicines Agency (EMA) issued GUIDANCE ON THE MANAGEMENT OF CLINICAL TRIALS DURING THE COVID-19 PANDEMIC in April 2020 and indicated that data collection should continue despite the absence of in-person interactions. This document gives guidance on how industry should evaluate the risk of protocol changes such as virtual visits and details how to document alternative methods of obtaining consent such as electronic consent (4).

The UK (United Kingdom) Medicines and Healthcare products Regulatory Agency (MHRA) issued initial guidance in March 2020 on managing clinical trials during the coronavirus pandemic, with several subsequent updates. The main item of interest in this guidance is that it allows for remote access to medical records if appropriate security is in place (5). The use of technology to obtain a response data directly from the patient has existed for many years, but this increased rapidly due to technological advances, and more recently out of necessity. Data from devices like the Apple Watch or continuous blood glucose monitoring systems are collected and viewed to better understand our current health endpoints. As of February 2020, clinicaltrials.gov shows that approximately 460 wearables studies are underway, and according to Kaiser Associates and Intel, 70% of clinical trials will incorporate sensors by 2025 (6).

When running a clinical trial, a single platform that collects data from all trial modalities and provides robust metrics is most efficient.

Many trials are collecting and processing data from wearables. Some, however, are collecting this data for exploratory endpoints or for use in proof-of-concept projects. In our industry’s next advance, we must leave this tentative approach behind. We must embrace these technologies for primary and secondary endpoints that directly affect the approval and disapproval of an Investigational Product (IP). To achieve this goal, a judicious selection of devices is necessary. There is a wide variety of approval approaches around the globe for devices. For direct primary and secondary endpoint collection, these devices must be validated for the specific data point and compared to the current gold standard collection of that endpoint to show they are equal or better.

This requires the industry knowledge to select mature vendors with solid scientific approaches that can show full validation of their devices and measurements.

We have seen the use of eSource with common endpoints like actigraphy and movement collection; also, the verticals within eCOA, wearables, and telehealth have grown vastly. Unfortunately, they have expanded within their own siloed verticals. However, there exists a new frontier of combining these data collection methods into one cohesive offering. These data platforms feed the hub of all the clinical data within a trial. The benefits of this approach are many, with being able to easily view metrics and analytics for all key risk indicators topping the list. A close second is the ability to see and resolve data discrepancies as they happen. Reducing the need for on-site monitoring by eliminating source data verification is third, followed by having signal detection and discovery enablement front and center when using one platform.

Questions always have existed around data from patients, in large part because many patients are eager to please clinicians, trials, and family. In the past, due to the siloed nature of clinical trial data cited above, detecting noise such as recall bias, observer expectancy effect, and response bias was more difficult. However, with the ability to collect data from many sources and view them from a single platform, it is far easier to identify, mitigate, and reduce these biases. This capability exists today and just needs adoption from industry to harvest the full benefit.

According to TransCelerate, eSource workstream categories fall into the four groups shown below (7).

The benefits of eSource are well documented as listed here from TransCelerate:

• Eliminate unnecessary duplication of data
• Reduce the possibility for transcription errors and eliminate transcription of source data prior to entry into an electronic case report form (eCRF)
• Facilitate remote monitoring of data; promote real-time access for data review

Facilitate the collection of accurate and complete data

TransCelerate lists many key metrics from industry projects that quantify the value of eSource:

• Manual data entry reduced by 20%
• Query rate reduced by 50%
• Data latency reduced from 20.4 to 3.5 days
• Transcription errors reduced from 6.7% to 0 Site effort reduced by 8
  hours/patient/study
• Monitoring activity reduced by 2 hours/patient/study
• Time saved 37% total time between traditional EDC vs eSource, 1 FTE (Full Time Equivalent), 65% total keystrokes
• Data Quality - reduced 9% error rate to 0

The position in the TransCelerate eSource paper is that the combination of electronic source data with data stored in a unified platform will further increase these efficiencies.

THREAD's data analysis saw these efficiencies rise in 2020. The chart below shows a direct increase in the number of ePRO entries in the THREAD platform during a large long-term research study that ran throughout the COVID-19 pandemic. This spike in entries appears correlated to the confirmed COVID-19 cases (8), implying that patients turned to electronic means such as registry studies to improve their medical care at that time.

Offering participants, sites, and sponsors a way to provide their data electronically changes the landscape. In order to be successful, we need to make this data collection simple. For example:

• Include the patient/participant – Data such as diary data for eCOA assessments coming directly from the participant needs to be collected electronically or via a clinician interacting directly with the patient. Participants today can use their own devices Bring Your Device (BYOD), which is more convenient than carrying a study-provided device. Telehealth offers the ability to gather information directly from the patient while entering it into the source system. This entry method is preferred by regulatory agencies, reduces transcription, and source data verification (SDV), and limits data checking.
• Avoid duplication – If the data already exists electronically, use that electronic source. An electronic source is always preferable. Today we can pull data from medical records, health devices such as blood pressure cuffs and thermometers, and many other
  activity modalities.
• Provide a single entry point – Eliminate multiple systems with multiple passwords. Today clinical sites maintain access to many systems: RTSM, eCONSENT, eTMF, EDC, CTMS, EKG, Lab Providers, etc. If we consolidate this data into a single platform, we can reduce the site burden and improve quality by reducing transcription.
• Provide a study diagram – A study data flow diagram is worth a thousand words. Provide all study team members with a study diagram showing all data streams, highlighting the ones for which they are responsible.

According to the Tufts-eClinical Solutions Data Strategies & Transformation Study, more than two-thirds of all sponsors are using or piloting at least four types of data in clinical trials (9). It is common today to collect patient/caregiver consent and assent electronically since this technology can ensure sites have the correct version and allows for multilingual deployment of multiple consents. Telehealth visits are scheduled to collect physician outcomes. Data is collected directly from the patient from electronic patient report outcomes (ePROs), or surveys or sensors.

Sponsors today are pre-qualifying a variety of sensors to develop a library of validated sensors prior to the study need. This is done by developing acceptance criteria, then piloting selected sensors while collecting scorecard results. Scorecard items may consist of:

• API and/or SDK availability
• Patient experience using the device
• Formatting of data for simple analysis
• Availability of data integration
• Device characteristics such as use, battery life, etc.

Today we could create a trial that includes telehealth to assess and collect eCOA, mobile apps to enable ePROs on a participant’s phone to collect validated patient data, and the addition of a wearable to collect data that is more independent of bias. So now we not only can assess how a patient reports they feel or how they present themselves to a clinician, we also can add in wearable data to see how they are performing and feeling. Thus, comparing movement and activity ePRO questionnaire type data to the fact that the patient’s actigraphy has plunged in the past two weeks relative to baseline is incredibly valuable.

The adoption of technology has transformed training. We have seen virtual learning skyrocket out of necessity. Training in a clinical trial is no exception. Participants have access to training videos on every aspect of their clinical trial, from information about the trial to consenting details to videos on how to set up their wearables to video updates and reminders. These advances in training are a welcome improvement over in-person or paper-based training methods of the past. Throughout the patient training and study video visits with clinical staff, patients are encouraged to ask any questions they have about the study.

Training can be separated at an elevated level between patients and clinicians, and while learning new systems and approaches can be a hurdle, that hurdle gets lower each year with the pervasive adoption of technologies such as smartphones and streaming multimedia and an increasingly tech-savvy population.

For sites and clinicians, the onset of commonplace Electronic Data Capture (EDC) (10) started to accelerate around 2010, and most clinical site personnel are now familiar with these systems, as well as customary practice management systems. As a result, the training hurdle is low for sites to complete and use effectively. To mitigate any risk in tech barriers, choosing to deploy a single platform, opposed to disparate systems, and having a single point of contact for support and training, makes the already low hurdle even lower.

Here, the challenges are providing the proper materials to train the trainer on educating the patient and having the combined systems appear as one integrated platform for the patient. Technologies today can provide learning options when they are needed, not as a bolus upfront. These learning strategies are required to be successful in technology adoption.

With billions of people using social media, banking apps, and now health apps such as Apple HealthKit, user training has become much easier.

The patient, of course, is the other hurdle and a more exponentially impactful part of the equation. However, this hurdle gets lower each year as tech becomes more pervasive. It is easy to forget how rapidly consumers have adopted digital technology. For example, the original iPhone was released in the summer of 2007 with 300,000 orders; as of 2019, the iPhone has a global installed base of more than 900 million. This is only half of the equation: 91% of the 4.66 billion unique internet users worldwide access the web via smartphones (11). With billions of people using social media, banking apps, and health apps like Apple HealthKit, user training has become much. User interfaces in app design are simplifying and drastically decreasing adoption barriers. It is far easier to teach an app-savvy senior how to use their health app when using Facebook and Facetime daily to interact with the outside world.

Worldwide clinical trials today have seen the reduced need for provisioned devices, which lowers trial costs. More importantly,
it lowers the patient burden. Now, instead of using unfamiliar devices, more patients can use their own device. This has many benefits:

1. The training bar is lower, as discussed
2. Since it is the patient’s own phone, the likelihood of their study device being left on the nightstand or kitchen table has
   drastically reduced
3. Providing study data directly using a phone is eSource and thus doesn’t require source data verification and improves data
   quality by eliminating transcription errors
4. Entry of critical study data can be tracked in real time allowing for increased compliance

Patient knowledge of technology, saturation of devices, and acceptance of sharing health data all have grown over the past ten years. Technology at clinical sites and institutions also has grown. All practices use some sort of practice management software and often use tablets for registration and EMR (Electronic Medical Record). Finally, eSource-related technology vendors have matured and are utilizing cloud-based storage and security linked to constantly connected devices and technology.

With the market dramatically consolidated, the need for provisioned devices is decreasing.

The last remaining need is for industry to accelerate adoption of proven technologies. The pandemic has forced our hand and we have seen the growth results. Industry adoption must grow at the rate of the customer – and the customer is the patient sitting in a clinical site waiting room using their smartphone for entertainment.

The data management role traditionally involves reviewing data to find transcription errors or site-reported data that is incomplete or has discrepancies. Data that comes directly from the source typically is left as it is received. There should be no need for queries or reconciliation since this data truly represents the activities where it was collected. The data management role shifts from finding incorrect data to orchestrating data of various data types into a single patient story. Data reviewchanges to check things such as whether all the data is in the correct timeframe, whether there are gaps in the data, and whether there are any data trends within the patient that need to be highlighted in the analysis. The role shifts from being a data manager issuing queries to becoming a data scientist looking at the validity and ensuring the quality of the data itself.

Integrity lets us see the data from the source to the final output, eliminating concerns about data manipulation. Data collected directly from the patient is considered the source and therefore cannot be changed by the participant. Certain data checks can be implemented to check for issues that may need to be explained in the analysis. The bottom line is that data collected directly from the electronic source has strong integrity and reduces the need for the type of data review in paper-based clinical trials.

When using an electronic source, the need for a verification step is eliminated, saving time and money, and improving quality. According to the U.S. Department of Health & Human Services, the typical SDV cost of a Phase 3 study is $400,173, or 3.5% of the overall cost of the trial (13). Eliminating or significantly reducing this SDV step will increase the database close and lock time while cutting costs and still providing high-quality data.

The adoption of Risk Based Quality Management is designed to improve data quality. Identifying study risks and proactively providing strategies to manage these risks will ensure a smoother data collection experience. For example, if a study is using a new sensor that the company is not familiar with, they may implement a proof-of-concept study to collect data from this sensor and comparator sensors for a brief period to close any gaps in processes, develop training materials, and ensure the output is as expected. Other risks may include improper storage of the clinical study drug. For this risk, the team can develop training and check-ins with the clinical site to evaluate that the procedures are being followed to safely store, distribute, and track the study drug.

is as expected. Other risks may include improper storage of the clinical study drug. For this risk, the team can develop training and check-ins with the clinical site to evaluate that the procedures are being followed to safely store, distribute, and track the study drug.

A big part of quality management is reviewing the data and metadata for items at risk throughout the study. Thresholds are provided around each area to help determine whether a risk is developing or if there is a quality issue. Take the example of enrollment: If a site is expected to enroll a certain number of patients per month, this is easily measured and identified as meeting expectations, becoming at risk for not meeting expectations, or not meeting expectations at all.

Typically, this risk-based approach is developed by a representative of each department in the clinical study process. Risks are identified, as well as proposed measurements and thresholds. Data around these risks updated on a regular basis, and the entire team meets to review all metrics throughout the life cycle of the study. This overall insight into the study benefits all departments and will result in fewer quality issues.

The analysis phase of a study uses data from the patient to determine if the medicine or device is safe and effective. This data is so critical that we have devised many ways to gather it, enter it, and make sure it is accurate. Providing the statisticians digital electronically sourced data directly from the patient eliminates multiple steps, all of which may cause data integrity issues.

According to the U.S. Department of Health & Human Services, the typical SDV costs of a Phase 3 study is $400,173, or 3.5% of the overall cost of the trial. (13)

According to Stat News, The Digital Medicine Society recently released a crowdsourced collection of “pharma companies that have publicly disclosed they’ve collected digital endpoints in their trials.” Stat News writes: “The library details 38 unique industry-sponsored studies that evaluated digital endpoints in the new medical products or new applications of existing medical products. The digital tools used range from accelerometers to spirometers and microphones. To date, digital endpoints are being used in trials by 15 different sponsors; they include eight primary endpoints and seven Phase 3 studies.” (14)

This summary only represents those companies that chose to participate in the survey; there are many more. The accuracy and quality of this primary efficacy data is our end goal that eSource data collection provides.

Paper collection requires data entry and human review to confirm validity. Both methods are impacted by error, delays, and rework. Deploying a trial model with ePRO, eCOA, and sensor data sources all on one platform allows for adoption of more robust AI and ML solutions. Merging paper diaries with EDC data is a manual task that takes time and validation to ensure accuracy.

Trial design focused on a single platform dramatically increases the power and capabilities of AI and ML models. Automatically notifying sites of visit inconsistencies, data anomalies, and compliance issues reduces trial duration and increases data quality. As adoption of these technologies grows, real-time notifications, data availability, and predictive risk modeling will increase, supporting cleaner data collected directly from the source with zero transcription error.

Dashboards and analytics provide clear and timely metrics of the health of a clinical trial and support real-time review and adjustments if required. Identifying and surfacing metrics of poor-performing sites will allow sponsors to amend recruitment strategies and optimize patient accrual timelines. Training algorithms across data sets will allow sites to monitor adverse events and concomitant medication logs across data sources in real-time. A patient may have identified difficulties with their daily activities in their ePRO but failed to mention them to site staff.

Fully integrated studies with eCOA, sensors, EDC, and real-world data in a single platform will allow computer algorithms to learn and evolve consistency evaluations, flag anomalies between ePRO and adverse events, and risk-score the reliability of data in real time as opposed to in retrospect. As the regulatory agencies increase acceptance of these technologies, and the industry deploys them to identify issues, learn patterns, and reduce human error, the more common these techniques will become, and the more efficient trials will become. AI and ML will allow sites to conduct more studies, Contract Research Organizations (CROs) to service more deliveries, sponsors to introduce more drugs, and more patients to benefit from an efficient and cost-effective data collection and management model.

This report highlighted the benefit of a completely paperless, electronic source ecosystem for clinical trials. In addition, the need for a single platform where all data and key performance indicators (KPIs) are displayed is critical for keeping a clinical study on track, monitoring compliance, and for the ability to identify any risks related to the study data. This industry has been historically slow to adopt any change in existing methodology due to the complexity of adding new processes, new Standard Operating Procedures (SOPs), providing training, and using innovative technology COVID-19 has pushed us into the digital age with the increased use of telehealth, virtual visits, data collected electronically by ePRO, electronic surveys and data provided directly by home health providers.

In the words of Scott Gottlieb, M.D., former FDA Commissioner, in a statement from the FDA: “Unfortunately, we have seen a continued reluctance to adopt innovative approaches among sponsors and clinical research organizations. In some cases, the business model adopted by the clinical trial establishment is incompatible with the positive but disruptive changes that specific innovations can enable. We appreciate that scientific and technical complexity is a real and ongoing challenge, but industry and academia also need to invest in and leverage these approaches and develop new incentives that reward collaboration and data sharing across the clinical research enterprise.” (15)

Citations

1 FDA Guidance for Industry, Electronic Source Data in Clinical Investigations, Sept 2013 https://www.fda.gov/media/85183/download

2 FDA Guidance for Industry, Electronic Source Data in Clinical Investigations, Sept 2013 https://www.fda.gov/media/85183/download

3 https://www.fda.gov/regulatory-information/search-fda-guidance-documents/fda-guidance-conduct-clinical-trials-medical-productsduring COVID-19-public-health-emergency

4 https://ec.europa.eu/health/sites/health/files/files/eudralex/vol-10/guidanceclinicaltrials_covid19_en.pdf

5 https://www.gov.uk/guidance/managing-clinical-trials-during-coronavirus-COVID-19

6 Wearables & big Data In Clinical Trials – Where do we Stand? By Yvette Jansen, Grant Thornton:
https://www.clinicalleader.com/doc/wearables-big-data-in-clinical-trials-where-do-we-stand-0001

7 https://www.transceleratebiopharmainc.com/assets/eSource-solutions/

8 COVID results: https://www.cdc.gov/coronavirus/2019-ncov/covid-data/covidview/10162020/specimens-tested.html

9 Tufts-eClinical Solutions Data Strategies & Transformation Study:
https://www.eclinicalsol.com/news/tufts-csdd-study-revealsincreased-external-data-sources-contributing-to-delays/

10 https://en.wikipedia.org/wiki/Electronic_data_capture

11 https://www.statista.com/statistics/617136/digital-population-worldwide/

12 https://www.statista.com/statistics/272698/global-market-share-held-by-mobile-operating-systems-since-2009

13 Examination of Clinical Trial Costs and Barriers for Drug Development:
https://aspe.hhs.gov/report/examination-clinical-trial-costsand-barriers-drug-development

14 Digital endpoints library can aid clinical trials for new medicines, by Jen Goldsack, Rachel A Chasse and William A. Wood.
https://www.statnews.com/2019/11/06/digital-endpoints-library-clinical-trials-drug-development/

15 Statement by FDA Commissioner Scott Gottlieb, M.D., on new strategies to modernize clinical trials to advance precision medicine, patient protections and more efficient product development -
fda.gov/news-events/press-announcements/statement-fdacommissioner-scott-gottlieb-md-new-strategies-modernize-clinical-trials-advance