Unraveling the eSource

It's easy to imagine a not-too-distant future when electronic health records (EHR) and electronic medical records (EMR) are an essential part of patient care.

In this future, doctors working in clinics, hospitals, and private practices work with computers, PDAs, or other devices to retrieve and enter information about their patients, whose records are updated in real-time.

On the pharma side, EHRs streamline the clinical trial process. Sponsors, for example, are able to tap into a national e-health system and identify investigators with appropriate patient pools. They're also able to more quickly develop the necessary clinical trial infrastructure—protocols, data-capture systems, and databases—by using computerized tools, electronic communication, and libraries of standard data elements and edit checks.

Learn the Lingo

The vision is that one day there will be industry-wide standards for EHRs. This will increase efficiencies by enabling sponsors to deploy clinical studies to trial sites in the form of independent software modules. Because the clinical modules will run on investigator sites' own EHR systems, they will look and feel familiar to physicians—sharply reducing training time.

Industry-wide standards will also allow an exchange of information between clinical researchers and providers of patient care. Physicians, as they enter information about an individual patient visit, will be informed if that patient is enrolled in a clinical trial, prompting them to collect additional information or assist them with scheduling and patient-visit reminders. At the same time, if a physician keys in wrong information—say, a prescription that is contraindicated for a particular patient—the system will send out a notification, helping to prevent medical errors.

Rewind to present day: The healthcare system and the pharmaceutical industry are just beginning to move toward using information technology in such a way that could make this scenario possible. But some substantial obstacles—rooted in the very nature of patient care and clinical trials—stand in the way. It is important for both healthcare and pharma leaders to understand the nature of the problems; if they can be solved, patients, physicians, and pharmaceutical companies all stand to gain substantial benefits.

The (Unrealized) EDC Dream

Adoption of EMRs in hospitals and private practice is on a steady incline, driven by the need to manage costs and deliver higher-quality healthcare while enhancing the safety of patients. Reports from 2002 (the last year data were collected) suggest that 20 to 25 percent of US healthcare practices, and close to 50 percent of hospitals, use EMR/EHR systems.

In Europe, there is great variation between countries: EMR use is as high as 90 percent or more in the Scandinavian countries, while less than 20 percent in some others (see "The European Divide").

The European Divide

At the same time, clinical trial sponsors are increasingly using electronic data capture (EDC) systems, which hold the potential to dramatically reduce the time and effort needed to acquire and process clinical trial data. Today, an estimated 27 to 30 percent of clinical trials in both hospitals and private practices use EDC.

Going forward, those systems will be particularly valuable to pharma if they can be used to populate EDC systems.To understand the problems associated with integrating EHRs and clinical trials, however, it is essential to focus on the concept of "source data," which lies at the heart of data collection and management for all clinical trials.

The ICH Harmonised Tripartite Guideline for Good Clinical Practice defines source data as all information in original records and certified copies of original records of clinical findings, observations, or other activities necessary for the reconstruction and evaluation of a clinical trial. Source data are contained in source documents, which include hospital records, patient charts, laboratory notes, and pharmacy dispensing records.

The information captured in clinical trial systems may be based upon an electronic source, such as an EMR. Unfortunately, clinical researchers cannot use many of today's EMRs and EHRs because of the variability among health record systems and the lack of conformance to clinical trial regulatory requirements. Therefore, very often, the data stored in EMR systems has to be printed or hand-transcribed and re-entered into the EDC system.

That may help explain recent survey results from the Clinical Data Interchange Standards Consortium: Despite the advantages of online data-discrepancy management, streamlined archiving, and other tools contained within major EDC systems, 25 percent of sites surveyed believe EDC is increasing their workload. And as the use of EHRs increases, the perception that EDC is an added burden also will increase.

There's also the problem of capturing trial-specific data that is not typically recorded in patients' charts or electronic health systems. This means, to stay compliant with regulations, investigators often must create a second "source," an alternative record-keeping system in addition to their EMR system.

The duplication of tasks, generation of paper, and associated costs and inefficiencies will only grow as clinics and hospitals adopt electronic data sources. This increasing burden is worrisome; if not controlled, it could compromise the quality of execution for both clinical trials and patient care. Small wonder that for many professionals, today's transitional environment feels like a step backward.

Not as Easy as It Looks

In many ways, the core challenge of integrating EHRs and clinical research—creating an EHR/CR system—is that of creating an electronic source record, or eSource, that fulfills the needs of caregivers and researchers while staying within regulatory requirements. Many regulations governing the use of source data (see "eSource: Rules & Regs"). However, these regulations fail to qualify EMRs and EHRs as eSource. If clinical trial investigators want to use EHRs as eSource for clinical trials, they need to tackle a number of issues.

eSource: Rules and Regs

Point-of-care data-collection processes Healthcare is under tremendous time and cost pressure. And while EHRs document and facilitate the management of individual patient care delivery, it may still be inefficient and costly for hospitals or physicians' offices to carry out the meticulous documentation processes that are common in the pharmaceutical industry.

If data in EMR/EHR systems are used as eSource for clinical trials, investigators will have to apply authority checks (as expected under 21 CFR Part 11) to ensure that only authorized persons can access the system, and an electronic audit trail will be required.

This will likely clash with common practices in small physicians' offices, such as using generic user names or sharing passwords with multiple staff. It will also force many companies to modify EMR/EHR products to include audit-trail capabilities compliant with clinical research. Many have speculated that, without appropriate encouragement and incentive, enforcement of requirements like audit trails may slow doctors' adoption of health records.

Source data are far from perfect Depending on the circumstances of healthcare delivery, the quality of the records may suffer. In emergency situations, for example, ambulance and emergency room staff will always focus more on rapid intervention than accurate record keeping.

Further complicating the picture, a variety of staff—physicians, nurses, phlebotomists, etc.—play a role in recording patient information. That data is then subject to the professional judgment of the physician responsible for writing the discharge summary.

But the question of source data must still be tackled: Should the notes from each of the consecutive caregivers be considered source data or just the summary from the overseeing physician? FDA's Division of Scientific Investigation (DSI) representatives acknowledged this ambiguity when they met with the PhRMA EDC/eSource Taskforce in January 2006. They noted that the regulations do not equate source data with initial data; instead, the discharge summary is considered the source data.

The traditional piecemeal record-keeping approach is notoriously flawed and leads to a great number of treatment errors. In fact, the improvement of this situation—and easy accessibility to the information—is one of the reasons for the introduction of the EMR and EHR systems. But without process and control changes similar to clinical research controls, the goal of improving the accuracy of healthcare information may remain an elusive one.

Control of data is difficult to achieve In clinical research, source data are transcribed onto the required paper case report forms (CRFs) or into the EDC system. It is the physician's responsibility to prepare and maintain the source data, and the sponsor's responsibility to ensure the reported trial data are accurate, complete, and verifiable from source documents. To ensure the accuracy of this process, sponsors usually carry out a process called source data verification, in which CRF records are manually compared to the corresponding source data.

But who retains physical control over data, whether on paper or electronic? FDA defined its current position in its February 2006 Draft Guidance on Patient Recorded Outcomes. This guidance states that sponsors must not have exclusive control of the source document, there must be more than one database, and investigators must be accountable for the accuracy of the data.

The issue at hand, however, is that current EDC systems are often hosted by sponsors or third parties over the Internet. If such EDC systems develop into or integrate with EMR/EHR systems, this model would no longer be viable, as it would make the sponsor the sole custodian of the dataset.

Pilots Proceed

Despite an unclear regulatory environment and the lack of accepted standards, organizations still see value in using eSource. The following examples illustrate how groups are coming to terms with the convergence of electronic healthcare and EDC through innovative pilots and experiments.

Eli Lilly conducted what is believed to be one of the first trials completely run over the Internet for an Investigational New Drug. In late 2001 and early 2002, with FDA's approval, the company used the Internet to gather data from patients, and formulated EHRs. The information was hosted by a third party, which transferred the depersonalized data to Lilly for statistical analysis and reporting.

Patients enrolling for the trial were required to have a home computer with access to the clinical trial Web site. While it is not clear how this might have biased the selected patient population, it did demonstrate that such online trials are feasible and can be compliant with regulations. However, it did not eliminate the problem for investigators who work with several sponsors—with each organization maintaining a different electronic record database, all separate from the investigator's non-clinical trial patient records.

Johnson & Johnson is performing Phase I trials with tablet PCs, taking advantage of the devices' mobility to make it easier for physicians to enter data directly—and thus eliminate a separate source. The data entered into the tablets do not reside there, but go directly to a local onsite server, and then are transmitted to J&J's central server.

The Karolinska Institute has reported on pilots involving retrospective mining of data. In these pilots, the organization retrieves large amounts of pre-existing data from EMR databases and analyzes them quickly. Although these pilots used retrospective collection of data, they did demonstrate that transfer of eSource data to a sponsor for clinical trial analysis is feasible, and that this process could realize efficiencies. It also showed, since the data was exported from the EMR, that it could be controlled to protect patients' privacy.

Siemens Medical Solutions, an EMR vendor, is currently conducting a pilot with the Technical University of Munich that includes back-end integration, enabling the automatic transfer of data from EMRs into an EDC system.

The Clinical Data Interchange Standards Consortium conducted a proof-of-concept pilot on generating both medical records and clinical trial databases using a single point of entry. Instead of entering data into the EMR and EDC system directly, clinicians relied on a dedicated "entry application," which maintains the proper regulatory control of the data and uses it to populate the appropriate databases.

Lundbeck Pharmaceutical is finding a way to power its EDC systems with eSource data. As eSource data are saved or updated in the EDC system, the company creates a controlled PDF copy that is automatically stored in a trusted third party's secure facility. These controlled PDFs may be viewed—but not modified or overwritten—by both the sponsor and the site, and used to verify data integrity should the need arise. In this way, Lundbeck meets regulatory guidelines: It maintains immediate access to the data, yet there is still a separate source for verification.

Challenges Identified

Significant benefits can be accrued through collaboration by both the healthcare and research worlds in effectively and efficiently sharing data. Without such collaboration, as the use of EMR/EHRs grows, both the healthcare sector and biotech and pharmaceutical companies will be obliged to spend valuable resources on duplicate tasks. The challenge is to develop systems and processes that will allow for direct use of patient electronic medical data for both prospective and retrospective clinical research, in a way that meets data protection, and regulatory and ethical research requirements. In meeting this challenge, all stakeholders must focus their efforts on:

• creating, or adapting from existing clinical research systems, a mechanism for satisfying regulatory and clinical research requirements

• determining data standards for EDC, interpretation, and exchange that will meet the needs of both the medical and clinical research communities

• developing controlled, secure processes for releasing and transferring data from and to EHRs that are consistent with personal data privacy, clinical trial regulations, and bioethical considerations

• establishing a solution that would satisfy the needs of the research community without violating anti-kickback regulations (e.g., US Stark legislation) or having direct ties to specific biopharmaceutical manufacturers.

The transition from disparate systems to a truly integrated EHR/CR environment is likely to be an evolutionary process that occurs over a number of years. However, the good news is that regulatory agencies are showing interest in furthering eSource. At the January 2006 meeting between the Division of Scientific Investigations and the PhRMA EDC/eSource Taskforce, DSI representatives explored several pressing issues facing the development of eSource today—and expressed their belief that those discussions needed to continue.

As the increasing number of recent initiatives demonstrates, the pharmaceutical industry has great interest in further exploring the potential of eSource for improving clinical trials. However, it is important that, as vendors, sponsors, and regulatory agencies work out the issues surrounding the use of eSource, the entire medical system work together to create data that can match the regulatory standards for clinical trials and the regulations for electronic health record systems.

This article is based on a white paper developed by the eClinical Forum and PhRMA EDC/eSource Taskforce. It has been presented at a few industry events, including DIA's Annual Meeting. For more information, including an extensive listing of citations and sources, visit www.eclinicalforum.com